Fix: WC highlights data from xemlaibongda.top/the-gioi/world-cup

#13
by bep40 - opened
.huggingface/rebuild DELETED
@@ -1 +0,0 @@
1
- rebuild
 
 
.rebuild DELETED
@@ -1,2 +0,0 @@
1
- Rebuild triggered: v4 - EPG multi-source (VTVGo API primary), VTV6 proxy handler, FPTPlay fallback
2
- Timestamp: 2026-07-05T12:00:00+00:00
 
 
 
.restart_trigger DELETED
@@ -1 +0,0 @@
1
- restart-20260407-vtv-fix
 
 
CHANGELOG.md CHANGED
@@ -1,71 +1,36 @@
1
- # VNEWS v6.5 - Resilient Shorts Auto-Updater
2
 
3
  ## Changes
4
 
5
- ### Critical Fix: Shorts timeout and homepage load stability
6
- **Root cause**: YouTube shorts fetching in `main.py` using `scrape_shorts()` and `_yt_channel_shorts_requests()` could hang indefinitely when YouTube blocks requests or yt-dlp times out, causing:
7
- - Homepage `/api/shorts` endpoint to time out (30s limit)
8
- - Space to appear unresponsive on first load
9
- - No fallback when sources fail
10
-
11
- **Fix applied**:
12
- 1. **shorts_updater.py** (NEW) Resilient background updater:
13
- - Hard timeout (25s) per channel using subprocess isolation
14
- - Stale-while-revalidate pattern: returns cached data immediately, updates in background
15
- - Automatic fallback to hardcoded short URLs when all sources fail
16
- - Persistent storage in `/data/shorts_cache.json` for cache across restarts
17
- - Background scheduler runs every 10 minutes automatically
18
- - No blocking on first homepage load
19
-
20
- 2. **_run.py** Integrated resilient shorts endpoint:
21
- - Overrides `/api/shorts` with non-blocking version
22
- - Returns cached/fallback data in <100ms guaranteed
23
- - Triggers background update if cache is stale or empty
24
- - Never hangs - always returns valid JSON response
25
-
26
- 3. **FALLBACK_SHORTS** — 6 hardcoded viral shorts as emergency fallback:
27
- - baodantri7941 (Dân trí) headlines
28
- - baosuckhoedoisongboyte (Sức khỏe & đời sống) stories
29
- - vtvnambo (VTV Nam Bộ) news
30
-
31
- ### Benefits
32
- - Homepage loads in <2 seconds always
33
- - Shorts data auto-updates every 10 minutes
34
- - Never times out - graceful degradation to fallback
35
- - Persistent cache survives Space restarts
36
- - Uses bucket `bep40/VNEWS-storage` for cache storage
37
-
38
- ### Channels monitored
39
- - baodantri7941 (Dân trí)
40
- - baosuckhoedoisongboyte (Sức khỏe & đời sống)
41
- - vtvnambo (VTV Nam Bộ)
42
-
43
- ---
44
-
45
- # VNEWS v5.1 - Rewrite Fix
46
-
47
- ## Changes
48
-
49
- ### Critical Fix: Rewrite button not creating posts on Tường AI
50
- **Root cause**: `_run.py` imports from `app_v2_entry.py`, but the `/api/rewrite_share` endpoint was only defined in `ai_runtime_patch_fast.py` (loaded through `app_entry.py` which is NOT used). The frontend called a non-existent endpoint → 404 → silent failure.
51
-
52
- **Fix applied**:
53
- 1. **app_v2_entry.py** — Added 3 new endpoints:
54
- - `POST /api/rewrite_slide` — Fast extractive summary (no AI needed), creates slides from article key points + images, saves to wall
55
- - `POST /api/rewrite_share` — AI-powered rewrite with extractive fallback, saves to wall
56
- - `POST /api/url_wall` — URL submission endpoint (alias for rewrite_share)
57
- - All endpoints use the same `_load_wall_posts()` / `_save_wall_posts()` and `WALL_FILE` path as the existing `/api/wall` endpoint
58
-
59
- 2. **static/index_v2.html** — Added `<script src="/static/rewrite_fix_v2.js"></script>` to load the rewrite fix
60
-
61
- 3. **static/rewrite_fix_v2.js** — New file that overrides `rewriteArticle()` to:
62
- - Call `/api/rewrite_slide` first (fast, no AI needed)
63
- - Fallback to `/api/rewrite_share` if slide fails
64
- - Show slide preview overlay after successful post
65
- - Use `prependWallPost()` to add the new post to Tường AI
66
-
67
- ### Previous changes (v5)
68
- - Rewrote match_detail_v2.py with correct event parsing
69
- - 2-tab layout for match detail (stats + timeline)
70
- - Fixed _run.py import
71
- - Dockerfile cache busting
 
1
+ # VNEWS v5 - Match Detail Fix
2
 
3
  ## Changes
4
 
5
+ ### 1. match_detail_v2.py Rewrote event parser with correct selectors
6
+ - Parse `.events > .period > .event` structure (not old `.timeline`)
7
+ - Extract event type from SVG icons in `.event-type` (goal/redcard/yellowcard/substitution)
8
+ - Parse player names from `.players > div` elements
9
+ - For goals: extract scorer + assist names
10
+ - For substitutions: extract player_out → player_in
11
+ - For cards: extract player name
12
+ - Normalize time format: `45' +2` → `45+2'`
13
+ - Fetch H2H stats from `/api/fixtures/h2h-stats` API
14
+ - Parse prediction card, recent matches, H2H standings
15
+
16
+ ### 2. static/match_detail.js Complete rewrite with 2-tab layout
17
+ - **Tab "Thống kê"**: H2H stats comparison, prediction vote, recent match results
18
+ - **Tab "Diễn biến"**: Detailed timeline with:
19
+ - ⚽ BÀN THẮNG — scorer name + assist
20
+ - 🟥 THỺ ĐỎ player name
21
+ - 🟨 THỺ VÀNG player name
22
+ - ↔️ THAY ĐỔI player_out → player_in
23
+ - Period grouping (H1, H2) with visual headers
24
+ - Team badges (HOME/AWAY) per event
25
+ - Color-coded event icons
26
+
27
+ ### 3. app_v2_entry.py Updated
28
+ - Module cache clearing for fresh match_detail_v2 import on each API call
29
+ - Single clean import for both `/detail` and `/live` endpoints
30
+ - Removed duplicate inline scraping code
31
+
32
+ ### 4. _run.py Fixed import
33
+ - Changed `import match_detail` to `import match_detail_v2`
34
+
35
+ ### 5. Dockerfile Cache busting
36
+ - Added `RUN date > /app/.build_timestamp` to force Docker rebuild
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Dockerfile CHANGED
@@ -12,5 +12,4 @@ RUN pip install --no-cache-dir -r requirements.txt || true
12
  COPY . .
13
  EXPOSE 7860
14
 
15
- # v9 - 2026-07-08 - FIX streams: removed expired ssaimh tokens. VTV6 only from sv2. Frontend uses yt_live.js CSS + EPG
16
  CMD ["uvicorn", "_run:app", "--host", "0.0.0.0", "--port", "7860", "--reload"]
 
12
  COPY . .
13
  EXPOSE 7860
14
 
 
15
  CMD ["uvicorn", "_run:app", "--host", "0.0.0.0", "--port", "7860", "--reload"]
README.md CHANGED
@@ -9,25 +9,17 @@ tags:
9
  - ml-intern
10
  ---
11
 
12
- # VNEWS - Tin Tức Việt Nam
13
-
14
- **v18 - FIXED VTV2/VTV3/VTV6/VTV9 stream hanging**
15
-
16
- ## 🔧 Changes in v18 (2026-07-06)
17
- - **VTV2, VTV3, VTV6, VTV9**: Skip expired ssaimh CDN token → immediately fall through to sv2.xemtivitop.com
18
- - **15+ extraction patterns** for m3u8 URL (up from 5), including: file:, src=, source:, player.src(), hls.loadSource(), href=, `<source src>`, url:, window.location, iframe follow (3 levels deep), base64 decode
19
- - **Backup CDN** `tv.mediacdn.vn` for VTV2/VTV3/VTV6/VTV9
20
- - **Fast timeout** 5s for CDN, 12s for PHP endpoints (was 15s each = 60s+ total)
21
- - **sv2.xemtivitop.com** re-prioritized to check BEFORE xemtv.us
22
- - **Iframe chain following**: if a PHP page returns an iframe → follow it up to 3 levels to find the m3u8
23
-
24
- ## Features:
25
- - 📰 News from VnExpress (10 categories) + GenK AI
26
- - ⚽ Livescore from bongda.com.vn (live, today, upcoming, results, standings)
27
- - 🎬 Football highlights from xemlaibongda.top (8 leagues)
28
- - 📺 VTV live channels (VTV1→VTV10, VTV Prime)
29
- - Priority: ssaimh CDN → sv2.xemtivitop.com → xemtv.us → xemtivitop blogspot → FPTPlay → VTVGo → mediacdn → xemtv.net
30
- - 🏆 World Cup 2026 (news, fixtures, standings, stats, highlights)
31
- - 🤖 AI article writing + TTS (multilingual, emotion-aware)
32
- - 🔍 Topic search (8 news sources)
33
- - 🎤 TTS: voice selector + emotion selector + speed control
 
9
  - ml-intern
10
  ---
11
 
12
+ # bep40/vnews
13
+ <!-- build: 2026-06-12T06:45:00 -->
14
+
15
+
16
+
17
+
18
+
19
+
20
+
21
+
22
+
23
+
24
+
25
+ <!-- rebuild: v6.5-final-1781173463 -->
 
 
 
 
 
 
 
 
TRIGGER_REBUILD_V6.md ADDED
@@ -0,0 +1 @@
 
 
1
+ # Rebuild trigger v6 - 1780971280.8544343
_static_build_trigger.txt ADDED
@@ -0,0 +1 @@
 
 
1
+ # Force rebuild - match_detail debug v2
ai_ext.py CHANGED
@@ -13,16 +13,7 @@ from bs4 import BeautifulSoup
13
  from fastapi import Request, Query
14
  from fastapi.responses import HTMLResponse, JSONResponse, FileResponse
15
 
16
- # Try to import main app, but don't fail if it doesn't exist
17
- try:
18
- from main import app
19
- except ImportError:
20
- # Create a minimal FastAPI app for standalone testing
21
- try:
22
- from fastapi import FastAPI
23
- app = FastAPI()
24
- except Exception:
25
- app = None
26
 
27
  # Import wall store from main.py so we read/write the SAME file
28
  try:
@@ -50,10 +41,6 @@ except ImportError:
50
  def _web_context(topic):
51
  return ""
52
 
53
- # ai_ext alias for backward compatibility
54
- _load_ai_wall = _load_wall
55
- _save_ai_wall = _save_wall
56
-
57
  try:
58
  from huggingface_hub import AsyncInferenceClient
59
  except Exception:
@@ -73,150 +60,21 @@ except Exception:
73
 
74
 
75
  def _hf_token():
76
- for k in ("HF_TOKEN", "HUGGINGFACE_HUB_API_TOKEN", "HUGGING_FACE_HUB_TOKEN", "HF_API_TOKEN"):
77
  v = os.getenv(k, "").strip()
78
  if v:
79
  return v
80
  return ""
81
 
82
-
83
- def _clean_text(s: str) -> str:
84
- """Clean text for processing."""
85
- s = html_lib.unescape(s or "")
86
- s = re.sub(r"\s+", " ", s)
87
- return s.strip()
88
-
89
-
90
- def _domain(url: str) -> str:
91
- """Extract domain from URL."""
92
- try:
93
- return urlparse(url or "").netloc.replace("www.", "")
94
- except Exception:
95
- return ""
96
-
97
-
98
- async def qwen_generate(prompt: str, image_url: str = None, max_tokens: int = 1200) -> str:
99
- """Generate text using Qwen models via Hugging Face Inference API.
100
-
101
- This function provides a resilient implementation that:
102
- 1. First tries the SDK-based inference client if available
103
- 2. Falls back to REST API calls to HF router endpoint
104
- 3. Returns a fallback summary if all else fails
105
- """
106
- token = _hf_token()
107
- errors = []
108
-
109
- # Try HF router API with multiple models
110
- if token:
111
- models = [
112
- os.getenv("QWEN_VL_MODEL", ""),
113
- "Qwen/Qwen2.5-VL-7B-Instruct",
114
- "Qwen/Qwen2.5-VL-3B-Instruct",
115
- "Qwen/Qwen2.5-7B-Instruct",
116
- "Qwen/Qwen2.5-3B-Instruct",
117
- "Qwen/Qwen2.5-1.5B-Instruct",
118
- "Qwen/Qwen2.5-72B-Instruct",
119
- "meta-llama/Llama-3.3-70B-Instruct",
120
- ]
121
- # Deduplicate while preserving order
122
- seen = set()
123
- models = [m for m in models if m and m not in seen and not seen.add(m)]
124
-
125
- headers = {"Authorization": f"Bearer {token}", "Content-Type": "application/json"}
126
-
127
- for model in models:
128
- try:
129
- is_vl = "VL" in model and image_url
130
- if is_vl:
131
- user_content = [
132
- {"type": "image_url", "image_url": {"url": image_url}},
133
- {"type": "text", "text": prompt}
134
- ]
135
- else:
136
- user_content = prompt
137
-
138
- payload = {
139
- "model": model,
140
- "messages": [
141
- {"role": "system", "content": "Bạn là trợ lý AI tiếng Việt. Trả lời tự nhiên, ngắn gọn, chính xác."},
142
- {"role": "user", "content": user_content},
143
- ],
144
- "max_tokens": min(int(max_tokens or 900), 1400),
145
- "temperature": 0.35,
146
- "top_p": 0.85,
147
- }
148
-
149
- r = requests.post(
150
- "https://router.huggingface.co/v1/chat/completions",
151
- headers=headers,
152
- json=payload,
153
- timeout=95
154
- )
155
-
156
- if r.status_code >= 300:
157
- errors.append(f"{model}: HTTP {r.status_code}")
158
- continue
159
-
160
- j = r.json()
161
- txt = (j.get("choices", [{}])[0].get("message", {}).get("content") or "").strip()
162
-
163
- if txt:
164
- return txt
165
-
166
- errors.append(f"{model}: empty response")
167
-
168
- except Exception as e:
169
- errors.append(f"{model}: {type(e).__name__}")
170
-
171
- # Fallback: extractive summary from prompt
172
- LAST_QWEN_ERROR = errors[-3:] if errors else "unknown error"
173
- return _fallback_summary_from_prompt(prompt, max_units=6)
174
-
175
-
176
- def _fallback_summary_from_prompt(prompt: str, max_units: int = 6) -> str:
177
- """Generate a simple fallback summary when AI is unavailable."""
178
- text = prompt or ""
179
- for marker in ["Nội dung nguồn:", "Nội dung bài:", "Nội dung gốc:", "Nội dung:", "Nguồn/bối cảnh internet:"]:
180
- if marker in text:
181
- text = text.split(marker, 1)[1]
182
- break
183
- text = re.sub(r"https?://\S+", "", text)
184
- text = re.sub(r"\s+", " ", text).strip()
185
-
186
- # Split into sentences - extract ALL valid sentences, not just first few
187
- sentences = re.split(r"(?<=[.!?])\s+(?=[A-ZÀ-Ỹ0-9])", text)
188
- units = []
189
- for s in sentences:
190
- s = _clean_text(s)
191
- if len(s) >= 30: # Lower threshold to capture more content
192
- units.append(s)
193
-
194
- if units:
195
- # Take up to max_units valid sentences
196
- result_units = units[:max_units]
197
- return "\n".join("• " + u for u in result_units)
198
- if text:
199
- # Fallback: take chunks if no sentence boundaries found
200
- chunks = []
201
- for i in range(0, min(len(text), max_units * 300), 280):
202
- chunk = _clean_text(text[i:i+300])
203
- if chunk and chunk not in chunks:
204
- chunks.append(chunk)
205
- if len(chunks) >= max_units:
206
- break
207
- if chunks:
208
- return "\n".join("• " + c for c in chunks)
209
- return "• Không có đủ nội dung để tóm tắt."
210
-
211
-
212
  HF_TOKEN = _hf_token()
213
  QWEN_VL_MODEL = os.getenv("QWEN_VL_MODEL", "Qwen/Qwen2.5-VL-7B-Instruct")
 
214
  QWEN_TEXT_MODELS = [m.strip() for m in os.getenv(
215
  "QWEN_TEXT_MODELS",
216
  "Qwen/Qwen2.5-72B-Instruct,meta-llama/Llama-3.3-70B-Instruct,Qwen/Qwen2.5-7B-Instruct"
217
  ).split(",") if m.strip()]
218
- _WORKING_MODEL_TEXT = None
219
- _WORKING_MODEL_VL = None
220
  DATA_DIR = "/data" if os.path.isdir("/data") else "/app/data"
221
  SHORTS_DIR = os.path.join(DATA_DIR, "ai_shorts")
222
  HEADERS = {
@@ -225,350 +83,1082 @@ HEADERS = {
225
  }
226
  LAST_QWEN_ERROR = ""
227
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
228
 
229
- # ===== MULTILINGUAL VOICES FOR TTS =====
230
- # Maps voice IDs to edge-tts voice names (only MultilingualNeural voices)
231
- MULTILINGUAL_VOICES = {
232
- # Vietnamese - Native voices
233
- "vi-vn-hoaimyneural": "vi-VN-HoaiMyNeural",
234
- "vi-vn-namminhneural": "vi-VN-NamMinhNeural",
235
- "hoaimy": "vi-VN-HoaiMyNeural",
236
- "namminh": "vi-VN-NamMinhNeural",
237
- "vi_female": "vi-VN-HoaiMyNeural",
238
- "vi_male": "vi-VN-NamMinhNeural",
239
- "nu": "vi-VN-HoaiMyNeural",
240
- "male": "vi-VN-NamMinhNeural",
241
- "female": "vi-VN-HoaiMyNeural",
242
- "mien-nam": "vi-VN-HoaiMyNeural",
243
- # English - Multilingual
244
- "en-us-andrewmultilingualneural": "en-US-AndrewMultilingualNeural",
245
- "en-au-williammultilingualneural": "en-AU-WilliamMultilingualNeural",
246
- "en_andrew": "en-US-AndrewMultilingualNeural",
247
- "andrew": "en-US-AndrewMultilingualNeural",
248
- "en_jenny": "en-US-AndrewMultilingualNeural",
249
- "jenny": "en-US-AndrewMultilingualNeural",
250
- # Portuguese - Thalita Multilingual ONLY
251
- "pt-br-thalitamultilingualneural": "pt-BR-ThalitaMultilingualNeural",
252
- "pt_thalita": "pt-BR-ThalitaMultilingualNeural",
253
- "thalita": "pt-BR-ThalitaMultilingualNeural",
254
- "pt_francisco": "pt-BR-ThalitaMultilingualNeural",
255
- "pt": "pt-BR-ThalitaMultilingualNeural",
256
- # French - Multilingual
257
- "fr-fr-viviennemultilingualneural": "fr-FR-VivienneMultilingualNeural",
258
- "fr-fr-remymultilingualneural": "fr-FR-RemyMultilingualNeural",
259
- "fr_denise": "fr-FR-VivienneMultilingualNeural",
260
- "denise": "fr-FR-VivienneMultilingualNeural",
261
- "fr": "fr-FR-VivienneMultilingualNeural",
262
- # German - Multilingual
263
- "de-de-seraphinamultilingualneural": "de-DE-SeraphinaMultilingualNeural",
264
- "de-de-florianmultilingualneural": "de-DE-FlorianMultilingualNeural",
265
- "de_katja": "de-DE-SeraphinaMultilingualNeural",
266
- "katja": "de-DE-SeraphinaMultilingualNeural",
267
- "de": "de-DE-SeraphinaMultilingualNeural",
268
- # Korean - Hyunsu Multilingual (NOT SunHee)
269
- "ko-kr-hyunsumultilingualneural": "ko-KR-HyunsuMultilingualNeural",
270
- "ko_sunhee": "ko-KR-HyunsuMultilingualNeural",
271
- "sunhee": "ko-KR-HyunsuMultilingualNeural",
272
- "ko": "ko-KR-HyunsuMultilingualNeural",
273
- # Italian - Multilingual
274
- "it-it-giuseppemultilingualneural": "it-IT-GiuseppeMultilingualNeural",
275
- # Spanish (fallback to English multilingual)
276
- "es_ela": "en-US-AndrewMultilingualNeural",
277
- "ela": "en-US-AndrewMultilingualNeural",
278
- "es_carlos": "en-US-AndrewMultilingualNeural",
279
- "es": "en-US-AndrewMultilingualNeural",
280
- # Japanese (fallback to English multilingual)
281
- "ja_nanami": "en-US-AndrewMultilingualNeural",
282
- "nanami": "en-US-AndrewMultilingualNeural",
283
- "ja": "en-US-AndrewMultilingualNeural",
284
- # Chinese (fallback to English multilingual)
285
- "zh_xiaochen": "en-US-AndrewMultilingualNeural",
286
- "xiaochen": "en-US-AndrewMultilingualNeural",
287
- "zh": "en-US-AndrewMultilingualNeural",
288
  }
289
 
290
 
291
- def _detect_voice_emotion(title, text):
292
- """Detect appropriate voice and emotion based on content for multilingual TTS."""
293
- content = ((title or "") + " " + (text or "")).lower()
294
-
295
- # World Cup / Football content - use Andrew multilingual
296
- if any(kw in content for kw in ["world cup", "wc 2026", "fifa", "bóng đá", "trận đấu", "bóng bóng", "đội tuyển", "cầu thủ"]):
297
- return ("andrew", "excited")
298
-
299
- # News categories - choose appropriate voice
300
- if any(kw in content for kw in ["kinh tế", "tài chính", "thị trường", "economics", "finance"]):
301
- return ("jenny", "calm")
302
- if any(kw in content for kw in ["thiên tai", "bão", "lũ lụt", "cháy nổ", "tai nạn", "disaster", "accident"]):
303
- return ("thalita", "serious")
304
- if any(kw in content for kw in ["giải trí", "showbiz", "entertainment", "hài hước"]):
305
- return ("ela", "happy")
306
- if any(kw in content for kw in ["công nghệ", "tech", "technology", "ai", "trí tuệ nhân tạo"]):
307
- return ("katja", "excited")
308
-
309
- # Default Vietnamese
310
- return ("hoaimy", "trung_tinh")
311
 
312
 
313
- def _safe_name(s: str) -> str:
314
- """Create safe filename from string."""
315
- s = re.sub(r"[^\w\-.]", "_", s)
316
- return s[:100] if len(s) > 100 else s
 
 
 
 
 
 
 
 
 
317
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
318
 
319
- def _download_image(url: str, fallback_title: str, out_path: str) -> bool:
320
- """Download image from URL to path."""
321
- if not url:
322
- return False
323
- try:
324
- r = requests.get(url, headers=HEADERS, timeout=15)
325
- if r.status_code == 200:
326
- os.makedirs(os.path.dirname(out_path), exist_ok=True)
327
- with open(out_path, "wb") as f:
328
- f.write(r.content)
329
- return True
330
- except Exception:
331
- pass
332
- return False
333
 
 
 
 
 
 
 
 
334
 
335
- def pollination_image_url(topic: str) -> str:
336
- """Generate image URL from Pollinations.ai."""
337
- return f"https://image.pollinations.ai/prompt/{quote(topic)}?width=1024&height=768&nologo=true&model=flux"
338
 
 
 
 
339
 
340
- # Use the same wall file as app_v2_entry.py for consistency
341
- WALL_FILE = os.path.join(DATA_DIR, "wall_posts.json")
342
 
343
- def _load_ai_wall():
344
- """Load AI wall posts from JSON file (uses wall_posts.json for consistency with app_v2_entry)."""
 
 
 
 
345
  try:
346
- if os.path.exists(WALL_FILE):
347
- with open(WALL_FILE, "r", encoding="utf-8") as f:
348
- return json.load(f)
349
  except Exception:
350
- pass
351
- return []
352
 
 
 
353
 
354
- def _save_ai_wall(posts):
355
- """Save AI wall posts to JSON file (uses wall_posts.json for consistency with app_v2_entry)."""
356
- try:
357
- os.makedirs(os.path.dirname(WALL_FILE), exist_ok=True)
358
- tmp = WALL_FILE + ".tmp"
359
- with open(tmp, "w", encoding="utf-8") as f:
360
- json.dump(posts[:100], f, ensure_ascii=False)
361
- os.replace(tmp, WALL_FILE)
362
- except Exception:
363
- pass
364
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
365
 
366
- # Helper functions for wall operations
367
- def _load_wall_posts():
368
- """Alias for _load_ai_wall for consistency with app_v2_entry.py."""
369
- return _load_ai_wall()
 
 
 
 
 
 
 
 
 
370
 
371
 
372
- def _save_wall_posts(posts):
373
- """Alias for _save_ai_wall for consistency with app_v2_entry.py."""
374
- return _save_ai_wall(posts)
 
 
 
 
 
 
 
375
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
376
 
377
- def make_post(title: str, text: str, img: str, url: str, kind: str, sources=None):
378
- """Create a post dict with standard fields."""
379
  return {
380
- "id": str(int(time.time() * 1000)),
381
- "title": title,
382
- "text": text,
383
- "img": img,
384
- "url": url,
385
- "kind": kind,
386
- "sources": sources or [],
387
- "ts": int(time.time())
388
  }
389
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
390
 
391
- def _short_script(post) -> str:
392
- """Extract clean text for TTS from post."""
393
- text = post.get("text", "") or post.get("title", "")
394
- text = re.sub(r"^[•\-\*]\s*", "", text, flags=re.M)
395
- text = re.sub(r"\s*\n\s*", ". ", text)
396
- return _clean_text(text)[:2000] # Increased from 1000 to 2000 for full content
397
 
 
 
 
 
398
 
399
- # ===== SCRAPER FUNCTIONS (required by ai_patch.py) =====
400
- def scrape_any_url(url: str) -> dict:
401
- """Scrape any URL and extract article content.
402
-
403
- Returns dict with: title, summary, text, image, og_image, via (domain)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
404
  """
405
- try:
406
- r = requests.get(url, headers=HEADERS, timeout=15, allow_redirects=True)
407
- r.encoding = 'utf-8'
408
- soup = BeautifulSoup(r.text, 'lxml')
409
-
410
- # Remove scripts, styles, nav, footer
411
- for tag in soup.find_all(['script', 'style', 'nav', 'footer', 'aside', 'form']):
412
- tag.decompose()
413
-
414
- # Extract title
415
- h1 = soup.find('h1')
416
- ogt = soup.find('meta', property='og:title')
417
- title = (h1.get_text(strip=True) if h1 else '') or (ogt.get('content', '') if ogt else url)
418
-
419
- # Extract OG image
420
- ogi = soup.find('meta', property='og:image')
421
- og_image = ogi.get('content', '') if ogi else ''
422
-
423
- # Extract article body
424
- block = None
425
- for sel in ['article', '.singular-content', '.detail-content', '.fck_detail', '.content-detail', '.knc-content', 'main', '.cms-body', '.article__body']:
426
- el = soup.select_one(sel)
427
- if el and len(el.find_all('p')) >= 2:
428
- block = el
429
- break
430
- if not block:
431
- block = soup.body or soup
432
-
433
- # Extract text from paragraphs
434
- paragraphs = []
435
- for el in block.find_all(['p', 'h2', 'h3'], recursive=True):
436
- t = _clean_text(el.get_text(strip=True))
437
- if t and len(t) > 40:
438
- paragraphs.append(t)
439
-
440
- # Extract images
441
- images = []
442
- for el in block.find_all(['figure', 'img'], recursive=True):
443
- im = el if el.name == 'img' else el.find('img')
444
- if im:
445
- src = im.get('data-src') or im.get('src') or im.get('data-original') or ''
446
- if src and 'base64' not in src:
447
- if src.startswith('//'):
448
- src = 'https:' + src
449
- images.append(src)
450
-
451
- # Prefer OG image as main image
452
- image = og_image or (images[0] if images else '')
453
-
454
- return {
455
- 'title': title,
456
- 'summary': paragraphs[0] if paragraphs else '',
457
- 'text': '\n'.join(paragraphs),
458
- 'image': image,
459
- 'og_image': og_image,
460
- 'via': _domain(url),
461
- 'images': images
462
- }
463
- except Exception as e:
464
- return {'title': url, 'summary': '', 'text': '', 'image': '', 'og_image': '', 'via': _domain(url), 'error': str(e)}
465
 
466
 
467
- def web_context(topic: str, limit: int = 5) -> tuple:
468
- """Get web context for a topic. Returns (context_text, sources_list)."""
469
- sources = []
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
470
  try:
471
- # Try Google News RSS
472
- rss_url = f"https://news.google.com/rss/search?q={quote_plus(topic)}&hl=vi&gl=VN&ceid=VN:vi"
473
- r = requests.get(rss_url, headers=HEADERS, timeout=15)
474
- r.encoding = 'utf-8'
475
- soup = BeautifulSoup(r.text, 'xml')
476
- for it in soup.find_all('item')[:limit]:
477
- title = it.find('title').get_text(' ', strip=True) if it.find('title') else ''
478
- link = it.find('link').get_text(strip=True) if it.find('link') else ''
479
- if title and link:
480
- sources.append({'title': title, 'url': link, 'via': _domain(link)})
481
  except Exception:
482
  pass
483
-
484
- context = f'Trên mạng có nhiều bài viết về "{topic}". Một số nguồn: ' + ', '.join([s.get('title', '') for s in sources[:3]])
485
- return context, sources
 
486
 
487
 
488
- # ===== SHORT FRAME FUNCTION (required by ai_patch.py) =====
489
- def _make_short_frame(post, img_path, out_path):
490
- """Create a short video frame from post and image.
491
-
492
- Called by ai_patch.py _make_short_frame_full when Image is available.
493
- """
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
494
  if Image is None:
495
- # Create a minimal frame without PIL - just return success
496
- # The caller should handle this case
497
- return False
498
-
499
  W, H = 1080, 1920
500
- bg = Image.new("RGB", (W, H), (14, 14, 14))
501
-
502
  try:
503
  im = Image.open(img_path).convert("RGB")
504
- target = (1080, 760)
505
- im_ratio = im.width / max(1, im.height)
506
- target_ratio = target[0] / target[1]
507
-
508
- if im_ratio > target_ratio:
509
- new_h = target[1]
510
- new_w = int(new_h * im_ratio)
511
  else:
512
- new_w = target[0]
513
- new_h = int(new_w / im_ratio)
514
-
515
- im = im.resize((new_w, new_h))
516
- left = (new_w - target[0]) // 2
517
- top = (new_h - target[1]) // 2
518
- im = im.crop((left, top, left + target[0], top + target[1]))
519
  bg.paste(im, (0, 0))
520
  except Exception:
521
  pass
522
-
523
  draw = ImageDraw.Draw(bg)
524
-
525
  try:
526
- font_title = ImageFont.truetype("/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf", 54)
527
- font_body = ImageFont.truetype("/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf", 38)
 
528
  except Exception:
529
- font_title = font_body = None
530
-
531
- draw.rectangle((0, 720, W, H), fill=(14, 14, 14))
532
- margin = 48
533
- maxw = W - margin * 2
534
-
535
- y = 830
536
- for ln in _wrap_text(draw, post.get("title", ""), font_title, maxw, 4):
537
- draw.text((margin, y), ln, fill=(255, 255, 255), font=font_title)
538
- y += 66
539
-
540
- y += 18
541
- text = post.get("text", "")
542
- text = re.sub(r"Nguồn tham khảo:.*", "", text, flags=re.S).strip()
543
- body_lines = _wrap_text(draw, text, font_body, maxw, 14)
544
- for ln in body_lines:
545
- draw.text((margin, y), ln, fill=(220, 220, 220), font=font_body)
546
- y += 50
547
- if y > 1640:
548
- break
549
-
550
  bg.save(out_path, quality=92)
551
- return True
552
 
553
 
554
- def _wrap_text(draw, text, font, max_width, max_lines):
555
- """Helper for wrapping text in frames."""
556
- words = _clean_text(text).split()
557
- lines, cur = [], ""
558
- for w in words:
559
- test = (cur + " " + w).strip()
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
560
  try:
561
- width = draw.textbbox((0, 0), test, font=font)[2]
562
- except Exception:
563
- width = len(test) * 20
564
- if width <= max_width:
565
- cur = test
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
566
  else:
567
- if cur:
568
- lines.append(cur)
569
- cur = w
570
- if len(lines) >= max_lines:
571
- break
572
- if cur and len(lines) < max_lines:
573
- lines.append(cur)
574
- return lines
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
13
  from fastapi import Request, Query
14
  from fastapi.responses import HTMLResponse, JSONResponse, FileResponse
15
 
16
+ from main import app
 
 
 
 
 
 
 
 
 
17
 
18
  # Import wall store from main.py so we read/write the SAME file
19
  try:
 
41
  def _web_context(topic):
42
  return ""
43
 
 
 
 
 
44
  try:
45
  from huggingface_hub import AsyncInferenceClient
46
  except Exception:
 
60
 
61
 
62
  def _hf_token():
63
+ for k in ("HF_TOKEN", "HUGGINGFACEHUB_API_TOKEN", "HUGGING_FACE_HUB_TOKEN", "HF_API_TOKEN"):
64
  v = os.getenv(k, "").strip()
65
  if v:
66
  return v
67
  return ""
68
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
69
  HF_TOKEN = _hf_token()
70
  QWEN_VL_MODEL = os.getenv("QWEN_VL_MODEL", "Qwen/Qwen2.5-VL-7B-Instruct")
71
+ # Fast TEXT models for summaries that don't need vision (much faster than the VL model).
72
  QWEN_TEXT_MODELS = [m.strip() for m in os.getenv(
73
  "QWEN_TEXT_MODELS",
74
  "Qwen/Qwen2.5-72B-Instruct,meta-llama/Llama-3.3-70B-Instruct,Qwen/Qwen2.5-7B-Instruct"
75
  ).split(",") if m.strip()]
76
+ _WORKING_MODEL_TEXT = None # cached last-working text model
77
+ _WORKING_MODEL_VL = None # cached last-working vision model
78
  DATA_DIR = "/data" if os.path.isdir("/data") else "/app/data"
79
  SHORTS_DIR = os.path.join(DATA_DIR, "ai_shorts")
80
  HEADERS = {
 
83
  }
84
  LAST_QWEN_ERROR = ""
85
 
86
+ # ===== TTS VOICE CONFIG =====
87
+ # Multilingual neural voices grouped by country/language
88
+ # Format: key -> {id, gender, name, country, lang, flag}
89
+ TTS_VOICES = {
90
+ # === VIETNAM (edge-tts) ===
91
+ "hoaimy": {"id": "vi-VN-HoaiMyNeural", "gender": "female", "name": "Hoài My", "country": "Việt Nam", "lang": "vi", "flag": "🇻🇳", "engine": "edge"},
92
+ "namminh": {"id": "vi-VN-NamMinhNeural", "gender": "male", "name": "Nam Minh", "country": "Việt Nam", "lang": "vi", "flag": "🇻🇳", "engine": "edge"},
93
+ # === gTTS (Google, tiếng Việt cơ bản) ===
94
+ "gtts_vi": {"id": "gtts", "gender": "female", "name": "gTTS Google", "country": "Việt Nam", "lang": "vi", "flag": "🇻🇳", "engine": "gtts"},
95
+ # === MULTILINGUAL (đa ngôn ngữ — đọc được tiếng Việt + nhiều thứ tiếng) ===
96
+ "en_au_william": {"id": "en-AU-WilliamMultilingualNeural", "gender": "male", "name": "William (Đa NN)", "country": "Đa ngôn ngữ", "lang": "multi", "flag": "🌐", "engine": "edge"},
97
+ "en_us_andrew": {"id": "en-US-AndrewMultilingualNeural", "gender": "male", "name": "Andrew (Đa NN)", "country": "Đa ngôn ngữ", "lang": "multi", "flag": "🌐", "engine": "edge"},
98
+ "en_us_ava": {"id": "en-US-AvaMultilingualNeural", "gender": "female", "name": "Ava (Đa NN)", "country": "Đa ngôn ngữ", "lang": "multi", "flag": "🌐", "engine": "edge"},
99
+ "en_us_brian": {"id": "en-US-BrianMultilingualNeural", "gender": "male", "name": "Brian (Đa NN)", "country": "Đa ngôn ngữ", "lang": "multi", "flag": "🌐", "engine": "edge"},
100
+ "en_us_emma": {"id": "en-US-EmmaMultilingualNeural", "gender": "female", "name": "Emma (Đa NN)", "country": "Đa ngôn ngữ", "lang": "multi", "flag": "🌐", "engine": "edge"},
101
+ "fr_vivienne": {"id": "fr-FR-VivienneMultilingualNeural","gender": "female", "name": "Vivienne (Đa NN)","country": "Đa ngôn ngữ", "lang": "multi", "flag": "🌐", "engine": "edge"},
102
+ "fr_remy": {"id": "fr-FR-RemyMultilingualNeural", "gender": "male", "name": "Rémy (Đa NN)", "country": "Đa ngôn ngữ", "lang": "multi", "flag": "🌐", "engine": "edge"},
103
+ "de_seraphina": {"id": "de-DE-SeraphinaMultilingualNeural","gender": "female","name": "Seraphina (Đa NN)","country": "Đa ngôn ngữ","lang": "multi", "flag": "🌐", "engine": "edge"},
104
+ "de_florian": {"id": "de-DE-FlorianMultilingualNeural", "gender": "male", "name": "Florian (Đa NN)", "country": "Đa ngôn ngữ", "lang": "multi", "flag": "🌐", "engine": "edge"},
105
+ "it_giuseppe": {"id": "it-IT-GiuseppeMultilingualNeural","gender": "male", "name": "Giuseppe (Đa NN)","country": "Đa ngôn ngữ", "lang": "multi", "flag": "🌐", "engine": "edge"},
106
+ "ko_hyunsu": {"id": "ko-KR-HyunsuMultilingualNeural", "gender": "male", "name": "Hyunsu (Đa NN)", "country": "Đa ngôn ngữ", "lang": "multi", "flag": "🌐", "engine": "edge"},
107
+ "pt_thalita": {"id": "pt-BR-ThalitaMultilingualNeural", "gender": "female", "name": "Thalita (Đa NN)", "country": "Đa ngôn ngữ", "lang": "multi", "flag": "🌐", "engine": "edge"},
108
+ }
109
+ TTS_DEFAULT_VOICE = "hoaimy"
110
+ TTS_DEFAULT_SPEED = 1.2 # 1.2x speed for faster reading
111
 
112
+ # Topic -> voice mapping (auto-detect based on topic keywords)
113
+ TOPIC_VOICE_MAP = {
114
+ # Sports -> male voice
115
+ "bóng đá": "namminh", "thể thao": "namminh", "world cup": "namminh",
116
+ "premier league": "namminh", "champions league": "namminh", "la liga": "namminh",
117
+ "serie a": "namminh", "bundesliga": "namminh", "v-league": "namminh",
118
+ "tennis": "namminh", "olympic": "namminh", "f1": "namminh", "moto": "namminh",
119
+ # Lifestyle/Health/Entertainment -> female voice
120
+ "sức khỏe": "hoaimy", "làm đẹp": "hoaimy", "giải trí": "hoaimy",
121
+ "âm nhạc": "hoaimy", "phim": "hoaimy", "thời trang": "hoaimy",
122
+ "ẩm thực": "hoaimy", "du lịch": "hoaimy", "gia đình": "hoaimy",
123
+ "tình yêu": "hoaimy", "hôn nhân": "hoaimy", "mẹ và bé": "hoaimy",
124
+ # Tech/Science -> male voice
125
+ "công nghệ": "namminh", "ai": "namminh", "robot": "namminh",
126
+ "khoa học": "namminh", "vũ trụ": "namminh", "điện thoại": "namminh",
127
+ "laptop": "namminh", "game": "namminh",
128
+ # News/Politics/Economy -> male voice
129
+ "chính trị": "namminh", "kinh tế": "namminh", "tài chính": "namminh",
130
+ "chứng khoán": "namminh", "ngân hàng": "namminh", "thị trường": "namminh",
131
+ "xã hội": "namminh", "pháp luật": "namminh", "giáo dục": "namminh",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
132
  }
133
 
134
 
135
+ def _detect_voice_for_topic(title: str, text: str) -> str:
136
+ """Auto-detect the best voice based on topic keywords."""
137
+ combined = (title + " " + text[:500]).lower()
138
+ for keyword, voice_id in TOPIC_VOICE_MAP.items():
139
+ if keyword in combined:
140
+ return voice_id
141
+ return TTS_DEFAULT_VOICE
 
 
 
 
 
 
 
 
 
 
 
 
 
142
 
143
 
144
+ # ===== EMOTION (CẢM XÚC) FOR TTS =====
145
+ # edge-tts does NOT support Azure express-as styles, so emotion is simulated
146
+ # with pitch + rate(speed multiplier) + volume. Each preset is a delta.
147
+ # rate_mul multiplies the user/auto speed; pitch is absolute Hz; volume is percent.
148
+ EMOTION_PRESETS = {
149
+ "vui": {"label": "Vui tươi", "emoji": "😊", "rate_mul": 1.06, "pitch": "+15Hz", "volume": "+6%"},
150
+ "hao_hung": {"label": "Hào hứng", "emoji": "🔥", "rate_mul": 1.12, "pitch": "+24Hz", "volume": "+12%"},
151
+ "nghiem": {"label": "Nghiêm túc", "emoji": "📰", "rate_mul": 1.00, "pitch": "-3Hz", "volume": "+0%"},
152
+ "tram": {"label": "Trầm ấm", "emoji": "🌙", "rate_mul": 0.94, "pitch": "-10Hz", "volume": "+0%"},
153
+ "buon": {"label": "Buồn/Xúc động", "emoji": "💧", "rate_mul": 0.88, "pitch": "-18Hz", "volume": "-4%"},
154
+ "trung_tinh":{"label": "Trung tính", "emoji": "🎙️", "rate_mul": 1.00, "pitch": "+0Hz", "volume": "+0%"},
155
+ }
156
+ EMOTION_DEFAULT = "trung_tinh"
157
 
158
+ # Topic keyword -> emotion. Checked in order; first match wins.
159
+ TOPIC_EMOTION_MAP = {
160
+ # Sports / wins -> excited
161
+ "chiến thắng": "hao_hung", "vô địch": "hao_hung", "world cup": "hao_hung",
162
+ "bóng đá": "hao_hung", "thể thao": "hao_hung", "ghi bàn": "hao_hung",
163
+ "champions league": "hao_hung", "premier league": "hao_hung", "chung kết": "hao_hung",
164
+ "olympic": "hao_hung", "kỷ lục": "hao_hung",
165
+ # Entertainment / lifestyle / good news -> cheerful
166
+ "giải trí": "vui", "âm nhạc": "vui", "phim": "vui", "lễ hội": "vui",
167
+ "du lịch": "vui", "ẩm thực": "vui", "thời trang": "vui", "ra mắt": "vui",
168
+ "khai trương": "vui", "tin vui": "vui", "hạnh phúc": "vui",
169
+ # Sad / accidents / loss -> sad
170
+ "tai nạn": "buon", "qua đời": "buon", "tử vong": "buon", "thiệt mạng": "buon",
171
+ "động đất": "buon", "lũ lụt": "buon", "thiên tai": "buon", "cháy": "buon",
172
+ "tang lễ": "buon", "mất tích": "buon", "thương tâm": "buon",
173
+ # Health / science / calm -> calm warm
174
+ "sức khỏe": "tram", "y tế": "tram", "bệnh": "tram", "dinh dưỡng": "tram",
175
+ "tâm lý": "tram", "thiền": "tram", "giấc ngủ": "tram",
176
+ # News / politics / economy / law -> serious
177
+ "chính trị": "nghiem", "kinh tế": "nghiem", "tài chính": "nghiem",
178
+ "chứng khoán": "nghiem", "pháp luật": "nghiem", "tòa án": "nghiem",
179
+ "ngân hàng": "nghiem", "thị trường": "nghiem", "lạm phát": "nghiem",
180
+ "công nghệ": "nghiem", "ai": "nghiem", "khoa học": "nghiem", "giáo dục": "nghiem",
181
+ }
182
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
183
 
184
+ def _detect_emotion_for_topic(title: str, text: str) -> str:
185
+ """Auto-detect emotion (cảm xúc) from topic/content keywords."""
186
+ combined = (title + " " + text[:600]).lower()
187
+ for keyword, emo in TOPIC_EMOTION_MAP.items():
188
+ if keyword in combined:
189
+ return emo
190
+ return EMOTION_DEFAULT
191
 
 
 
 
192
 
193
+ def _detect_voice_emotion(title: str, text: str) -> tuple:
194
+ """Return (voice_id, emotion_id) auto-chosen for the article's topic."""
195
+ return _detect_voice_for_topic(title, text), _detect_emotion_for_topic(title, text)
196
 
 
 
197
 
198
+ # ===== TEXT HELPERS =====
199
+ def _clean_text(s: str) -> str:
200
+ s = html_lib.unescape(s or "")
201
+ return re.sub(r"\s+", " ", s).strip()
202
+
203
+ def _domain(u):
204
  try:
205
+ return urlparse(u).netloc.replace("www.", "")
 
 
206
  except Exception:
207
+ return ""
 
208
 
209
+ def _safe_name(s):
210
+ return re.sub(r"[^a-zA-Z0-9_-]+", "_", str(s))[:80]
211
 
 
 
 
 
 
 
 
 
 
 
212
 
213
+ # ===== CLEAN AI OUTPUT =====
214
+ def _clean_ai_output(text: str) -> str:
215
+ """Remove markdown artifacts, instruction leakage, and aggressively dedup content."""
216
+ if not text:
217
+ return ""
218
+ # Remove markdown headings, bold, italic, horizontal rules
219
+ text = re.sub(r'^#{1,6}\s+', '', text, flags=re.MULTILINE)
220
+ text = re.sub(r'\*\*([^*]+)\*\*', r'\1', text)
221
+ text = re.sub(r'\*([^*]+)\*', r'\1', text)
222
+ text = re.sub(r'^---+\s*$', '', text, flags=re.MULTILINE)
223
+ text = re.sub(r'^[-*_]{3,}\s*$', '', text, flags=re.MULTILINE)
224
+ # Remove common AI instruction leakage phrases (entire line)
225
+ leakage = [
226
+ r'Dưới đây là', r'Theo yêu cầu', r'Tôi sẽ viết', r'Tôi sẽ tóm tắt',
227
+ r'Đây là bài', r'Đây là nội dung', r'Bài viết sau đây',
228
+ r'Nội dung (tóm tắt|chính)', r'Nhiệm vụ', r'Vai trò', r'Tôi là',
229
+ r'Dựa trên.*tôi sẽ', r'Hãy', r'Bạn cần', r'Đọc bài viết',
230
+ r'Tôi xin', r'Xin chào', r'Trân trọng', r'Kính thưa',
231
+ r'Dựa trên.*dưới đây', r'Sau đây là', r'Dưới đây là bài',
232
+ ]
233
+ for phrase in leakage:
234
+ text = re.sub(r'^' + phrase + r'[^\n]*\n?', '', text, flags=re.MULTILINE | re.IGNORECASE)
235
+ text = re.sub(r'\n{3,}', '\n\n', text)
236
+ # --- Aggressive dedup: split into sentences, remove any that repeats ---
237
+ # Normalize: collapse whitespace, strip
238
+ def _norm(s):
239
+ return re.sub(r'\s+', ' ', s.strip().lower())
240
+
241
+ # Split by sentence-ending punctuation (keep delimiters)
242
+ raw_parts = re.split(r'(?<=[.!?])\s+', text.strip())
243
+ seen_sentences = set()
244
+ unique_parts = []
245
+ for part in raw_parts:
246
+ n = _norm(part)
247
+ # Skip near-duplicate: if >70% of an existing seen sentence matches
248
+ is_dup = False
249
+ if n:
250
+ if n in seen_sentences:
251
+ is_dup = True
252
+ else:
253
+ partial = re.sub(r'\W+', '', n)
254
+ for seen in seen_sentences:
255
+ seen_clean = re.sub(r'\W+', '', seen)
256
+ # Check substring match for very similar sentences
257
+ if partial and seen_clean and (
258
+ partial in seen_clean or seen_clean in partial
259
+ ):
260
+ shorter = min(len(partial), len(seen_clean))
261
+ longer = max(len(partial), len(seen_clean))
262
+ if shorter > 20 and shorter / longer > 0.75:
263
+ is_dup = True
264
+ break
265
+ if is_dup:
266
+ continue
267
+ if n:
268
+ seen_sentences.add(n)
269
+ unique_parts.append(part)
270
 
271
+ result = ' '.join(unique_parts).strip()
272
+ # Final pass: remove any remaining consecutive duplicate lines
273
+ lines = result.split('\n')
274
+ final_lines = []
275
+ prev_line = ""
276
+ for line in lines:
277
+ stripped = line.strip()
278
+ if stripped and stripped == prev_line:
279
+ continue
280
+ final_lines.append(line)
281
+ prev_line = stripped
282
+ result = '\n'.join(final_lines).strip()
283
+ return result
284
 
285
 
286
+ # ===== EXTRACT ALL IMAGES FROM ARTICLE =====
287
+ def _extract_all_images(soup, base_url: str) -> List[Dict]:
288
+ """Extract ALL content images from an article page using multi-strategy approach."""
289
+ images = []
290
+ seen_urls = set()
291
+ skip_patterns = [
292
+ "avatar", "icon", "logo", "button", "banner-ad", "tracking",
293
+ "beacon", "pixel", "1x1", "spacer", "emoji", "sprite", "placeholder",
294
+ "advertisement", "ads", "widget", "sidebar", "footer-logo",
295
+ ]
296
 
297
+ def _add_image(src: str, alt: str = "", source_tag: str = "img"):
298
+ if not src or src.startswith("data:"):
299
+ return
300
+ abs_url = urljoin(base_url, src.strip())
301
+ if abs_url in seen_urls:
302
+ return
303
+ # Skip non-content images by URL pattern
304
+ if any(p in abs_url.lower() for p in skip_patterns):
305
+ return
306
+ # Skip very small images (likely icons)
307
+ try:
308
+ parsed = urlparse(abs_url)
309
+ path = parsed.path.lower()
310
+ if any(path.endswith(ext) for ext in ['.svg', '.ico', '.gif']):
311
+ return
312
+ except Exception:
313
+ pass
314
+ seen_urls.add(abs_url)
315
+ images.append({"url": abs_url, "alt": alt, "source": source_tag})
316
+
317
+ # Strategy 1: Standard <img> tags with all lazy-load attributes
318
+ for img in soup.find_all("img"):
319
+ src = (img.get("src") or img.get("data-src") or img.get("data-lazy-src") or
320
+ img.get("data-original") or img.get("data-srcset", "").split(",")[0].strip().split(" ")[0])
321
+ _add_image(src, alt=img.get("alt", ""), source_tag="img")
322
+
323
+ # Strategy 2: srcset on <img>
324
+ for img in soup.find_all("img", srcset=True):
325
+ for part in img["srcset"].split(","):
326
+ part = part.strip()
327
+ if part:
328
+ _add_image(part.split(" ")[0], alt=img.get("alt", ""), source_tag="srcset")
329
+
330
+ # Strategy 3: <picture> with <source>
331
+ for picture in soup.find_all("picture"):
332
+ for source in picture.find_all("source"):
333
+ srcset = source.get("srcset", "")
334
+ for part in srcset.split(","):
335
+ part = part.strip()
336
+ if part:
337
+ _add_image(part.split(" ")[0], source_tag="picture/srcset")
338
+ fallback_img = picture.find("img")
339
+ if fallback_img:
340
+ _add_image(
341
+ fallback_img.get("src") or fallback_img.get("data-src"),
342
+ alt=fallback_img.get("alt", ""),
343
+ source_tag="picture/img"
344
+ )
345
+
346
+ # Strategy 4: WordPress CMS patterns
347
+ for img in soup.find_all("img", class_=re.compile(r"wp-image|size-large|size-full|aligncenter")):
348
+ _add_image(img.get("data-src") or img.get("src"),
349
+ alt=img.get("alt", ""), source_tag="wp-image")
350
+
351
+ # Strategy 5: Background images in style attributes
352
+ for tag in soup.find_all(style=re.compile(r"background-image")):
353
+ for m in re.findall(r'url\(["\']?(.*?)["\']?\)', tag.get("style", "")):
354
+ _add_image(m, source_tag="background-style")
355
+
356
+ # Strategy 6: og:image (featured/hero image)
357
+ og_image = soup.find("meta", property="og:image")
358
+ if og_image and og_image.get("content"):
359
+ _add_image(og_image["content"], source_tag="og:image")
360
+
361
+ # Strategy 7: twitter:image
362
+ tw_image = soup.find("meta", attrs={"name": "twitter:image"})
363
+ if tw_image and tw_image.get("content"):
364
+ _add_image(tw_image["content"], source_tag="twitter:image")
365
+
366
+ # Strategy 8: <figure> with <figcaption>
367
+ for figure in soup.find_all("figure"):
368
+ img = figure.find("img")
369
+ if img:
370
+ src = img.get("data-src") or img.get("src")
371
+ figcaption = figure.find("figcaption")
372
+ alt = figcaption.get_text(strip=True) if figcaption else img.get("alt", "")
373
+ _add_image(src, alt=alt, source_tag="figure")
374
+
375
+ # Strategy 9: <a> tags linking to images
376
+ for a in soup.find_all("a", href=True):
377
+ href = a["href"]
378
+ if any(href.lower().endswith(ext) for ext in [".jpg", ".jpeg", ".png", ".webp", ".gif"]):
379
+ _add_image(href, alt=a.get_text(strip=True)[:80], source_tag="link")
380
+
381
+ return images
382
+
383
+
384
+ # ===== JINA READER =====
385
+ def _reader_url(target_url: str) -> str:
386
+ safe = quote(target_url, safe=":/?#[]@!$&'()*+,;=%")
387
+ return "https://r.jina.ai/http://" + safe
388
+
389
+ def jina_reader_markdown(url: str) -> str:
390
+ jr = _reader_url(url)
391
+ r = requests.get(jr, headers={"Accept": "text/markdown,text/plain,*/*", "X-Return-Format": "markdown", "User-Agent": "Mozilla/5.0"}, timeout=35)
392
+ r.raise_for_status()
393
+ return r.text or ""
394
+
395
+ def _parse_jina_markdown(md: str, url: str):
396
+ lines = [x.rstrip() for x in (md or "").splitlines()]
397
+ title = ""; first_image = ""; all_images = []; content_lines = []; in_content = False
398
+ for ln in lines:
399
+ if ln.startswith("Title:") and not title:
400
+ title = _clean_text(ln.replace("Title:", "", 1)); continue
401
+ if ln.startswith("URL Source:"):
402
+ continue
403
+ if ln.startswith("Markdown Content:"):
404
+ in_content = True; continue
405
+ # Extract ALL images from markdown ![alt](url)
406
+ for mimg in re.finditer(r'!\[[^\]]*\]\((https?://[^)]+)\)', ln):
407
+ img_url = mimg.group(1)
408
+ if img_url not in all_images:
409
+ all_images.append(img_url)
410
+ if not first_image:
411
+ first_image = img_url
412
+ if in_content or (title and not ln.startswith("Title:")):
413
+ if ln.strip():
414
+ content_lines.append(ln)
415
+ text = "\n".join(content_lines)
416
+ text = re.sub(r'!\[[^\]]*\]\([^)]+\)', '', text)
417
+ paras = []
418
+ for part in re.split(r'\n{2,}|\n(?=#{1,3}\s)', text):
419
+ t = _clean_text(re.sub(r'^#{1,6}\s*', '', part))
420
+ if len(t) >= 40:
421
+ paras.append(t)
422
+ if len(paras) >= 35:
423
+ break
424
+ if not title and paras:
425
+ title = paras[0][:90]
426
+ return {"url": url, "title": title or url, "summary": paras[0] if paras else "",
427
+ "text": "\n".join(paras), "image": first_image,
428
+ "images": all_images, "via": "jina"}
429
+
430
+
431
+ # ===== WEB SCRAPE (with full image extraction) =====
432
+ def _best_content_block(soup):
433
+ best, best_score = None, 0
434
+ for el in soup.find_all(["article", "main", "section", "div"]):
435
+ ps = el.find_all("p")
436
+ txt = " ".join(p.get_text(" ", strip=True) for p in ps)
437
+ score = len(ps) * 100 + len(txt)
438
+ cls = " ".join(el.get("class", []))
439
+ if any(k in cls.lower() for k in ["content", "article", "detail", "body", "post", "entry"]):
440
+ score += 800
441
+ if score > best_score:
442
+ best, best_score = el, score
443
+ return best
444
+
445
+ def scrape_any_url_direct(url: str):
446
+ r = requests.get(url, headers=HEADERS, timeout=18)
447
+ if r.status_code in {401, 403, 406, 409, 429, 451, 503}:
448
+ raise RuntimeError(f"blocked status {r.status_code}")
449
+ r.encoding = "utf-8"
450
+ soup = BeautifulSoup(r.text, "lxml")
451
+ for tag in soup.find_all(["script", "style", "nav", "footer", "aside", "form", "noscript"]):
452
+ tag.decompose()
453
+
454
+ # Title
455
+ title = soup.find("h1").get_text(" ", strip=True) if soup.find("h1") else ""
456
+ if not title:
457
+ ogt = soup.find("meta", property="og:title") or soup.find("meta", attrs={"name": "title"})
458
+ title = ogt.get("content", "") if ogt else (soup.title.get_text(strip=True) if soup.title else "")
459
+
460
+ # Summary
461
+ desc_tag = soup.find("meta", property="og:description") or soup.find("meta", attrs={"name": "description"})
462
+ summary = desc_tag.get("content", "") if desc_tag else ""
463
+
464
+ # Featured image (og:image)
465
+ img_tag = soup.find("meta", property="og:image") or soup.find("meta", attrs={"name": "twitter:image"})
466
+ image = img_tag.get("content", "") if img_tag else ""
467
+ if image and image.startswith("//"):
468
+ image = "https:" + image
469
+
470
+ # Extract ALL images from the article
471
+ all_images = _extract_all_images(soup, url)
472
+ image_urls = [img["url"] for img in all_images]
473
+
474
+ # Ensure featured image is first
475
+ if image and image not in image_urls:
476
+ image_urls.insert(0, image)
477
+ elif image in image_urls:
478
+ image_urls.remove(image)
479
+ image_urls.insert(0, image)
480
+
481
+ # Content paragraphs
482
+ block = _best_content_block(soup) or soup
483
+ paras, seen_p = [], set()
484
+ for p in block.find_all("p"):
485
+ t = _clean_text(p.get_text(" ", strip=True))
486
+ if len(t) >= 40 and t not in seen_p:
487
+ seen_p.add(t)
488
+ paras.append(t)
489
+ if len(paras) >= 35:
490
+ break
491
+
492
+ if not title and paras:
493
+ title = paras[0][:90]
494
 
 
 
495
  return {
496
+ "url": url, "title": title or url, "summary": paras[0] if paras else "",
497
+ "text": "\n".join(paras), "image": image_urls[0] if image_urls else "",
498
+ "images": image_urls, "via": _domain(url)
 
 
 
 
 
499
  }
500
 
501
+ def scrape_any_url(url: str):
502
+ """Try direct scrape first, fall back to Jina Reader."""
503
+ data = scrape_any_url_direct(url)
504
+ raw_text = (data.get("summary", "") + "\n" + data.get("text", "")).strip()
505
+ if len(raw_text) >= 120:
506
+ return data
507
+ try:
508
+ md = jina_reader_markdown(url)
509
+ if md:
510
+ jr = _parse_jina_markdown(md, url)
511
+ if jr.get("text"):
512
+ if data.get("title") and data["title"] != url:
513
+ jr["title"] = data["title"]
514
+ if data.get("image"):
515
+ jr["image"] = data["image"]
516
+ if data.get("images"):
517
+ jr["images"] = data["images"]
518
+ jr["via"] = data.get("via", _domain(url)) + " + jina"
519
+ return jr
520
+ except Exception:
521
+ pass
522
+ return data
523
 
 
 
 
 
 
 
524
 
525
+ # ===== POLLINATIONS IMAGE =====
526
+ def pollinations_image_url(topic: str) -> str:
527
+ prompt = "editorial illustration, Vietnamese news, " + topic
528
+ return "https://image.pollinations.ai/prompt/" + quote(prompt, safe="") + "?width=1024&height=576&nologo=true"
529
 
530
+
531
+ @app.get("/api/ai/probe")
532
+ async def api_ai_probe():
533
+ """Diagnostic: test which chat models actually work on this token + their latency."""
534
+ import time as _t
535
+ tok = _hf_token()
536
+ out = []
537
+ extra = os.getenv("PROBE_MODELS", "").split(",")
538
+ cand = [m.strip() for m in extra if m.strip()] + [
539
+ "Qwen/Qwen2.5-VL-7B-Instruct", "Qwen/Qwen2.5-VL-3B-Instruct", "Qwen/Qwen2-VL-7B-Instruct",
540
+ "Qwen/Qwen2.5-VL-72B-Instruct", "Qwen/Qwen3-8B", "Qwen/Qwen3-4B", "Qwen/Qwen3-32B",
541
+ "Qwen/Qwen2.5-7B-Instruct-1M", "meta-llama/Llama-3.2-3B-Instruct"]
542
+ seen = set()
543
+ for m in cand:
544
+ if not m or m in seen:
545
+ continue
546
+ seen.add(m)
547
+ t0 = _t.time()
548
+ try:
549
+ c = AsyncInferenceClient(provider="auto", api_key=tok, timeout=40)
550
+ r = await c.chat_completion(model=m, messages=[{"role": "user", "content": "Trả lời đúng 1 từ: xin chào"}], max_tokens=10)
551
+ out.append({"model": m, "ok": True, "sec": round(_t.time() - t0, 1), "txt": (r.choices[0].message.content or "")[:30]})
552
+ except Exception as e:
553
+ out.append({"model": m, "ok": False, "sec": round(_t.time() - t0, 1), "err": (type(e).__name__ + ": " + str(e))[-300:]})
554
+ return JSONResponse({"results": out})
555
+
556
+
557
+ # ===== QWEN AI (strict, concise) =====
558
+ async def qwen_generate(prompt: str, image_url: Optional[str] = None, max_tokens: int = 500, image_urls: Optional[List[str]] = None):
559
+ global LAST_QWEN_ERROR, HF_TOKEN
560
+ HF_TOKEN = _hf_token()
561
+ if not HF_TOKEN:
562
+ LAST_QWEN_ERROR = "Không tìm thấy token"
563
+ return None
564
+ if not AsyncInferenceClient:
565
+ LAST_QWEN_ERROR = "Thiếu huggingface_hub"
566
+ return None
567
+ errors = []; models = []
568
+ has_images = bool(image_urls) or bool(image_url)
569
+ if has_images:
570
+ # Vision needed -> VL models
571
+ candidate = [QWEN_VL_MODEL, "Qwen/Qwen2.5-VL-7B-Instruct", "Qwen/Qwen2.5-VL-3B-Instruct"]
572
+ else:
573
+ # Text-only summary -> FAST text models first, VL only as last resort.
574
+ candidate = QWEN_TEXT_MODELS + [QWEN_VL_MODEL]
575
+ # Use the last-known-working model first to avoid wasting time on unavailable models.
576
+ global _WORKING_MODEL_TEXT, _WORKING_MODEL_VL
577
+ cached_ok = _WORKING_MODEL_VL if has_images else _WORKING_MODEL_TEXT
578
+ if cached_ok and cached_ok in candidate:
579
+ candidate = [cached_ok] + [m for m in candidate if m != cached_ok]
580
+ for m in candidate:
581
+ if m and m not in models:
582
+ models.append(m)
583
+ for model in models:
584
+ try:
585
+ client = AsyncInferenceClient(provider="auto", api_key=HF_TOKEN, timeout=60)
586
+ content = []
587
+ # Collect all images: image_urls list takes priority, fall back to single image_url
588
+ all_img_urls = []
589
+ if image_urls:
590
+ all_img_urls = image_urls[:6] # max 6 images to avoid context overflow
591
+ elif image_url:
592
+ all_img_urls = [image_url]
593
+ for img_u in all_img_urls:
594
+ if img_u and img_u.startswith("http"):
595
+ content.append({"type": "image_url", "image_url": {"url": img_u}})
596
+ content.append({"type": "text", "text": prompt})
597
+ messages = [
598
+ {"role": "system", "content": (
599
+ "Bạn là biên tập viên báo điện tử tiếng Việt. "
600
+ "NHIỆM VỤ: Chỉ TÓM TẮT nội dung, KHÔNG viết lại bài đầy đủ. "
601
+ "QUY TẮC CỨNG: "
602
+ "(1) KHÔNG lặp lại bất kỳ nội dung nào — mỗi ý chỉ xuất hiện ĐÚNG 1 LẦN. "
603
+ "(2) Nếu 2 câu diễn đạt cùng 1 ý → bỏ cây thứ 2. "
604
+ "(3) KHÔNG dùng Markdown (##, **, ---, *). "
605
+ "(4) KHÔNG viết 'Dưới đây là', 'Tôi sẽ', 'Theo yêu cầu', 'Nhiệm vụ', 'Vai trò', 'Đây là bài tóm tắt'. "
606
+ "(5) KHÔNG bịa thông tin ngoài nguồn. "
607
+ "(6) Chỉ viết ĐOẠN VĂN THUẦN, không bullet points. "
608
+ "(7) Tối đa 200 từ. Ngắn gọn, súc tích."
609
+ )},
610
+ {"role": "user", "content": content}
611
+ ]
612
+ resp = await client.chat_completion(model=model, messages=messages, max_tokens=max_tokens, temperature=0.3, top_p=0.8)
613
+ txt = (resp.choices[0].message.content or "").strip()
614
+ if txt:
615
+ LAST_QWEN_ERROR = ""
616
+ if has_images: _WORKING_MODEL_VL = model
617
+ else: _WORKING_MODEL_TEXT = model
618
+ return txt
619
+ except Exception as e:
620
+ errors.append(f"{model}: {type(e).__name__}: {str(e)[:220]}")
621
+ LAST_QWEN_ERROR = " | ".join(errors) or "Qwen không trả nội dung."
622
+ print("[qwen errors]", LAST_QWEN_ERROR)
623
+ return None
624
+
625
+
626
+ # ===== TTS GENERATION =====
627
+ async def _generate_tts_edge(text: str, voice_id: str, speed: float, out_path: str, emotion: str = None):
628
+ """Generate TTS using edge-tts (or gTTS if engine=gtts) with voice, speed & emotion control.
629
+
630
+ Emotion (cảm xúc) is simulated via pitch + rate + volume (edge-tts has no express-as).
631
  """
632
+ vcfg = TTS_VOICES.get(voice_id, TTS_VOICES[TTS_DEFAULT_VOICE])
633
+ # gTTS engine (no voice/speed/emotion control)
634
+ if vcfg.get("engine") == "gtts":
635
+ _generate_tts_gtts(text, out_path)
636
+ return
637
+ if edge_tts is None:
638
+ raise RuntimeError("edge-tts chưa cài đặt")
639
+ voice = vcfg["id"]
640
+ emo = EMOTION_PRESETS.get(emotion or EMOTION_DEFAULT, EMOTION_PRESETS[EMOTION_DEFAULT])
641
+ # Apply emotion rate multiplier on top of base speed
642
+ eff_speed = speed * emo.get("rate_mul", 1.0)
643
+ pct = int(round((eff_speed - 1.0) * 100))
644
+ rate = f"+{pct}%" if pct >= 0 else f"{pct}%"
645
+ pitch = emo.get("pitch", "+0Hz")
646
+ volume = emo.get("volume", "+0%")
647
+ communicate = edge_tts.Communicate(text, voice, rate=rate, pitch=pitch, volume=volume)
648
+ await communicate.save(out_path)
649
+
650
+ def _generate_tts_gtts(text: str, out_path: str):
651
+ """Fallback TTS using gTTS (no voice/speed control)."""
652
+ if gTTS is None:
653
+ raise RuntimeError("gTTS chưa cài đặt")
654
+ gTTS(text, lang="vi").save(out_path)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
655
 
656
 
657
+ # ===== SHORT VIDEO GENERATION (multi-segment: each key point with its own image) =====
658
+ def _download_image(url, fallback_topic, out_path):
659
+ """Download an image (un-proxying our own /api/proxy/img). Falls back to generated image."""
660
+ if url:
661
+ u = url
662
+ m = re.search(r'/api/proxy/img\?url=(.+)$', u)
663
+ if m:
664
+ from urllib.parse import unquote
665
+ u = unquote(m.group(1))
666
+ try:
667
+ r = requests.get(u, headers={**HEADERS, "Referer": "https://dantri.com.vn/"}, timeout=15)
668
+ if r.status_code == 200 and len(r.content) > 1000:
669
+ with open(out_path, "wb") as f:
670
+ f.write(r.content)
671
+ if Image:
672
+ Image.open(out_path).verify()
673
+ return out_path
674
+ except Exception:
675
+ pass
676
+ gen = pollinations_image_url(fallback_topic)
677
  try:
678
+ r = requests.get(gen, headers=HEADERS, timeout=25)
679
+ if r.status_code == 200 and len(r.content) > 1000:
680
+ with open(out_path, "wb") as f:
681
+ f.write(r.content)
682
+ return out_path
 
 
 
 
 
683
  except Exception:
684
  pass
685
+ if Image:
686
+ Image.new("RGB", (1080, 980), (30, 55, 42)).save(out_path)
687
+ return out_path
688
+ raise RuntimeError("Không tạo được ảnh")
689
 
690
 
691
+ def _split_keypoint_sentences(text, max_points=6):
692
+ """Split summary text into key points: prefer bullet markers, else sentences."""
693
+ text = _clean_text(text)
694
+ parts = re.split(r'\s*•\s*', text)
695
+ good = [p.strip() for p in parts if len(p.strip()) > 20]
696
+ if len(good) >= 2:
697
+ # Explicit bullet points: keep each one as-is (never merge).
698
+ return good[:max_points]
699
+ # Fallback: split into sentences and merge orphan short fragments.
700
+ pts = [p.strip() for p in re.split(r'(?<=[.!?])\s+', text) if len(p.strip()) > 20]
701
+ out = []
702
+ for p in pts:
703
+ if out and len(p) < 40:
704
+ out[-1] = (out[-1] + " " + p).strip()
705
+ else:
706
+ out.append(p)
707
+ return out[:max_points] if out else ([text] if text else [])
708
+
709
+
710
+ def _build_keypoints(post, max_points=6):
711
+ """Return [{text, image}] pairing each key point with its own image."""
712
+ slides = post.get("slides") or []
713
+ images = post.get("images") or ([post.get("img")] if post.get("img") else [])
714
+ images = [i for i in images if i]
715
+ if slides:
716
+ kps = []
717
+ for i, s in enumerate(slides[:max_points]):
718
+ t = _clean_text(s.get("text", ""))
719
+ img = s.get("image") or (images[i] if i < len(images) else (images[-1] if images else ""))
720
+ if t:
721
+ kps.append({"text": t, "image": img})
722
+ if kps:
723
+ return kps
724
+ points = _split_keypoint_sentences(post.get("text", ""), max_points)
725
+ kps = []
726
+ for i, t in enumerate(points):
727
+ img = images[i] if i < len(images) else (images[-1] if images else "")
728
+ kps.append({"text": t, "image": img})
729
+ if not kps:
730
+ kps = [{"text": _clean_text(post.get("title", "")) or "VNEWS", "image": images[0] if images else ""}]
731
+ return kps
732
+
733
+
734
+ def _wrap_text(draw, text, font, max_w):
735
+ words = text.split()
736
+ lines, cur = [], ""
737
+ for w in words:
738
+ test = (cur + " " + w).strip()
739
+ if draw.textlength(test, font=font) <= max_w:
740
+ cur = test
741
+ else:
742
+ if cur:
743
+ lines.append(cur)
744
+ cur = w
745
+ if cur:
746
+ lines.append(cur)
747
+ return lines
748
+
749
+
750
+ def _make_segment_frame(title, point_text, img_path, idx, total, out_path):
751
+ """Render a 1080x1920 vertical frame: image on top, key point text below."""
752
  if Image is None:
753
+ raise RuntimeError("Pillow chưa sẵn sàng")
 
 
 
754
  W, H = 1080, 1920
755
+ IMG_H = 980
756
+ bg = Image.new("RGB", (W, H), (12, 14, 18))
757
  try:
758
  im = Image.open(img_path).convert("RGB")
759
+ tr = W / IMG_H
760
+ ir = im.width / im.height
761
+ if ir > tr:
762
+ nh = IMG_H; nw = int(nh * ir)
 
 
 
763
  else:
764
+ nw = W; nh = int(nw / ir)
765
+ im = im.resize((nw, nh))
766
+ left = (nw - W) // 2; top = (nh - IMG_H) // 2
767
+ im = im.crop((left, top, left + W, top + IMG_H))
 
 
 
768
  bg.paste(im, (0, 0))
769
  except Exception:
770
  pass
 
771
  draw = ImageDraw.Draw(bg)
772
+ draw.rectangle((0, IMG_H, W, H), fill=(12, 14, 18))
773
  try:
774
+ f_label = ImageFont.truetype("/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf", 34)
775
+ f_title = ImageFont.truetype("/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf", 46)
776
+ f_point = ImageFont.truetype("/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf", 52)
777
  except Exception:
778
+ f_label = f_title = f_point = ImageFont.load_default()
779
+ draw.text((54, IMG_H + 24), "VNEWS · Tường AI", fill=(92, 184, 122), font=f_label)
780
+ cnt = f"{idx + 1}/{total}"
781
+ draw.text((W - 54 - draw.textlength(cnt, font=f_label), IMG_H + 24), cnt, fill=(240, 192, 64), font=f_label)
782
+ y = IMG_H + 86
783
+ for ln in _wrap_text(draw, _clean_text(title), f_title, W - 108)[:2]:
784
+ draw.text((54, y), ln, fill=(255, 255, 255), font=f_title)
785
+ y += 56
786
+ y += 16
787
+ for ln in _wrap_text(draw, _clean_text(point_text), f_point, W - 108)[:11]:
788
+ draw.text((54, y), ln, fill=(225, 230, 235), font=f_point)
789
+ y += 64
 
 
 
 
 
 
 
 
 
790
  bg.save(out_path, quality=92)
791
+ return out_path
792
 
793
 
794
+ def _ffmpeg_bin():
795
+ return os.environ.get("FFMPEG_BIN", "ffmpeg")
796
+
797
+
798
+ def _audio_duration(path):
799
+ try:
800
+ out = subprocess.run([_ffmpeg_bin(), "-i", path], capture_output=True, text=True, timeout=30).stderr
801
+ m = re.search(r"Duration:\s*(\d+):(\d+):(\d+\.\d+)", out)
802
+ if m:
803
+ h, mi, s = m.groups()
804
+ return int(h) * 3600 + int(mi) * 60 + float(s)
805
+ except Exception:
806
+ pass
807
+ return 0.0
808
+
809
+
810
+ async def _generate_short_video(post, post_id: str, voice_id: str = None, speed: float = None, emotion: str = None) -> str:
811
+ """Generate a multi-segment MP4 short: each key point shown with its OWN image + narration."""
812
+ try:
813
+ os.makedirs(SHORTS_DIR, exist_ok=True)
814
+ out_mp4 = os.path.join(SHORTS_DIR, _safe_name(post_id) + ".mp4")
815
+ if os.path.exists(out_mp4) and voice_id is None and speed is None and emotion is None:
816
+ return "/api/ai/short-file/" + post_id
817
+
818
+ work = os.path.join(SHORTS_DIR, _safe_name(post_id) + "_work")
819
+ os.makedirs(work, exist_ok=True)
820
+
821
+ title = _clean_text(post.get("title", "")) or "VNEWS"
822
+ kps = _build_keypoints(post)
823
+
824
+ # Resolve voice + emotion (auto from topic, or from post, or explicit args)
825
+ auto_voice, auto_emotion = _detect_voice_emotion(post.get("title", ""), post.get("text", ""))
826
+ if voice_id is None:
827
+ voice_id = post.get("voice") or auto_voice
828
+ if emotion is None:
829
+ emotion = post.get("emotion") or auto_emotion
830
+ if speed is None:
831
+ speed = TTS_DEFAULT_SPEED
832
+ vcfg = TTS_VOICES.get(voice_id, TTS_VOICES[TTS_DEFAULT_VOICE])
833
+
834
+ seg_files = []
835
+ ff = _ffmpeg_bin()
836
+ for i, kp in enumerate(kps):
837
+ img_path = os.path.join(work, f"img{i}.jpg")
838
+ frame_path = os.path.join(work, f"frame{i}.jpg")
839
+ audio_path = os.path.join(work, f"voice{i}.mp3")
840
+ seg_mp4 = os.path.join(work, f"seg{i}.mp4")
841
+ _download_image(kp.get("image", ""), title, img_path)
842
+ _make_segment_frame(title, kp["text"], img_path, i, len(kps), frame_path)
843
+ narration = (title + ". " + kp["text"]) if i == 0 else kp["text"]
844
+ try:
845
+ await _generate_tts_edge(narration, voice_id, speed, audio_path, emotion=emotion)
846
+ except Exception as e:
847
+ print(f"[TTS edge-tts error] {e}, falling back to gTTS")
848
+ if gTTS:
849
+ _generate_tts_gtts(narration, audio_path)
850
+ else:
851
+ return ""
852
+ dur = _audio_duration(audio_path)
853
+ if dur < 1.0:
854
+ dur = 2.0
855
+ cmd = [ff, "-y", "-loop", "1", "-i", frame_path, "-i", audio_path,
856
+ "-c:v", "libx264", "-tune", "stillimage", "-pix_fmt", "yuv420p",
857
+ "-t", f"{dur + 0.4:.2f}", "-c:a", "aac", "-b:a", "128k", "-ar", "44100",
858
+ "-vf", "scale=1080:1920", "-r", "25", seg_mp4]
859
+ subprocess.run(cmd, check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, timeout=180)
860
+ seg_files.append(seg_mp4)
861
+
862
+ if not seg_files:
863
+ return ""
864
+ if len(seg_files) == 1:
865
+ os.replace(seg_files[0], out_mp4)
866
+ return "/api/ai/short-file/" + post_id
867
+
868
+ listfile = os.path.join(work, "concat.txt")
869
+ with open(listfile, "w", encoding="utf-8") as f:
870
+ f.write("\n".join(f"file '{s}'" for s in seg_files))
871
+ cmd = [ff, "-y", "-f", "concat", "-safe", "0", "-i", listfile,
872
+ "-c:v", "libx264", "-pix_fmt", "yuv420p", "-c:a", "aac", "-b:a", "128k", out_mp4]
873
+ subprocess.run(cmd, check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, timeout=300)
874
+ return "/api/ai/short-file/" + post_id
875
+ except Exception as e:
876
+ print(f"[short video error] {e}")
877
+ return ""
878
+
879
+ import threading as _threading
880
+ def _spawn_background_video(post):
881
+ """Generate the short video in a BACKGROUND thread so the rewrite/topic endpoint
882
+ can return immediately. When done, persist the video URL onto the wall post.
883
+ This is the main fix for 'rewrite tu cac nguon tin qua lau' — the user gets the
884
+ text post instantly; the video appears shortly after (or via the 'Tao Video' button)."""
885
+ pid = post.get("id")
886
+ if not pid:
887
+ return
888
+ def _run():
889
  try:
890
+ loop = asyncio.new_event_loop()
891
+ asyncio.set_event_loop(loop)
892
+ video_url = loop.run_until_complete(_generate_short_video(post, pid))
893
+ loop.close()
894
+ if video_url:
895
+ posts = _load_wall()
896
+ for i, p in enumerate(posts):
897
+ if str(p.get("id")) == str(pid):
898
+ posts[i]["video"] = video_url
899
+ break
900
+ _save_wall(posts)
901
+ print(f"[bg-video] done {pid} -> {video_url}")
902
+ except Exception as e:
903
+ print(f"[bg-video] error {pid}: {e}")
904
+ _threading.Thread(target=_run, daemon=True).start()
905
+
906
+ # ===== MAKE POST =====
907
+ def make_post(title, text, image, source_url, kind, sources=None, images=None, voice=None, emotion=None):
908
+ # Auto-pick voice + emotion from the article topic when not provided
909
+ auto_voice, auto_emotion = _detect_voice_emotion(title or "", text or "")
910
+ return {
911
+ "id": str(int(time.time() * 1000)) + str(random.randint(100, 999)),
912
+ "title": title, "text": text, "img": image, "url": source_url,
913
+ "kind": kind, "sources": sources or [], "video": "",
914
+ "images": images or [], "ts": int(time.time()),
915
+ "voice": voice or auto_voice,
916
+ "emotion": emotion or auto_emotion,
917
+ }
918
+
919
+
920
+ # ===== SHARED PROMPT BUILDER =====
921
+ def _build_rewrite_prompt(title: str, raw: str, images: List[str] = None) -> str:
922
+ image_info = ""
923
+ if images:
924
+ num = len(images)
925
+ if num == 1:
926
+ image_info = "\n\nBài viết có 1 ảnh minh họa. Hãy tham khảo ảnh để hiểu ngữ cảnh (nếu phù hợp)."
927
  else:
928
+ image_info = f"\n\nBài viết có {num} ảnh minh họa. Hãy tham khảo tất cả ảnh để hiểu ngữ cảnh và bổ sung thông tin cho bài viết (nếu phù hợp)."
929
+
930
+ return f"""Tóm tắt bài viết sau thành bài TÓM TẮT đăng Tường AI.
931
+
932
+ QUY TẮC BẮT BUỘC:
933
+ 1. Chỉ viết TÓM TẮT các ý chính. KHÔNG sao chép nguyên văn từ bài gốc.
934
+ 2. KHÔNG lặp lại bất kỳ nội dung nào. Mỗi thông tin chỉ xuất hiện ĐÚNG 1 LẦN.
935
+ 3. Nếu 2 câu nói cùng 1 ý → chỉ giữ 1 câu, bỏ cây còn lại.
936
+ 4. KHÔNG dùng Markdown (##, **, ---, *).
937
+ 5. KHÔNG viết "Dưới đây là", "Tôi sẽ", "Theo yêu cầu", "Nhiệm vụ", "Vai trò", "Đây là bài tóm tắt".
938
+ 6. Viết thành ĐOẠN VĂN THUẦN, mạch lạc, dễ đọc. Không dùng bullet points.
939
+ 7. Giữ sự thật, KHÔNG bịa thông tin.
940
+ 8. Tối đa 200 từ. Ngắn gọn, đủ ý.{image_info}
941
+
942
+ Tiêu đề gốc: {title}
943
+
944
+ Nội dung gốc:
945
+ {raw[:14000]}"""
946
+
947
+
948
+ def _build_topic_prompt(topic: str, ctx: str) -> str:
949
+ return f"""Viết bài TÓM TẮT NGẮN GỌN về chủ đề: "{topic}".
950
+
951
+ QUY TẮC BẮT BUỘC:
952
+ 1. Chỉ viết TÓM TẮT các ý chính từ nguồn. KHÔNG sao chép nguyên văn.
953
+ 2. KHÔNG lặp lại bất kỳ nội dung nào. Mỗi thông tin chỉ xuất hiện ĐÚNG 1 LẦN.
954
+ 3. Nếu 2 câu nói cùng 1 ý → chỉ giữ 1 câu.
955
+ 4. KHÔNG dùng Markdown (##, **, ---, *).
956
+ 5. KHÔNG viết "Dưới đây là", "Tôi sẽ", "Theo yêu cầu", "Nhiệm vụ", "Vai trò".
957
+ 6. Viết thành ĐOẠN VĂN THU���N, mạch lạc. Không dùng bullet points.
958
+ 7. Giữ sự thật, KHÔNG bịa.
959
+ 8. Tối đa 200 từ. Ngắn gọn, đủ ý.
960
+
961
+ Nguồn thực tế:
962
+ {ctx[:12000]}"""
963
+
964
+
965
+ # ===== WRITE ENDPOINTS =====
966
+ @app.post("/api/rewrite_share")
967
+ async def api_rewrite_share(request: Request):
968
+ body = await request.json()
969
+ url = _clean_text(body.get("url", ""))
970
+ if not url.startswith("http"):
971
+ return JSONResponse({"error": "missing url"}, status_code=400)
972
+ try:
973
+ data = scrape_any_url(url)
974
+ except Exception as e:
975
+ return JSONResponse({"error": "Không đọc được bài viết: " + str(e)[:180]}, status_code=422)
976
+ raw = (data.get("summary", "") + "\n" + data.get("text", "")).strip()
977
+ if len(raw) < 60:
978
+ return JSONResponse({"error": "Bài viết quá ngắn để tóm tắt"}, status_code=422)
979
+
980
+ images = data.get("images", [])
981
+ prompt = _build_rewrite_prompt(data.get("title", ""), raw, images)
982
+ # Text-only summary for SPEED (images are kept on the post for display + short video).
983
+ text = await qwen_generate(prompt, max_tokens=500)
984
+ if not text:
985
+ return JSONResponse({"error": "Qwen2.5-VL chưa sẵn sàng: " + LAST_QWEN_ERROR}, status_code=503)
986
+ text = _clean_ai_output(text)
987
+ post = make_post(data.get("title") or "Bài viết", text,
988
+ images[0] if images else data.get("image", ""),
989
+ url, "rewrite", images=images)
990
+
991
+ # Save post and return IMMEDIATELY; generate the short video in the background
992
+ # (so rewrite is fast). Video appears on the wall when ready / via 'Tao Video' button.
993
+ posts = _load_wall()
994
+ posts.insert(0, post)
995
+ _save_wall(posts)
996
+ _spawn_background_video(post)
997
+ return JSONResponse({"post": post})
998
+
999
+
1000
+ @app.post("/api/url_wall")
1001
+ async def api_url_wall(request: Request):
1002
+ body = await request.json()
1003
+ url = _clean_text(body.get("url", ""))
1004
+ if not url.startswith("http"):
1005
+ return JSONResponse({"error": "missing url"}, status_code=400)
1006
+ try:
1007
+ data = scrape_any_url(url)
1008
+ except Exception as e:
1009
+ return JSONResponse({"error": "Không scrape được URL: " + str(e)[:180]}, status_code=422)
1010
+ raw = (data.get("summary", "") + "\n" + data.get("text", "")).strip()
1011
+ if len(raw) < 60:
1012
+ return JSONResponse({"error": "URL không có đủ nội dung"}, status_code=422)
1013
+
1014
+ images = data.get("images", [])
1015
+ prompt = _build_rewrite_prompt(data.get("title", ""), raw, images)
1016
+ # Text-only summary for SPEED (images are kept on the post for display + short video).
1017
+ text = await qwen_generate(prompt, max_tokens=500)
1018
+ if not text:
1019
+ return JSONResponse({"error": "Qwen2.5-VL chưa sẵn sàng: " + LAST_QWEN_ERROR}, status_code=503)
1020
+ text = _clean_ai_output(text)
1021
+ post = make_post(data.get("title") or "Bài viết", text,
1022
+ images[0] if images else data.get("image", ""),
1023
+ url, "url", images=images)
1024
+
1025
+ posts = _load_wall()
1026
+ posts.insert(0, post)
1027
+ _save_wall(posts)
1028
+ _spawn_background_video(post)
1029
+ return JSONResponse({"post": post})
1030
+
1031
+
1032
+ @app.post("/api/topic_post")
1033
+ async def api_topic_post(request: Request):
1034
+ body = await request.json()
1035
+ topic = _clean_text(body.get("topic", ""))
1036
+ if not topic:
1037
+ return JSONResponse({"error": "missing topic"}, status_code=400)
1038
+
1039
+ ctx = _web_context(topic)
1040
+ if not ctx:
1041
+ return JSONResponse({"error": "Không lấy được dữ liệu cho chủ đề này"}, status_code=422)
1042
+
1043
+ image = pollinations_image_url(topic)
1044
+ prompt = _build_topic_prompt(topic, ctx)
1045
+ # NOTE: do NOT pass the decorative pollinations image to the VL model — feeding an
1046
+ # image makes Qwen2.5-VL much slower (it must download+process it) with no benefit for
1047
+ # a text summary. We keep the image only for display on the post. This is a major
1048
+ # speed-up for 'rewrite tong hop'. (Text-only inference is several times faster.)
1049
+ text = await qwen_generate(prompt, max_tokens=500)
1050
+ if not text:
1051
+ return JSONResponse({"error": "Qwen2.5-VL chưa sẵn sàng: " + LAST_QWEN_ERROR}, status_code=503)
1052
+ text = _clean_ai_output(text)
1053
+ post = make_post(topic, text, image, "", "topic")
1054
+
1055
+ posts = _load_wall()
1056
+ posts.insert(0, post)
1057
+ _save_wall(posts)
1058
+ _spawn_background_video(post)
1059
+ return JSONResponse({"post": post})
1060
+
1061
+
1062
+ # ===== WALL ENDPOINTS =====
1063
+ @app.get("/api/ai_wall")
1064
+ def api_ai_wall():
1065
+ return JSONResponse({"posts": _load_wall()[:80]})
1066
+
1067
+ @app.get("/api/wall")
1068
+ def api_wall():
1069
+ return JSONResponse({"posts": _load_wall()[:80]})
1070
+
1071
+
1072
+ # ===== SHORT VIDEO ENDPOINT (with voice + speed params) =====
1073
+ @app.post("/api/ai/short/{post_id}")
1074
+ async def api_ai_short(post_id: str, voice: str = Query(default=None), speed: float = Query(default=None), emotion: str = Query(default=None)):
1075
+ """Generate (or retrieve cached) short video for a wall post.
1076
+
1077
+ Query params:
1078
+ - voice: 'hoaimy' (female) | 'namminh' (male) | auto-detect if not specified
1079
+ - speed: float (default 1.2), e.g. 1.0=normal, 1.2=fast, 0.8=slow
1080
+ - emotion: 'vui'|'hao_hung'|'nghiem'|'tram'|'buon'|'trung_tinh' | auto by topic
1081
+ """
1082
+ posts = _load_wall()
1083
+ post = next((p for p in posts if str(p.get("id")) == str(post_id)), None)
1084
+ if not post:
1085
+ return JSONResponse({"error": "post not found"}, status_code=404)
1086
+
1087
+ os.makedirs(SHORTS_DIR, exist_ok=True)
1088
+ out_mp4 = os.path.join(SHORTS_DIR, _safe_name(post_id) + ".mp4")
1089
+
1090
+ # If cached and no custom voice/speed/emotion requested, return cached
1091
+ if os.path.exists(out_mp4) and voice is None and speed is None and emotion is None:
1092
+ video_url = "/api/ai/short-file/" + post_id
1093
+ for i, p in enumerate(posts):
1094
+ if str(p.get("id")) == str(post_id):
1095
+ posts[i]["video"] = video_url
1096
+ break
1097
+ _save_wall(posts)
1098
+ return JSONResponse({"video": video_url})
1099
+
1100
+ # Validate params
1101
+ if voice is not None and voice not in TTS_VOICES:
1102
+ return JSONResponse({"error": f"voice không hợp lệ. Chọn: {list(TTS_VOICES.keys())}"}, status_code=400)
1103
+ if emotion is not None and emotion not in EMOTION_PRESETS:
1104
+ return JSONResponse({"error": f"emotion không hợp lệ. Chọn: {list(EMOTION_PRESETS.keys())}"}, status_code=400)
1105
+
1106
+ video_url = await _generate_short_video(post, post_id, voice_id=voice, speed=speed, emotion=emotion)
1107
+ if video_url:
1108
+ for i, p in enumerate(posts):
1109
+ if str(p.get("id")) == str(post_id):
1110
+ posts[i]["video"] = video_url
1111
+ if voice:
1112
+ posts[i]["voice"] = voice
1113
+ if emotion:
1114
+ posts[i]["emotion"] = emotion
1115
+ break
1116
+ _save_wall(posts)
1117
+ return JSONResponse({"video": video_url})
1118
+ return JSONResponse({"error": "Không tạo được shorts"}, status_code=500)
1119
+
1120
+
1121
+ @app.get("/api/ai/short-file/{post_id}")
1122
+ def api_ai_short_file(post_id: str):
1123
+ path = os.path.join(SHORTS_DIR, _safe_name(post_id) + ".mp4")
1124
+ if not os.path.exists(path):
1125
+ return JSONResponse({"error": "not found"}, status_code=404)
1126
+ return FileResponse(path, media_type="video/mp4", filename=f"vnews-ai-{post_id}.mp4")
1127
+
1128
+
1129
+ @app.get("/api/ai/status")
1130
+ def api_ai_status():
1131
+ return JSONResponse({
1132
+ "has_token": bool(_hf_token()),
1133
+ "client_imported": AsyncInferenceClient is not None,
1134
+ "model": QWEN_VL_MODEL,
1135
+ "last_error": LAST_QWEN_ERROR,
1136
+ "tts_ready": gTTS is not None or edge_tts is not None,
1137
+ "tts_engine": "edge-tts" if edge_tts else ("gtts" if gTTS else "none"),
1138
+ "tts_voices": {k: v["flag"] + " " + v["name"] for k, v in TTS_VOICES.items()},
1139
+ "tts_voice_count": len(TTS_VOICES),
1140
+ "tts_default_speed": TTS_DEFAULT_SPEED,
1141
+ })
1142
+
1143
+
1144
+ @app.get("/api/ai/voices")
1145
+ def api_ai_voices():
1146
+ """Return available TTS voices with country/group info."""
1147
+ voices_out = {}
1148
+ for k, v in TTS_VOICES.items():
1149
+ voices_out[k] = {
1150
+ "name": v["name"],
1151
+ "gender": v["gender"],
1152
+ "country": v["country"],
1153
+ "lang": v["lang"],
1154
+ "flag": v["flag"],
1155
+ "label": f"{v['flag']} {v['name']} ({v['gender']})",
1156
+ }
1157
+ return JSONResponse({
1158
+ "voices": voices_out,
1159
+ "default_voice": TTS_DEFAULT_VOICE,
1160
+ "default_speed": TTS_DEFAULT_SPEED,
1161
+ "topic_voice_map": TOPIC_VOICE_MAP,
1162
+ "emotions": {k: {"label": v["label"], "emoji": v["emoji"]} for k, v in EMOTION_PRESETS.items()},
1163
+ "default_emotion": EMOTION_DEFAULT,
1164
+ })
ai_fix2.py CHANGED
@@ -321,7 +321,7 @@ async def ai_short_full(post_id: str, request: Request):
321
  subprocess.run(['ffmpeg','-y','-i',audio,'-filter:a',f'atempo={speed}','-vn',audio_fast], check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, timeout=220)
322
  duration = 45.0
323
  try:
324
- pr = subprocess.run(['ffprobe','-v','error','-show_entries','format=duration','-of','default=noprint_wrappers=1:no_key=1',audio_fast], stdout=subprocess.PIPE, stderr=subprocess.PIPE, timeout=20)
325
  duration = float((pr.stdout or b'45').decode().strip() or 45)
326
  except Exception:
327
  pass
 
321
  subprocess.run(['ffmpeg','-y','-i',audio,'-filter:a',f'atempo={speed}','-vn',audio_fast], check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, timeout=220)
322
  duration = 45.0
323
  try:
324
+ pr = subprocess.run(['ffprobe','-v','error','-show_entries','format=duration','-of','default=noprint_wrappers=1:nokey=1',audio_fast], stdout=subprocess.PIPE, stderr=subprocess.PIPE, timeout=20)
325
  duration = float((pr.stdout or b'45').decode().strip() or 45)
326
  except Exception:
327
  pass
ai_patch.py CHANGED
@@ -6,7 +6,6 @@ import json
6
  import html as html_lib
7
  import subprocess
8
  import requests
9
- import hashlib
10
  import ai_ext as base
11
  from ai_ext import app
12
  from fastapi import Request
@@ -42,17 +41,17 @@ def _similar(a, b):
42
  return len(ta & tb) / max(1, min(len(ta), len(tb))) >= 0.72
43
 
44
 
45
- def _dedupe_units(units, max_units=25):
46
- """Deduplicate units - only skip exact matches to ensure all bullet points are read."""
47
  out, seen = [], set()
48
  for u in units:
49
  u = _clean(re.sub(r"^[-•*\d\.\)\s]+", "", u))
50
  if len(u) < 18:
51
  continue
52
  nu = _norm(u)
53
- # Only skip exact matches, NOT similar content (to avoid skipping valid bullet points)
54
  if nu in seen:
55
  continue
 
 
56
  seen.add(nu)
57
  out.append(u)
58
  if len(out) >= max_units:
@@ -60,7 +59,7 @@ def _dedupe_units(units, max_units=25):
60
  return out
61
 
62
 
63
- def _postprocess_ai_text(text, max_units=20):
64
  text = _clean(text)
65
  if not text:
66
  return text
@@ -79,9 +78,10 @@ def _postprocess_ai_text(text, max_units=20):
79
  raw_lines.append(line)
80
  units = []
81
  for line in raw_lines:
82
- # KEEP FULL bullet point - don't truncate or split into segments
83
- if len(line) >= 18:
84
- units.append(_clean(re.sub(r"^[-•*\d\.\)\s]+", "", line)))
 
85
  units = _dedupe_units(units, max_units=max_units)
86
  if not units:
87
  return text[:900]
@@ -268,7 +268,7 @@ async def qwen_generate_resilient(prompt: str, image_url=None, max_tokens: int =
268
  errors.append("missing HF_TOKEN")
269
  base.LAST_QWEN_ERROR = " | ".join(errors[-6:]) or "Qwen unavailable; used extractive fallback"
270
  print("[qwen resilient fallback]", base.LAST_QWEN_ERROR)
271
- return _fallback_summary_from_prompt(prompt, max_units=12)
272
 
273
 
274
  if not hasattr(base, "_original_qwen_generate"):
@@ -321,8 +321,8 @@ Yêu cầu bắt buộc:
321
 
322
  Nội dung bài:
323
  {art['raw'][:14000]}"""
324
- text = await base.qwen_generate(prompt, image_url=art.get('image') or None, max_tokens=1500)
325
- text = _postprocess_ai_text(text, max_units=20)
326
  src = [art['source']]
327
  if 'Nguồn tham khảo:' not in text:
328
  text += "\n\n" + _source_line(src)
@@ -347,8 +347,8 @@ async def compat_url_wall(request: Request):
347
  if len(raw) < 120:
348
  return JSONResponse({'error': 'URL không có đủ nội dung để tóm tắt'}, status_code=422)
349
  prompt = _make_summary_prompt(data.get('title', ''), raw, data.get('via', '') or base._domain(url))
350
- text = await base.qwen_generate(prompt, image_url=data.get('image') or None, max_tokens=1500)
351
- text = _postprocess_ai_text(text, max_units=20)
352
  src = [{'title': data.get('title'), 'url': url, 'excerpt': raw[:500], 'via': data.get('via') or base._domain(url)}]
353
  if 'Nguồn tham khảo:' not in text:
354
  text += "\n\n" + _source_line(src)
@@ -357,116 +357,6 @@ async def compat_url_wall(request: Request):
357
  return JSONResponse({'post': post})
358
 
359
 
360
- def _is_relevant_image(img_url, title, text):
361
- """Check if an image is relevant to the article content."""
362
- if not img_url:
363
- return False
364
- skip_patterns = ['pixel', 'analytics', 'tracking', '1x1.gif', 'spacer.gif',
365
- 'logo', 'icon', 'avatar', 'emoji', 'smiley', 'sprite',
366
- 'advertisement', 'ad-banner', 'sponsored', 'banner-ads']
367
- img_lower = img_url.lower()
368
- for p in skip_patterns:
369
- if p in img_lower:
370
- return False
371
- if not any(img_lower.endswith(ext) for ext in ['.jpg', '.jpeg', '.png', '.webp', '.gif']):
372
- return False
373
- return True
374
-
375
-
376
- def _filter_relevant_images(images, title, text, max_images=8):
377
- """Filter and rank images by relevance to article content."""
378
- if not images:
379
- return []
380
- seen = set()
381
- relevant = []
382
- for img in images:
383
- if img in seen:
384
- continue
385
- seen.add(img)
386
- if _is_relevant_image(img, title, text):
387
- relevant.append(img)
388
- return relevant[:max_images]
389
-
390
-
391
- def _extract_key_points_for_slides(paragraphs, max_points=12):
392
- """Extract key points from paragraphs for slides - extracts ALL sentences, not just first one."""
393
- points = []
394
- for p in paragraphs:
395
- if len(points) >= max_points:
396
- break
397
- p = _clean(p)
398
- if not p:
399
- continue
400
- # Split paragraph into sentences using Vietnamese + English punctuation - GET ALL SENTENCES
401
- sentences = re.split(r'(?<=[.!?])\s+(?=[A-ZÀ-Ỹ0-9])', p)
402
- sentences = [s.strip() for s in sentences if s.strip()]
403
-
404
- for sentence in sentences:
405
- if len(points) >= max_points:
406
- break
407
- sentence = _clean(sentence)
408
- if len(sentence) < 30:
409
- continue
410
- if any(sentence[:60] in existing for existing in points):
411
- continue
412
- if not sentence.endswith(('.', '!', '?')):
413
- sentence = sentence + '.'
414
- points.append(sentence)
415
- return points
416
-
417
-
418
- def _scrape_article_images(url):
419
- """Scrape article page and return only relevant images."""
420
- try:
421
- headers = {"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36",
422
- "Accept-Language": "vi-VN,vi;q=0.9,en;q=0.8"}
423
- r = requests.get(url, headers=headers, timeout=15, allow_redirects=True)
424
- r.encoding = 'utf-8'
425
- soup = BeautifulSoup(r.text, 'lxml')
426
- for tag in soup.find_all(['script', 'style', 'nav', 'footer', 'aside', 'form']):
427
- tag.decompose()
428
- h1 = soup.find('h1')
429
- ogt = soup.find('meta', property='og:title')
430
- title = (h1.get_text(strip=True) if h1 else '') or (ogt.get('content', '') if ogt else '')
431
- ogi = soup.find('meta', property='og:image')
432
- og_img = ogi.get('content', '') if ogi else ''
433
- if og_img and og_img.startswith('//'):
434
- og_img = 'https:' + og_img
435
- block = None
436
- for sel in ['article', '.singular-content', '.detail-content', '.fck_detail', '.content-detail', '.knc-content', 'main', '.cms-body', '.article__body']:
437
- el = soup.select_one(sel)
438
- if el and len(el.find_all('p')) >= 2:
439
- block = el
440
- break
441
- if not block:
442
- block = soup.body or soup
443
- paragraphs = []
444
- all_images = []
445
- seen_imgs = set()
446
- if og_img and og_img not in seen_imgs:
447
- all_images.append(og_img)
448
- seen_imgs.add(og_img)
449
- for el in block.find_all(['p', 'h2', 'h3', 'figure', 'img'], recursive=True):
450
- if el.name == 'p':
451
- t = _clean(el.get_text(strip=True))
452
- if t and len(t) > 40:
453
- paragraphs.append(t)
454
- elif el.name in ('figure', 'img'):
455
- im = el if el.name == 'img' else el.find('img')
456
- if im:
457
- src = im.get('data-src') or im.get('src') or im.get('data-original') or ''
458
- if src and 'base64' not in src:
459
- if src.startswith('//'):
460
- src = 'https:' + src
461
- if src not in seen_imgs:
462
- all_images.append(src)
463
- seen_imgs.add(src)
464
- relevant_images = _filter_relevant_images(all_images, title, ' '.join(paragraphs[:5]))
465
- return {'title': _clean(title), 'paragraphs': paragraphs, 'images': relevant_images, 'og_img': og_img}
466
- except Exception:
467
- return None
468
-
469
-
470
  @app.post('/api/rewrite_share')
471
  async def compat_rewrite_share(request: Request):
472
  body = await request.json()
@@ -481,53 +371,40 @@ async def compat_rewrite_share(request: Request):
481
  if len(raw) < 120:
482
  return JSONResponse({'error': 'Bài viết không đủ nội dung để tóm tắt'}, status_code=422)
483
  prompt = _make_summary_prompt(data.get('title', ''), raw, data.get('via', '') or base._domain(url))
484
- text = await base.qwen_generate(prompt, image_url=data.get('image') or None, max_tokens=1500)
485
- text = _postprocess_ai_text(text, max_units=20)
486
  src = [{'title': data.get('title'), 'url': url, 'excerpt': raw[:500], 'via': data.get('via') or base._domain(url)}]
487
  if 'Nguồn tham khảo:' not in text:
488
  text += "\n\n" + _source_line(src)
489
  post = base.make_post(data.get('title') or 'Bài viết', text, data.get('image') or '', url, 'summary', sources=src)
490
  posts = base._load_ai_wall(); posts.insert(0, post); base._save_ai_wall(posts)
491
-
492
- # Generate slides with relevant images only
493
- slides = []
494
- page_data = _scrape_article_images(url)
495
- if page_data and page_data.get('paragraphs'):
496
- key_points = _extract_key_points_for_slides(page_data['paragraphs'], max_points=12)
497
- if key_points:
498
- relevant_imgs = page_data.get('images', [])
499
- if not relevant_imgs and page_data.get('og_img'):
500
- relevant_imgs = [page_data['og_img']]
501
- for i, point in enumerate(key_points):
502
- img = relevant_imgs[i] if i < len(relevant_imgs) else (relevant_imgs[-1] if relevant_imgs else '')
503
- slides.append({'text': point, 'image': img, 'index': i + 1})
504
-
505
- return JSONResponse({'post': post, 'slides': slides})
506
 
507
 
508
  def _emotion_script(text, emotion):
509
- """Prepend emotion-appropriate prefix to text based on emotion type.
510
-
511
- NOTE: Prefix is NOT added to avoid cluttering Short AI speech.
512
- The emotion is still used for voice selection but content is read cleanly.
513
- """
514
  text = _clean(text)
515
- # REMOVED: No prefix added to keep content clean and natural
 
 
 
 
 
 
 
516
  return text
517
 
518
 
519
  def _tts_script_smart(post, emotion):
520
- raw = base._short_script(post) if hasattr(base, '_short_script') else _clean(post.get('text', '') or post.get('title', ''))
521
  raw = re.sub(r"^[•\-\*]\s*", "", raw, flags=re.M)
522
  raw = re.sub(r"\s*\n\s*", ". ", raw)
523
  raw = re.sub(r"([\.\!\?])\s*", r"\1\n", raw)
524
  raw = re.sub(r"\n{2,}", "\n", raw).strip()
525
- # REMOVED: _emotion_script call - read content cleanly without prefix
526
- # INCREASED to 3000 to read full content of all bullet points
527
- if len(raw) > 3000:
528
- raw = raw[:3000]
529
  cut = max(raw.rfind("."), raw.rfind("!"), raw.rfind("?"))
530
- if cut > 700:
531
  raw = raw[:cut + 1]
532
  return raw
533
 
@@ -642,7 +519,7 @@ def _make_short_frame_full(post, img_path, out_path):
642
 
643
 
644
 
645
- def _summary_segments_from_post(post, max_segments=25):
646
  raw = _clean(post.get('text') or post.get('title') or '')
647
  raw = re.sub(r'^Bản tin AI viết lại:\s*', '', raw, flags=re.I)
648
  raw = re.sub(r'Nguồn tham khảo:.*$', '', raw, flags=re.I|re.S).strip()
@@ -653,7 +530,7 @@ def _summary_segments_from_post(post, max_segments=25):
653
  low=ln.lower()
654
  if low.startswith(('điểm chính','tiêu đề','sapo','nguồn tham khảo')): continue
655
  if len(ln)>=18: lines.append(ln)
656
- if len(lines)<3:
657
  lines=[]
658
  for s in re.split(r'(?<=[\.\!\?])\s+', raw):
659
  s=_clean(s)
@@ -705,8 +582,7 @@ def _make_scene_frame(post, segment, idx, total, img_path, out_path, emotion='ne
705
  draw.rounded_rectangle((48,834,260,880), radius=20, fill=(28,70,45))
706
  draw.text((66,842),f'Đoạn {idx+1}/{total}',fill=(235,235,235),font=font_small)
707
  y=940; maxw=W-96
708
- # INCREASED from 12 to 18 for full content display - each key point can span multiple lines
709
- for ln in _wrap_text_px(draw, segment, font_seg, maxw, 18):
710
  draw.text((48,y),ln,fill=(255,255,255),font=font_seg)
711
  y+=74
712
  if y>1500: break
@@ -718,11 +594,10 @@ def _make_scene_frame(post, segment, idx, total, img_path, out_path, emotion='ne
718
  bg.save(out_path, quality=92)
719
 
720
 
721
- def _estimate_audio_duration(path, fallback=15.0):
722
- """Estimate audio duration with 15s minimum per segment for complete bullet reading."""
723
  try:
724
- pr=subprocess.run(['ffprobe','-v','error','-show_entries','format=duration','-of','default=noprint_wrappers=1:no_key=1',path], stdout=subprocess.PIPE, stderr=subprocess.PIPE, timeout=20)
725
- return max(12.0, float((pr.stdout or b'').decode().strip() or fallback))
726
  except Exception:
727
  return fallback
728
 
@@ -735,7 +610,7 @@ async def patched_ai_short(post_id: str, request: Request):
735
  body = {}
736
  voice = str(body.get('voice', 'nu')).strip().lower()
737
  emotion = str(body.get('emotion', 'neutral')).strip().lower()
738
- speed = float(body.get('speed', 1.0) or 1.0)
739
  speed = max(0.85, min(1.35, speed))
740
 
741
  posts = base._load_ai_wall()
@@ -743,7 +618,7 @@ async def patched_ai_short(post_id: str, request: Request):
743
  if not post:
744
  return JSONResponse({'error': 'post not found'}, status_code=404)
745
 
746
- segments = _summary_segments_from_post(post, max_segments=25)
747
  seg_hash = hashlib.md5(('|'.join(segments)+voice+emotion+str(speed)).encode('utf-8')).hexdigest()[:8]
748
  os.makedirs(base.SHORTS_DIR, exist_ok=True)
749
  suffix = f"_{voice}_{emotion}_{str(speed).replace('.', 'p')}_{seg_hash}_scenes_nosub"
@@ -766,63 +641,11 @@ async def patched_ai_short(post_id: str, request: Request):
766
  try:
767
  base._download_image(post.get('img'), post.get('title', 'AI news'), img)
768
  edge_voice = {
769
- # Vietnamese
770
- 'vi-vn-hoaimyneural': 'vi-VN-HoaiMyNeural',
771
- 'vi-vn-namminhneural': 'vi-VN-NamMinhNeural',
772
- 'hoaimy': 'vi-VN-HoaiMyNeural',
773
- 'namminh': 'vi-VN-NamMinhNeural',
774
  'nam': 'vi-VN-NamMinhNeural',
775
  'male': 'vi-VN-NamMinhNeural',
776
  'nu': 'vi-VN-HoaiMyNeural',
777
  'female': 'vi-VN-HoaiMyNeural',
778
  'mien-nam': 'vi-VN-HoaiMyNeural',
779
- # English - Multilingual
780
- 'en-us-andrewmultilingualneural': 'en-US-AndrewMultilingualNeural',
781
- 'en-au-williammultilingualneural': 'en-AU-WilliamMultilingualNeural',
782
- 'andrew': 'en-US-AndrewMultilingualNeural',
783
- 'en_andrew': 'en-US-AndrewMultilingualNeural',
784
- 'jenny': 'en-US-AndrewMultilingualNeural',
785
- 'en_jenny': 'en-US-AndrewMultilingualNeural',
786
- # Portuguese - Multilingual (ONLY Thalita)
787
- 'pt-br-thalitamultilingualneural': 'pt-BR-ThalitaMultilingualNeural',
788
- 'thalita': 'pt-BR-ThalitaMultilingualNeural',
789
- 'pt_thalita': 'pt-BR-ThalitaMultilingualNeural',
790
- 'pt_br_thalita': 'pt-BR-ThalitaMultilingualNeural',
791
- 'pt': 'pt-BR-ThalitaMultilingualNeural',
792
- 'pt_francisco': 'pt-BR-ThalitaMultilingualNeural',
793
- # French - Multilingual
794
- 'fr-fr-viviennemultilingualneural': 'fr-FR-VivienneMultilingualNeural',
795
- 'fr-fr-remymultilingualneural': 'fr-FR-RemyMultilingualNeural',
796
- 'denise': 'fr-FR-VivienneMultilingualNeural',
797
- 'fr': 'fr-FR-VivienneMultilingualNeural',
798
- 'fr_denise': 'fr-FR-VivienneMultilingualNeural',
799
- # German - Multilingual
800
- 'de-de-seraphinamultilingualneural': 'de-DE-SeraphinaMultilingualNeural',
801
- 'de-de-florianmultilingualneural': 'de-DE-FlorianMultilingualNeural',
802
- 'katja': 'de-DE-SeraphinaMultilingualNeural',
803
- 'de': 'de-DE-SeraphinaMultilingualNeural',
804
- 'de_katja': 'de-DE-SeraphinaMultilingualNeural',
805
- # Korean - Multilingual (Hyunsu, NOT SunHee)
806
- 'ko-kr-hyusumultilingualneural': 'ko-KR-HyunsuMultilingualNeural',
807
- 'ko-kr-hyunsuneural': 'ko-KR-HyunsuMultilingualNeural',
808
- 'sunhee': 'ko-KR-HyunsuMultilingualNeural',
809
- 'ko': 'ko-KR-HyunsuMultilingualNeural',
810
- 'ko_sunhee': 'ko-KR-HyunsuMultilingualNeural',
811
- # Italian - Multilingual
812
- 'it-it-giuseppemultilingualneural': 'it-IT-GiuseppeMultilingualNeural',
813
- # Spanish (keep for backward compat)
814
- 'ela': 'en-US-AndrewMultilingualNeural',
815
- 'es_ela': 'en-US-AndrewMultilingualNeural',
816
- 'es': 'en-US-AndrewMultilingualNeural',
817
- 'es_carlos': 'en-US-AndrewMultilingualNeural',
818
- # Japanese (keep for backward compat)
819
- 'nanami': 'en-US-AndrewMultilingualNeural',
820
- 'ja': 'en-US-AndrewMultilingualNeural',
821
- 'ja_nanami': 'en-US-AndrewMultilingualNeural',
822
- # Chinese (keep for backward compat)
823
- 'xiaochen': 'en-US-AndrewMultilingualNeural',
824
- 'zh': 'en-US-AndrewMultilingualNeural',
825
- 'zh_xiaochen': 'en-US-AndrewMultilingualNeural',
826
  }.get(voice, 'vi-VN-HoaiMyNeural')
827
  part_files=[]
828
  for idx, seg in enumerate(segments):
@@ -835,13 +658,13 @@ async def patched_ai_short(post_id: str, request: Request):
835
  try:
836
  subprocess.run(['python','-m','edge_tts','--voice',edge_voice,'--text',spoken,'--write-media',aud], check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, timeout=120)
837
  except Exception:
838
- tld='com.vn' if voice in ('nu','female','mien-nam','hoaimy') else 'com'
839
  try:
840
  base.gTTS(spoken, lang='vi', tld=tld, slow=False).save(aud)
841
  except TypeError:
842
  base.gTTS(spoken, lang='vi', slow=False).save(aud)
843
  subprocess.run(['ffmpeg','-y','-i',aud,'-filter:a',f'atempo={speed}','-vn',aud_fast], check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, timeout=90)
844
- dur=_estimate_audio_duration(aud_fast, fallback=15.0)+0.35
845
  subprocess.run(['ffmpeg','-y','-loop','1','-t',str(dur),'-i',frame,'-i',aud_fast,'-shortest','-c:v','libx264','-tune','stillimage','-pix_fmt','yuv420p','-c:a','aac','-b:a','128k',part], check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, timeout=150)
846
  part_files.append(part)
847
  concat=os.path.join(work,'concat.txt')
@@ -876,3 +699,53 @@ def api_ai_shorts():
876
 
877
 
878
  app.router.routes = [r for r in app.router.routes if not (getattr(r, 'path', None) == '/' and 'GET' in getattr(r, 'methods', set()))]
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
  import html as html_lib
7
  import subprocess
8
  import requests
 
9
  import ai_ext as base
10
  from ai_ext import app
11
  from fastapi import Request
 
41
  return len(ta & tb) / max(1, min(len(ta), len(tb))) >= 0.72
42
 
43
 
44
+ def _dedupe_units(units, max_units=7):
 
45
  out, seen = [], set()
46
  for u in units:
47
  u = _clean(re.sub(r"^[-•*\d\.\)\s]+", "", u))
48
  if len(u) < 18:
49
  continue
50
  nu = _norm(u)
 
51
  if nu in seen:
52
  continue
53
+ if any(_similar(u, old) for old in out):
54
+ continue
55
  seen.add(nu)
56
  out.append(u)
57
  if len(out) >= max_units:
 
59
  return out
60
 
61
 
62
+ def _postprocess_ai_text(text, max_units=7):
63
  text = _clean(text)
64
  if not text:
65
  return text
 
78
  raw_lines.append(line)
79
  units = []
80
  for line in raw_lines:
81
+ if len(line) > 260:
82
+ units.extend(re.split(r"(?<=[\.\!\?])\s+(?=[A-ZÀ-Ỹ0-9])", line))
83
+ else:
84
+ units.append(line)
85
  units = _dedupe_units(units, max_units=max_units)
86
  if not units:
87
  return text[:900]
 
268
  errors.append("missing HF_TOKEN")
269
  base.LAST_QWEN_ERROR = " | ".join(errors[-6:]) or "Qwen unavailable; used extractive fallback"
270
  print("[qwen resilient fallback]", base.LAST_QWEN_ERROR)
271
+ return _fallback_summary_from_prompt(prompt, max_units=6)
272
 
273
 
274
  if not hasattr(base, "_original_qwen_generate"):
 
321
 
322
  Nội dung bài:
323
  {art['raw'][:14000]}"""
324
+ text = await base.qwen_generate(prompt, image_url=art.get('image') or None, max_tokens=900)
325
+ text = _postprocess_ai_text(text, max_units=6)
326
  src = [art['source']]
327
  if 'Nguồn tham khảo:' not in text:
328
  text += "\n\n" + _source_line(src)
 
347
  if len(raw) < 120:
348
  return JSONResponse({'error': 'URL không có đủ nội dung để tóm tắt'}, status_code=422)
349
  prompt = _make_summary_prompt(data.get('title', ''), raw, data.get('via', '') or base._domain(url))
350
+ text = await base.qwen_generate(prompt, image_url=data.get('image') or None, max_tokens=850)
351
+ text = _postprocess_ai_text(text, max_units=6)
352
  src = [{'title': data.get('title'), 'url': url, 'excerpt': raw[:500], 'via': data.get('via') or base._domain(url)}]
353
  if 'Nguồn tham khảo:' not in text:
354
  text += "\n\n" + _source_line(src)
 
357
  return JSONResponse({'post': post})
358
 
359
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
360
  @app.post('/api/rewrite_share')
361
  async def compat_rewrite_share(request: Request):
362
  body = await request.json()
 
371
  if len(raw) < 120:
372
  return JSONResponse({'error': 'Bài viết không đủ nội dung để tóm tắt'}, status_code=422)
373
  prompt = _make_summary_prompt(data.get('title', ''), raw, data.get('via', '') or base._domain(url))
374
+ text = await base.qwen_generate(prompt, image_url=data.get('image') or None, max_tokens=850)
375
+ text = _postprocess_ai_text(text, max_units=6)
376
  src = [{'title': data.get('title'), 'url': url, 'excerpt': raw[:500], 'via': data.get('via') or base._domain(url)}]
377
  if 'Nguồn tham khảo:' not in text:
378
  text += "\n\n" + _source_line(src)
379
  post = base.make_post(data.get('title') or 'Bài viết', text, data.get('image') or '', url, 'summary', sources=src)
380
  posts = base._load_ai_wall(); posts.insert(0, post); base._save_ai_wall(posts)
381
+ return JSONResponse({'post': post})
 
 
 
 
 
 
 
 
 
 
 
 
 
 
382
 
383
 
384
  def _emotion_script(text, emotion):
 
 
 
 
 
385
  text = _clean(text)
386
+ if emotion == 'urgent':
387
+ return 'Tin nhanh. ' + text
388
+ if emotion == 'warm':
389
+ return 'Câu chuyện đáng chú ý. ' + text
390
+ if emotion == 'serious':
391
+ return 'Bản tin nghiêm túc. ' + text
392
+ if emotion == 'energetic':
393
+ return 'Cập nhật nổi bật. ' + text
394
  return text
395
 
396
 
397
  def _tts_script_smart(post, emotion):
398
+ raw = base._short_script(post)
399
  raw = re.sub(r"^[•\-\*]\s*", "", raw, flags=re.M)
400
  raw = re.sub(r"\s*\n\s*", ". ", raw)
401
  raw = re.sub(r"([\.\!\?])\s*", r"\1\n", raw)
402
  raw = re.sub(r"\n{2,}", "\n", raw).strip()
403
+ raw = _emotion_script(raw, emotion)
404
+ if len(raw) > 1000:
405
+ raw = raw[:1000]
 
406
  cut = max(raw.rfind("."), raw.rfind("!"), raw.rfind("?"))
407
+ if cut > 350:
408
  raw = raw[:cut + 1]
409
  return raw
410
 
 
519
 
520
 
521
 
522
+ def _summary_segments_from_post(post, max_segments=7):
523
  raw = _clean(post.get('text') or post.get('title') or '')
524
  raw = re.sub(r'^Bản tin AI viết lại:\s*', '', raw, flags=re.I)
525
  raw = re.sub(r'Nguồn tham khảo:.*$', '', raw, flags=re.I|re.S).strip()
 
530
  low=ln.lower()
531
  if low.startswith(('điểm chính','tiêu đề','sapo','nguồn tham khảo')): continue
532
  if len(ln)>=18: lines.append(ln)
533
+ if len(lines)<2:
534
  lines=[]
535
  for s in re.split(r'(?<=[\.\!\?])\s+', raw):
536
  s=_clean(s)
 
582
  draw.rounded_rectangle((48,834,260,880), radius=20, fill=(28,70,45))
583
  draw.text((66,842),f'Đoạn {idx+1}/{total}',fill=(235,235,235),font=font_small)
584
  y=940; maxw=W-96
585
+ for ln in _wrap_text_px(draw, segment, font_seg, maxw, 8):
 
586
  draw.text((48,y),ln,fill=(255,255,255),font=font_seg)
587
  y+=74
588
  if y>1500: break
 
594
  bg.save(out_path, quality=92)
595
 
596
 
597
+ def _estimate_audio_duration(path, fallback=4.0):
 
598
  try:
599
+ pr=subprocess.run(['ffprobe','-v','error','-show_entries','format=duration','-of','default=noprint_wrappers=1:nokey=1',path], stdout=subprocess.PIPE, stderr=subprocess.PIPE, timeout=20)
600
+ return max(1.5, float((pr.stdout or b'').decode().strip() or fallback))
601
  except Exception:
602
  return fallback
603
 
 
610
  body = {}
611
  voice = str(body.get('voice', 'nu')).strip().lower()
612
  emotion = str(body.get('emotion', 'neutral')).strip().lower()
613
+ speed = float(body.get('speed', 1.2) or 1.2)
614
  speed = max(0.85, min(1.35, speed))
615
 
616
  posts = base._load_ai_wall()
 
618
  if not post:
619
  return JSONResponse({'error': 'post not found'}, status_code=404)
620
 
621
+ segments = _summary_segments_from_post(post, max_segments=7)
622
  seg_hash = hashlib.md5(('|'.join(segments)+voice+emotion+str(speed)).encode('utf-8')).hexdigest()[:8]
623
  os.makedirs(base.SHORTS_DIR, exist_ok=True)
624
  suffix = f"_{voice}_{emotion}_{str(speed).replace('.', 'p')}_{seg_hash}_scenes_nosub"
 
641
  try:
642
  base._download_image(post.get('img'), post.get('title', 'AI news'), img)
643
  edge_voice = {
 
 
 
 
 
644
  'nam': 'vi-VN-NamMinhNeural',
645
  'male': 'vi-VN-NamMinhNeural',
646
  'nu': 'vi-VN-HoaiMyNeural',
647
  'female': 'vi-VN-HoaiMyNeural',
648
  'mien-nam': 'vi-VN-HoaiMyNeural',
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
649
  }.get(voice, 'vi-VN-HoaiMyNeural')
650
  part_files=[]
651
  for idx, seg in enumerate(segments):
 
658
  try:
659
  subprocess.run(['python','-m','edge_tts','--voice',edge_voice,'--text',spoken,'--write-media',aud], check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, timeout=120)
660
  except Exception:
661
+ tld='com.vn' if voice in ('nu','female','mien-nam') else 'com'
662
  try:
663
  base.gTTS(spoken, lang='vi', tld=tld, slow=False).save(aud)
664
  except TypeError:
665
  base.gTTS(spoken, lang='vi', slow=False).save(aud)
666
  subprocess.run(['ffmpeg','-y','-i',aud,'-filter:a',f'atempo={speed}','-vn',aud_fast], check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, timeout=90)
667
+ dur=_estimate_audio_duration(aud_fast, fallback=4.0)+0.35
668
  subprocess.run(['ffmpeg','-y','-loop','1','-t',str(dur),'-i',frame,'-i',aud_fast,'-shortest','-c:v','libx264','-tune','stillimage','-pix_fmt','yuv420p','-c:a','aac','-b:a','128k',part], check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, timeout=150)
669
  part_files.append(part)
670
  concat=os.path.join(work,'concat.txt')
 
699
 
700
 
701
  app.router.routes = [r for r in app.router.routes if not (getattr(r, 'path', None) == '/' and 'GET' in getattr(r, 'methods', set()))]
702
+
703
+ PATCH_INJECT = r'''
704
+ <style>
705
+ .ai-wall-patched{margin:6px 4px;background:#1a1a1a;border:1px solid #2a2a2a;border-radius:8px;overflow:hidden}
706
+ .ai-wall-card{flex:0 0 250px;background:#141414;border:1px solid #2b2b2b;border-radius:10px;padding:8px}
707
+ .ai-wall-img{width:100%;aspect-ratio:16/9;background:#222;border-radius:8px;overflow:hidden;margin-bottom:6px}
708
+ .ai-wall-img img{width:100%;height:100%;object-fit:cover}
709
+ .ai-wall-title{font-size:12px;color:#5cb87a;font-weight:800;line-height:1.3;margin-bottom:4px}
710
+ .ai-wall-text{font-size:11px;color:#bbb;line-height:1.45;white-space:pre-wrap;display:-webkit-box;-webkit-line-clamp:5;-webkit-box-orient:vertical;overflow:hidden}
711
+ .ai-wall-actions{display:flex;gap:6px;margin-top:8px}
712
+ .ai-wall-actions button,.ai-wall-actions select{flex:1;border:1px solid #333;background:#222;color:#ddd;border-radius:14px;padding:6px 8px;font-size:10px;min-width:0}
713
+ .ai-wall-actions button.primary{background:#2d8659;border-color:#2d8659;color:#fff}
714
+ .ai-short-card{flex:0 0 145px}
715
+ .ai-short-video{width:100%;aspect-ratio:9/16;background:#000;border-radius:8px;overflow:hidden}
716
+ .ai-short-video video{width:100%;height:100%;object-fit:cover}
717
+ .ai-short-progress{position:fixed;inset:0;background:rgba(0,0,0,.78);z-index:99999;display:none;align-items:center;justify-content:center;padding:20px}
718
+ .ai-short-progress.active{display:flex}
719
+ .ai-short-box{max-width:420px;width:100%;background:#141414;border:2px solid #2d8659;border-radius:14px;padding:18px;color:#eee;box-shadow:0 0 30px rgba(45,134,89,.35)}
720
+ .ai-short-box h3{color:#5cb87a;margin-bottom:10px}
721
+ .ai-short-step{font-size:13px;line-height:1.55;color:#ccc}
722
+ .ai-short-spinner{width:34px;height:34px;border:4px solid #333;border-top-color:#5cb87a;border-radius:50%;animation:spin 1s linear infinite;margin:10px auto}
723
+ @keyframes spin{to{transform:rotate(360deg)}}
724
+ </style>
725
+ <div id="ai-short-progress" class="ai-short-progress"><div class="ai-short-box"><h3>🎬 Đang tạo Short AI</h3><div class="ai-short-spinner"></div><div class="ai-short-step" id="ai-short-step">Đang chuẩn bị...</div></div></div>
726
+ <script>
727
+ (function(){
728
+ function esc(s){return String(s||'').replace(/[&<>"']/g,m=>({'&':'&amp;','<':'&lt;','>':'&gt;','"':'&quot;',"'":'&#39;'}[m]));}
729
+ let patchedWall=[];let aiShorts=[];
730
+ function showProgress(msg){let box=document.getElementById('ai-short-progress');let st=document.getElementById('ai-short-step');if(st)st.innerHTML=msg;if(box)box.classList.add('active');}
731
+ function hideProgress(){document.getElementById('ai-short-progress')?.classList.remove('active');}
732
+ function updateAiLabels(){document.querySelectorAll('.ai-compose-title').forEach(e=>e.textContent='🤖 Tường AI: lọc từng bài theo chủ đề, tóm tắt nội dung bài');document.querySelectorAll('button').forEach(b=>{if((b.textContent||'').includes('AI viết lại'))b.textContent='🤖 Tóm tắt AI & đăng tường';});}
733
+ async function loadPatchedWall(){try{const r=await fetch('/api/ai_wall');const j=await r.json();patchedWall=j.posts||[];renderPatchedWall();updateAiLabels();}catch(e){}try{const r2=await fetch('/api/ai_shorts');const j2=await r2.json();aiShorts=j2.posts||[];renderAiShorts();}catch(e){}}
734
+ function renderAiShorts(){const home=document.getElementById('view-home');if(!home)return;document.getElementById('ai-shorts-patched')?.remove();if(!aiShorts.length)return;let wrap=document.createElement('div');wrap.id='ai-shorts-patched';wrap.className='ai-wall-patched';let h='<div class="slider-header"><span class="slider-label">🎬 Short AI</span><span class="slider-note">Video đã tạo</span></div><div class="slider-track">';aiShorts.slice(0,30).forEach((p,i)=>{h+=`<div class="ai-short-card" onclick="aiReadShortPatched(${i})"><div class="ai-short-video"><video src="${p.video}" muted playsinline preload="metadata"></video></div><div class="slider-title">${esc(p.title)}</div></div>`});h+='</div>';wrap.innerHTML=h;let wall=document.getElementById('ai-wall-patched');if(wall)wall.after(wrap);else home.prepend(wrap);}
735
+ function renderPatchedWall(){const home=document.getElementById('view-home');if(!home)return;document.getElementById('ai-wall-patched')?.remove();if(!patchedWall.length)return;let wrap=document.createElement('div');wrap.id='ai-wall-patched';wrap.className='ai-wall-patched';let h='<div class="slider-header"><span class="slider-label">🧱 Tường AI</span><span class="slider-note">Mỗi nguồn = một bài tóm tắt</span></div><div class="slider-track">';patchedWall.slice(0,30).forEach((p,i)=>{h+=`<div class="ai-wall-card"><div class="ai-wall-img">${p.img?`<img src="${p.img}">`:''}</div><div class="ai-wall-title">${esc(p.title)}</div><div class="ai-wall-text">${esc(p.text)}</div><div class="ai-wall-actions"><button onclick="aiReadWallPatched(${i})">Xem</button><button class="primary" onclick="aiMakeShortPatched(${i})">Shorts</button></div></div>`});h+='</div>';wrap.innerHTML=h;let after=document.querySelector('.ai-compose');if(after)after.after(wrap);else home.prepend(wrap);}
736
+ window.aiReadShortPatched=function(i){const p=aiShorts[i];if(!p)return;showView('view-article');let h=`<button class="back-btn" onclick="switchCat('home')">← Quay lại</button><div class="article-view"><span class="badge badge-ai">Short AI</span><h1 class="article-title">${esc(p.title)}</h1><video class="article-img" src="${p.video}" controls playsinline autoplay></video><p class="article-p" style="white-space:pre-wrap">${esc(p.text||'')}</p><div class="article-actions"><button onclick="window.open('${p.video}','_blank')">⬇ Mở video</button>${p.url?`<button onclick="window.open('${p.url}','_blank')">🔗 Nguồn</button>`:''}</div></div>`;document.getElementById('view-article').innerHTML=h;window.scrollTo(0,0)};
737
+ window.aiReadWallPatched=function(i){const p=patchedWall[i];if(!p)return;showView('view-article');let sources='';if(p.sources&&p.sources.length){sources='<div class="article-summary"><b>Nguồn tham khảo:</b><br>'+p.sources.slice(0,5).map(s=>`• ${esc(s.title||s.url||'Nguồn')} ${s.url?`(${esc(new URL(s.url).hostname.replace('www.',''))})`:''}`).join('<br>')+'</div>'}let voiceBox=`<div class="article-actions"><select id="ai-short-voice"><option value="nu">Giọng nữ Việt</option><option value="nam">Giọng nam Việt</option><option value="mien-nam">Giọng miền Nam</option></select><select id="ai-short-emotion"><option value="neutral">Trung tính</option><option value="urgent">Tin nhanh</option><option value="warm">Ấm áp</option><option value="serious">Nghiêm túc</option><option value="energetic">Sôi nổi</option></select><button onclick="aiMakeShortPatched(${i})">🎬 Tạo video shorts</button></div>`;let h=`<button class="back-btn" onclick="switchCat('home')">← Quay lại</button><div class="article-view"><span class="badge badge-ai">AI</span><h1 class="article-title">${esc(p.title)}</h1>${p.img?`<img class="article-img" src="${p.img}">`:''}${sources}<p class="article-p" style="white-space:pre-wrap">${esc(p.text)}</p>${p.video?`<video class="article-img" src="${p.video}" controls playsinline></video>`:''}<div class="article-actions">${p.url?`<button onclick="window.open('${p.url}','_blank')">🔗 Nguồn</button>`:''}</div>${voiceBox}</div>`;document.getElementById('view-article').innerHTML=h;window.scrollTo(0,0)};
738
+ window.aiMakeShortPatched=async function(i){const p=patchedWall[i];if(!p)return;let voice=document.getElementById('ai-short-voice')?.value||'nu';let emotion=document.getElementById('ai-short-emotion')?.value||'neutral';let voiceName={nu:'Giọng nữ Việt',nam:'Giọng nam Việt','mien-nam':'Giọng miền Nam'}[voice]||voice;let emotionName={neutral:'Trung tính',urgent:'Tin nhanh',warm:'Ấm áp',serious:'Nghiêm túc',energetic:'Sôi nổi'}[emotion]||emotion;let ok=confirm(`Quy trình tạo short AI:\n\n1) Dùng ảnh đại diện của bài hoặc tạo ảnh minh họa nếu thiếu.\n2) Rút gọn nội dung tóm tắt thành kịch bản đọc ngắn.\n3) Tự ngắt câu theo dấu câu và xuống dòng hợp lý.\n4) Tạo giọng đọc tiếng Việt: ${voiceName}.\n5) Áp dụng cảm xúc/kịch bản: ${emotionName}.\n6) Tăng tốc giọng đọc 1.2 lần.\n7) Mỗi đoạn tóm tắt sẽ là một cảnh riêng theo thời lượng đọc.\n8) Không thêm phụ đề; video chỉ có chữ cảnh và giọng đọc.\n9) Sau khi xong, video xuất hiện ở slide "Short AI".\n\nQuá trình có thể mất 1-3 phút. Bạn muốn bắt đầu?`);if(!ok)return;try{showProgress(`Bước 1/5: Chuẩn bị ảnh và căn chữ full width...<br>Bước 2/5: Tạo kịch bản, tự ngắt câu/xuống dòng...<br>Bước 3/5: Tạo giọng đọc ${voiceName}, cảm xúc ${emotionName}.<br>Bước 4/5: Tăng tốc 1.2x và ghép từng cảnh riêng, không phụ đề.<br>Bước 5/5: Lưu vào slide "Short AI".`);const r=await fetch('/api/ai/short/'+p.id,{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({voice,emotion,speed:1.2})});const j=await r.json();if(!r.ok||j.error)throw new Error(j.error||'Lỗi');p.video=j.video;hideProgress();alert('Hoàn tất: video shorts đã được tạo và thêm vào slide "Short AI".');aiReadWallPatched(i);loadPatchedWall();}catch(e){hideProgress();alert('Không tạo được shorts: '+e.message)}};
739
+ window.createTopicPost=function(){let inp=document.getElementById('ai-topic-input');let topic=(inp&&inp.value||'').trim();if(!topic)return alert('Nhập chủ đề trước');fetch('/api/topic_post',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({topic})}).then(r=>r.json().then(j=>({ok:r.ok,j}))).then(({ok,j})=>{if(ok&&(j.posts||j.post)){let arr=j.posts||[j.post];patchedWall=arr.concat(patchedWall.filter(x=>!arr.find(y=>y.id===x.id)));renderPatchedWall();if(inp)inp.value='';alert(`Đã lọc và tóm tắt ${arr.length} bài viết theo chủ đề lên Tường AI`);}else alert(j.error||'Lỗi tạo bài')}).catch(e=>alert(e.message||'Lỗi tạo bài'));};
740
+ window.createUrlPost=function(){let inp=document.getElementById('ai-url-input');let url=(inp&&inp.value||'').trim();if(!url)return alert('Dán URL trước');if(!/^https?:\/\//i.test(url))return alert('URL cần bắt đầu bằng http:// hoặc https://');fetch('/api/url_wall',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({url})}).then(r=>r.json().then(j=>({ok:r.ok,j}))).then(({ok,j})=>{if(ok&&j.post){patchedWall=[j.post].concat(patchedWall.filter(x=>x.id!==j.post.id));renderPatchedWall();if(inp)inp.value='';alert('Đã tóm tắt URL và đăng lên Tường AI');}else alert(j.error||'Lỗi URL')}).catch(e=>alert(e.message||'Lỗi URL'));};
741
+ window.rewriteCurrentArticle=function(){if(!window._currentArticle&&typeof _currentArticle!=='undefined')window._currentArticle=_currentArticle;let cur=window._currentArticle||_currentArticle;if(!cur)return;let btn=document.querySelector('.article-actions button.primary');if(btn){btn.textContent='Đang tóm tắt...';btn.disabled=true}fetch('/api/rewrite_share',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({url:cur.url})}).then(r=>r.json().then(j=>({ok:r.ok,j}))).then(({ok,j})=>{if(ok&&j.post){document.getElementById('rewrite-result').innerHTML=`<div class="rewrite-box"><div class="rewrite-title">Đã tóm tắt và đăng Tường AI</div><div class="rewrite-text">${esc(j.post.text||'')}</div></div>`;patchedWall=[j.post].concat(patchedWall.filter(x=>x.id!==j.post.id));renderPatchedWall();alert('Đã tóm tắt lên Tường AI');}else alert(j.error||'Không tóm tắt được')}).catch(e=>alert(e.message||'Lỗi tóm tắt')).finally(()=>{if(btn){btn.textContent='🤖 Tóm tắt AI & đăng tường';btn.disabled=false}})};
742
+ setTimeout(loadPatchedWall,1500);setInterval(updateAiLabels,2000);
743
+ })();
744
+ </script>
745
+ '''
746
+
747
+ @app.get('/')
748
+ async def index_patched():
749
+ with open('/app/static/index.html','r',encoding='utf-8') as f:
750
+ html=f.read()
751
+ return HTMLResponse(html.replace('</body>', PATCH_INJECT+'\n</body>'))
ai_runtime_final6.py CHANGED
@@ -840,10 +840,486 @@ FINAL6E_INJECT = """
840
  function escE(s){return String(s||'').replace(/[&<>\"']/g,m=>({'&':'&amp;','<':'&lt;','>':'&gt;','\"':'&quot;',"'":'&#39;'}[m]));}
841
  window.__topicWallE=[];
842
  function sourceDetailsHtml(p){let arr=p.source_details||[];if(!arr.length)return '';let h='<div class="source-detail-box"><h3>📚 Nội dung từng nguồn đã dùng</h3>';arr.forEach((s,i)=>{h+=`<div class="source-detail-item"><div class="source-detail-title">${i+1}. ${escE(s.title)}</div><div class="source-detail-meta">${escE(s.via||'Nguồn')}</div><div class="source-detail-content">${escE(s.content||'')}</div>${s.url?`<a href="${escE(s.url)}" target="_blank">Mở nguồn gốc</a>`:''}</div>`});h+='</div>';return h;}
843
- function renderTopicWallE(){let home=document.getElementById('view-home');if(!home||!window.__topicWallE.length)return;document.getElementById('ai-wall-topic-live')?.remove();let wrap=document.createElement('div');wrap.id='ai-wall-topic-live';wrap.className='ai-wall-topic-live';let h='<div class="slider-header"><span class="slider-label">🧱 Tường AI mới</span><span class="slider-note">Tổng hợp từ web</span></div><div class="slider-track">';window.__topicWallE.slice(0,20).forEach((p,i)=>{h+=`<div class="wall-item"><div class="wall-thumb">${p.img?`<img src="${escE(p.img)}">`:''}</div><div class="wall-title">${escE(p.title)}</div><div class="wall-text">${escE(p.text)}</div><div class="wall-actions"><button class="primary" onclick="readTopicWallE(${i})">Xem</button></div></div>`});h+='</div>';wrap.innerHTML=h;let comp=document.querySelector('.ai-compose');if(comp)comp.after(wrap);else home.prepend(wrap);}
844
- window.readTopicWallE=function(i){let p=window.__topicWallE[i];if(!p)return;showView('view-article');let imgs=(p.images||[]).filter(Boolean);let gal=imgs.length?'<div class="ai-wall-gallery">'+imgs.slice(0,12).map(u=>`<img src="${escE(u)}" loading="lazy">`).join('')+'</div>':(p.img?`<img class="article-img" src="${escE(p.img)}">`:'');let srcDetails=sourceDetailsHtml(p);document.getElementById('view-article').innerHTML=`<button class="back-btn" onclick="switchCat('home')">← Quay lại</button><div class="article-view"><span class="badge badge-ai">AI</span><h1 class="article-title">${escE(p.title)}</h1>${gal}<p class="article-p" style="white-space:pre-wrap">${escE(p.text)}</p>${srcDetails}<div class="article-actions"><button onclick="shareAI?shareAI(${JSON.stringify(p).replace(/"/g,'&quot;')},false):navigator.clipboard.writeText(location.href)">📤 Chia sẻ</button></div></div>`;window.scrollTo(0,0)};
845
- window.createTopicPostFinal5=async function(){let inp=document.getElementById('ai-topic-input-final5');let topic=(inp&&inp.value||'').trim();if(!topic)return alert('Nhập chủ đề trước');let btn=document.getElementById('ai-topic-btn-final5');if(btn){btn.disabled=true;btn.textContent='Đang tìm nguồn...'}try{let src=await fetch('/api/topic_sources?topic='+encodeURIComponent(topic)).then(r=>r.json()).catch(()=>null);if(btn&&src)btn.textContent='Đã tìm '+(src.count||0)+' nguồn, đang tổng hợp...';let r=await fetch('/api/topic_post',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({topic})});let j=await r.json();if(!r.ok||j.error)throw new Error(j.error||'Lỗi');window.__topicWallE.unshift(j.post);if(inp)inp.value='';renderTopicWallE();readTopicWallE(0);alert('Đã tạo bài tổng hợp từ nội dung web và đăng lên Tường AI.');}catch(e){alert(e.message)}finally{if(btn){btn.disabled=false;btn.textContent='✨ Tạo bài tổng hợp từ web bằng Qwen'}}};
846
- setInterval(()=>{document.querySelectorAll('#ai-topic-input-final3,.topic-final3,#ai-topic-input-final4,.topic-final4').forEach(e=>(e.closest('.topic-final3,.topic-final4,.ai-compose-row')||e).remove());let b=document.getElementById('ai-topic-btn-final5');if(b){b.style.display='block';b.textContent='✨ Tạo bài tổng hợp từ web bằng Qwen';}},1200);
847
  })();
848
  </script>
849
- '''
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
840
  function escE(s){return String(s||'').replace(/[&<>\"']/g,m=>({'&':'&amp;','<':'&lt;','>':'&gt;','\"':'&quot;',"'":'&#39;'}[m]));}
841
  window.__topicWallE=[];
842
  function sourceDetailsHtml(p){let arr=p.source_details||[];if(!arr.length)return '';let h='<div class="source-detail-box"><h3>📚 Nội dung từng nguồn đã dùng</h3>';arr.forEach((s,i)=>{h+=`<div class="source-detail-item"><div class="source-detail-title">${i+1}. ${escE(s.title)}</div><div class="source-detail-meta">${escE(s.via||'Nguồn')}</div><div class="source-detail-content">${escE(s.content||'')}</div>${s.url?`<a href="${escE(s.url)}" target="_blank">Mở nguồn gốc</a>`:''}</div>`});h+='</div>';return h;}
843
+ function renderTopicWallE(){let home=document.getElementById('view-home');if(!home||!window.__topicWallE.length)return;document.getElementById('ai-wall-topic-live')?.remove();let wrap=document.createElement('div');wrap.id='ai-wall-topic-live';wrap.className='ai-wall-topic-live';let h='<div class="slider-header"><span class="slider-label">🧱 Tường AI mới</span><span class="slider-note"> nội dung nguồn</span></div><div class="slider-track">';window.__topicWallE.slice(0,20).forEach((p,i)=>{h+=`<div class="wall-item"><div class="wall-thumb">${p.img?`<img src="${escE(p.img)}">`:''}</div><div class="wall-title">${escE(p.title)}</div><div class="wall-text">${escE(p.text)}</div><div class="wall-actions"><button class="primary" onclick="readTopicWallE(${i})">Xem</button></div></div>`});h+='</div>';wrap.innerHTML=h;let comp=document.querySelector('.ai-compose');if(comp)comp.after(wrap);else home.prepend(wrap);}
844
+ window.readTopicWallE=function(i){let p=window.__topicWallE[i];if(!p)return;showView('view-article');let imgs=(p.images||[]).filter(Boolean);let gal=imgs.length?'<div class="ai-wall-gallery">'+imgs.slice(0,12).map(u=>`<img src="${escE(u)}" loading="lazy">`).join('')+'</div>':(p.img?`<img class="article-img" src="${escE(p.img)}">`:'');document.getElementById('view-article').innerHTML=`<button class="back-btn" onclick="switchCat('home')">← Quay lại</button><div class="article-view"><span class="badge badge-ai">AI</span><h1 class="article-title">${escE(p.title)}</h1>${gal}<p class="article-p" style="white-space:pre-wrap">${escE(p.text)}</p>${sourceDetailsHtml(p)}<div class="article-actions"><button onclick="shareAI?shareAI(${JSON.stringify(p).replace(/\"/g,'&quot;')},false):navigator.clipboard.writeText(location.href)">📤 Chia sẻ</button></div></div>`;window.scrollTo(0,0)};
845
+ window.createTopicPostFinal5=async function(){let inp=document.getElementById('ai-topic-input-final5');let topic=(inp&&inp.value||'').trim();if(!topic)return alert('Nhập chủ đề trước');let btn=document.getElementById('ai-topic-btn-final5');if(btn){btn.disabled=true;btn.textContent='Đang lấy nguồn...'}try{let src=await fetch('/api/topic_sources?topic='+encodeURIComponent(topic)).then(r=>r.json()).catch(()=>null);if(btn&&src)btn.textContent='Đã tìm '+(src.count||0)+' nguồn, đang viết...';let r=await fetch('/api/topic_post',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({topic})});let j=await r.json();if(!r.ok||j.error)throw new Error(j.error||'Lỗi');window.__topicWallE.unshift(j.post);if(inp)inp.value='';renderTopicWallE();readTopicWallE(0);}catch(e){alert(e.message)}finally{if(btn){btn.disabled=false;btn.textContent='✨ Tạo bài tổng hợp từ web bằng Qwen'}}};
 
846
  })();
847
  </script>
848
+ """
849
+
850
+ # Override root one last time to append source-details UI.
851
+ app.router.routes=[r for r in app.router.routes if not (getattr(r,'path',None)=='/' and 'GET' in getattr(r,'methods',set()))]
852
+ @app.get('/')
853
+ async def index_final6_source_details():
854
+ html=f5.f4.f3.f2.f1._load_index_html()
855
+ body=getattr(rt.old,'PATCH_INJECT','')+f5.f4.f3.f2.f1.FINAL_INJECT+f5.f4.f3.FINAL3_INJECT+f5.f4.FINAL4_INJECT+f5.FINAL5_INJECT+FINAL6_INJECT+globals().get('FINAL6_FAST_HOME_INJECT','')+FINAL6E_INJECT
856
+ return HTMLResponse(html.replace('</body>',body+'\n</body>') if '</body>' in html else html+body)
857
+
858
+
859
+ # ===== FINAL6F: CLEAN TOPIC OUTPUT + IN-APP SOURCE READER =====
860
+ def _clean_generated_article(text, topic=''):
861
+ """Remove prompt/instruction leakage from generated topic articles."""
862
+ text=str(text or '').strip()
863
+ bad_patterns=[
864
+ r'^\s*[•\-]*\s*Hãy viết .*',
865
+ r'^\s*[•\-]*\s*Dưới đây là .*',
866
+ r'^\s*[•\-]*\s*Dữ liệu .*',
867
+ r'^\s*[•\-]*\s*NỘI DUNG NGUỒN.*',
868
+ r'^\s*[•\-]*\s*Yêu cầu\s*:.*',
869
+ r'^\s*[•\-]*\s*Tiêu đề mới.*',
870
+ r'^\s*[•\-]*\s*Sapo\s*2.*',
871
+ r'^\s*[•\-]*\s*5\s*[-–]\s*8\s*đoạn.*',
872
+ r'^\s*[•\-]*\s*Không dùng câu.*',
873
+ r'^\s*[•\-]*\s*Cuối bài.*',
874
+ r'^\s*[•\-]*\s*Không liệt kê.*',
875
+ r'^\s*[•\-]*\s*Tổng hợp thành.*',
876
+ r'^\s*[•\-]*\s*Diễn đạt lại.*',
877
+ r'^\s*[•\-]*\s*Tuyệt đối.*',
878
+ ]
879
+ out=[]
880
+ for ln in text.splitlines():
881
+ s=ln.strip()
882
+ if not s:
883
+ out.append(ln);continue
884
+ if any(re.search(p,s,re.I) for p in bad_patterns):
885
+ continue
886
+ out.append(ln)
887
+ cleaned='\n'.join(out).strip()
888
+ # If model returned a markdown code/prompt-like block, keep content after first plausible title line.
889
+ cleaned=re.sub(r'^(?:Bài viết|Nội dung bài viết)\s*[::]\s*','',cleaned,flags=re.I).strip()
890
+ # Remove duplicated leading topic instruction if it appears inline.
891
+ cleaned=re.sub(r'Hãy viết MỘT BÀI VIẾT HOÀN CHỈNH[^\n\.]*[\.\n]*','',cleaned,flags=re.I).strip()
892
+ return cleaned or text
893
+
894
+ def _source_article_data(url):
895
+ try:
896
+ r=requests.get(url,headers=UA,timeout=14);r.encoding='utf-8'
897
+ soup=BeautifulSoup(r.text,'lxml')
898
+ h1=soup.find('h1')
899
+ ogt=soup.find('meta',property='og:title')
900
+ ogd=soup.find('meta',property='og:description')
901
+ ogi=soup.find('meta',property='og:image')
902
+ title=clean(h1.get_text(' ',strip=True) if h1 else (ogt.get('content','') if ogt else ''))
903
+ summary=clean(ogd.get('content','') if ogd else '')
904
+ img=ogi.get('content','') if ogi else ''
905
+ except Exception:
906
+ title='';summary='';img=''
907
+ text=_scrape_article_text(url,12000)
908
+ body=[]
909
+ for para in re.split(r'\n+',text or ''):
910
+ para=clean(para)
911
+ if len(para)>35:
912
+ body.append({'type':'p','text':para})
913
+ if len(body)>=80:break
914
+ if not title:title=url
915
+ if not body and summary:body=[{'type':'p','text':summary}]
916
+ return {'title':title,'summary':summary,'og_image':img,'body':body,'source':'topic-source','url':url}
917
+
918
+ @app.get('/api/topic_source_article')
919
+ def api_topic_source_article(url:str=Query(...)):
920
+ if not url.startswith('http'):
921
+ return JSONResponse({'error':'bad url'},status_code=400)
922
+ return JSONResponse(_source_article_data(url))
923
+
924
+ # Override topic generation one last time with output cleaning and source details.
925
+ app.router.routes=[r for r in app.router.routes if not (getattr(r,'path',None)=='/api/topic_post' and 'POST' in getattr(r,'methods',set()))]
926
+
927
+ @app.post('/api/topic_post')
928
+ async def topic_post_clean_final(request:Request):
929
+ body=await request.json();topic=clean(body.get('topic',''))
930
+ if not topic:return JSONResponse({'error':'missing topic'},status_code=400)
931
+ img=_topic_image(topic)
932
+ research=_fast_context(topic) if '_fast_context' in globals() else _web_research_context(topic)
933
+ context=research.get('context','');sources=research.get('sources',[])
934
+ details=_extract_source_details_from_context(context,sources) if '_extract_source_details_from_context' in globals() else []
935
+ if not details:
936
+ # Build details directly from sources/snippets if helper unavailable or empty.
937
+ for s in sources[:8]:
938
+ details.append({'title':s.get('title',''),'url':s.get('url',''),'via':s.get('via',''),'content':s.get('excerpt','') or s.get('snippet','') or ''})
939
+ if not context and not details:
940
+ return JSONResponse({'error':'Không tìm/crawl được đủ nội dung về chủ đề này. Hãy thử chủ đề cụ thể hơn hoặc dùng hashtag gợi ý.'},status_code=422)
941
+ source_brief='\n\n'.join([f"[{i+1}] {d.get('title','')} ({d.get('via','')})\n{d.get('content','')[:1400]}" for i,d in enumerate(details[:8])])
942
+ prompt=f"""Vai trò: biên tập viên VNEWS.
943
+ Nhiệm vụ: viết một bài báo tiếng Việt hoàn chỉnh về chủ đề "{topic}" dựa trên các nguồn bên dưới.
944
+
945
+ Nguồn thu thập:
946
+ {source_brief[:18000]}
947
+
948
+ Quy tắc biên tập:
949
+ 1. Chỉ xuất bản bài viết cuối cùng, không nhắc lại yêu cầu, không liệt kê chỉ dẫn.
950
+ 2. Không sao chép nguyên văn; hãy tổng hợp và diễn đạt lại.
951
+ 3. Bài có tiêu đề, sapo, các đoạn phân tích/bối cảnh/tác động, và mục Nguồn tham khảo ngắn.
952
+ 4. Không dùng các câu như "Dưới đây là", "Tôi sẽ", "Yêu cầu".
953
+ """
954
+ text=None
955
+ try:
956
+ import asyncio
957
+ text=await asyncio.wait_for(f5.base.qwen_generate(prompt,image_url=img,max_tokens=1700),timeout=35)
958
+ except Exception:
959
+ text=None
960
+ if not text or len(text)<350:
961
+ bullets='\n'.join([f"• {d.get('title','')}: {d.get('content','')[:320]}" for d in details[:6]])
962
+ vias=', '.join(sorted({d.get('via','') for d in details if d.get('via')}))
963
+ text=(f"{topic}: tổng hợp những điểm đáng chú ý\n\n"
964
+ f"{topic} đang được nhiều nguồn tin đề cập với các góc nhìn khác nhau. Dựa trên nội dung đã thu thập, có thể rút ra một số điểm chính để người đọc nắm nhanh bối cảnh.\n\n"
965
+ f"{bullets}\n\n"
966
+ f"Nhìn chung, chủ đề này cần được theo dõi thêm ở các khía cạnh: bối cảnh, tác động thực tế, phản ứng của các bên liên quan và các diễn biến mới trong thời gian tới.\n\n"
967
+ f"Nguồn tham khảo: {vias}")
968
+ text=_clean_generated_article(text,topic)
969
+ post=f5.base.make_post(topic,text,img,'','topic_clean_with_sources',sources=[s for s in sources if s.get('url')])
970
+ post['images']=[img]
971
+ post['source_details']=details[:8]
972
+ posts=f5.base._load_ai_wall();posts.insert(0,post);f5.base._save_ai_wall(posts)
973
+ return JSONResponse({'post':post,'mode':'clean_with_source_details','sources_count':len(details)})
974
+
975
+ FINAL6F_INJECT = """
976
+ <script>
977
+ (function(){
978
+ function escF(s){return String(s||'').replace(/[&<>\\"']/g,m=>({'&':'&amp;','<':'&lt;','>':'&gt;','\\"':'&quot;',"'":'&#39;'}[m]));}
979
+ window.readTopicSourceE=async function(url){
980
+ showView('view-article');
981
+ const el=document.getElementById('view-article');
982
+ el.innerHTML='<div class="loading">Đang tải nguồn...</div>';
983
+ try{
984
+ let data=await fetch('/api/topic_source_article?url='+encodeURIComponent(url)).then(r=>r.json());
985
+ if(!data||data.error||!data.body||!data.body.length){throw new Error('Không đọc được nguồn')}
986
+ let h=`<button class="back-btn" onclick="switchCat('home')">← Quay lại</button><div class="article-view"><h1 class="article-title">${escF(data.title)}</h1>`;
987
+ if(data.summary)h+=`<div class="article-summary">${escF(data.summary)}</div>`;
988
+ if(data.og_image)h+=`<img class="article-img" src="${escF(data.og_image)}">`;
989
+ data.body.forEach(b=>{if(b.type==='p')h+=`<p class="article-p">${escF(b.text)}</p>`;else if(b.type==='heading')h+=`<h2 class="article-h2">${escF(b.text)}</h2>`;});
990
+ h+=`<div class="article-actions"><button onclick="window.open('${escF(url)}','_blank')">🔗 Mở gốc</button></div></div>`;
991
+ el.innerHTML=h;window.scrollTo(0,0);
992
+ }catch(e){el.innerHTML=`<button class="back-btn" onclick="switchCat('home')">← Quay lại</button><div class="loading">Không đọc được nguồn.<br><a href="${escF(url)}" target="_blank" style="color:#5cb87a">Mở link gốc</a></div>`}
993
+ };
994
+ // Upgrade existing source detail boxes: replace external open behavior by in-app button.
995
+ function enhanceSourceButtons(){document.querySelectorAll('.source-detail-item a[href]').forEach(a=>{let u=a.getAttribute('href');if(!u||a.dataset.vnews)return;a.dataset.vnews='1';a.textContent='Xem trực tiếp trên VNEWS';a.setAttribute('href','javascript:void(0)');a.onclick=function(){readTopicSourceE(u);return false;};});}
996
+ setInterval(enhanceSourceButtons,1000);setTimeout(enhanceSourceButtons,500);
997
+ })();
998
+ </script>
999
+ """
1000
+
1001
+ app.router.routes=[r for r in app.router.routes if not (getattr(r,'path',None)=='/' and 'GET' in getattr(r,'methods',set()))]
1002
+ @app.get('/')
1003
+ async def index_final6_clean_links():
1004
+ html=f5.f4.f3.f2.f1._load_index_html()
1005
+ body=getattr(rt.old,'PATCH_INJECT','')+f5.f4.f3.f2.f1.FINAL_INJECT+f5.f4.f3.FINAL3_INJECT+f5.f4.FINAL4_INJECT+f5.FINAL5_INJECT+FINAL6_INJECT+globals().get('FINAL6_FAST_HOME_INJECT','')+globals().get('FINAL6E_INJECT','')+FINAL6F_INJECT
1006
+ return HTMLResponse(html.replace('</body>',body+'\n</body>') if '</body>' in html else html+body)
1007
+
1008
+
1009
+ # ===== FINAL6G: SELECTED FAST SOURCES =====
1010
+ # Restrict hot hashtags + topic articles to the requested sources only:
1011
+ # Thethaovanhoa, Dantri, VTV, VnExpress, Vatvostudio, GenK, VNReview.
1012
+ SELECTED_SOURCE_FEEDS=[
1013
+ ('VnExpress','https://vnexpress.net/rss/tin-moi-nhat.rss'),
1014
+ ('VnExpress Thời sự','https://vnexpress.net/rss/thoi-su.rss'),
1015
+ ('VnExpress Thế giới','https://vnexpress.net/rss/the-gioi.rss'),
1016
+ ('VnExpress Kinh doanh','https://vnexpress.net/rss/kinh-doanh.rss'),
1017
+ ('VnExpress Công nghệ','https://vnexpress.net/rss/so-hoa.rss'),
1018
+ ('VnExpress Thể thao','https://vnexpress.net/rss/the-thao.rss'),
1019
+ ('Dân trí','https://dantri.com.vn/rss/home.rss'),
1020
+ ('Dân trí Xã hội','https://dantri.com.vn/rss/xa-hoi.rss'),
1021
+ ('Dân trí Kinh doanh','https://dantri.com.vn/rss/kinh-doanh.rss'),
1022
+ ('Dân trí Thể thao','https://dantri.com.vn/rss/the-thao.rss'),
1023
+ ('Dân trí Công nghệ','https://dantri.com.vn/rss/suc-manh-so.rss'),
1024
+ ('VTV','https://vtv.vn/rss/trang-chu.rss'),
1025
+ ('VTV Thời sự','https://vtv.vn/rss/thoi-su.rss'),
1026
+ ('VTV Công nghệ','https://vtv.vn/rss/cong-nghe.rss'),
1027
+ ('Thể thao văn hóa','https://thethaovanhoa.vn/rss/home.rss'),
1028
+ ('Thể thao văn hóa Bóng đá','https://thethaovanhoa.vn/rss/bong-da.rss'),
1029
+ ('GenK','https://genk.vn/home.rss'),
1030
+ ('GenK AI','https://genk.vn/ai.rss'),
1031
+ ('VNReview','https://vnreview.vn/rss/home.rss'),
1032
+ ('VNReview Công nghệ','https://vnreview.vn/rss/cong-nghe.rss'),
1033
+ ]
1034
+ SELECTED_HOMEPAGES=[
1035
+ ('VTV','https://vtv.vn/'),
1036
+ ('Thể thao văn hóa','https://thethaovanhoa.vn/'),
1037
+ ('GenK','https://genk.vn/'),
1038
+ ('VNReview','https://vnreview.vn/'),
1039
+ ('Vatvostudio','https://vatvostudio.vn/'),
1040
+ ]
1041
+ SELECTED_DOMAINS=['vnexpress.net','dantri.com.vn','vtv.vn','thethaovanhoa.vn','genk.vn','vnreview.vn','vatvostudio.vn']
1042
+
1043
+ def _selected_fetch_rss(feed_name, feed_url, max_items=10):
1044
+ items=[]
1045
+ try:
1046
+ r=requests.get(feed_url,headers=UA,timeout=4);r.encoding='utf-8'
1047
+ if r.status_code>=400:return []
1048
+ soup=BeautifulSoup(r.text,'xml')
1049
+ for it in soup.find_all('item')[:max_items]:
1050
+ title=clean(it.find('title').get_text(' ',strip=True) if it.find('title') else '')
1051
+ link=clean(it.find('link').get_text(strip=True) if it.find('link') else '')
1052
+ desc_raw=it.find('description').get_text(' ',strip=True) if it.find('description') else ''
1053
+ ds=BeautifulSoup(desc_raw,'lxml')
1054
+ desc=clean(ds.get_text(' ',strip=True))
1055
+ if title and link:
1056
+ items.append({'title':title,'url':link,'source':feed_name,'snippet':desc})
1057
+ except Exception:pass
1058
+ return items
1059
+
1060
+ def _selected_scrape_homepage(name, url, max_items=10):
1061
+ items=[];seen=set()
1062
+ try:
1063
+ r=requests.get(url,headers=UA,timeout=4);r.encoding='utf-8'
1064
+ if r.status_code>=400:return []
1065
+ soup=BeautifulSoup(r.text,'lxml')
1066
+ base=url.rstrip('/')
1067
+ for a in soup.find_all('a',href=True):
1068
+ href=a.get('href','').strip();title=clean(a.get('title','') or a.get_text(' ',strip=True))
1069
+ if not href or not title or len(title)<18:continue
1070
+ if href.startswith('/'):
1071
+ p=urlparse(url); href=f'{p.scheme}://{p.netloc}{href}'
1072
+ if not href.startswith('http') or href in seen:continue
1073
+ dom=_domain(href)
1074
+ if not any(d in dom for d in SELECTED_DOMAINS):continue
1075
+ if any(x in href.lower() for x in ['#','javascript:','facebook','youtube','tiktok']):continue
1076
+ seen.add(href)
1077
+ items.append({'title':title,'url':href,'source':name,'snippet':''})
1078
+ if len(items)>=max_items:break
1079
+ except Exception:pass
1080
+ return items
1081
+
1082
+ def _fast_rss_pool():
1083
+ now=time.time();key='selected_fast_pool'
1084
+ if key in _FAST_TOPIC_CACHE and now-_FAST_TOPIC_CACHE[key]['t']<600:return _FAST_TOPIC_CACHE[key]['d']
1085
+ pool=[];seen=set()
1086
+ # RSS first: fast and reliable.
1087
+ for name,url in SELECTED_SOURCE_FEEDS:
1088
+ for it in _selected_fetch_rss(name,url,10):
1089
+ if it['url'] not in seen:
1090
+ seen.add(it['url']);pool.append(it)
1091
+ # Homepage fallback for sources with weak/no RSS, especially Vatvostudio.
1092
+ for name,url in SELECTED_HOMEPAGES:
1093
+ for it in _selected_scrape_homepage(name,url,10):
1094
+ if it['url'] not in seen:
1095
+ seen.add(it['url']);pool.append(it)
1096
+ _FAST_TOPIC_CACHE[key]={'t':now,'d':pool}
1097
+ return pool
1098
+
1099
+ def _hot_topics():
1100
+ now=time.time()
1101
+ if _HOT_CACHE['d'] and now-_HOT_CACHE['t']<600:return _HOT_CACHE['d']
1102
+ pool=_fast_rss_pool()
1103
+ freq={};display={}
1104
+ for it in pool[:220]:
1105
+ title=re.sub(r'\s+-\s+.*$','',it.get('title',''))
1106
+ kws=[]
1107
+ for m in re.findall(r'([A-ZĐÀ-Ỹ][A-Za-zÀ-ỹ0-9]+(?:\s+[A-ZĐÀ-ỸA-Za-zÀ-ỹ0-9][A-Za-zÀ-ỹ0-9]+){1,4})',title):
1108
+ if len(m)>=6:kws.append(m)
1109
+ kws+=_keywords_from_title(title)
1110
+ for kw in kws[:5]:
1111
+ words=[w for w in clean(kw).split() if w.lower() not in STOP_WORDS]
1112
+ if len(words)<2:continue
1113
+ kw=' '.join(words[:5])
1114
+ if len(kw)<6 or len(kw)>55:continue
1115
+ key=kw.lower();freq[key]=freq.get(key,0)+1;display[key]=kw
1116
+ topics=[];seen=set()
1117
+ for key,_ in sorted(freq.items(),key=lambda x:x[1],reverse=True):
1118
+ kw=display[key]
1119
+ if key in seen:continue
1120
+ seen.add(key);topics.append({'label':'#'+re.sub(r'\s+','',kw.title()),'topic':kw})
1121
+ if len(topics)>=24:break
1122
+ for kw in ['AI tại Việt Nam','Công nghệ Việt Nam','VTV thời sự','VnExpress kinh doanh','Dân trí xã hội','GenK AI','VNReview công nghệ','Vatvostudio smartphone','Thể thao văn hóa World Cup']:
1123
+ if kw.lower() not in seen:topics.append({'label':'#'+re.sub(r'\s+','',kw.title()),'topic':kw})
1124
+ _HOT_CACHE.update({'t':now,'d':topics[:24]})
1125
+ return _HOT_CACHE['d']
1126
+
1127
+ def _candidate_urls(topic):
1128
+ seen=set();items=[]
1129
+ scored=[]
1130
+ for it in _fast_rss_pool():
1131
+ sc=_fast_score(topic,it)
1132
+ if sc>0:scored.append((sc,it))
1133
+ for sc,it in sorted(scored,key=lambda x:(x[0],len(x[1].get('snippet',''))),reverse=True)[:14]:
1134
+ if it['url'] not in seen:
1135
+ seen.add(it['url']);items.append(it)
1136
+ # Search only selected sources when RSS lacks matches.
1137
+ if len(items)<6:
1138
+ for dom in SELECTED_DOMAINS:
1139
+ for it in _ddg_search(f'{topic} site:{dom}',4):
1140
+ if it['url'] not in seen:
1141
+ seen.add(it['url']);items.append(it)
1142
+ if len(items)>=12:break
1143
+ return items[:20]
1144
+
1145
+
1146
+ # ===== FINAL6G: SOURCE-LIMITED FAST TOPICS AND FAST HOME =====
1147
+ # Limit hot hashtags/topic context to requested sources and make homepage APIs return quickly.
1148
+ _SOURCE_FEEDS = [
1149
+ ('VnExpress','https://vnexpress.net/rss/tin-moi-nhat.rss','vnexpress.net'),
1150
+ ('VnExpress Thời sự','https://vnexpress.net/rss/thoi-su.rss','vnexpress.net'),
1151
+ ('VnExpress Thế giới','https://vnexpress.net/rss/the-gioi.rss','vnexpress.net'),
1152
+ ('VnExpress Kinh doanh','https://vnexpress.net/rss/kinh-doanh.rss','vnexpress.net'),
1153
+ ('VnExpress Công nghệ','https://vnexpress.net/rss/so-hoa.rss','vnexpress.net'),
1154
+ ('VnExpress Thể thao','https://vnexpress.net/rss/the-thao.rss','vnexpress.net'),
1155
+ ('Dân trí','https://dantri.com.vn/rss/home.rss','dantri.com.vn'),
1156
+ ('Dân trí Xã hội','https://dantri.com.vn/rss/xa-hoi.rss','dantri.com.vn'),
1157
+ ('Dân trí Kinh doanh','https://dantri.com.vn/rss/kinh-doanh.rss','dantri.com.vn'),
1158
+ ('Dân trí Thể thao','https://dantri.com.vn/rss/the-thao.rss','dantri.com.vn'),
1159
+ ('Dân trí Công nghệ','https://dantri.com.vn/rss/suc-manh-so.rss','dantri.com.vn'),
1160
+ ('VTV','https://vtv.vn/rss/trang-chu.rss','vtv.vn'),
1161
+ ('VTV Thời sự','https://vtv.vn/rss/thoi-su.rss','vtv.vn'),
1162
+ ('GenK','https://genk.vn/rss/home.rss','genk.vn'),
1163
+ ('GenK AI','https://genk.vn/ai.rss','genk.vn'),
1164
+ ('VnReview','https://vnreview.vn/rss/tin-moi-nhat.rss','vnreview.vn'),
1165
+ ('VnReview Công nghệ','https://vnreview.vn/rss/cong-nghe.rss','vnreview.vn'),
1166
+ ('Vật Vờ Studio','https://vatvostudio.vn/feed/','vatvostudio.vn'),
1167
+ ('Thể thao văn hóa','https://thethaovanhoa.vn/rss/home.rss','thethaovanhoa.vn'),
1168
+ ('Thể thao văn hóa World Cup','https://thethaovanhoa.vn/rss/world-cup-2026.rss','thethaovanhoa.vn'),
1169
+ ]
1170
+ _SOURCE_CACHE={'t':0,'items':[]}
1171
+ _FAST_ROUTE_CACHE={}
1172
+
1173
+ def _feed_items_source_limited(max_per_feed=10):
1174
+ now=time.time()
1175
+ if _SOURCE_CACHE['items'] and now-_SOURCE_CACHE['t']<600:return _SOURCE_CACHE['items']
1176
+ items=[];seen=set()
1177
+ def one(feed):
1178
+ name,url,dom=feed;out=[]
1179
+ try:
1180
+ r=requests.get(url,headers=UA,timeout=3.5);r.encoding='utf-8'
1181
+ soup=BeautifulSoup(r.text,'xml')
1182
+ for it in soup.find_all('item')[:max_per_feed*2]:
1183
+ title=clean(it.find('title').get_text(' ',strip=True) if it.find('title') else '')
1184
+ link=clean(it.find('link').get_text(strip=True) if it.find('link') else '')
1185
+ desc_raw=it.find('description').get_text(' ',strip=True) if it.find('description') else ''
1186
+ ds=BeautifulSoup(desc_raw,'lxml')
1187
+ img='';im=ds.find('img')
1188
+ if im:img=im.get('src','') or im.get('data-src','')
1189
+ desc=clean(ds.get_text(' ',strip=True))[:700]
1190
+ if title and link:
1191
+ out.append({'title':title,'url':link,'link':link,'source':name,'via':name,'domain':dom,'snippet':desc,'img':img})
1192
+ if len(out)>=max_per_feed:break
1193
+ except Exception:pass
1194
+ return out
1195
+ try:
1196
+ from concurrent.futures import ThreadPoolExecutor, as_completed
1197
+ with ThreadPoolExecutor(max_workers=8) as ex:
1198
+ futs=[ex.submit(one,f) for f in _SOURCE_FEEDS]
1199
+ for f in as_completed(futs,timeout=5.5):
1200
+ try:
1201
+ for it in f.result() or []:
1202
+ if it['url'] not in seen:
1203
+ seen.add(it['url']);items.append(it)
1204
+ except Exception:pass
1205
+ except Exception:
1206
+ for f in _SOURCE_FEEDS[:8]:
1207
+ for it in one(f):
1208
+ if it['url'] not in seen:
1209
+ seen.add(it['url']);items.append(it)
1210
+ _SOURCE_CACHE.update({'t':now,'items':items})
1211
+ return items
1212
+
1213
+ def _score_topic_source(topic,it):
1214
+ toks=[w.lower() for w in re.findall(r'[A-Za-zÀ-ỹ0-9]+',topic or '') if len(w)>1 and w.lower() not in STOP_WORDS]
1215
+ hay=(it.get('title','')+' '+it.get('snippet','')+' '+it.get('source','')).lower()
1216
+ if not toks:return 0
1217
+ score=sum((3 if len(t)>3 else 1) for t in toks if t in hay)
1218
+ if topic.lower().strip() and topic.lower().strip() in hay:score+=12
1219
+ return score
1220
+
1221
+ def _hot_topics():
1222
+ now=time.time()
1223
+ if _HOT_CACHE['d'] and now-_HOT_CACHE['t']<600:return _HOT_CACHE['d']
1224
+ freq={};display={}
1225
+ for it in _feed_items_source_limited(8)[:180]:
1226
+ title=re.sub(r'\s+-\s+.*$','',it.get('title',''))
1227
+ kws=[]
1228
+ for m in re.findall(r'([A-ZĐÀ-Ỹ][A-Za-zÀ-ỹ0-9]+(?:\s+[A-ZĐÀ-ỸA-Za-zÀ-ỹ0-9][A-Za-zÀ-ỹ0-9]+){1,4})',title):
1229
+ if len(m)>=6:kws.append(m)
1230
+ kws += _keywords_from_title(title)
1231
+ for kw in kws[:4]:
1232
+ words=[w for w in clean(kw).split() if w.lower() not in STOP_WORDS]
1233
+ if len(words)<2:continue
1234
+ kw=' '.join(words[:5])
1235
+ if 6<=len(kw)<=55:
1236
+ key=kw.lower();freq[key]=freq.get(key,0)+1;display[key]=kw
1237
+ topics=[]
1238
+ for key,_ in sorted(freq.items(),key=lambda x:x[1],reverse=True):
1239
+ kw=display[key];topics.append({'label':'#'+re.sub(r'\s+','',kw.title()),'topic':kw})
1240
+ if len(topics)>=24:break
1241
+ for kw in ['Giá vàng trong nước','AI tại Việt Nam','Bóng đá Việt Nam','Kinh tế Việt Nam','Công nghệ AI','Vật Vờ Studio','World Cup 2026','Sức khỏe cộng đồng']:
1242
+ if not any(t['topic'].lower()==kw.lower() for t in topics):topics.append({'label':'#'+re.sub(r'\s+','',kw.title()),'topic':kw})
1243
+ _HOT_CACHE.update({'t':now,'d':topics[:24]})
1244
+ return _HOT_CACHE['d']
1245
+
1246
+ def _fast_context(topic):
1247
+ now=time.time();key='source_limited_ctx:'+topic.lower().strip()
1248
+ if key in _FAST_TOPIC_CACHE and now-_FAST_TOPIC_CACHE[key]['t']<600:return _FAST_TOPIC_CACHE[key]['d']
1249
+ pool=_feed_items_source_limited(12)
1250
+ scored=[]
1251
+ for it in pool:
1252
+ sc=_score_topic_source(topic,it)
1253
+ if sc>0:scored.append((sc,it))
1254
+ if not scored:
1255
+ # Try broader matching by first token only before giving up.
1256
+ first=(_fast_topic_tokens(topic) or [''])[0]
1257
+ if first:
1258
+ for it in pool:
1259
+ if first in (it.get('title','')+' '+it.get('snippet','')).lower():scored.append((1,it))
1260
+ picked=[it for sc,it in sorted(scored,key=lambda x:(x[0],len(x[1].get('snippet',''))),reverse=True)[:8]]
1261
+ if not picked:picked=pool[:6]
1262
+ blocks=[];src=[]
1263
+ for it in picked:
1264
+ content=it.get('snippet','') or it.get('title','')
1265
+ blocks.append(f"NGUỒN: {it.get('source','')}\nTIÊU ĐỀ: {it.get('title','')}\nTÓM TẮT RSS:\n{content}")
1266
+ src.append({'title':it.get('title',''),'url':it.get('url',''),'via':it.get('source',''),'snippet':content})
1267
+ data={'context':'\n\n---\n\n'.join(blocks),'sources':src,'count':len(blocks)}
1268
+ _FAST_TOPIC_CACHE[key]={'t':now,'d':data}
1269
+ return data
1270
+
1271
+ # Override slow search functions to never crawl open web during topic generation.
1272
+ def _web_research_context(topic):
1273
+ return _fast_context(topic)
1274
+
1275
+ def _candidate_urls(topic):
1276
+ return _fast_context(topic).get('sources',[])
1277
+
1278
+ # Fast homepage endpoints from requested source RSS; no slow HTML scrapers.
1279
+ def _fast_homepage_sources():
1280
+ now=time.time();key='home_sources'
1281
+ if key in _FAST_ROUTE_CACHE and now-_FAST_ROUTE_CACHE[key]['t']<600:return _FAST_ROUTE_CACHE[key]['d']
1282
+ groups=[];seen=set()
1283
+ group_map=[('Tin mới','https://vnexpress.net/rss/tin-moi-nhat.rss','vne'),('Thời Sự','https://vnexpress.net/rss/thoi-su.rss','vne'),('Kinh Doanh','https://vnexpress.net/rss/kinh-doanh.rss','vne'),('Công Nghệ','https://vnexpress.net/rss/so-hoa.rss','vne'),('Dân Trí','https://dantri.com.vn/rss/home.rss','dantri'),('GenK','https://genk.vn/rss/home.rss','genk'),('VnReview','https://vnreview.vn/rss/tin-moi-nhat.rss','vnreview')]
1284
+ for g,u,s in group_map:
1285
+ for it in _rss_articles_fast(u,g,s,6) if '_rss_articles_fast' in globals() else []:
1286
+ if it['link'] not in seen:
1287
+ seen.add(it['link']);groups.append(it)
1288
+ _FAST_ROUTE_CACHE[key]={'t':now,'d':groups}
1289
+ return groups
1290
+
1291
+ for _p in ['/api/homepage','/api/dantri_hot','/api/vne_video','/api/highlights','/api/hot_topics','/api/topic_sources']:
1292
+ app.router.routes=[r for r in app.router.routes if not (getattr(r,'path',None)==_p and 'GET' in getattr(r,'methods',set()))]
1293
+ @app.get('/api/homepage')
1294
+ def api_homepage_source_fast():return JSONResponse(_fast_homepage_sources())
1295
+ @app.get('/api/dantri_hot')
1296
+ def api_dantri_hot_source_fast():
1297
+ data=[{**it,'source':'dantri','link':it.get('url') or it.get('link')} for it in _feed_items_source_limited(8) if it.get('domain')=='dantri.com.vn'][:12]
1298
+ return JSONResponse(data)
1299
+ @app.get('/api/vne_video')
1300
+ def api_vne_video_source_fast():
1301
+ return JSONResponse([]) # do not block homepage if VnEgo is slow
1302
+ @app.get('/api/highlights')
1303
+ def api_highlights_source_fast():return JSONResponse([])
1304
+ @app.get('/api/hot_topics')
1305
+ def api_hot_topics_source_fast():return JSONResponse({'topics':_hot_topics(),'sources':'vn_only'})
1306
+ @app.get('/api/topic_sources')
1307
+ def api_topic_sources_source_fast(topic:str=Query(...)):
1308
+ data=_fast_context(clean(topic));return JSONResponse({'count':data.get('count',0),'sources':data.get('sources',[]),'has_context':bool(data.get('context')),'mode':'source_limited_rss'})
1309
+
1310
+ # Override root: include all existing UI but add a script that prevents forced shorts refresh on initial load.
1311
+ ROOT_FAST_INJECT="""
1312
+ <script>
1313
+ (function(){
1314
+ const oldFetch=window.fetch;window.__allowShortRefresh=false;
1315
+ window.fetch=function(url,opts){try{let u=String(url||'');if(u.includes('/api/shorts?refresh=1')&&!window.__allowShortRefresh)url='/api/shorts';}catch(e){}return oldFetch.call(this,url,opts)};
1316
+ setTimeout(()=>{window.__allowShortRefresh=true;},8000);
1317
+ })();
1318
+ </script>
1319
+ """
1320
+ app.router.routes=[r for r in app.router.routes if not (getattr(r,'path',None)=='/' and 'GET' in getattr(r,'methods',set()))]
1321
+ @app.get('/')
1322
+ async def index_final_fast_sources():
1323
+ html=f5.f4.f3.f2.f1._load_index_html()
1324
+ body=getattr(rt.old,'PATCH_INJECT','')+f5.f4.f3.f2.f1.FINAL_INJECT+f5.f4.f3.FINAL3_INJECT+f5.f4.FINAL4_INJECT+f5.FINAL5_INJECT+FINAL6_INJECT+globals().get('FINAL6_FAST_HOME_INJECT','')+globals().get('FINAL6E_INJECT','')+globals().get('FINAL6F_INJECT','')+ROOT_FAST_INJECT
1325
+ return HTMLResponse(html.replace('</body>',body+'\n</body>') if '</body>' in html else html+body)
app_v2_entry.py CHANGED
@@ -1,4 +1,4 @@
1
- """VNEWS v2 Entry Point - with fast bongda proxy + rewrite endpoints + multilingual TTS"""
2
  import sys, os
3
  from main import app, HEADERS, BONGDA_HEADERS, fetch_bongda_api, HL_LEAGUES
4
 
@@ -7,18 +7,13 @@ try:
7
  except Exception as e:
8
  print(f"[WARN] ai_ext import failed: {e}")
9
 
10
- try:
11
- import ai_patch
12
- except Exception as e:
13
- print(f"[WARN] ai_patch import failed: {e}")
14
-
15
  from fastapi.responses import HTMLResponse, JSONResponse, FileResponse, Response
16
  from fastapi.staticfiles import StaticFiles
17
  from starlette.routing import Mount
18
- from fastapi import Query, Request, UploadFile, File, Form
19
  import requests as req
20
  from bs4 import BeautifulSoup
21
- import re, html as html_lib, json, threading, time, uuid
22
  from concurrent.futures import ThreadPoolExecutor, as_completed
23
  from urllib.parse import quote
24
 
@@ -36,17 +31,22 @@ _match_cache = {}
36
 
37
  # === FAST BONGDA PROXY ENDPOINT ===
38
  def _get_match_detail(event_id, slug=None):
 
39
  headers = {"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36", "Accept": "text/html", "Referer": "https://bongda.com.vn/"}
 
40
  if slug:
41
  url = f"https://bongda.com.vn/tran-dau/{event_id}/centre/{slug}"
42
  else:
43
  url = f"https://bongda.com.vn/tran-dau/{event_id}"
 
44
  resp = req.get(url, headers=headers, timeout=15, allow_redirects=True)
45
  if resp.status_code != 200:
46
  return None
 
47
  soup = BeautifulSoup(resp.text, 'html.parser')
48
  result = {"event_id": event_id, "found": False, "sections": []}
49
  info = {}
 
50
  tel = soup.select_one('.teams')
51
  if tel:
52
  he = tel.select_one('.team.home')
@@ -68,22 +68,27 @@ def _get_match_detail(event_id, slug=None):
68
  if len(parts) >= 2: info['score'] = f"{parts[0]} - {parts[1]}"
69
  lb = sc.select_one('.label')
70
  if lb: info['status_label'] = _clean(lb.get_text())
 
71
  if info.get('home_team') and info.get('away_team'):
72
  result['info'] = info
73
  result['found'] = True
74
  result['sections'].append('info')
 
75
  events = []
76
  for ev in soup.select('.events .period .event'):
77
  ev_cls = ' '.join(ev.get('class', []))
78
  ev_data = {'team': 'home' if 'home' in ev_cls else 'away', 'period': '', 'type': 'unknown', 'time': '', 'players': ''}
 
79
  parent = ev.parent
80
  if parent:
81
  h2 = parent.find('h2')
82
  if h2: ev_data['period'] = _clean(h2.get_text())
 
83
  if ev.select_one('[class*="goal"]'): ev_data['type'] = 'goal'
84
  elif ev.select_one('[class*="redcard"]'): ev_data['type'] = 'redcard'
85
  elif ev.select_one('[class*="yellowcard"]'): ev_data['type'] = 'yellowcard'
86
  elif ev.select_one('[class*="substitution"]'): ev_data['type'] = 'substitution'
 
87
  players_el = ev.select_one('.players')
88
  if players_el:
89
  pl_text = _clean(players_el.get_text(' ', strip=True))
@@ -94,9 +99,11 @@ def _get_match_detail(event_id, slug=None):
94
  else:
95
  ev_data['players'] = pl_text
96
  events.append(ev_data)
 
97
  if events:
98
  result['events'] = events
99
  result['sections'].append('events')
 
100
  pred = soup.select_one('.prediction-card')
101
  if pred:
102
  team_info = pred.select_one('.team-info')
@@ -111,6 +118,7 @@ def _get_match_detail(event_id, slug=None):
111
  vc = pred.select_one('.vote-count')
112
  if vc: pred_data['vote_count'] = _clean(vc.get_text())
113
  result['prediction'] = pred_data
 
114
  recent = []
115
  ml = soup.select_one('.matches-list')
116
  if ml:
@@ -125,6 +133,7 @@ def _get_match_detail(event_id, slug=None):
125
  if recent:
126
  result['recent_matches'] = recent
127
  result['sections'].append('recent')
 
128
  try:
129
  api_h = {"User-Agent": "Mozilla/5.0", "Accept": "application/json", "X-Requested-With": "XMLHttpRequest", "Referer": "https://bongda.com.vn/"}
130
  ar = req.get(f"https://bongda.com.vn/api/fixtures/h2h-stats?event_id={event_id}", headers=api_h, timeout=10)
@@ -142,17 +151,20 @@ def _get_match_detail(event_id, slug=None):
142
  result['h2h_stats_parsed'] = ast
143
  result['sections'].append('h2h_stats')
144
  except: pass
 
145
  return result
146
 
147
  @app.get('/api/proxy/bongda')
148
  def proxy_bongda(event_id: int = Query(default=None), slug: str = Query(default=None)):
149
  if event_id is None:
150
  return JSONResponse({'error': 'event_id required'}, status_code=400)
 
151
  cache_key = f"{event_id}_{slug}"
152
  now = time.time()
153
  cached = _match_cache.get(cache_key)
154
  if cached and now - cached.get('_ts', 0) < 300:
155
  return JSONResponse(cached)
 
156
  try:
157
  result = _get_match_detail(event_id, slug)
158
  if result:
@@ -163,21 +175,26 @@ def proxy_bongda(event_id: int = Query(default=None), slug: str = Query(default=
163
  err = {"event_id": event_id, "found": False, "error": str(e), "_ts": now}
164
  _match_cache[cache_key] = err
165
  return JSONResponse(err)
 
166
  return JSONResponse({"event_id": event_id, "found": False})
167
 
168
  @app.get('/api/match/{event_id}/detail')
169
  def api_match_detail(event_id: int, url: str = Query(default=None)):
 
170
  slug = None
171
  if url:
172
  m = re.match(r'.+/tran-dau/\d+/(?:centre|preview)/(.+)', url)
173
  if m:
174
  slug = m.group(1)
 
175
  cache_key = f"{event_id}_{slug or ''}"
176
  now = time.time()
177
  cached = _match_cache.get(cache_key)
178
  if cached and now - cached.get('_ts', 0) < 300:
179
  return JSONResponse(cached)
 
180
  try:
 
181
  if not slug:
182
  try:
183
  home_r = req.get("https://bongda.com.vn/", headers={"User-Agent": "Mozilla/5.0"}, timeout=10)
@@ -191,6 +208,7 @@ def api_match_detail(event_id: int, url: str = Query(default=None)):
191
  cache_key = f"{event_id}_{slug}"
192
  break
193
  except: pass
 
194
  result = _get_match_detail(event_id, slug)
195
  if result:
196
  result['_ts'] = now
@@ -200,8 +218,10 @@ def api_match_detail(event_id: int, url: str = Query(default=None)):
200
  err = {"event_id": event_id, "found": False, "error": str(e), "_ts": now}
201
  _match_cache[cache_key] = err
202
  return JSONResponse(err)
 
203
  return JSONResponse({"event_id": event_id, "found": False})
204
 
 
205
  _STOP=set('và của các những một được trong với cho tại sau trước khi không người việt nam hôm nay mới nhất nóng tin tức cập nhật theo từ đến là có thì này đã để'.split())
206
 
207
  def _has_kw(topic,title):
@@ -241,7 +261,7 @@ def _s_vietnamnet(topic,limit=6):
241
  try:
242
  r=req.get(f"https://vietnamnet.vn/tim-kiem?q={quote(topic)}",headers={'User-Agent':'Mozilla/5.0'},timeout=10);soup=BeautifulSoup(r.text,'lxml')
243
  for a in soup.select('h3 a[href], .vnn-title a')[:limit*2]:
244
- t=_clean(a.get('title','') or a.get_text(strip=True));href=a.get('href','')
245
  if t and len(t)>15 and _has_kw(topic,t):
246
  if not href.startswith('http'):href='https://vietnamnet.vn'+href
247
  items.append({'title':t,'url':href,'via':'VietNamNet'})
@@ -327,133 +347,43 @@ def _search_all(topic,limit=36):
327
  if i<len(s) and s[i].get('url') and s[i]['url'] not in seen:seen.add(s[i]['url']);out.append(s[i])
328
  return out[:limit]
329
 
330
- for _path in ['/api/article', '/api/hot_topics', '/api/categories', '/api/storage_status', '/s']:
331
- app.router.routes=[r for r in app.router.routes if not(getattr(r,'path',None)==_path and 'GET' in getattr(r,'methods',set()))]
332
-
333
- _article_cache = {}
334
- _article_cache_ttl = 1800
335
-
336
- _art_session = None
337
- _art_lock = threading.Lock()
338
- def _get_art_session():
339
- global _art_session
340
- if _art_session is None:
341
- with _art_lock:
342
- if _art_session is None:
343
- _art_session = req.Session()
344
- _art_session.headers.update({
345
- "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36",
346
- "Accept-Language": "vi-VN,vi;q=0.9,en;q=0.8",
347
- "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8",
348
- })
349
- return _art_session
350
 
351
- def _scrape_article_fast(url):
352
- from urllib.parse import urlparse
353
- domain = urlparse(url).netloc
354
- sess = _get_art_session()
355
- uas = [
356
- {"User-Agent": "Mozilla/5.0 (iPhone; CPU iPhone OS 17_0 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/17.0 Mobile/15E148 Safari/604.1"},
357
- {"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36"},
358
- ]
359
- for ua in uas:
360
- try:
361
- r = sess.get(url, headers=ua, timeout=6, allow_redirects=True)
362
- if not r or r.status_code != 200:
363
- continue
364
- r.encoding = 'utf-8'
365
- soup = BeautifulSoup(r.text, 'lxml')
366
- for tag in soup.find_all(['script','style','nav','footer','aside','form','noscript','iframe','.ads','.ad','.banner-ads','.fb-comments','.fb-root','.social-share','.related-news','.tag','.breadcrumb']):
367
- tag.decompose()
368
- title = summary = og_img = ""
369
- ogt = soup.find('meta', property='og:title')
370
- if ogt: title = ogt.get('content', '')
371
- ogd = soup.find('meta', property='og:description') or soup.find('meta', attrs={'name': 'description'})
372
- if ogd: summary = ogd.get('content', '')[:500]
373
- ogi = soup.find('meta', property='og:image')
374
- if ogi:
375
- og_img = ogi.get('content', '')
376
- if og_img.startswith('//'): og_img = 'https:' + og_img
377
- h1 = soup.find('h1')
378
- if not title and h1: title = h1.get_text(strip=True)[:200]
379
- body = []
380
- selectors = [
381
- '.fck_detail', '.sidebar-1',
382
- '.singular-content', '.dt__content', '.article-content', '.content-detail', '#divNewsContent',
383
- '.content-detail', '.main-content-detail', '.box-content',
384
- '.knc-content', '.article-body', '.detail-body',
385
- '.article-detail', '.detail-content',
386
- 'article', 'main', '.cms-body', '.article__body', '.post-content',
387
- '.entry-content', '#content', '.article-text', '.story-body',
388
- ]
389
- for sel in selectors:
390
- el = soup.select_one(sel)
391
- if el and len(el.find_all('p')) >= 2:
392
- seen_imgs = set()
393
- for child in el.find_all(['p','h2','h3','figure','img'], recursive=True):
394
- if child.name == 'p':
395
- t = child.get_text(strip=True)
396
- if t and len(t) > 15:
397
- body.append({'type': 'p', 'text': t})
398
- elif child.name in ('h2','h3'):
399
- t = child.get_text(strip=True)
400
- if t:
401
- body.append({'type': 'heading', 'text': t})
402
- elif child.name in ('figure','img'):
403
- im = child if child.name == 'img' else child.find('img')
404
- if im:
405
- src = im.get('data-src') or im.get('src') or im.get('data-lazy') or ''
406
- if src and 'base64' not in src and src not in seen_imgs:
407
- seen_imgs.add(src)
408
- if src.startswith('//'): src = 'https:' + src
409
- body.append({'type': 'img', 'src': src})
410
- if child.name == 'figure':
411
- cap = child.find('figcaption')
412
- if cap:
413
- ct = cap.get_text(strip=True)
414
- if ct: body.append({'type': 'p', 'text': ct})
415
- if len(body) >= 2:
416
- return {'title': _clean(title), 'summary': _clean(summary), 'og_image': og_img,
417
- 'body': body[:50], 'source': domain, 'url': url}
418
- if title and (summary or og_img):
419
- fallback = []
420
- if og_img: fallback.append({'type': 'img', 'src': og_img})
421
- if summary: fallback.append({'type': 'p', 'text': summary})
422
- if fallback:
423
- return {'title': _clean(title), 'summary': _clean(summary), 'og_image': og_img,
424
- 'body': fallback, 'source': domain, 'url': url, 'fallback': True}
425
- if title:
426
- return {'title': _clean(title), 'summary': '', 'og_image': '',
427
- 'body': [{'type': 'p', 'text': 'Nội dung đang được tải...'}],
428
- 'source': domain, 'url': url, 'fallback': True}
429
- break
430
- except Exception:
431
- continue
432
- return None
433
 
434
  @app.get('/api/article')
435
- def api_article_v2(url: str = Query(...)):
436
- from urllib.parse import unquote
437
- safe_url = unquote(url)
438
- try:
439
- now = time.time()
440
- cached = _article_cache.get(safe_url)
441
- if cached and now - cached['t'] < _article_cache_ttl:
442
- resp = JSONResponse(cached['d'])
443
- resp.headers["Cache-Control"] = "public, max-age=1800"
444
- return resp
445
- data = _scrape_article_fast(safe_url)
446
- if data and data.get('body'):
447
- _article_cache[safe_url] = {'d': data, 't': now}
448
- resp = JSONResponse(data)
449
- resp.headers["Cache-Control"] = "public, max-age=1800"
450
- return resp
451
- result = {'error': 'Không đọc được', 'url': safe_url}
452
- resp = JSONResponse(result)
453
- resp.headers["Cache-Control"] = "public, max-age=60"
454
- return resp
455
- except Exception as e:
456
- return JSONResponse({'error': f'Server error: {str(e)[:100]}', 'url': safe_url}, status_code=200)
457
 
458
  _hot_cache={'t':0,'d':[]}
459
  def _get_hot_topics():
@@ -486,10 +416,7 @@ def _get_hot_topics():
486
  _hot_cache.update({'t':now,'d':topics[:24]});return topics[:24]
487
 
488
  @app.get('/api/hot_topics')
489
- def api_hot_topics():
490
- resp = JSONResponse({'topics':_get_hot_topics()})
491
- resp.headers["Cache-Control"] = "public, max-age=120"
492
- return resp
493
  @app.get('/')
494
  async def serve_index():
495
  p=os.path.join(STATIC_DIR,'index_v2.html')
@@ -506,12 +433,14 @@ def _st():return JSONResponse({'persistent':os.path.isdir('/data') and os.access
506
  @app.get('/s')
507
  async def _sh(url:str='',title:str='',img:str=''):return HTMLResponse(f'<!DOCTYPE html><html><head><meta property="og:title" content="{_clean(title)}"><meta property="og:image" content="{_clean(img)}"><meta http-equiv="refresh" content="0;url={_clean(url) or "/"}"></head><body></body></html>')
508
 
509
- from wc2026_scraper import scrape_summary,scrape_fixtures,scrape_standings,scrape_stats,scrape_wc_news,scrape_road_to_wc,get_wc2026_all,scrape_history,scrape_h2h,scrape_lineups,scrape_match_detail
510
 
 
511
  _xlb_cache = {}
512
  _xlb_lock = threading.Lock()
513
 
514
  def _xlb_scrape(path):
 
515
  url = f"https://xemlaibongda.top/{path}"
516
  r = req.get(url, headers={"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36"}, timeout=15, allow_redirects=True)
517
  if r.status_code != 200:
@@ -526,53 +455,36 @@ def _xlb_scrape(path):
526
  seen.add(href)
527
  if not href.startswith('http'):
528
  href = 'https://xemlaibongda.top' + href
529
- img = a.select_one('img')
530
- p = a.parent
531
- for _ in range(4):
532
- if img:
533
- break
534
- if p:
535
- img = p.select_one('img')
536
- p = p.parent
537
- img_src = ''
538
- if img:
539
- img_src = img.get('data-src','') or img.get('src','') or img.get('data-lazy','') or img.get('data-original','')
540
- if img_src.startswith('//'):
541
- img_src = 'https:' + img_src
542
- elif img_src.startswith('/'):
543
- img_src = 'https://xemlaibongda.top' + img_src
544
  title = ''
545
- for sel in ['.title', 'h3', 'h2', '.name', '.post-title', '.entry-title', '.video-title']:
546
  t = a.select_one(sel)
547
  if t:
548
  title = _clean(t.get_text())
549
  break
550
  if not title:
551
- title = _clean(a.get('title',''))
552
- if not title:
553
- img_alt = a.select_one('img')
554
- if img_alt:
555
- title = _clean(img_alt.get('alt',''))
556
- if not title:
557
- parent = a.parent
558
- if parent:
559
- pt = _clean(parent.get_text(' ',strip=True))
560
- if 5 < len(pt) < 120:
561
- title = pt
562
- if not title or len(title) < 3:
563
- continue
564
- vids.append({"link": href, "img": img_src, "title": title})
565
  if len(vids) >= 30:
566
  break
567
  return vids
568
 
569
  @app.get('/api/proxy/xlb')
570
  def proxy_xlb(path: str = Query(default="")):
 
571
  now = time.time()
572
  cache_key = f"xlb:{path}"
573
  with _xlb_lock:
574
  cached = _xlb_cache.get(cache_key)
575
- if cached and now - cached['t'] < 120:
576
  return JSONResponse(cached['d'])
577
  try:
578
  vids = _xlb_scrape(path)
@@ -605,14 +517,8 @@ def _wl(eid:int):return JSONResponse(scrape_lineups(eid))
605
  def _wm(eid:int):return JSONResponse(scrape_match_detail(eid))
606
 
607
  DATA_DIR='/data' if os.path.isdir('/data') else os.path.join(os.path.dirname(os.path.abspath(__file__)),'data')
608
- os.makedirs(DATA_DIR,exist_ok=True)
609
- IF=os.path.join(DATA_DIR,'interactions_v2.json')
610
- CF=os.path.join(DATA_DIR,'comments_v2.json')
611
- WALL_FILE=os.path.join(DATA_DIR,'wall_posts.json')
612
- WALL_VIDEO_DIR=os.path.join(DATA_DIR,'wall_videos')
613
- os.makedirs(WALL_VIDEO_DIR,exist_ok=True)
614
-
615
- _il=threading.Lock();_cl=threading.Lock();_wl_lock=threading.Lock()
616
  def _lj(p):
617
  try:
618
  if os.path.exists(p):return json.load(open(p,'r',encoding='utf-8'))
@@ -645,635 +551,6 @@ async def _pc(request:Request):
645
  with _il:idb=_lj(IF);idb.setdefault(v,{'views':0,'likes':0,'comments':0});idb[v]['comments']=len(cms);_sj(IF,idb)
646
  return JSONResponse({'comments':cms})
647
 
648
- def _load_wall_posts():
649
- with _wl_lock:
650
- return _lj(WALL_FILE)
651
-
652
- def _save_wall_posts(posts):
653
- with _wl_lock:
654
- _sj(WALL_FILE, posts)
655
-
656
- @app.get('/api/wall')
657
- def api_wall():
658
- posts = _load_wall_posts()
659
- if not posts:
660
- return JSONResponse({"posts": []})
661
- return JSONResponse({"posts": posts})
662
-
663
- @app.post('/api/wall')
664
- async def api_wall_post(request: Request):
665
- content_type = request.headers.get('content-type', '')
666
- if 'multipart/form-data' in content_type:
667
- try:
668
- form = await request.form()
669
- except Exception as e:
670
- return JSONResponse({"error": f"Form parse error: {str(e)}"}, status_code=400)
671
- title = form.get('title', 'Video mới') or 'Video mới'
672
- text = form.get('text', '') or ''
673
- source = form.get('source', 'vtv_recorder') or 'vtv_recorder'
674
- video_file = form.get('video')
675
- post_id = str(uuid.uuid4())[:12]
676
- video_url = None
677
- if video_file and hasattr(video_file, 'filename') and video_file.filename:
678
- fname = video_file.filename.lower()
679
- if fname.endswith('.mp4'):
680
- ext = '.mp4'
681
- elif fname.endswith('.webm'):
682
- ext = '.webm'
683
- else:
684
- ext = '.webm'
685
- video_filename = f"wall_{post_id}{ext}"
686
- video_path = os.path.join(WALL_VIDEO_DIR, video_filename)
687
- try:
688
- content = await video_file.read()
689
- if not content:
690
- return JSONResponse({"error": "Empty video file"}, status_code=400)
691
- with open(video_path, 'wb') as f:
692
- f.write(content)
693
- file_size_mb = len(content) / 1024 / 1024
694
- if file_size_mb > 50:
695
- os.remove(video_path)
696
- return JSONResponse({"error": f"Video quá lớn ({file_size_mb:.1f}MB). Tối đa 50MB."}, status_code=400)
697
- video_url = f"/api/wall/video/{video_filename}"
698
- except Exception as e:
699
- return JSONResponse({"error": f"Lỗi lưu video: {str(e)}"}, status_code=500)
700
- post = {
701
- "id": post_id,
702
- "title": title[:200],
703
- "text": text[:2000],
704
- "source": source,
705
- "video": video_url,
706
- "img": None,
707
- "images": [],
708
- "created": int(time.time()),
709
- "created_str": time.strftime('%H:%M %d/%m/%Y', time.localtime()),
710
- }
711
- posts = _load_wall_posts()
712
- if not isinstance(posts, list):
713
- posts = []
714
- posts.insert(0, post)
715
- posts = posts[:200]
716
- _save_wall_posts(posts)
717
- return JSONResponse({"post": post, "ok": True})
718
- try:
719
- body = await request.json()
720
- except:
721
- body = {}
722
- title = body.get('title', 'Bài mới') or 'Bài mới'
723
- text = body.get('text', '') or ''
724
- img = body.get('img', None)
725
- source = body.get('source', 'user') or 'user'
726
- post_id = str(uuid.uuid4())[:12]
727
- post = {
728
- "id": post_id,
729
- "title": title[:200],
730
- "text": text[:2000],
731
- "source": source,
732
- "video": None,
733
- "img": img,
734
- "images": [],
735
- "created": int(time.time()),
736
- "created_str": time.strftime('%H:%M %d/%m/%Y', time.localtime()),
737
- }
738
- posts = _load_wall_posts()
739
- if not isinstance(posts, list):
740
- posts = []
741
- posts.insert(0, post)
742
- posts = posts[:200]
743
- _save_wall_posts(posts)
744
- return JSONResponse({"post": post, "ok": True})
745
-
746
- @app.get('/api/wall/video/{filename}')
747
- def api_wall_video(filename: str):
748
- if '..' in filename or '/' in filename:
749
- return Response(status_code=403)
750
- video_path = os.path.join(WALL_VIDEO_DIR, filename)
751
- if not os.path.exists(video_path):
752
- return Response(status_code=404)
753
- ext = os.path.splitext(filename)[1].lower()
754
- media_type = 'video/mp4' if ext == '.mp4' else 'video/webm'
755
- return FileResponse(video_path, media_type=media_type)
756
-
757
- @app.delete('/api/wall/{post_id}')
758
- def api_wall_delete(post_id: str):
759
- posts = _load_wall_posts()
760
- if not isinstance(posts, list):
761
- return JSONResponse({"error": "No posts"}, status_code=404)
762
- for i, p in enumerate(posts):
763
- if p.get('id') == post_id:
764
- if p.get('video'):
765
- video_name = p['video'].split('/')[-1]
766
- video_path = os.path.join(WALL_VIDEO_DIR, video_name)
767
- if os.path.exists(video_path):
768
- os.remove(video_path)
769
- posts.pop(i)
770
- _save_wall_posts(posts)
771
- return JSONResponse({"ok": True})
772
- return JSONResponse({"error": "Post not found"}, status_code=404)
773
-
774
- # ===== LANGUAGE & EMOTION DETECTION =====
775
- import random as _random2
776
- from urllib.parse import quote as _quote2
777
-
778
- _UA_RW = {'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36', 'Accept-Language': 'vi-VN,vi;q=0.9'}
779
-
780
- # Unique character markers for language detection
781
- _UNIQUE_CHARS = {
782
- 'vietnamese': set('đăâêôơưàảãạáằẳẵặắầẩẫậấèẻẽẹéềễểệếìỉĩịíòỏõọóồổỗộốờởỡợớùủũụúừửữựứỳỷỹỵý'),
783
- 'spanish': set('ñáéíóúü¿¡'),
784
- 'portuguese': set('ãõçáéíóúâêôà'),
785
- }
786
-
787
- _STOPWORDS = {
788
- 'english': {'the', 'is', 'at', 'which', 'on', 'a', 'an', 'and', 'or', 'but', 'in', 'with', 'to', 'for', 'of', 'not', 'no', 'can', 'had', 'have', 'has', 'was', 'were', 'are', 'be', 'been', 'this', 'that', 'it', 'he', 'she', 'they', 'his', 'her', 'my', 'your', 'our', 'we', 'you', 'i'},
789
- 'vietnamese': {'là', 'của', 'và', 'có', 'được', 'cho', 'không', 'với', 'này', 'đó', 'từ', 'trong', 'đã', 'sẽ', 'một', 'các', 'những', 'về', 'tại', 'người', 'năm', 'đến', 'ra', 'lại', 'như', 'khi', 'để', 'rất', 'cũng', 'mà', 'nếu', 'sau', 'trên', 'theo', 'vì', 'do', 'nên', 'thì', 'mình', 'tôi', 'bạn', 'anh', 'chị', 'em'},
790
- 'portuguese': {'de', 'um', 'que', 'e', 'do', 'da', 'em', 'para', 'com', 'não', 'uma', 'os', 'no', 'se', 'na', 'por', 'mais', 'as', 'dos', 'como', 'mas', 'ao', 'ele', 'das', 'tem', 'seu', 'sua', 'ou', 'quando', 'muito', 'nos', 'já', 'eu', 'também', 'só', 'pelo', 'pela', 'até', 'isso', 'ela', 'entre', 'depois', 'sem', 'mesmo', 'aos', 'são', 'está', 'ter', 'ser', 'foi', 'era', 'há', 'estão', 'você', 'nós', 'eles', 'elas'},
791
- 'spanish': {'de', 'que', 'el', 'en', 'y', 'a', 'los', 'del', 'se', 'las', 'por', 'un', 'para', 'con', 'no', 'una', 'su', 'al', 'es', 'lo', 'como', 'más', 'pero', 'sus', 'le', 'ya', 'o', 'fue', 'este', 'ha', 'si', 'porque', 'esta', 'son', 'entre', 'está', 'cuando', 'muy', 'sin', 'sobre', 'ser', 'también', 'me', 'hasta', 'hay', 'donde', 'han', 'quien', 'están', 'desde', 'todo', 'nos', 'durante', 'todos', 'uno', 'les', 'ni', 'contra', 'otros', 'fueron', 'ese', 'eso', 'ante', 'ellos', 'yo', 'tú', 'él', 'ella', 'nosotros', 'usted', 'ustedes'},
792
- }
793
-
794
- def detect_language(text):
795
- """Detect language from text content using stopword + character analysis."""
796
- if not text:
797
- return 'vietnamese'
798
- text_lower = text.lower()
799
- text_chars = set(text_lower)
800
-
801
- # Strong signal: Vietnamese unique characters
802
- vn_chars = len(text_chars & _UNIQUE_CHARS['vietnamese'])
803
- if vn_chars >= 2:
804
- return 'vietnamese'
805
-
806
- # Spanish unique chars (ñ, ¿, ¡)
807
- es_chars = len(text_chars & _UNIQUE_CHARS['spanish'])
808
- pt_chars = len(text_chars & _UNIQUE_CHARS['portuguese'])
809
-
810
- # Stopword scoring
811
- words = set(re.findall(r'\b\w+\b', text_lower))
812
- scores = {}
813
- for lang, stops in _STOPWORDS.items():
814
- scores[lang] = len(words & stops) / max(len(stops), 1)
815
-
816
- # Disambiguate Portuguese vs Spanish
817
- pt_markers = {'não', 'pelo', 'pela', 'isso', 'há', 'estão', 'num', 'numa', 'tenho', 'posso', 'você', 'nós', 'eles', 'elas', 'também', 'muito', 'já', 'só', 'até', 'entre', 'depois', 'sem', 'mesmo', 'aos', 'serão'}
818
- es_markers = {'pero', 'está', 'están', 'porque', 'también', 'hasta', 'donde', 'quien', 'fue', 'son', 'fueron', 'ese', 'eso', 'ante', 'ellos', 'ella', 'nosotros', 'usted', 'ustedes', 'tú', 'él', 'desde', 'todo', 'durante', 'todos', 'uno', 'les', 'ni', 'contra', 'otros', 'fueron'}
819
-
820
- pt_overlap = len(words & pt_markers)
821
- es_overlap = len(words & es_markers)
822
-
823
- if scores.get('portuguese', 0) > 0 and pt_overlap > es_overlap:
824
- return 'portuguese'
825
- if scores.get('spanish', 0) > 0 and es_overlap > pt_overlap:
826
- return 'spanish'
827
- if scores.get('english', 0) > 0.15:
828
- return 'english'
829
-
830
- best = max(scores, key=scores.get)
831
- return best if scores[best] > 0.05 else 'vietnamese'
832
-
833
- # Emotion keyword-based detection
834
- _EMOTION_KEYWORDS = {
835
- 'happy': {
836
- 'en': ['happy', 'joy', 'wonderful', 'great', 'amazing', 'fantastic', 'love', 'excellent', 'beautiful', 'glad', 'delighted', 'pleased', 'cheerful', 'celebrate', 'victory', 'win', 'success'],
837
- 'pt': ['feliz', 'alegria', 'maravilhoso', 'ótimo', 'incrível', 'fantástico', 'amor', 'excelente', 'lindo', 'contente', 'encantado', 'vitória', 'sucesso'],
838
- 'es': ['feliz', 'alegria', 'maravilloso', 'genial', 'increíble', 'fantástico', 'amor', 'excelente', 'hermoso', 'contento', 'encantado', 'victoria', 'éxito'],
839
- 'vi': ['vui', 'hạnh phúc', 'tuyệt vời', 'tuyệt', 'ý nghĩa', 'đẹp', 'thích', 'yêu', 'vui vẻ', 'hân hoan', 'phấn khích', 'chiến thắng', 'thành công'],
840
- },
841
- 'sad': {
842
- 'en': ['sad', 'unhappy', 'terrible', 'awful', 'horrible', 'miserable', 'depressed', 'grief', 'sorrow', 'tragic', 'unfortunate', 'painful', 'death', 'die', 'kill'],
843
- 'pt': ['triste', 'infeliz', 'terrível', 'horrível', 'miserável', 'deprimido', 'dor', 'trágico', 'infelizmente', 'penoso', 'morte', 'morrer'],
844
- 'es': ['triste', 'infeliz', 'terrible', 'horrible', 'miserable', 'deprimido', 'dolor', 'trágico', 'desafortunado', 'penoso', 'muerte', 'morir'],
845
- 'vi': ['buồn', 'không vui', 'tồi tệ', 'kinh khủng', 'đau khổ', 'đau buồn', 'bi thương', 'khốn nạn', 'đau đớn', 'thảm họa', 'chết', 'mất'],
846
- },
847
- 'excited': {
848
- 'en': ['excited', 'thrilling', 'amazing', 'wow', 'incredible', 'unbelievable', 'awesome', 'exhilarating', 'electrifying', 'breathtaking', 'breakthrough', 'record'],
849
- 'pt': ['animado', 'emocionante', 'incrível', 'impressionante', 'sensacional', 'eletrizante', 'empolgante', 'recorde'],
850
- 'es': ['emocionante', 'increíble', 'impresionante', 'sensacional', 'electrizante', 'emocionado', 'entusiasmado', 'récord'],
851
- 'vi': ['hào hứng', 'phấn khích', 'thú vị', 'tuyệt cú mèo', 'đỉnh cao', 'ngoạn mục', 'sục sôi', 'kỷ lục', 'đột phá'],
852
- },
853
- 'humorous': {
854
- 'en': ['funny', 'hilarious', 'joke', 'laugh', 'comedy', 'humor', 'amusing', 'witty', 'sarcastic', 'ironic', 'ridiculous', 'absurd', 'lol', 'haha'],
855
- 'pt': ['engraçado', 'hilário', 'piada', 'rir', 'comédia', 'humor', 'divertido', 'irônico', 'ridículo', 'absurdo', 'kkk'],
856
- 'es': ['gracioso', 'hilarante', 'broma', 'risa', 'comedia', 'humor', 'divertido', 'irónico', 'ridículo', 'absurdo', 'jaja'],
857
- 'vi': ['hài hước', 'buồn cười', 'đùa', 'cười', 'hài', 'vui nhộn', 'hóm hỉnh', 'mỉa mai', 'lố bịch', 'vô lý', 'haha'],
858
- },
859
- 'serious': {
860
- 'en': ['serious', 'critical', 'important', 'urgent', 'severe', 'grave', 'significant', 'crucial', 'vital', 'essential', 'alarming', 'concerning', 'crisis', 'war', 'conflict'],
861
- 'pt': ['sério', 'crítico', 'importante', 'urgente', 'grave', 'significativo', 'crucial', 'vital', 'essencial', 'preocupante', 'crise', 'guerra', 'conflito'],
862
- 'es': ['serio', 'crítico', 'importante', 'urgente', 'grave', 'significativo', 'crucial', 'vital', 'esencial', 'preocupante', 'crisis', 'guerra', 'conflicto'],
863
- 'vi': ['nghiêm trọng', 'quan trọng', 'khẩn cấp', 'nghiêm túc', 'đáng kể', 'thiết yếu', 'cần thiết', 'báo động', 'lo ngại', 'khủng hoảng', 'chiến tranh', 'xung đột'],
864
- },
865
- }
866
-
867
- def detect_emotion(text, language='vietnamese'):
868
- """Detect emotion from text using keyword matching."""
869
- if not text:
870
- return 'neutral'
871
- text_lower = text.lower()
872
-
873
- scores = {}
874
- for emotion, lang_keywords in _EMOTION_KEYWORDS.items():
875
- keywords = lang_keywords.get(language, lang_keywords.get('en', []))
876
- score = sum(1 for kw in keywords if kw in text_lower)
877
- scores[emotion] = score
878
-
879
- if max(scores.values()) == 0:
880
- return 'neutral'
881
-
882
- return max(scores, key=scores.get)
883
-
884
- def detect_language_and_emotion(title, text):
885
- """Detect both language and emotion from article content."""
886
- combined = f"{title} {text}"
887
- lang = detect_language(combined)
888
- emotion = detect_emotion(combined, lang)
889
- return lang, emotion
890
-
891
- # Voice selection based on language and emotion (using MultilingualNeural voices)
892
- VOICE_BY_LANG_EMOTION = {
893
- 'vietnamese': {
894
- 'happy': ('vi-VN-HoaiMyNeural', 'vui'),
895
- 'sad': ('vi-VN-NamMinhNeural', 'buồn'),
896
- 'excited': ('vi-VN-HoaiMyNeural', 'hào hứng'),
897
- 'humorous': ('vi-VN-HoaiMyNeural', 'vui'),
898
- 'serious': ('vi-VN-NamMinhNeural', 'nghiêm túc'),
899
- 'neutral': ('vi-VN-HoaiMyNeural', 'trung_tinh'),
900
- },
901
- 'portuguese': {
902
- 'happy': ('pt-BR-ThalitaMultilingualNeural', 'feliz'),
903
- 'sad': ('pt-BR-ThalitaMultilingualNeural', 'triste'),
904
- 'excited': ('pt-BR-ThalitaMultilingualNeural', 'animado'),
905
- 'humorous': ('pt-BR-ThalitaMultilingualNeural', 'engraçado'),
906
- 'serious': ('pt-BR-ThalitaMultilingualNeural', 'sério'),
907
- 'neutral': ('pt-BR-ThalitaMultilingualNeural', 'neutro'),
908
- },
909
- 'english': {
910
- 'happy': ('en-US-AndrewMultilingualNeural', 'happy'),
911
- 'sad': ('en-AU-WilliamMultilingualNeural', 'sad'),
912
- 'excited': ('en-US-AndrewMultilingualNeural', 'excited'),
913
- 'humorous': ('en-US-AndrewMultilingualNeural', 'funny'),
914
- 'serious': ('en-AU-WilliamMultilingualNeural', 'serious'),
915
- 'neutral': ('en-US-AndrewMultilingualNeural', 'neutral'),
916
- },
917
- 'french': {
918
- 'happy': ('fr-FR-VivienneMultilingualNeural', 'heureux'),
919
- 'sad': ('fr-FR-RemyMultilingualNeural', 'triste'),
920
- 'excited': ('fr-FR-VivienneMultilingualNeural', 'excité'),
921
- 'humorous': ('fr-FR-VivienneMultilingualNeural', 'drôle'),
922
- 'serious': ('fr-FR-RemyMultilingualNeural', 'sérieux'),
923
- 'neutral': ('fr-FR-VivienneMultilingualNeural', 'neutre'),
924
- },
925
- 'german': {
926
- 'happy': ('de-DE-SeraphinaMultilingualNeural', 'glücklich'),
927
- 'sad': ('de-DE-FlorianMultilingualNeural', 'traurig'),
928
- 'excited': ('de-DE-SeraphinaMultilingualNeural', 'aufgeregt'),
929
- 'humorous': ('de-DE-SeraphinaMultilingualNeural', 'lustig'),
930
- 'serious': ('de-DE-FlorianMultilingualNeural', 'ernst'),
931
- 'neutral': ('de-DE-SeraphinaMultilingualNeural', 'neutral'),
932
- },
933
- 'korean': {
934
- 'happy': ('ko-KR-HyunsuMultilingualNeural', '행복'),
935
- 'sad': ('ko-KR-HyunsuMultilingualNeural', '슬픔'),
936
- 'excited': ('ko-KR-HyunsuMultilingualNeural', '흥분'),
937
- 'humorous': ('ko-KR-HyunsuMultilingualNeural', '유쾌'),
938
- 'serious': ('ko-KR-HyunsuMultilingualNeural', '진지'),
939
- 'neutral': ('ko-KR-HyunsuMultilingualNeural', '중립'),
940
- },
941
- 'italian': {
942
- 'happy': ('it-IT-GiuseppeMultilingualNeural', 'felice'),
943
- 'sad': ('it-IT-GiuseppeMultilingualNeural', 'triste'),
944
- 'excited': ('it-IT-GiuseppeMultilingualNeural', 'emozionato'),
945
- 'humorous': ('it-IT-GiuseppeMultilingualNeural', 'divertente'),
946
- 'serious': ('it-IT-GiuseppeMultilingualNeural', 'serio'),
947
- 'neutral': ('it-IT-GiuseppeMultilingualNeural', 'neutro'),
948
- },
949
- }
950
-
951
- # All valid voice IDs (new MultilingualNeural format)
952
- VALID_VOICES = {
953
- 'vi-VN-HoaiMyNeural', 'vi-VN-NamMinhNeural',
954
- 'en-US-AndrewMultilingualNeural', 'en-AU-WilliamMultilingualNeural',
955
- 'pt-BR-ThalitaMultilingualNeural',
956
- 'fr-FR-VivienneMultilingualNeural', 'fr-FR-RemyMultilingualNeural',
957
- 'de-DE-SeraphinaMultilingualNeural', 'de-DE-FlorianMultilingualNeural',
958
- 'ko-KR-HyunsuMultilingualNeural',
959
- 'it-IT-GiuseppeMultilingualNeural',
960
- }
961
-
962
- def get_voice_for_content(title, text, preferred_voice=None):
963
- """Get appropriate voice based on content language and emotion."""
964
- # Accept the new MultilingualNeural voices directly
965
- if preferred_voice and preferred_voice in VALID_VOICES:
966
- return preferred_voice
967
-
968
- # Also accept old shorthand voice IDs and map them to new format
969
- old_voice_map = {
970
- 'hoaimy': 'vi-VN-HoaiMyNeural',
971
- 'namminh': 'vi-VN-NamMinhNeural',
972
- 'andrew': 'en-US-AndrewMultilingualNeural',
973
- 'jenny': 'en-US-AndrewMultilingualNeural',
974
- 'thalita': 'pt-BR-ThalitaMultilingualNeural',
975
- 'pt_thalita': 'pt-BR-ThalitaMultilingualNeural',
976
- 'pt_francisco': 'pt-BR-ThalitaMultilingualNeural',
977
- 'ela': 'en-US-AndrewMultilingualNeural',
978
- 'es_carlos': 'en-US-AndrewMultilingualNeural',
979
- 'denise': 'fr-FR-VivienneMultilingualNeural',
980
- 'katja': 'de-DE-SeraphinaMultilingualNeural',
981
- 'nanami': 'en-US-AndrewMultilingualNeural',
982
- 'sunhee': 'ko-KR-HyunsuMultilingualNeural',
983
- 'xiaochen': 'en-US-AndrewMultilingualNeural',
984
- }
985
- if preferred_voice and preferred_voice in old_voice_map:
986
- return old_voice_map[preferred_voice]
987
-
988
- lang, emotion = detect_language_and_emotion(title, text)
989
- lang_map = VOICE_BY_LANG_EMOTION.get(lang, VOICE_BY_LANG_EMOTION['vietnamese'])
990
- voice, _ = lang_map.get(emotion, lang_map['neutral'])
991
- return voice
992
-
993
-
994
- def _is_relevant_image(img_url, title, text):
995
- """Check if an image is relevant to the article content."""
996
- if not img_url:
997
- return False
998
- skip_patterns = ['pixel', 'analytics', 'tracking', '1x1.gif', 'spacer.gif',
999
- 'logo', 'icon', 'avatar', 'emoji', 'smiley', 'sprite',
1000
- 'advertisement', 'ad-banner', 'sponsored', 'banner-ads']
1001
- img_lower = img_url.lower()
1002
- for p in skip_patterns:
1003
- if p in img_lower:
1004
- return False
1005
- if not any(img_lower.endswith(ext) for ext in ['.jpg', '.jpeg', '.png', '.webp', '.gif']):
1006
- return False
1007
- return True
1008
-
1009
-
1010
- def _filter_relevant_images(images, title, text, max_images=8):
1011
- """Filter and rank images by relevance to article content."""
1012
- if not images:
1013
- return []
1014
- seen = set()
1015
- relevant = []
1016
- for img in images:
1017
- if img in seen:
1018
- continue
1019
- seen.add(img)
1020
- if _is_relevant_image(img, title, text):
1021
- relevant.append(img)
1022
- return relevant[:max_images]
1023
-
1024
-
1025
- def _scrape_article_for_rewrite(url):
1026
- """Scrape article: extract title, paragraphs, RELEVANT images, OG image."""
1027
- try:
1028
- r = req.get(url, headers=_UA_RW, timeout=15, allow_redirects=True)
1029
- r.encoding = 'utf-8'
1030
- soup = BeautifulSoup(r.text, 'lxml')
1031
- for tag in soup.find_all(['script', 'style', 'nav', 'footer', 'aside', 'form']):
1032
- tag.decompose()
1033
- h1 = soup.find('h1')
1034
- ogt = soup.find('meta', property='og:title')
1035
- title = (h1.get_text(strip=True) if h1 else '') or (ogt.get('content', '') if ogt else '')
1036
- ogi = soup.find('meta', property='og:image')
1037
- og_img = ogi.get('content', '') if ogi else ''
1038
- if og_img and og_img.startswith('//'):
1039
- og_img = 'https:' + og_img
1040
- block = None
1041
- for sel in ['article', '.singular-content', '.detail-content', '.fck_detail', '.content-detail', '.knc-content', 'main', '.cms-body', '.article__body']:
1042
- el = soup.select_one(sel)
1043
- if el and len(el.find_all('p')) >= 2:
1044
- block = el
1045
- break
1046
- if not block:
1047
- block = soup.body or soup
1048
- paragraphs = []
1049
- all_images = []
1050
- seen_imgs = set()
1051
- if og_img and og_img not in seen_imgs:
1052
- all_images.append(og_img)
1053
- seen_imgs.add(og_img)
1054
- for el in block.find_all(['p', 'h2', 'h3', 'figure', 'img'], recursive=True):
1055
- if el.name == 'p':
1056
- t = _clean(el.get_text(strip=True))
1057
- if t and len(t) > 40:
1058
- paragraphs.append(t)
1059
- elif el.name in ('figure', 'img'):
1060
- im = el if el.name == 'img' else el.find('img')
1061
- if im:
1062
- src = im.get('data-src') or im.get('src') or im.get('data-original') or ''
1063
- if src and 'base64' not in src:
1064
- if src.startswith('//'):
1065
- src = 'https:' + src
1066
- if src not in seen_imgs:
1067
- all_images.append(src)
1068
- seen_imgs.add(src)
1069
- # Filter to relevant images only
1070
- relevant_images = _filter_relevant_images(all_images, title, ' '.join(paragraphs[:5]))
1071
- return {'title': _clean(title), 'paragraphs': paragraphs, 'images': relevant_images, 'og_img': og_img}
1072
- except Exception:
1073
- return None
1074
-
1075
-
1076
- def _extract_key_points_rw(paragraphs, max_points=5):
1077
- """Extract key points from paragraphs - extracts ALL sentences, not just first one.
1078
-
1079
- Fixes: Original regex `^(.+?[.!?])\s` only captured first sentence per paragraph.
1080
- Now splits on all sentence boundaries and takes valid sentences until max_points.
1081
- """
1082
- points = []
1083
-
1084
- for p in paragraphs:
1085
- if len(points) >= max_points:
1086
- break
1087
-
1088
- p = _clean(p)
1089
- if not p:
1090
- continue
1091
-
1092
- # Split paragraph into sentences using Vietnamese + English punctuation
1093
- sentences = re.split(r'(?<=[.!?])\s+(?=[A-ZÀ-Ỹ0-9])', p)
1094
- sentences = [s.strip() for s in sentences if s.strip()]
1095
-
1096
- for sentence in sentences:
1097
- if len(points) >= max_points:
1098
- break
1099
-
1100
- # Clean sentence - remove extra whitespace
1101
- sentence = _clean(sentence)
1102
-
1103
- if len(sentence) < 30:
1104
- continue
1105
-
1106
- # Check for duplicates
1107
- if any(sentence[:60] in existing for existing in points):
1108
- continue
1109
-
1110
- # Ensure sentence ends with punctuation
1111
- if not sentence.endswith(('.', '!', '?')):
1112
- sentence = sentence + '.'
1113
-
1114
- points.append(sentence)
1115
-
1116
- # If no valid sentences found, take chunks from raw text
1117
- if not points:
1118
- raw = '\n'.join(paragraphs)
1119
- for i in range(0, min(len(raw), max_points * 300), 280):
1120
- chunk = _clean(raw[i:i+280])
1121
- if len(chunk) >= 30 and chunk not in points:
1122
- points.append(chunk + ('.' if not chunk.endswith('.') else ''))
1123
- if len(points) >= max_points:
1124
- break
1125
-
1126
- return points
1127
-
1128
-
1129
- @app.post("/api/rewrite_slide")
1130
- async def api_rewrite_slide(request: Request):
1131
- """Fast rewrite as SLIDES - no AI needed, instant response."""
1132
- body = await request.json()
1133
- url = _clean(body.get("url", ""))
1134
- context = body.get("context", "")
1135
- preferred_voice = body.get("voice", "") # Accept custom voice selection
1136
- if not url and not context:
1137
- return JSONResponse({"error": "Cần URL hoặc nội dung"}, status_code=400)
1138
- data = None
1139
- if url and url.startswith("http"):
1140
- data = _scrape_article_for_rewrite(url)
1141
- if not data and context:
1142
- paragraphs = [_clean(p) for p in context.split('\n') if len(_clean(p)) > 40]
1143
- data = {'title': paragraphs[0][:80] if paragraphs else 'Bài viết', 'paragraphs': paragraphs, 'images': [], 'og_img': ''}
1144
- if not data or not data.get('paragraphs'):
1145
- return JSONResponse({"error": "Không đọc được bài viết"}, status_code=422)
1146
- points = _extract_key_points_rw(data['paragraphs'], max_points=12)
1147
- if not points:
1148
- return JSONResponse({"error": "Không tìm được ý chính"}, status_code=422)
1149
- images = data.get('images', [])
1150
- slides = []
1151
- for i, point in enumerate(points):
1152
- img = images[i] if i < len(images) else (images[-1] if images else '')
1153
- if img and 'cdnphoto.dantri' in img:
1154
- img = '/api/proxy/img?url=' + _quote2(img, safe='')
1155
- slides.append({'text': point, 'image': img, 'index': i + 1})
1156
- summary_text = '\n\n'.join([f"• {s['text']}" for s in slides])
1157
-
1158
- # Auto-detect language and emotion
1159
- lang, emotion = detect_language_and_emotion(data['title'], summary_text)
1160
- # Use preferred voice if provided, otherwise auto-detect
1161
- voice = preferred_voice if preferred_voice else get_voice_for_content(data['title'], summary_text)
1162
-
1163
- post = {
1164
- "id": str(int(time.time() * 1000)) + str(_random2.randint(100, 999)),
1165
- "title": data['title'],
1166
- "text": summary_text,
1167
- "img": images[0] if images else '',
1168
- "url": url,
1169
- "kind": "slide_summary",
1170
- "slides": slides,
1171
- "images": images[:10],
1172
- "video": "",
1173
- "voice": voice,
1174
- "emotion": emotion,
1175
- "language": lang,
1176
- "ts": int(time.time())
1177
- }
1178
- posts = _load_wall_posts()
1179
- posts.insert(0, post)
1180
- _save_wall_posts(posts)
1181
- return JSONResponse({"post": post, "slides": slides})
1182
-
1183
-
1184
- @app.post("/api/rewrite_share")
1185
- async def api_rewrite_share(request: Request):
1186
- """Rewrite article and post to Tường AI with SLIDES + AI text."""
1187
- body = await request.json()
1188
- url = _clean(body.get("url", ""))
1189
- ctx = _clean(body.get("context", ""))
1190
- preferred_voice = body.get("voice", "") # Accept custom voice selection
1191
- if not url and not ctx:
1192
- return JSONResponse({"error": "Cần URL hoặc nội dung"}, status_code=400)
1193
- data = None
1194
- if url and url.startswith("http"):
1195
- data = _scrape_article_for_rewrite(url)
1196
- if not data and ctx:
1197
- paragraphs = [_clean(p) for p in ctx.split('\n') if len(_clean(p)) > 40]
1198
- data = {'title': paragraphs[0][:80] if paragraphs else 'Bài viết', 'paragraphs': paragraphs, 'images': [], 'og_img': ''}
1199
- if not data or not data.get('paragraphs'):
1200
- return JSONResponse({"error": "Không đọc được bài viết"}, status_code=422)
1201
- raw_text = '\n'.join(data['paragraphs'])
1202
- if len(raw_text) < 50:
1203
- raw_text = ctx[:14000]
1204
- if len(raw_text) < 50:
1205
- return JSONResponse({"error": "Bài viết quá ngắn"}, status_code=422)
1206
- domain = ''
1207
- try:
1208
- from urllib.parse import urlparse
1209
- domain = urlparse(url).netloc.replace('www.', '')
1210
- except:
1211
- pass
1212
-
1213
- # Generate AI summary text
1214
- ai_text = None
1215
- try:
1216
- import ai_ext
1217
- if hasattr(ai_ext, 'qwen_generate'):
1218
- prompt = f'Tóm tắt đăng Tường AI:\nTiêu đề: {data["title"]}\n{raw_text[:14000]}\n\n4-6 ý chính. Cuối ghi nguồn.'
1219
- ai_text = await ai_ext.qwen_generate(prompt, max_tokens=1000)
1220
- except Exception:
1221
- pass
1222
- if not ai_text or len(ai_text) < 80:
1223
- key_pts = _extract_key_points_rw(data['paragraphs'], max_points=12)
1224
- if key_pts:
1225
- ai_text = '\n\n'.join([f"• {p}" for p in key_pts])
1226
- else:
1227
- ai_text = f"Tóm tắt: {data['title']}\n\n{raw_text[:1200]}\n\nNguồn: {domain}"
1228
-
1229
- # Build slides from key points (FIX: include slides in rewrite_share too!)
1230
- points = _extract_key_points_rw(data['paragraphs'], max_points=12)
1231
- images = data.get('images', [])
1232
- slides = []
1233
- for i, point in enumerate(points):
1234
- img = images[i] if i < len(images) else (images[-1] if images else '')
1235
- if img and 'cdnphoto.dantri' in img:
1236
- img = '/api/proxy/img?url=' + _quote2(img, safe='')
1237
- slides.append({'text': point, 'image': img, 'index': i + 1})
1238
-
1239
- # Auto-detect language and emotion
1240
- lang, emotion = detect_language_and_emotion(data['title'], ai_text)
1241
- # Use preferred voice if provided, otherwise auto-detect
1242
- voice = preferred_voice if preferred_voice else get_voice_for_content(data['title'], ai_text)
1243
-
1244
- post = {
1245
- "id": str(int(time.time() * 1000)) + str(_random2.randint(100, 999)),
1246
- "title": data['title'],
1247
- "text": ai_text,
1248
- "img": images[0] if images else '',
1249
- "url": url,
1250
- "kind": "rewrite",
1251
- "slides": slides,
1252
- "images": images[:10],
1253
- "video": "",
1254
- "voice": voice,
1255
- "emotion": emotion,
1256
- "language": lang,
1257
- "ts": int(time.time())
1258
- }
1259
- posts = _load_wall_posts()
1260
- posts.insert(0, post)
1261
- _save_wall_posts(posts)
1262
- return JSONResponse({"post": post, "slides": slides})
1263
-
1264
-
1265
- @app.post("/api/url_wall")
1266
- async def api_url_wall(request: Request):
1267
- """Submit URL to add to Tường AI."""
1268
- body = await request.json()
1269
- url = _clean(body.get("url", ""))
1270
- if not url or not url.startswith('http'):
1271
- return JSONResponse({"error": "URL không hợp lệ"}, status_code=400)
1272
- # Reuse rewrite_share logic
1273
- req._body = json.dumps({"url": url}).encode()
1274
- return await api_rewrite_share(request)
1275
-
1276
-
1277
  def _bg():
1278
  time.sleep(15)
1279
  while True:
 
1
+ """VNEWS v2 Entry Point - with fast bongda proxy"""
2
  import sys, os
3
  from main import app, HEADERS, BONGDA_HEADERS, fetch_bongda_api, HL_LEAGUES
4
 
 
7
  except Exception as e:
8
  print(f"[WARN] ai_ext import failed: {e}")
9
 
 
 
 
 
 
10
  from fastapi.responses import HTMLResponse, JSONResponse, FileResponse, Response
11
  from fastapi.staticfiles import StaticFiles
12
  from starlette.routing import Mount
13
+ from fastapi import Query, Request
14
  import requests as req
15
  from bs4 import BeautifulSoup
16
+ import re, html as html_lib, json, threading, time
17
  from concurrent.futures import ThreadPoolExecutor, as_completed
18
  from urllib.parse import quote
19
 
 
31
 
32
  # === FAST BONGDA PROXY ENDPOINT ===
33
  def _get_match_detail(event_id, slug=None):
34
+ """Internal function to scrape match detail from bongda.com.vn"""
35
  headers = {"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36", "Accept": "text/html", "Referer": "https://bongda.com.vn/"}
36
+
37
  if slug:
38
  url = f"https://bongda.com.vn/tran-dau/{event_id}/centre/{slug}"
39
  else:
40
  url = f"https://bongda.com.vn/tran-dau/{event_id}"
41
+
42
  resp = req.get(url, headers=headers, timeout=15, allow_redirects=True)
43
  if resp.status_code != 200:
44
  return None
45
+
46
  soup = BeautifulSoup(resp.text, 'html.parser')
47
  result = {"event_id": event_id, "found": False, "sections": []}
48
  info = {}
49
+
50
  tel = soup.select_one('.teams')
51
  if tel:
52
  he = tel.select_one('.team.home')
 
68
  if len(parts) >= 2: info['score'] = f"{parts[0]} - {parts[1]}"
69
  lb = sc.select_one('.label')
70
  if lb: info['status_label'] = _clean(lb.get_text())
71
+
72
  if info.get('home_team') and info.get('away_team'):
73
  result['info'] = info
74
  result['found'] = True
75
  result['sections'].append('info')
76
+
77
  events = []
78
  for ev in soup.select('.events .period .event'):
79
  ev_cls = ' '.join(ev.get('class', []))
80
  ev_data = {'team': 'home' if 'home' in ev_cls else 'away', 'period': '', 'type': 'unknown', 'time': '', 'players': ''}
81
+
82
  parent = ev.parent
83
  if parent:
84
  h2 = parent.find('h2')
85
  if h2: ev_data['period'] = _clean(h2.get_text())
86
+
87
  if ev.select_one('[class*="goal"]'): ev_data['type'] = 'goal'
88
  elif ev.select_one('[class*="redcard"]'): ev_data['type'] = 'redcard'
89
  elif ev.select_one('[class*="yellowcard"]'): ev_data['type'] = 'yellowcard'
90
  elif ev.select_one('[class*="substitution"]'): ev_data['type'] = 'substitution'
91
+
92
  players_el = ev.select_one('.players')
93
  if players_el:
94
  pl_text = _clean(players_el.get_text(' ', strip=True))
 
99
  else:
100
  ev_data['players'] = pl_text
101
  events.append(ev_data)
102
+
103
  if events:
104
  result['events'] = events
105
  result['sections'].append('events')
106
+
107
  pred = soup.select_one('.prediction-card')
108
  if pred:
109
  team_info = pred.select_one('.team-info')
 
118
  vc = pred.select_one('.vote-count')
119
  if vc: pred_data['vote_count'] = _clean(vc.get_text())
120
  result['prediction'] = pred_data
121
+
122
  recent = []
123
  ml = soup.select_one('.matches-list')
124
  if ml:
 
133
  if recent:
134
  result['recent_matches'] = recent
135
  result['sections'].append('recent')
136
+
137
  try:
138
  api_h = {"User-Agent": "Mozilla/5.0", "Accept": "application/json", "X-Requested-With": "XMLHttpRequest", "Referer": "https://bongda.com.vn/"}
139
  ar = req.get(f"https://bongda.com.vn/api/fixtures/h2h-stats?event_id={event_id}", headers=api_h, timeout=10)
 
151
  result['h2h_stats_parsed'] = ast
152
  result['sections'].append('h2h_stats')
153
  except: pass
154
+
155
  return result
156
 
157
  @app.get('/api/proxy/bongda')
158
  def proxy_bongda(event_id: int = Query(default=None), slug: str = Query(default=None)):
159
  if event_id is None:
160
  return JSONResponse({'error': 'event_id required'}, status_code=400)
161
+
162
  cache_key = f"{event_id}_{slug}"
163
  now = time.time()
164
  cached = _match_cache.get(cache_key)
165
  if cached and now - cached.get('_ts', 0) < 300:
166
  return JSONResponse(cached)
167
+
168
  try:
169
  result = _get_match_detail(event_id, slug)
170
  if result:
 
175
  err = {"event_id": event_id, "found": False, "error": str(e), "_ts": now}
176
  _match_cache[cache_key] = err
177
  return JSONResponse(err)
178
+
179
  return JSONResponse({"event_id": event_id, "found": False})
180
 
181
  @app.get('/api/match/{event_id}/detail')
182
  def api_match_detail(event_id: int, url: str = Query(default=None)):
183
+ # Try to extract slug from url if provided
184
  slug = None
185
  if url:
186
  m = re.match(r'.+/tran-dau/\d+/(?:centre|preview)/(.+)', url)
187
  if m:
188
  slug = m.group(1)
189
+
190
  cache_key = f"{event_id}_{slug or ''}"
191
  now = time.time()
192
  cached = _match_cache.get(cache_key)
193
  if cached and now - cached.get('_ts', 0) < 300:
194
  return JSONResponse(cached)
195
+
196
  try:
197
+ # If no slug, try to find it from homepage
198
  if not slug:
199
  try:
200
  home_r = req.get("https://bongda.com.vn/", headers={"User-Agent": "Mozilla/5.0"}, timeout=10)
 
208
  cache_key = f"{event_id}_{slug}"
209
  break
210
  except: pass
211
+
212
  result = _get_match_detail(event_id, slug)
213
  if result:
214
  result['_ts'] = now
 
218
  err = {"event_id": event_id, "found": False, "error": str(e), "_ts": now}
219
  _match_cache[cache_key] = err
220
  return JSONResponse(err)
221
+
222
  return JSONResponse({"event_id": event_id, "found": False})
223
 
224
+ # === Rest of endpoints (existing) ===
225
  _STOP=set('và của các những một được trong với cho tại sau trước khi không người việt nam hôm nay mới nhất nóng tin tức cập nhật theo từ đến là có thì này đã để'.split())
226
 
227
  def _has_kw(topic,title):
 
261
  try:
262
  r=req.get(f"https://vietnamnet.vn/tim-kiem?q={quote(topic)}",headers={'User-Agent':'Mozilla/5.0'},timeout=10);soup=BeautifulSoup(r.text,'lxml')
263
  for a in soup.select('h3 a[href], .vnn-title a')[:limit*2]:
264
+ t=_clean(a.get_text(strip=True));href=a.get('href','')
265
  if t and len(t)>15 and _has_kw(topic,t):
266
  if not href.startswith('http'):href='https://vietnamnet.vn'+href
267
  items.append({'title':t,'url':href,'via':'VietNamNet'})
 
347
  if i<len(s) and s[i].get('url') and s[i]['url'] not in seen:seen.add(s[i]['url']);out.append(s[i])
348
  return out[:limit]
349
 
350
+ app.router.routes=[r for r in app.router.routes if not(getattr(r,'path',None)=='/api/article' and 'GET' in getattr(r,'methods',set()))]
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
351
 
352
+ def _scrape_generic(url):
353
+ try:
354
+ r=req.get(url,headers={'User-Agent':'Mozilla/5.0','Accept-Language':'vi-VN,vi;q=0.9'},timeout=15,allow_redirects=True);r.encoding='utf-8';soup=BeautifulSoup(r.text,'lxml')
355
+ for tag in soup.find_all(['script','style','nav','footer','aside','form','noscript']):tag.decompose()
356
+ h1=soup.find('h1');ogt=soup.find('meta',property='og:title');title=(h1.get_text(strip=True) if h1 else '')or(ogt.get('content','') if ogt else '')
357
+ ogd=soup.find('meta',property='og:description');summary=ogd.get('content','') if ogd else ''
358
+ ogi=soup.find('meta',property='og:image');og_img=ogi.get('content','') if ogi else ''
359
+ if og_img and og_img.startswith('//'):og_img='https:'+og_img
360
+ block=None
361
+ for sel in['article','.singular-content','.detail-content','.fck_detail','.content-detail','.knc-content','main','.cms-body','.article__body']:
362
+ el=soup.select_one(sel)
363
+ if el and len(el.find_all('p'))>=2:block=el;break
364
+ if not block:block=soup.body or soup
365
+ body=[]
366
+ for el in block.find_all(['p','h2','h3','figure','img'],recursive=True):
367
+ if el.name=='p':t=el.get_text(strip=True);(body.append({'type':'p','text':t}) if t and len(t)>30 else None)
368
+ elif el.name in('h2','h3'):t=el.get_text(strip=True);(body.append({'type':'heading','text':t}) if t else None)
369
+ elif el.name in('figure','img'):
370
+ im=el if el.name=='img' else el.find('img')
371
+ if im:src=im.get('data-src') or im.get('src') or'';(body.append({'type':'img','src':'https:'+src if src.startswith('//') else src}) if src and'base64' not in src else None)
372
+ if not body and summary:body=[{'type':'p','text':summary}]
373
+ return{'title':_clean(title),'summary':_clean(summary),'og_image':og_img,'body':body[:50],'source':'generic','url':url}
374
+ except:return None
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
375
 
376
  @app.get('/api/article')
377
+ def api_article_v2(url:str=Query(...)):
378
+ from main import scrape_vne_article,scrape_bbc_article,scrape_dantri_article,scrape_genk_article,scrape_ttvh_article
379
+ if 'vnexpress.net' in url:data=scrape_vne_article(url)
380
+ elif 'bbc.com' in url:data=scrape_bbc_article(url)
381
+ elif 'dantri.com.vn' in url:data=scrape_dantri_article(url)
382
+ elif 'genk.vn' in url:data=scrape_genk_article(url)
383
+ elif 'thethaovanhoa.vn' in url:data=scrape_ttvh_article(url)
384
+ else:data=_scrape_generic(url)
385
+ if data and data.get('body'):return JSONResponse(data)
386
+ return JSONResponse(data if data else{'error':'Không đọc được','url':url})
 
 
 
 
 
 
 
 
 
 
 
 
387
 
388
  _hot_cache={'t':0,'d':[]}
389
  def _get_hot_topics():
 
416
  _hot_cache.update({'t':now,'d':topics[:24]});return topics[:24]
417
 
418
  @app.get('/api/hot_topics')
419
+ def api_hot_topics():return JSONResponse({'topics':_get_hot_topics()})
 
 
 
420
  @app.get('/')
421
  async def serve_index():
422
  p=os.path.join(STATIC_DIR,'index_v2.html')
 
433
  @app.get('/s')
434
  async def _sh(url:str='',title:str='',img:str=''):return HTMLResponse(f'<!DOCTYPE html><html><head><meta property="og:title" content="{_clean(title)}"><meta property="og:image" content="{_clean(img)}"><meta http-equiv="refresh" content="0;url={_clean(url) or "/"}"></head><body></body></html>')
435
 
436
+ from wc2026_scraper import(scrape_summary,scrape_fixtures,scrape_standings,scrape_stats,scrape_wc_news,scrape_road_to_wc,get_wc2026_all,scrape_history,scrape_h2h,scrape_lineups,scrape_match_detail)
437
 
438
+ # === XEMLAIBONGDA.PROXY (CORS workaround for WC highlights) ===
439
  _xlb_cache = {}
440
  _xlb_lock = threading.Lock()
441
 
442
  def _xlb_scrape(path):
443
+ """Scrape xemlaibongda.top and extract video articles."""
444
  url = f"https://xemlaibongda.top/{path}"
445
  r = req.get(url, headers={"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36"}, timeout=15, allow_redirects=True)
446
  if r.status_code != 200:
 
455
  seen.add(href)
456
  if not href.startswith('http'):
457
  href = 'https://xemlaibongda.top' + href
458
+ img = ''
459
+ img_el = a.select_one('img')
460
+ if img_el:
461
+ img = img_el.get('src', '') or img_el.get('data-src', '')
462
+ if img.startswith('//'):
463
+ img = 'https:' + img
464
+ elif img.startswith('/'):
465
+ img = 'https://xemlaibongda.top' + img
 
 
 
 
 
 
 
466
  title = ''
467
+ for sel in ['.title', 'h3', 'h2', '.name', '.post-title']:
468
  t = a.select_one(sel)
469
  if t:
470
  title = _clean(t.get_text())
471
  break
472
  if not title:
473
+ title = _clean(a.get('title', '') or a.get_text())
474
+ if title and len(title) > 3:
475
+ vids.append({"link": href, "img": img, "title": title})
 
 
 
 
 
 
 
 
 
 
 
476
  if len(vids) >= 30:
477
  break
478
  return vids
479
 
480
  @app.get('/api/proxy/xlb')
481
  def proxy_xlb(path: str = Query(default="")):
482
+ """Proxy for xemlaibongda.top to avoid CORS in browser."""
483
  now = time.time()
484
  cache_key = f"xlb:{path}"
485
  with _xlb_lock:
486
  cached = _xlb_cache.get(cache_key)
487
+ if cached and now - cached['t'] < 120: # 2 min cache
488
  return JSONResponse(cached['d'])
489
  try:
490
  vids = _xlb_scrape(path)
 
517
  def _wm(eid:int):return JSONResponse(scrape_match_detail(eid))
518
 
519
  DATA_DIR='/data' if os.path.isdir('/data') else os.path.join(os.path.dirname(os.path.abspath(__file__)),'data')
520
+ os.makedirs(DATA_DIR,exist_ok=True);IF=os.path.join(DATA_DIR,'interactions_v2.json');CF=os.path.join(DATA_DIR,'comments_v2.json')
521
+ _il=threading.Lock();_cl=threading.Lock()
 
 
 
 
 
 
522
  def _lj(p):
523
  try:
524
  if os.path.exists(p):return json.load(open(p,'r',encoding='utf-8'))
 
551
  with _il:idb=_lj(IF);idb.setdefault(v,{'views':0,'likes':0,'comments':0});idb[v]['comments']=len(cms);_sj(IF,idb)
552
  return JSONResponse({'comments':cms})
553
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
554
  def _bg():
555
  time.sleep(15)
556
  while True:
bongda_proxy.py CHANGED
@@ -4,6 +4,7 @@ from bs4 import BeautifulSoup
4
  import re
5
  import json
6
 
 
7
  def _cl(s):
8
  return re.sub(r'\s+', ' ', str(s or '')).strip()
9
 
@@ -14,18 +15,23 @@ def _normalize_time(raw):
14
  return t
15
 
16
  def scrape_match_html(event_id, url=None):
 
17
  result = {"event_id": event_id, "found": False, "sections": []}
 
 
18
  headers = {
19
  "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36",
20
  "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8",
21
  "Referer": "https://bongda.com.vn/",
22
  }
 
23
  html = None
24
  urls_to_try = [url] if url else []
25
  urls_to_try += [
26
  f"https://bongda.com.vn/tran-dau/{event_id}/centre/",
27
  f"https://bongda.com.vn/tran-dau/{event_id}/preview/",
28
  ]
 
29
  for u in urls_to_try:
30
  if not u:
31
  continue
@@ -36,78 +42,203 @@ def scrape_match_html(event_id, url=None):
36
  break
37
  except Exception:
38
  continue
 
39
  if not html:
40
  return result
 
41
  try:
42
  soup = BeautifulSoup(html, 'html.parser')
43
  info = {}
 
 
44
  tel = soup.select_one('.teams')
45
  if tel:
46
  he = tel.select_one('.team.home')
47
  if he:
48
  ne = he.select_one('p:not(.logo)') or he.find('p')
49
- if ne: info['home_team'] = _cl(ne.get_text())
 
50
  lo = he.select_one('img')
51
- if lo: info['home_logo'] = lo.get('src', '')
 
 
52
  ae = tel.select_one('.team.away')
53
  if ae:
54
  ne = ae.select_one('p:not(.logo)') or ae.find('p')
55
- if ne: info['away_team'] = _cl(ne.get_text())
 
56
  lo = ae.select_one('img')
57
- if lo: info['away_logo'] = lo.get('src', '')
 
 
58
  sc = tel.select_one('.score')
59
  if sc:
60
  parts = [_cl(p.get_text()) for p in sc.select('p')]
61
- if len(parts) >= 2: info['score'] = f"{parts[0]} - {parts[1]}"
 
62
  lb = sc.select_one('.label')
63
- if lb: info['status_label'] = _cl(lb.get_text())
 
 
64
  if info.get('home_team') and info.get('away_team'):
65
  result['info'] = info
66
  result['found'] = True
67
  result['sections'].append('info')
68
  else:
69
  return result
 
 
70
  events = []
71
  events_div = soup.select_one('.events')
72
  if events_div:
73
  period = ''
74
  for child in events_div.children:
75
- if not hasattr(child, 'name') or not child.name: continue
 
76
  cls = ' '.join(child.get('class', []))
77
  if 'period' in cls:
78
  h2 = child.find('h2')
79
- if h2: period = _cl(h2.get_text())
 
80
  for ev in child.children:
81
- if not hasattr(ev, 'name') or not ev.name: continue
 
82
  ev_cls = ' '.join(ev.get('class', []))
83
- if 'event' not in ev_cls: continue
84
- ev_data = {'team': 'home' if 'home' in ev_cls else 'away', 'period': period, 'type': 'unknown', 'time': ''}
 
 
 
 
 
 
 
 
 
85
  type_el = ev.select_one('.event-type')
86
  if type_el:
87
- if type_el.select_one('[class*="redcard"]'): ev_data['type'] = 'redcard'
88
- elif type_el.select_one('[class*="yellowcard"]'): ev_data['type'] = 'yellowcard'
89
- elif type_el.select_one('[class*="goal"]'): ev_data['type'] = 'goal'
90
- elif type_el.select_one('[class*="substitution"]'): ev_data['type'] = 'substitution'
 
 
 
 
 
91
  players_el = ev.select_one('.players')
92
  if players_el:
93
  time_el = players_el.select_one('.event-time')
94
- if time_el: ev_data['time'] = _normalize_time(time_el.get_text())
 
 
 
95
  text = _cl(players_el.get_text(' ', strip=True).replace(ev_data['time'], '').strip())
96
  ev_data['players'] = text
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
97
  events.append(ev_data)
 
98
  if events:
99
  result['events'] = events
100
  result['sections'].append('events')
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
101
  except Exception as e:
102
  result['error'] = str(e)
 
103
  return result
104
 
 
 
105
  from fastapi import Query
106
  from fastapi.responses import JSONResponse
107
 
108
  def add_bongda_proxy_endpoint(app):
109
  @app.get('/api/proxy/bongda')
110
  def proxy_bongda(event_id: int = Query(default=None), url: str = Query(default=None)):
 
111
  if event_id is None:
112
  return JSONResponse({'error': 'event_id required'}, status_code=400)
113
- return JSONResponse(scrape_match_html(event_id, url))
 
4
  import re
5
  import json
6
 
7
+ # Import-safe parsing functions
8
  def _cl(s):
9
  return re.sub(r'\s+', ' ', str(s or '')).strip()
10
 
 
15
  return t
16
 
17
  def scrape_match_html(event_id, url=None):
18
+ """Fast scrape bongda.com.vn for match detail - no external CORS needed."""
19
  result = {"event_id": event_id, "found": False, "sections": []}
20
+
21
+ # Fetch HTML
22
  headers = {
23
  "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36",
24
  "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8",
25
  "Referer": "https://bongda.com.vn/",
26
  }
27
+
28
  html = None
29
  urls_to_try = [url] if url else []
30
  urls_to_try += [
31
  f"https://bongda.com.vn/tran-dau/{event_id}/centre/",
32
  f"https://bongda.com.vn/tran-dau/{event_id}/preview/",
33
  ]
34
+
35
  for u in urls_to_try:
36
  if not u:
37
  continue
 
42
  break
43
  except Exception:
44
  continue
45
+
46
  if not html:
47
  return result
48
+
49
  try:
50
  soup = BeautifulSoup(html, 'html.parser')
51
  info = {}
52
+
53
+ # Teams + score
54
  tel = soup.select_one('.teams')
55
  if tel:
56
  he = tel.select_one('.team.home')
57
  if he:
58
  ne = he.select_one('p:not(.logo)') or he.find('p')
59
+ if ne:
60
+ info['home_team'] = _cl(ne.get_text())
61
  lo = he.select_one('img')
62
+ if lo:
63
+ info['home_logo'] = lo.get('src', '')
64
+
65
  ae = tel.select_one('.team.away')
66
  if ae:
67
  ne = ae.select_one('p:not(.logo)') or ae.find('p')
68
+ if ne:
69
+ info['away_team'] = _cl(ne.get_text())
70
  lo = ae.select_one('img')
71
+ if lo:
72
+ info['away_logo'] = lo.get('src', '')
73
+
74
  sc = tel.select_one('.score')
75
  if sc:
76
  parts = [_cl(p.get_text()) for p in sc.select('p')]
77
+ if len(parts) >= 2:
78
+ info['score'] = f"{parts[0]} - {parts[1]}"
79
  lb = sc.select_one('.label')
80
+ if lb:
81
+ info['status_label'] = _cl(lb.get_text())
82
+
83
  if info.get('home_team') and info.get('away_team'):
84
  result['info'] = info
85
  result['found'] = True
86
  result['sections'].append('info')
87
  else:
88
  return result
89
+
90
+ # Events parsing
91
  events = []
92
  events_div = soup.select_one('.events')
93
  if events_div:
94
  period = ''
95
  for child in events_div.children:
96
+ if not hasattr(child, 'name') or not child.name:
97
+ continue
98
  cls = ' '.join(child.get('class', []))
99
  if 'period' in cls:
100
  h2 = child.find('h2')
101
+ if h2:
102
+ period = _cl(h2.get_text())
103
  for ev in child.children:
104
+ if not hasattr(ev, 'name') or not ev.name:
105
+ continue
106
  ev_cls = ' '.join(ev.get('class', []))
107
+ if 'event' not in ev_cls:
108
+ continue
109
+
110
+ ev_data = {
111
+ 'team': 'home' if 'home' in ev_cls else 'away',
112
+ 'period': period,
113
+ 'type': 'unknown',
114
+ 'time': '',
115
+ }
116
+
117
+ # Type detection
118
  type_el = ev.select_one('.event-type')
119
  if type_el:
120
+ if type_el.select_one('[class*="redcard"]'):
121
+ ev_data['type'] = 'redcard'
122
+ elif type_el.select_one('[class*="yellowcard"]'):
123
+ ev_data['type'] = 'yellowcard'
124
+ elif type_el.select_one('[class*="goal"]'):
125
+ ev_data['type'] = 'goal'
126
+ elif type_el.select_one('[class*="substitution"]'):
127
+ ev_data['type'] = 'substitution'
128
+
129
  players_el = ev.select_one('.players')
130
  if players_el:
131
  time_el = players_el.select_one('.event-time')
132
+ if time_el:
133
+ ev_data['time'] = _normalize_time(time_el.get_text())
134
+
135
+ # Parse player names
136
  text = _cl(players_el.get_text(' ', strip=True).replace(ev_data['time'], '').strip())
137
  ev_data['players'] = text
138
+
139
+ if ev_data['type'] == 'goal':
140
+ words = text.split()
141
+ if len(words) >= 2:
142
+ ev_data['scorer'] = ' '.join(words[:2])
143
+ elif len(words) == 1:
144
+ ev_data['scorer'] = words[0]
145
+ elif ev_data['type'] == 'substitution':
146
+ words = text.split()
147
+ if len(words) >= 4:
148
+ ev_data['player_out'] = ' '.join(words[:len(words)//2])
149
+ ev_data['player_in'] = ' '.join(words[len(words)//2:])
150
+ elif ev_data['type'] in ('redcard', 'yellowcard'):
151
+ ev_data['player'] = text
152
+
153
  events.append(ev_data)
154
+
155
  if events:
156
  result['events'] = events
157
  result['sections'].append('events')
158
+
159
+ # Prediction
160
+ pred = soup.select_one('.prediction-card')
161
+ if pred:
162
+ pred_data = {}
163
+ team_info = pred.select_one('.team-info')
164
+ if team_info:
165
+ teams = team_info.select('.team')
166
+ if len(teams) >= 2:
167
+ pred_data['home_name'] = _cl(teams[0].select_one('.team-name').get_text() or '') if teams[0].select_one('.team-name') else ''
168
+ pred_data['away_name'] = _cl(teams[1].select_one('.team-name').get_text() or '') if teams[1].select_one('.team-name') else ''
169
+ divider = team_info.select_one('.divider')
170
+ if divider:
171
+ pred_data['result'] = _cl(divider.get_text())
172
+ vc = pred.select_one('.vote-count')
173
+ if vc:
174
+ pred_data['vote_count'] = _cl(vc.get_text())
175
+ result['prediction'] = pred_data
176
+
177
+ # Recent matches
178
+ recent = []
179
+ ml = soup.select_one('.matches-list')
180
+ if ml:
181
+ for item in ml.select('.match-detail, .match-item, li'):
182
+ de = item.select_one('.date, .time')
183
+ le = item.select_one('.league')
184
+ he = item.select_one('.home, .team-home')
185
+ ae = item.select_one('.away, .team-away')
186
+ se = item.select_one('.score, .result')
187
+ if he or ae:
188
+ recent.append({
189
+ 'date': _cl(de.get_text()) if de else '',
190
+ 'league': _cl(le.get_text()) if le else '',
191
+ 'home': _cl(he.get_text()) if he else '',
192
+ 'away': _cl(ae.get_text()) if ae else '',
193
+ 'score': _cl(se.get_text()) if se else 'vs',
194
+ })
195
+ if recent:
196
+ result['recent_matches'] = recent
197
+ result['sections'].append('recent')
198
+
199
+ # H2H stats API
200
+ try:
201
+ api_headers = {
202
+ "User-Agent": "Mozilla/5.0",
203
+ "Accept": "application/json",
204
+ "X-Requested-With": "XMLHttpRequest",
205
+ "Referer": "https://bongda.com.vn/",
206
+ }
207
+ r = requests.get(
208
+ f"https://bongda.com.vn/api/fixtures/h2h-stats?event_id={event_id}",
209
+ headers=api_headers, timeout=10
210
+ )
211
+ if r.status_code == 200:
212
+ ad = r.json()
213
+ if ad.get('status') == 'success' and ad.get('html'):
214
+ asp = BeautifulSoup(ad['html'], 'html.parser')
215
+ ast = {}
216
+ for row in asp.select('li, tr'):
217
+ cells = row.select('td, span, p')
218
+ if len(cells) >= 3:
219
+ lb = _cl(cells[0].get_text())
220
+ if lb:
221
+ ast[lb] = {'home': _cl(cells[1].get_text()), 'away': _cl(cells[2].get_text())}
222
+ if ast:
223
+ result['h2h_stats_parsed'] = ast
224
+ result['sections'].append('h2h_stats')
225
+ except Exception:
226
+ pass
227
+
228
  except Exception as e:
229
  result['error'] = str(e)
230
+
231
  return result
232
 
233
+
234
+ # FastAPI endpoint (to be added to app_v2_entry.py)
235
  from fastapi import Query
236
  from fastapi.responses import JSONResponse
237
 
238
  def add_bongda_proxy_endpoint(app):
239
  @app.get('/api/proxy/bongda')
240
  def proxy_bongda(event_id: int = Query(default=None), url: str = Query(default=None)):
241
+ """Proxy bongda.com.vn match data - fast server-side scraping."""
242
  if event_id is None:
243
  return JSONResponse({'error': 'event_id required'}, status_code=400)
244
+ return JSONResponse(scrape_match_html(event_id, url))
main.py CHANGED
@@ -1,263 +1,374 @@
1
- """VNEWS - FastAPI backend with livescore + xemlaibongda highlights + VTV channels"""
2
- import re, time, subprocess, json, os, threading
3
  import html as html_lib
4
  from datetime import datetime, timezone, timedelta
5
- from collections import defaultdict
6
 
7
- VN_TZ = timezone(timedelta(hours=0)) # GMT+0 (UTC)
8
  from concurrent.futures import ThreadPoolExecutor, as_completed
9
  from fastapi import FastAPI, Query, Request
10
  from fastapi.responses import HTMLResponse, JSONResponse, StreamingResponse, Response
11
- from urllib.parse import quote
 
12
  import requests
13
  from bs4 import BeautifulSoup
14
 
15
  app = FastAPI()
16
 
17
- # ===== WORLD CUP 2026 SCRAPER =====
18
- from wc2026_scraper import get_wc2026_all, scrape_fixtures, scrape_standings, scrape_stats, scrape_wc_news
19
-
20
- # ===== RATE LIMITING =====
21
- _rate_limit_data = defaultdict(list)
22
- _rate_limit_lock = threading.Lock()
23
- RATE_LIMIT_MAX = 60
24
- RATE_LIMIT_WINDOW = 60
25
-
26
- def _check_rate_limit(ip: str) -> bool:
27
- with _rate_limit_lock:
28
- now = time.time()
29
- _rate_limit_data[ip] = [t for t in _rate_limit_data[ip] if now - t < RATE_LIMIT_WINDOW]
30
- if len(_rate_limit_data[ip]) >= RATE_LIMIT_MAX: return False
31
- _rate_limit_data[ip].append(now)
32
- return True
33
-
34
- @app.middleware("http")
35
- async def rate_limit_middleware(request: Request, call_next):
36
- if request.url.path.startswith("/api/"):
37
- ip = request.client.host
38
- if not _check_rate_limit(ip): return JSONResponse({"error": "rate limit exceeded"}, status_code=429)
39
- return await call_next(request)
40
-
41
- # ===== VTV CHANNELS API =====
42
  from vtv_api import router as vtv_router
43
  app.include_router(vtv_router)
44
 
45
  HEADERS = {"User-Agent":"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36","Accept-Language":"vi-VN,vi;q=0.9,en;q=0.8"}
46
  BONGDA_HEADERS = {"User-Agent":"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36","Accept-Language":"vi-VN,vi;q=0.9","Referer":"https://bongda.com.vn/lich-thi-dau","X-Requested-With":"XMLHttpRequest"}
47
  BASE_BDP = "https://bongdaplus.vn"
 
48
  _cache = {}
49
  _cache_ttl = 300
50
  _cache_ttl_live = 60
51
  _cache_ttl_yt = 1800
52
-
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
53
  PRIORITY_LEAGUES = ["Ngoại Hạng Anh","FA Cup","Champions League","LaLiga","Copa del Rey","Serie A","Bundesliga","Ligue 1","V-League"]
54
  LEAGUE_IDS = {"nha":27110,"laliga":27233,"seriea":27044,"bundesliga":26891,"ligue1":27212}
55
- HL_LEAGUES = {
56
- "premier-league":{"path":"anh/premier-league","name":"Premier League","emoji":"🏴󠁧󠁢󠁥󠁮󠁧󠁿"},
57
- "fa-cup":{"path":"anh/fa-cup","name":"FA Cup","emoji":"🏆"},
58
- "bundesliga":{"path":"duc/bundesliga","name":"Bundesliga","emoji":"🇩🇪"},
59
- "serie-a":{"path":"italy/serie-a","name":"Serie A","emoji":"🇮🇹"},
60
- "la-liga":{"path":"tay-ban-nha/la-liga","name":"La Liga","emoji":"🇪🇸"},
61
- "champions-league":{"path":"cup-chau-au/uefa-champions-league","name":"Champions League","emoji":"⭐"},
62
- "europa-league":{"path":"cup-chau-au/uefa-europa-league","name":"Europa League","emoji":"🟠"},
63
- "world-cup":{"path":"the-gioi/world-cup","name":"World Cup 2026","emoji":"🌍"},
64
- }
65
  def _cached(key, fn, ttl=None):
66
- now=time.time(); t=ttl or _cache_ttl
67
- if key in _cache and now-_cache[key]["t"]<t: return _cache[key]["d"]
68
- try: data=fn()
69
- except: data=_cache.get(key,{}).get("d",[])
70
- _cache[key]={"d":data,"t":now}; return data
71
- def _get(url, headers=None):
72
- h=headers or HEADERS; r=requests.get(url, headers=h, timeout=15); r.encoding="utf-8"
73
  return BeautifulSoup(r.text,"lxml")
74
  def fetch_bongda_api(endpoint):
75
  try:
76
- r=requests.get(f"https://bongda.com.vn{endpoint}", headers=BONGDA_HEADERS, timeout=10)
77
  if r.status_code==200:
78
  data=r.json()
79
- if data.get("status")=="success": return data.get("html","")
80
  return ""
81
- except: return ""
82
-
83
  def _parse_match_from_li(li, status_type="live"):
84
  match_div=li.select_one("div.match")
85
- if not match_div: return None
86
- home_el=match_div.select_one(".home-team .name"); away_el=match_div.select_one(".away-team .name")
87
- if not home_el or not away_el: return None
88
- status_el=match_div.select_one(".status a"); league_el=li.find_previous("strong"); time_el=match_div.select_one(".match-time")
89
- home_logo=match_div.select_one(".home-team .logo img"); away_logo=match_div.select_one(".away-team .logo img")
90
  event_id=""
91
  if status_el:
92
- href=status_el.get("href",""); m=re.search(r'/tran-dau/(\d+)/',href)
93
- if m: event_id=m.group(1)
94
- spans=status_el.find_all("span") if status_el else []; score=""; minute=""
95
- if len(spans)>=3: score=f"{spans[0].get_text(strip=True)} - {spans[2].get_text(strip=True)}"
96
- if len(spans)>=4: minute=spans[3].get_text(strip=True)
97
- if not score and status_el and status_el.select_one(".vs"): score="VS"
98
  league=league_el.get_text(strip=True) if league_el else ""
99
- return {"home":home_el.get_text(strip=True),"away":away_el.get_text(strip=True),"score":score or"VS","minute":minute,"league":league,"time":time_el.get_text(strip=True) if time_el else "","event_id":event_id,"home_logo":home_logo.get("src","") if home_logo else "","away_logo":away_logo.get("src","") if away_logo else "","status":status_type}
100
 
101
  # ===== VIDEO PROXY =====
102
  @app.get("/api/proxy/m3u8")
103
  def proxy_m3u8(url: str = Query(...)):
104
  try:
105
  r = requests.get(url, headers=HEADERS, timeout=15)
106
- if r.status_code != 200: return Response(status_code=502, content="upstream error")
107
- lines = r.text.strip().split('\n'); rewritten = []
108
  for line in lines:
109
- if line.startswith('#') or not line.strip(): rewritten.append(line)
110
- else: rewritten.append("/api/proxy/seg?url=" + quote(line.strip(), safe=""))
111
- return Response(content='\n'.join(rewritten).encode('utf-8'), media_type="application/vnd.apple.mpegurl", headers={"Access-Control-Allow-Origin":"*","Cache-Control":"public, max-age=300"})
112
- except: return Response(status_code=502, content="proxy error")
113
 
114
  @app.get("/api/proxy/seg")
115
  def proxy_segment(url: str = Query(...)):
116
  try:
117
  r = requests.get(url, headers=HEADERS, timeout=30)
118
- if r.status_code != 200: return Response(status_code=502, content="upstream error")
119
  data = r.content
120
- if len(data) > 188 and data[0:4] == b'\x89PNG' and data[188] == 0x47: data = data[188:]
121
- return Response(content=data, media_type="video/mp2t", headers={"Access-Control-Allow-Origin":"*","Cache-Control":"public, max-age=3600"})
122
- except: return Response(status_code=502, content="proxy error")
123
 
124
  @app.get("/api/proxy/video")
125
  def proxy_video(url: str = Query(...), request: Request = None):
126
  try:
127
  req_headers = dict(HEADERS)
128
- if request and request.headers.get("range"): req_headers["Range"] = request.headers["range"]
129
  r = requests.get(url, headers=req_headers, timeout=30, stream=True)
130
  resp_headers = {"Access-Control-Allow-Origin":"*","Accept-Ranges":"bytes","Content-Type":r.headers.get("Content-Type","video/mp4")}
131
- if "Content-Range" in r.headers: resp_headers["Content-Range"] = r.headers["Content-Range"]
132
- if "Content-Length" in r.headers: resp_headers["Content-Length"] = r.headers["Content-Length"]
133
- return StreamingResponse(r.iter_content(chunk_size=256*1024), status_code=r.status_code, headers=resp_headers)
134
- except: return Response(status_code=502, content="proxy error")
135
 
136
  @app.get("/api/proxy/img")
137
  def proxy_img(url: str = Query(...)):
 
138
  try:
139
- from urllib.parse import urlparse
140
- _u = urlparse(url); _host = _u.netloc.lower()
141
- _referer = "https://dantri.com.vn/"
142
- if "refooty" in _host or "xemlaibongda" in _host: _referer = "https://xemlaibongda.top/"
143
- elif "ytimg" in _host or "youtube" in _host: _referer = "https://www.youtube.com/"
144
- elif "vncecdn" in _host or "vnexpress" in _host: _referer = "https://vnexpress.net/"
145
- r = requests.get(url, headers={**HEADERS, "Referer": _referer}, timeout=10)
146
- if r.status_code != 200: return Response(status_code=502)
147
- return Response(content=r.content, media_type=r.headers.get("Content-Type", "image/jpeg"), headers={"Cache-Control": "public, max-age=86400", "Access-Control-Allow-Origin": "*"})
148
- except: return Response(status_code=502)
149
 
150
  # ===== XEMLAIBONGDA HIGHLIGHTS =====
151
  def _scrape_xemlaibongda_page(page_path, limit=20):
 
152
  try:
153
  url = f"https://xemlaibongda.top/{page_path}" if page_path else "https://xemlaibongda.top/"
154
  r = requests.get(url, headers=HEADERS, timeout=15)
155
- if r.status_code != 200: return []
 
156
  r.encoding = "utf-8"
157
  soup = BeautifulSoup(r.text, "lxml")
158
- videos = []; seen = set()
159
- for a in soup.find_all("a", href=True):
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
160
  href = a.get("href", "")
161
- if "/video/" not in href and "/xem-lai/" not in href: continue
162
- if not href.startswith("http"): href = "https://xemlaibongda.top" + href
163
- clean_href = href.split("?")[0].split("#")[0]
164
- if clean_href in seen: continue
165
- seen.add(clean_href)
166
- img_src = ""
 
 
 
 
 
 
 
 
 
 
 
167
  img = a.find("img")
168
- if not img and a.parent: img = a.parent.find("img")
 
169
  if not img:
 
170
  p = a.parent
171
- for _ in range(4):
172
- if p and p.find("img"): img = p.find("img"); break
173
- p = p.parent if p else None
174
- if img:
175
- img_src = (img.get("data-src", "") or img.get("src", "") or img.get("data-lazy", "") or img.get("data-original", "") or img.get("data-thumb", "") or img.get("data-image", ""))
176
- if img_src.startswith("//"): img_src = "https:" + img_src
177
- elif img_src.startswith("/"): img_src = "https://xemlaibongda.top" + img_src
178
- if not img_src:
179
- p = a.parent
180
- for _ in range(5):
181
- if p is None: break
182
- style = p.get("style", "")
183
- bg_match = re.search(r'url\(["\']?(.*?)["\']?\)', style)
184
- if bg_match:
185
- img_src = bg_match.group(1)
186
- if img_src.startswith("//"): img_src = "https:" + img_src
187
- elif img_src.startswith("/"): img_src = "https://xemlaibongda.top" + img_src
188
  break
189
  p = p.parent if p else None
 
 
 
 
 
 
 
 
 
190
  title = ""
191
- for attr in ["title", "aria-label"]:
192
- val = a.get(attr, "")
193
- if val and len(val) >= 5: title = val; break
 
 
 
194
  if not title:
195
- for selector in ["h3", "h2", "h4", ".title", ".video-title", "strong"]:
196
- try:
197
- el = a.select_one(selector)
198
- if el: t = el.get_text(strip=True)
199
- if t and len(t) >= 5: title = t; break
200
- except: pass
201
  if not title:
202
- text = a.get_text(strip=True)
203
- if text and len(text) >= 5: title = text[:100]
204
- if not title or len(title) < 3:
205
- slug = clean_href.split("/video/")[-1].rstrip("/").split("/xem-lai/")[-1].rstrip("/")
206
- title = slug.replace("-", " ").replace("_", " ").title()
 
 
 
 
 
207
  title = re.sub(r'\d{4}-\d{2}-\d{2}', '', title).strip()
208
- if not title or len(title) < 3: continue
209
- if not img_src:
210
- slug = clean_href.split("/video/")[-1].rstrip("/").split("/xem-lai/")[-1].rstrip("/")
211
- img_src = f"https://xemlaibongda.top/uploads/thumb/{slug}.jpg"
212
- videos.append({"title": title[:100], "link": clean_href, "img": img_src, "source": "xemlaibongda"})
213
- if len(videos) >= limit: break
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
214
  return videos
215
- except Exception as e:
216
- print(f"[xemlaibongda] Error: {e}"); return []
217
 
218
- def scrape_xemlaibongda(): return _scrape_xemlaibongda_page("", 20)
219
  def scrape_highlights_by_league(league_key):
220
- if league_key not in HL_LEAGUES: return []
221
- return _scrape_xemlaibongda_page(HL_LEAGUES[league_key]["path"], 20)
 
222
  def scrape_all_league_highlights():
223
  results = {}
224
- def _fetch(key): return key, scrape_highlights_by_league(key)
225
  with ThreadPoolExecutor(8) as ex:
226
  futs = [ex.submit(_fetch, k) for k in HL_LEAGUES]
227
- for f in as_completed(futs, timeout=25):
228
- try: key, vids = f.result()
229
- except: continue
230
- if vids: results[key] = vids
 
231
  return results
232
 
233
  def extract_xemlaibongda_video(url):
234
  try:
235
- r=requests.get(url, headers=HEADERS, timeout=15)
236
- if r.status_code!=200: return None
237
- r.encoding="utf-8"; soup=BeautifulSoup(r.text,"lxml")
238
- og=soup.find("meta",property="og:image")
239
- og_poster=og.get("content","") if og else ""
240
- if og_poster.startswith("//"): og_poster="https:"+og_poster
241
- video=soup.find("video")
242
  if video:
243
- src=video.get("src",""); poster=video.get("poster","")
244
  if not src:
245
  source=video.find("source")
246
- if source: src=source.get("src","")
247
- if not poster: poster=og_poster
248
- if src: return{"src":src,"poster":poster,"type":"hls" if".m3u8" in src else"video"}
249
  m3u8s=re.findall(r'(https?://[^\s"\'<>]+\.m3u8)',r.text)
250
- if m3u8s: return{"src":m3u8s[0],"poster":og_poster,"type":"hls"}
251
- yt_iframe = soup.find("iframe", src=re.compile(r"youtube\.com/embed|youtube-nocookie\.com/embed"))
252
- if yt_iframe: return{"src":yt_iframe.get("src",""),"poster":og_poster,"type":"youtube"}
253
  return None
254
- except: return None
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
255
 
256
  # ===== LIVESCORE =====
257
  @app.get("/api/livescore/live")
258
- def api_livescore_live(): return JSONResponse({"html":_cached("ls_live",lambda:fetch_bongda_api("/api/fixtures/live"),ttl=_cache_ttl_live)})
259
  @app.get("/api/livescore/incoming")
260
- def api_livescore_incoming(): return JSONResponse({"html":_cached("ls_incoming",lambda:fetch_bongda_api("/api/fixtures/incoming"),ttl=_cache_ttl_live)})
261
  @app.get("/api/livescore/today")
262
  def api_livescore_today():
263
  today=datetime.now(VN_TZ).strftime("%Y-%m-%d");return JSONResponse({"html":_cached("ls_today",lambda:fetch_bongda_api(f"/api/fixtures/get-by-date?date={today}"),ttl=_cache_ttl)})
@@ -274,15 +385,20 @@ def api_match_commentaries(event_id:int):return JSONResponse({"html":fetch_bongd
274
  @app.get("/api/match/{event_id}/stats")
275
  def api_match_stats(event_id:int):return JSONResponse({"html":fetch_bongda_api(f"/api/event-standing/player-performance?event_id={event_id}")})
276
 
 
277
  from match_detail_v2 import fetch_match_detail, fetch_match_detail_by_url
278
 
279
  @app.get("/api/match/{event_id}/detail")
280
  def api_match_detail(event_id: int, url: str = Query(default="")):
 
281
  try:
282
- if url: data = fetch_match_detail_by_url(url)
283
- else: data = fetch_match_detail(event_id)
 
 
284
  return JSONResponse(data)
285
- except Exception as e: return JSONResponse({"event_id": event_id, "found": False, "error": str(e)})
 
286
 
287
  @app.get("/api/livescore/featured")
288
  def api_livescore_featured():
@@ -305,216 +421,548 @@ def api_livescore_featured():
305
  return None
306
  return JSONResponse(_cached("ls_featured",_f,ttl=30))
307
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
308
  @app.get("/api/highlights")
309
- def api_highlights(): return JSONResponse(_cached("xemlaibongda_hl",scrape_xemlaibongda,ttl=_cache_ttl))
310
  @app.get("/api/highlights/leagues")
311
- def api_highlights_leagues(): return JSONResponse(_cached("hl_leagues",scrape_all_league_highlights,ttl=_cache_ttl))
312
  @app.get("/api/highlights/{league}")
313
  def api_highlights_league(league:str):
314
- if league not in HL_LEAGUES: return JSONResponse({"error":"league not found"})
315
  return JSONResponse(_cached(f"hl_{league}",lambda:scrape_highlights_by_league(league),ttl=_cache_ttl))
316
-
 
317
  @app.get("/api/video_url")
318
- def api_video_url(url:str=Query(...), img:str=Query(default="")):
319
  if "youtube.com" in url or "youtu.be" in url:
320
  m=re.search(r'(?:v=|shorts/|youtu\.be/)([a-zA-Z0-9_-]{11})',url)
321
- if m: vid=m.group(1); return JSONResponse({"src":f"https://www.youtube.com/embed/{vid}?autoplay=1&rel=0&enablejsapi=1","poster":f"https://i.ytimg.com/vi/{vid}/hqdefault.jpg","type":"youtube"})
322
  if "xemlaibongda.top" in url:
323
  v=extract_xemlaibongda_video(url)
324
  if v:
325
- if v["type"]=="hls": v["src"]="/api/proxy/m3u8?url="+quote(v["src"],safe="")
326
- if not v.get("poster") and img: v["poster"] = img
327
  return JSONResponse(v)
 
 
 
 
 
 
 
 
 
 
328
  return JSONResponse({"error":"not found"})
329
-
330
- # ===== WORLD CUP 2026 API =====
331
- _wc_request_times = []; _wc_rate_limit_lock = threading.Lock()
332
- _WC_RATE_LIMIT = 10
333
- def _wc_rate_limit():
334
- global _wc_request_times
335
- with _wc_rate_limit_lock:
336
- now = time.time()
337
- _wc_request_times = [t for t in _wc_request_times if now - t < 60]
338
- if len(_wc_request_times) >= _WC_RATE_LIMIT: return False
339
- _wc_request_times.append(now)
340
- return True
341
-
342
- @app.get("/api/wc2026")
343
- def api_wc2026():
344
- return JSONResponse(_cached("wc2026", get_wc2026_all, ttl=_cache_ttl))
345
-
346
- @app.get("/api/wc2026/{tab}")
347
- def api_wc2026_tab(tab: str):
348
- valid_tabs = ["news", "fixtures", "standings", "stats", "highlights"]
349
- if tab not in valid_tabs: return JSONResponse({"error": "invalid tab"}, status_code=400)
350
- def _fetch_tab():
351
- if tab == "highlights": return scrape_highlights_by_league("world-cup")
352
- elif tab == "news": return scrape_wc_news()
353
- elif tab == "fixtures": return scrape_fixtures()
354
- elif tab == "standings": return scrape_standings()
355
- elif tab == "stats": return scrape_stats()
356
- return []
357
- return JSONResponse(_cached(f"wc2026_{tab}", _fetch_tab, ttl=_cache_ttl))
358
-
359
  @app.get("/api/bdp_videos")
360
  def api_bdp_videos():
361
  def _f():
362
  try:
363
- soup=_get(f"{BASE_BDP}/video"); arts=[]; seen=set()
364
  for a in soup.find_all("a",href=True):
365
  href=a.get("href","")
366
  if"/video/" not in href or href in("/video/","/video/ban-thang-dep","/video/highlight"):continue
367
- if not href.startswith("http"): href=BASE_BDP+href
368
- if href in seen: continue
369
  title=re.sub(r'^\d{2}:\d{2}','',a.get_text(strip=True)).strip()
370
- if not title or len(title)<5: continue
371
  img_tag=a.find("img") or(a.parent.find("img") if a.parent else None)
372
  img=(img_tag.get("data-src") or img_tag.get("src","")) if img_tag else ""
373
- seen.add(href); arts.append({"title":title,"link":href,"img":img,"source":"bdp"})
374
  return arts[:20]
375
- except: return []
376
  return JSONResponse(_cached("bdp_videos",_f))
377
-
378
  # ===== NEWS =====
379
- VNE_CATS={"thoi-su":("https://vnexpress.net/thoi-su","Thời Sự"),"the-gioi":("https://vnexpress.net/the-gioi","Thế Giới"),"kinh-doanh":("https://vnexpress.net/kinh-doanh","Kinh Doanh"),"the-thao":("https://vnexpress.net/the-thao","Thể Thao"),"giai-tri":("https://vnexpress.net/giai-tri","Giải Trí"),"suc-khoe":("https://vnexpress.net/suc-khoe","Sức Khỏe"),"phap-luat":("https://vnexpress.net/phap-luat","Pháp Luật"),"giao-duc":("https://vnexpress.net/giao-duc","Giáo Dục"),"du-lich":("https://vnexpress.net/du-lich","Du Lịch"),"doi-song":("https://vnexpress.net/doi-song","Đời Sống")}
380
-
381
  def scrape_vne(cat_url):
382
  try:
383
- soup=_get(cat_url); arts=[]
384
  for it in soup.select("article.item-news")[:15]:
385
  a=it.select_one("h2.title-news a") or it.select_one("h3.title-news a")
386
- if not a: continue
387
- t=a.get("title","") or a.get_text(strip=True); lk=a.get("href","")
388
- if not t or not lk: continue
389
- im=it.find("img"); img=(im.get("data-src") or im.get("src","")) if im else ""
390
- if img and 'blank' in img:
391
  src=it.find("source")
392
- if src: img=src.get("srcset","").split(",")[0].strip().split(" ")[0]
393
  arts.append({"title":t,"link":lk,"img":img,"source":"vne"})
394
  return arts
395
- except: return []
396
-
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
397
  def scrape_genk_ai():
 
398
  try:
399
  r=requests.get("https://genk.vn/ai.chn",headers=HEADERS,timeout=15)
400
- if r.status_code!=200: return []
401
- r.encoding="utf-8"; soup=BeautifulSoup(r.text,"lxml"); articles=[]; seen=set()
 
402
  for a in soup.find_all("a",href=True):
403
  href=a.get("href","")
404
- if not href.endswith(".chn") or href=="/ai.chn": continue
405
- if href.startswith("/"): href="https://genk.vn"+href
406
- if href in seen or "genk.vn" not in href: continue
407
  title=a.get("title","") or a.get_text(strip=True)
408
- if not title or len(title)<20: continue
409
- container=a.parent; img_src=""
410
  for _ in range(6):
411
- if container is None: break
412
  for img in container.find_all("img"):
413
  s=img.get("data-src","") or img.get("src","")
414
- if s and "mediacdn" in s and "avatar" not in s and "logo" not in s: img_src=s; break
415
- if img_src: break; container=container.parent
 
 
416
  seen.add(href)
417
  if not img_src:
418
  try:
419
- og_r=requests.get(href,headers=HEADERS,timeout=8); og_r.encoding="utf-8"
420
- og_soup=BeautifulSoup(og_r.text,"lxml"); og_tag=og_soup.find("meta",property="og:image")
421
- if og_tag: img_src=og_tag.get("content","")
422
- except: pass
423
  articles.append({"title":title,"link":href,"img":img_src,"source":"genk"})
424
- if len(articles)>=30: break
425
  return articles
426
- except: return []
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
427
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
428
  @app.get("/api/homepage")
429
  def api_homepage():
430
  def _f():
431
  articles=[]
432
  with ThreadPoolExecutor(12) as ex:
433
  futs={ex.submit(scrape_vne,VNE_CATS[k][0]):VNE_CATS[k][1] for k in["thoi-su","the-gioi","kinh-doanh","the-thao","giai-tri","phap-luat","giao-duc","du-lich","doi-song"]}
 
434
  for f in as_completed(futs):
435
  try:
436
- for a in f.result(): a["group"]=futs[f]; articles.append(a)
437
- except: pass
438
  return articles
439
  return JSONResponse(_cached("homepage",_f))
440
-
441
  @app.get("/api/category/{cat_id}")
442
  def api_category(cat_id:str):
443
  def _f():
444
- if cat_id=="cong-nghe": return scrape_genk_ai()
445
- if cat_id in VNE_CATS:
446
- arts=scrape_vne(VNE_CATS[cat_id][0])
447
- [a.update({"group":VNE_CATS[cat_id][1]}) for a in arts]
448
- return arts
449
- return []
450
  return JSONResponse(_cached(f"cat_{cat_id}",_f))
451
-
452
  @app.get("/api/categories")
453
  def api_categories():
454
- cats=[{"id":"cong-nghe","name":"Công Nghệ","source":"genk"}]
455
- for k,(u,n) in VNE_CATS.items(): cats.append({"id":k,"name":n,"source":"vne"})
456
  return JSONResponse(cats)
457
-
458
- @app.get("/api/proxy/xlb")
459
- def api_xlb(path: str = Query(default=""), limit: int = Query(default=20)):
 
 
 
 
460
  try:
461
- url = f"https://xemlaibongda.top/{path}" if path else "https://xemlaibongda.top/"
462
- r = requests.get(url, headers=HEADERS, timeout=15)
463
- if r.status_code != 200: return JSONResponse({"videos": []})
464
- r.encoding = "utf-8"
465
- soup = BeautifulSoup(r.text, "lxml")
466
- videos, seen = [], set()
467
- for a in soup.find_all("a", href=True):
468
- href = a.get("href", "")
469
- if "/video/" not in href and "/xem-lai/" not in href: continue
470
- if not href.startswith("http"): href = "https://xemlaibongda.top" + href
471
- clean = href.split("?")[0].split("#")[0]
472
- if clean in seen: continue
473
- seen.add(clean)
474
- img_src = ""
475
- img = a.find("img") or (a.parent.find("img") if a.parent else None)
476
- if not img:
477
- p = a.parent
478
- for _ in range(5):
479
- if p and p.find("img"): img = p.find("img"); break
480
- p = p.parent if p else None
481
- if img:
482
- img_src = (img.get("data-src", "") or img.get("src", "") or img.get("data-lazy", "") or img.get("data-original", ""))
483
- if img_src.startswith("//"): img_src = "https:" + img_src
484
- elif img_src.startswith("/"): img_src = "https://xemlaibongda.top" + img_src
485
- title = a.find("h3")
486
- if not title: title = a.find("h2")
487
- if not title: title = a.find("strong")
488
- t = title.get_text(strip=True) if title else ""
489
- if not t:
490
- slug = clean.split("/video/")[-1].rstrip("/")
491
- t = slug.replace("-", " ").title()
492
- videos.append({"title": t[:100], "link": clean, "img": img_src, "source": "xemlaibongda"})
493
- if len(videos) >= limit: break
494
- return JSONResponse({"videos": videos})
495
- except Exception as e:
496
- return JSONResponse({"videos": [], "error": str(e)})
497
 
498
  @app.get("/api/article")
499
  def api_article(url:str=Query(...)):
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
500
  try:
501
- r2 = requests.get(url, headers=HEADERS, timeout=10)
502
- if r2.status_code == 200:
503
- r2.encoding = "utf-8"
504
- soup = BeautifulSoup(r2.text, "lxml")
505
- og = soup.find("meta", property="og:image")
506
- return JSONResponse({"og_image": og.get("content", "") if og else ""})
507
- except: pass
508
- return JSONResponse({"og_image": ""})
509
-
510
- @app.get("/api/storage_status")
511
- def api_storage_status():
512
- return JSONResponse({"persistent":os.path.isdir("/data")})
513
-
514
- @app.get("/api/hot_topics")
515
- def api_hot_topics():
516
- return JSONResponse({"topics":[]})
517
-
518
- @app.get("/", response_class=HTMLResponse)
519
- async def root():
520
- return HTMLResponse("<h1>VNEWS v17</h1><p>VTV sv2.xemtivitop.com · GMT+0 · EPG vtv.vn</p>")
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """VNEWS - FastAPI backend with livescore + xemlaibongda highlights + YouTube FPT shorts"""
2
+ import hashlib, re, time, subprocess, json, os, threading
3
  import html as html_lib
4
  from datetime import datetime, timezone, timedelta
 
5
 
6
+ VN_TZ = timezone(timedelta(hours=7))
7
  from concurrent.futures import ThreadPoolExecutor, as_completed
8
  from fastapi import FastAPI, Query, Request
9
  from fastapi.responses import HTMLResponse, JSONResponse, StreamingResponse, Response
10
+ from fastapi.staticfiles import StaticFiles
11
+ from urllib.parse import unquote, quote, urlencode
12
  import requests
13
  from bs4 import BeautifulSoup
14
 
15
  app = FastAPI()
16
 
17
+ # ===== VTV CHANNELS API (VTV1-VTV10 + VTVPrime) =====
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
18
  from vtv_api import router as vtv_router
19
  app.include_router(vtv_router)
20
 
21
  HEADERS = {"User-Agent":"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36","Accept-Language":"vi-VN,vi;q=0.9,en;q=0.8"}
22
  BONGDA_HEADERS = {"User-Agent":"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36","Accept-Language":"vi-VN,vi;q=0.9","Referer":"https://bongda.com.vn/lich-thi-dau","X-Requested-With":"XMLHttpRequest"}
23
  BASE_BDP = "https://bongdaplus.vn"
24
+ SPACE_URL = "https://bep40-vnews.hf.space"
25
  _cache = {}
26
  _cache_ttl = 300
27
  _cache_ttl_live = 60
28
  _cache_ttl_yt = 1800
29
+ SHORTS_FALLBACK = [
30
+ {"id":"Lu_iCQ5YwNM","title":"Công an lập hồ sơ xử lý người phụ nữ chửi bới, tát nam tài xế ô tô ở Hà Nội | #shorts","channel":"baodantri7941"},
31
+ {"id":"CwWvijF8BOA","title":"Chú rể Ninh Bình bật khóc nhận món quà bí mật người cha quá cố gửi 26 năm trước | #shorts","channel":"baodantri7941"},
32
+ {"id":"tvPewsc2ph4","title":"Tính năng ẩn trên iPhone giúp giảm mỏi mắt | #shorts","channel":"baodantri7941"},
33
+ {"id":"b1Nxzv9ixlU","title":"Y án 3 năm tù với nữ tài xế uống 8 lon bia lái xe tông chủ tịch xã tử vong | #shorts","channel":"baodantri7941"},
34
+ {"id":"Xp5eTwAZAis","title":"Người đánh hàng xóm tại chung cư ở Hà Nội bị tuyên hơn 4 tháng tù | #shorts","channel":"baodantri7941"},
35
+ {"id":"Htzvwg6iOBM","title":"Xe điện Audi S6 Sportback e-tron có gì đặc biệt? | #shorts","channel":"baodantri7941"},
36
+ {"id":"iMdFmWvYdlo","title":"Cô gái người Nga yêu thời trang và đất nước Việt Nam | #shorts","channel":"baodantri7941"},
37
+ {"id":"IVaRc6moEv8","title":"Người nông dân Trung Quốc đột quỵ, bệnh viện giúp bán sạch 4 tấn táo | #shorts","channel":"baodantri7941"},
38
+ {"id":"uVxqPxToItU","title":"Công an vào cuộc vụ người phụ nữ chửi bới, hành hung tài xế ô tô ở Hà Nội | #shorts","channel":"baodantri7941"},
39
+ {"id":"VAfgNNgZDRs","title":"Khởi tố 4 đối tượng ném bom xăng vào nhà dân ở Đồng Nai | #shorts","channel":"baodantri7941"},
40
+ {"id":"sBH_-zGh0Xw","title":"Vì sao Times New Roman vẫn nổi tiếng sau hàng chục năm? | #shorts","channel":"baodantri7941"},
41
+ {"id":"woKn5f2bLHM","title":"Quảng Ninh ngập sâu diện rộng sau đợt mưa lớn | #shorts","channel":"baodantri7941"},
42
+ {"id":"bcpgRoxbLPw","title":"Giông lốc quật bay mái tôn ở TP.HCM | #shorts","channel":"baodantri7941"},
43
+ {"id":"ZIIC5osy544","title":"Bé trai Trung Quốc rơi từ tầng 11 vẫn sống sót kỳ diệu | #shorts","channel":"baodantri7941"},
44
+ {"id":"uTMJ49NQpyc","title":"Sau lớp mascot 40kg: Câu chuyện mưu sinh của người trẻ ở TPHCM | #shorts","channel":"baodantri7941"},
45
+ {"id":"7Pd6vZ2Lz1M","title":"Hành động ấm lòng của người đàn ông tham gia tìm kiếm 5 học sinh tử vong ở sông Lô | SKĐS","channel":"baosuckhoedoisongboyte"},
46
+ {"id":"SlHLt_ZyPiE","title":"Xử phạt người đàn ông xóa số điện thoại cứu hộ trên cao tốc Bắc - Nam | SK��S","channel":"baosuckhoedoisongboyte"},
47
+ {"id":"IUOprcJyYr4","title":"Phụ nữ táo bón có phải do lười ăn rau? | SKĐS #shorts","channel":"baosuckhoedoisongboyte"},
48
+ {"id":"YY8ojFNE-AU","title":"Quái xế tự quay clip nẹt pô, đánh võng đăng TikTok bị xử lý | SKĐS","channel":"baosuckhoedoisongboyte"},
49
+ {"id":"OV7_oGdQGII","title":"Bố cô dâu khóc sụt sùi rồi quẩy cực sung gây bão mạng | SKĐS","channel":"baosuckhoedoisongboyte"},
50
+ {"id":"FoxhFyz2skY","title":"Người đàn ông nước ngoài đập phá ô tô, bẻ cần gạt nước ở Đà Nẵng | SKĐS","channel":"baosuckhoedoisongboyte"},
51
+ {"id":"R1oC_I8dFPU","title":"Thanh niên buông tay lái, đứng trên xe máy khi đổ đèo ở Đắk Lắk | SKĐS","channel":"baosuckhoedoisongboyte"},
52
+ {"id":"U0Ft6ChWAIo","title":"Cô giáo kể phút tháo chạy khỏi xe khách trước khi bị lũ vò nát ở Cao Bằng | SKĐS","channel":"baosuckhoedoisongboyte"},
53
+ {"id":"hH0ANeze_4E","title":"Liên tiếp hàng chục con bò bị sét đánh chết trong ngày mưa dông | SKĐS","channel":"baosuckhoedoisongboyte"},
54
+ {"id":"pXWt0QbAzRQ","title":"Va chạm giao thông, người phụ nữ lăng mạ tài xế ô tô | SKĐS","channel":"baosuckhoedoisongboyte"},
55
+ {"id":"UWWLPY1OYt4","title":"CSGT chặn xe khách khống chế đối tượng cướp dây chuyền tại Gia Lai | SKĐS","channel":"baosuckhoedoisongboyte"},
56
+ {"id":"AxhVTQutsuo","title":"Xuất tinh sớm và những hiểu lầm thường gặp | SKĐS #shorts","channel":"baosuckhoedoisongboyte"},
57
+ {"id":"cNy6FgaNxYM","title":"Cô dâu khóc sưng mắt vì 6 chỉ vàng không cánh mà bay trong ngày cưới | SKĐS","channel":"baosuckhoedoisongboyte"},
58
+ {"id":"IDt_S6q59Ro","title":"Chở bạn gái không đội mũ bảo hiểm, thanh niên đấm CSGT | SKĐS","channel":"baosuckhoedoisongboyte"},
59
+ {"id":"LFxJ9Ik6W0A","title":"Mệnh lệnh từ trái tim: CSGT Hà Nội mở đường đưa bé 5 tháng tuổi đi cấp cứu | SKĐS","channel":"baosuckhoedoisongboyte"}
60
+ ]
61
+ for _v in SHORTS_FALLBACK:
62
+ _v["link"]="https://www.youtube.com/watch?v="+_v["id"]
63
+ _v["img"]="https://i.ytimg.com/vi/"+_v["id"]+"/hqdefault.jpg"
64
+ _v["source"]="yt"
65
+ SHORT_STATS_FILE = "/data/short_stats.json" if os.path.isdir("/data") else "/app/short_stats.json"
66
+ _short_lock = threading.Lock()
67
+ def _load_short_db():
68
+ try:
69
+ if os.path.exists(SHORT_STATS_FILE):
70
+ with open(SHORT_STATS_FILE,"r",encoding="utf-8") as f:return json.load(f)
71
+ except:pass
72
+ return {}
73
+ def _save_short_db(db):
74
+ try:
75
+ os.makedirs(os.path.dirname(SHORT_STATS_FILE),exist_ok=True)
76
+ tmp=SHORT_STATS_FILE+".tmp"
77
+ with open(tmp,"w",encoding="utf-8") as f:json.dump(db,f,ensure_ascii=False)
78
+ os.replace(tmp,SHORT_STATS_FILE)
79
+ except:pass
80
+
81
+ def _short_default():return {"views":0,"likes":0,"shares":0,"comments":[]}
82
+ WALL_FILE = "/data/wall_posts.json" if os.path.isdir("/data") else "/app/wall_posts.json"
83
+ def _load_wall():
84
+ try:
85
+ if os.path.exists(WALL_FILE):
86
+ with open(WALL_FILE,"r",encoding="utf-8") as f:return json.load(f)
87
+ except:pass
88
+ return []
89
+ def _save_wall(posts):
90
+ try:
91
+ os.makedirs(os.path.dirname(WALL_FILE),exist_ok=True)
92
+ tmp=WALL_FILE+".tmp"
93
+ with open(tmp,"w",encoding="utf-8") as f:json.dump(posts[:100],f,ensure_ascii=False)
94
+ os.replace(tmp,WALL_FILE)
95
+ except:pass
96
  PRIORITY_LEAGUES = ["Ngoại Hạng Anh","FA Cup","Champions League","LaLiga","Copa del Rey","Serie A","Bundesliga","Ligue 1","V-League"]
97
  LEAGUE_IDS = {"nha":27110,"laliga":27233,"seriea":27044,"bundesliga":26891,"ligue1":27212}
98
+ HL_LEAGUES = {"premier-league":{"path":"anh/premier-league","name":"Premier League","emoji":"🏴󠁧󠁢󠁥󠁮󠁧󠁿"},"fa-cup":{"path":"anh/fa-cup","name":"FA Cup","emoji":"🏆"},"bundesliga":{"path":"duc/bundesliga","name":"Bundesliga","emoji":"🇩🇪"},"serie-a":{"path":"italy/serie-a","name":"Serie A","emoji":"🇮🇹"},"la-liga":{"path":"tay-ban-nha/la-liga","name":"La Liga","emoji":"🇪🇸"},"champions-league":{"path":"cup-chau-au/uefa-champions-league","name":"Champions League","emoji":"⭐"},"europa-league":{"path":"cup-chau-au/uefa-europa-league","name":"Europa League","emoji":"🟠"},"world-cup":{"path":"the-gioi/world-cup","name":"World Cup 2026","emoji":"🌍"}}
 
 
 
 
 
 
 
 
 
99
  def _cached(key, fn, ttl=None):
100
+ now=time.time();t=ttl or _cache_ttl
101
+ if key in _cache and now-_cache[key]["t"]<t:return _cache[key]["d"]
102
+ try:data=fn()
103
+ except:data=_cache.get(key,{}).get("d",[])
104
+ _cache[key]={"d":data,"t":now};return data
105
+ def _get(url,headers=None):
106
+ h=headers or HEADERS;r=requests.get(url,headers=h,timeout=15);r.encoding="utf-8"
107
  return BeautifulSoup(r.text,"lxml")
108
  def fetch_bongda_api(endpoint):
109
  try:
110
+ r=requests.get(f"https://bongda.com.vn{endpoint}",headers=BONGDA_HEADERS,timeout=10)
111
  if r.status_code==200:
112
  data=r.json()
113
+ if data.get("status")=="success":return data.get("html","")
114
  return ""
115
+ except:return ""
 
116
  def _parse_match_from_li(li, status_type="live"):
117
  match_div=li.select_one("div.match")
118
+ if not match_div:return None
119
+ home_el=match_div.select_one(".home-team .name");away_el=match_div.select_one(".away-team .name")
120
+ if not home_el or not away_el:return None
121
+ status_el=match_div.select_one(".status a");league_el=li.find_previous("strong");time_el=match_div.select_one(".match-time")
122
+ home_logo=match_div.select_one(".home-team .logo img");away_logo=match_div.select_one(".away-team .logo img")
123
  event_id=""
124
  if status_el:
125
+ href=status_el.get("href","");m=re.search(r'/tran-dau/(\d+)/',href)
126
+ if m:event_id=m.group(1)
127
+ spans=status_el.find_all("span") if status_el else [];score="";minute=""
128
+ if len(spans)>=3:score=f"{spans[0].get_text(strip=True)} - {spans[2].get_text(strip=True)}"
129
+ if len(spans)>=4:minute=spans[3].get_text(strip=True)
130
+ if not score and status_el and status_el.select_one(".vs"):score="VS"
131
  league=league_el.get_text(strip=True) if league_el else ""
132
+ return{"home":home_el.get_text(strip=True),"away":away_el.get_text(strip=True),"score":score or"VS","minute":minute,"league":league,"time":time_el.get_text(strip=True) if time_el else "","event_id":event_id,"home_logo":home_logo.get("src","") if home_logo else "","away_logo":away_logo.get("src","") if away_logo else "","status":status_type}
133
 
134
  # ===== VIDEO PROXY =====
135
  @app.get("/api/proxy/m3u8")
136
  def proxy_m3u8(url: str = Query(...)):
137
  try:
138
  r = requests.get(url, headers=HEADERS, timeout=15)
139
+ if r.status_code != 200:return Response(status_code=502, content="upstream error")
140
+ lines = r.text.strip().split('\n');rewritten = []
141
  for line in lines:
142
+ if line.startswith('#') or not line.strip():rewritten.append(line)
143
+ else:rewritten.append("/api/proxy/seg?url=" + quote(line.strip(), safe=""))
144
+ return Response(content='\n'.join(rewritten).encode('utf-8'),media_type="application/vnd.apple.mpegurl",headers={"Access-Control-Allow-Origin":"*","Cache-Control":"public, max-age=300"})
145
+ except:return Response(status_code=502, content="proxy error")
146
 
147
  @app.get("/api/proxy/seg")
148
  def proxy_segment(url: str = Query(...)):
149
  try:
150
  r = requests.get(url, headers=HEADERS, timeout=30)
151
+ if r.status_code != 200:return Response(status_code=502, content="upstream error")
152
  data = r.content
153
+ if len(data) > 188 and data[0:4] == b'\x89PNG' and data[188] == 0x47:data = data[188:]
154
+ return Response(content=data,media_type="video/mp2t",headers={"Access-Control-Allow-Origin":"*","Cache-Control":"public, max-age=3600"})
155
+ except:return Response(status_code=502, content="proxy error")
156
 
157
  @app.get("/api/proxy/video")
158
  def proxy_video(url: str = Query(...), request: Request = None):
159
  try:
160
  req_headers = dict(HEADERS)
161
+ if request and request.headers.get("range"):req_headers["Range"] = request.headers["range"]
162
  r = requests.get(url, headers=req_headers, timeout=30, stream=True)
163
  resp_headers = {"Access-Control-Allow-Origin":"*","Accept-Ranges":"bytes","Content-Type":r.headers.get("Content-Type","video/mp4")}
164
+ if "Content-Range" in r.headers:resp_headers["Content-Range"] = r.headers["Content-RRange"]
165
+ if "Content-Length" in r.headers:resp_headers["Content-Length"] = r.headers["Content-Length"]
166
+ return StreamingResponse(r.iter_content(chunk_size=256*1024),status_code=r.status_code,headers=resp_headers)
167
+ except:return Response(status_code=502, content="proxy error")
168
 
169
  @app.get("/api/proxy/img")
170
  def proxy_img(url: str = Query(...)):
171
+ """Proxy images from sources that block hotlinking (DanTri CDN)."""
172
  try:
173
+ r = requests.get(url, headers={**HEADERS, "Referer": "https://dantri.com.vn/"}, timeout=10)
174
+ if r.status_code != 200:return Response(status_code=502)
175
+ ct = r.headers.get("Content-Type", "image/jpeg")
176
+ return Response(content=r.content, media_type=ct, headers={"Cache-Control": "public, max-age=86400", "Access-Control-Allow-Origin": "*"})
177
+ except:return Response(status_code=502)
 
 
 
 
 
178
 
179
  # ===== XEMLAIBONGDA HIGHLIGHTS =====
180
  def _scrape_xemlaibongda_page(page_path, limit=20):
181
+ """Scrape video articles from xemlaibongda.top. Tries multiple CSS selectors for robust parsing."""
182
  try:
183
  url = f"https://xemlaibongda.top/{page_path}" if page_path else "https://xemlaibongda.top/"
184
  r = requests.get(url, headers=HEADERS, timeout=15)
185
+ if r.status_code != 200:
186
+ return []
187
  r.encoding = "utf-8"
188
  soup = BeautifulSoup(r.text, "lxml")
189
+ videos = []
190
+ seen = set()
191
+
192
+ # Strategy 1: Look for article/video cards with links containing /video/
193
+ # xemlaibongda.top uses various HTML structures, try multiple selectors
194
+ selectors = [
195
+ 'a[href*="/video/"]',
196
+ 'a[href*="/xem-lai/"]',
197
+ '.video-item a',
198
+ '.post-item a',
199
+ '.item a',
200
+ 'article a',
201
+ ]
202
+ links = []
203
+ for sel in selectors:
204
+ links.extend(soup.select(sel))
205
+
206
+ for a in links:
207
  href = a.get("href", "")
208
+ if not href:
209
+ continue
210
+ # Normalize: pick the main slug part
211
+ is_video = "/video/" in href or "/xem-lai/" in href
212
+ if not is_video:
213
+ # Check if parent has video indicator
214
+ parent = a.parent
215
+ has_video_icon = parent.find(class_=re.compile(r'vid|play|highlight')) if parent else False
216
+ if not has_video_icon:
217
+ continue
218
+ if not href.startswith("http"):
219
+ href = "https://xemlaibongda.top" + href
220
+ if href in seen:
221
+ continue
222
+ seen.add(href)
223
+
224
+ # Extract image: check <img> inside the link or its parent
225
  img = a.find("img")
226
+ if not img and a.parent:
227
+ img = a.parent.find("img")
228
  if not img:
229
+ # search up to 3 levels up
230
  p = a.parent
231
+ for _ in range(3):
232
+ if p and p.find("img"):
233
+ img = p.find("img")
 
 
 
 
 
 
 
 
 
 
 
 
 
 
234
  break
235
  p = p.parent if p else None
236
+ img_src = ""
237
+ if img:
238
+ img_src = img.get("data-src", "") or img.get("src", "") or img.get("data-lazy", "") or img.get("data-original", "")
239
+ if img_src.startswith("//"):
240
+ img_src = "https:" + img_src
241
+ elif img_src.startswith("/"):
242
+ img_src = "https://xemlaibongda.top" + img_src
243
+
244
+ # Extract title: try multiple strategies
245
  title = ""
246
+ # Try heading tags first
247
+ for tag in ["h3", "h2", "h4", ".title", ".post-title", ".entry-title", ".video-title"]:
248
+ t = a.select_one(tag) if hasattr(a, 'select_one') else None
249
+ if t:
250
+ title = t.get_text(" ", strip=True)
251
+ break
252
  if not title:
253
+ title = a.get("title", "")
 
 
 
 
 
254
  if not title:
255
+ # Try aria-label or alt
256
+ title = a.get("aria-label", "")
257
+ if not title:
258
+ img_for_alt = a.find("img")
259
+ if img_for_alt:
260
+ title = img_for_alt.get("alt", "")
261
+ if not title:
262
+ # Last resort: extract from URL slug but clean it up
263
+ slug = href.split("/video/")[-1].rstrip("/").split("/xem-lai/")[-1].rstrip("/")
264
+ title = slug.replace("-", " ").replace("_", " ")
265
  title = re.sub(r'\d{4}-\d{2}-\d{2}', '', title).strip()
266
+ title = title.title()
267
+ if not title or len(title) < 3:
268
+ continue
269
+
270
+ videos.append({"title": title, "link": href, "img": img_src, "source": "xemlaibongda"})
271
+ if len(videos) >= limit:
272
+ break
273
+
274
+ # Strategy 2: If no videos found via links, look for video elements
275
+ if not videos:
276
+ for vid_el in soup.find_all("video"):
277
+ src = vid_el.get("src", "")
278
+ poster = vid_el.get("poster", "")
279
+ source_el = vid_el.find("source")
280
+ if not src and source_el:
281
+ src = source_el.get("src", "")
282
+ if src:
283
+ title = poster.rsplit("/", 1)[-1].rsplit(".", 1)[0].replace("-", " ").title() if poster else "Video"
284
+ videos.append({"title": title, "link": src, "img": poster, "source": "xemlaibongda"})
285
+ if len(videos) >= limit:
286
+ break
287
+
288
  return videos
289
+ except:
290
+ return []
291
 
292
+ def scrape_xemlaibongda():return _scrape_xemlaibongda_page("",20)
293
  def scrape_highlights_by_league(league_key):
294
+ if league_key not in HL_LEAGUES:return[]
295
+ return _scrape_xemlaibongda_page(HL_LEAGUES[league_key]["path"],20)
296
+
297
  def scrape_all_league_highlights():
298
  results = {}
299
+ def _fetch(key):return key, scrape_highlights_by_league(key)
300
  with ThreadPoolExecutor(8) as ex:
301
  futs = [ex.submit(_fetch, k) for k in HL_LEAGUES]
302
+ for f in as_completed(futs):
303
+ try:
304
+ key, vids = f.result()
305
+ if vids:results[key] = vids
306
+ except:pass
307
  return results
308
 
309
  def extract_xemlaibongda_video(url):
310
  try:
311
+ r=requests.get(url,headers=HEADERS,timeout=15)
312
+ if r.status_code!=200:return None
313
+ r.encoding="utf-8";soup=BeautifulSoup(r.text,"lxml");video=soup.find("video")
 
 
 
 
314
  if video:
315
+ src=video.get("src","");poster=video.get("poster","")
316
  if not src:
317
  source=video.find("source")
318
+ if source:src=source.get("src","")
319
+ if src:return{"src":src,"poster":poster,"type":"hls" if".m3u8" in src else"video"}
 
320
  m3u8s=re.findall(r'(https?://[^\s"\'<>]+\.m3u8)',r.text)
321
+ if m3u8s:
322
+ og=soup.find("meta",property="og:image");poster=og.get("content","") if og else ""
323
+ return{"src":m3u8s[0],"poster":poster,"type":"hls"}
324
  return None
325
+ except:return None
326
+
327
+ # ===== YOUTUBE SHORTS =====
328
+ def _yt_channel_shorts(channel, count=15):
329
+ """Fast scrape YouTube shorts tab without yt-dlp. Returns newest-first IDs/titles."""
330
+ try:
331
+ url=f"https://www.youtube.com/@{channel}/shorts"
332
+ r=requests.get(url,headers={**HEADERS,"Accept-Language":"vi,en;q=0.8"},timeout=15)
333
+ if r.status_code!=200:return[]
334
+ html=r.text
335
+ ids=[];items=[]
336
+ for m in re.finditer(r'"videoId":"([A-Za-z0-9_-]{11})"',html):
337
+ vid=m.group(1)
338
+ if vid in ids:continue
339
+ ids.append(vid)
340
+ snip=html[max(0,m.start()-900):m.start()+1600]
341
+ title=""
342
+ mt=re.search(r'"title":\{"runs":\[\{"text":"([^"]+)"',snip)
343
+ if not mt:mt=re.search(r'"accessibilityText":"([^"]+)"',snip)
344
+ if mt:title=html_lib.unescape(mt.group(1)).replace('\n',' ').strip()
345
+ if not title:title="YouTube Short"
346
+ items.append({"title":title,"link":f"https://www.youtube.com/watch?v={vid}","img":f"https://i.ytimg.com/vi/{vid}/hqdefault.jpg","source":"yt","id":vid,"channel":channel})
347
+ if len(items)>=count:break
348
+ return items
349
+ except:return[]
350
+ def scrape_shorts():
351
+ """Stable shorts feed: fast HTML scrape + static fallback so slide never disappears."""
352
+ vids=[]
353
+ with ThreadPoolExecutor(2) as ex:
354
+ futs=[ex.submit(_yt_channel_shorts,ch,24) for ch in ["baodantri7941","baosuckhoedoisongboyte","vtvnambo"]]
355
+ for f in as_completed(futs):
356
+ try:
357
+ r=f.result()
358
+ if r:vids.extend(r)
359
+ except:pass
360
+ merged=[];seen=set()
361
+ for v in vids+SHORTS_FALLBACK:
362
+ vid=v.get("id")
363
+ if not vid or vid in seen:continue
364
+ seen.add(vid);merged.append(v)
365
+ return merged[:40]
366
 
367
  # ===== LIVESCORE =====
368
  @app.get("/api/livescore/live")
369
+ def api_livescore_live():return JSONResponse({"html":_cached("ls_live",lambda:fetch_bongda_api("/api/fixtures/live"),ttl=_cache_ttl_live)})
370
  @app.get("/api/livescore/incoming")
371
+ def api_livescore_incoming():return JSONResponse({"html":_cached("ls_incoming",lambda:fetch_bongda_api("/api/fixtures/incoming"),ttl=_cache_ttl_live)})
372
  @app.get("/api/livescore/today")
373
  def api_livescore_today():
374
  today=datetime.now(VN_TZ).strftime("%Y-%m-%d");return JSONResponse({"html":_cached("ls_today",lambda:fetch_bongda_api(f"/api/fixtures/get-by-date?date={today}"),ttl=_cache_ttl)})
 
385
  @app.get("/api/match/{event_id}/stats")
386
  def api_match_stats(event_id:int):return JSONResponse({"html":fetch_bongda_api(f"/api/event-standing/player-performance?event_id={event_id}")})
387
 
388
+ # ===== MATCH DETAIL (server-side scrape from bongda.com.vn) =====
389
  from match_detail_v2 import fetch_match_detail, fetch_match_detail_by_url
390
 
391
  @app.get("/api/match/{event_id}/detail")
392
  def api_match_detail(event_id: int, url: str = Query(default="")):
393
+ """Get full match detail by scraping bongda.com.vn server-side."""
394
  try:
395
+ if url:
396
+ data = fetch_match_detail_by_url(url)
397
+ else:
398
+ data = fetch_match_detail(event_id)
399
  return JSONResponse(data)
400
+ except Exception as e:
401
+ return JSONResponse({"event_id": event_id, "found": False, "error": str(e)})
402
 
403
  @app.get("/api/livescore/featured")
404
  def api_livescore_featured():
 
421
  return None
422
  return JSONResponse(_cached("ls_featured",_f,ttl=30))
423
 
424
+ # ===== VIDEO APIs =====
425
+ @app.get("/api/shorts")
426
+ def api_shorts():return JSONResponse(_cached("yt_shorts_v3",scrape_shorts,ttl=_cache_ttl_yt))
427
+ @app.get("/api/short-stats")
428
+ def api_short_stats(ids:str=Query(default="")):
429
+ arr=[x for x in ids.split(",") if x]
430
+ with _short_lock:
431
+ db=_load_short_db();out={}
432
+ for vid in arr:
433
+ st=db.get(vid) or _short_default()
434
+ out[vid]={"views":int(st.get("views",0)),"likes":int(st.get("likes",0)),"shares":int(st.get("shares",0)),"comments":st.get("comments",[])[:80]}
435
+ return JSONResponse({"stats":out})
436
+
437
+ @app.post("/api/short-action")
438
+ async def api_short_action(request:Request):
439
+ try:body=await request.json()
440
+ except:body={}
441
+ vid=str(body.get("id","")).strip();action=str(body.get("action","")).strip();txt=str(body.get("text","")).strip()
442
+ if not vid:return JSONResponse({"error":"missing id"},status_code=400)
443
+ with _short_lock:
444
+ db=_load_short_db();st=db.get(vid) or _short_default()
445
+ if action=="view":st["views"]=int(st.get("views",0))+1
446
+ elif action=="like":st["likes"]=int(st.get("likes",0))+1
447
+ elif action=="share":st["shares"]=int(st.get("shares",0))+1
448
+ elif action=="comment" and txt:
449
+ comments=st.get("comments",[])
450
+ comments.insert(0,{"text":txt[:180],"ts":int(time.time())})
451
+ st["comments"]=comments[:80]
452
+ st["updated"]=int(time.time());db[vid]=st;_save_short_db(db)
453
+ out={"views":int(st.get("views",0)),"likes":int(st.get("likes",0)),"shares":int(st.get("shares",0)),"comments":st.get("comments",[])[:80]}
454
+ return JSONResponse({"stats":out})
455
+
456
  @app.get("/api/highlights")
457
+ def api_highlights():return JSONResponse(_cached("xemlaibongda_hl",scrape_xemlaibongda,ttl=_cache_ttl))
458
  @app.get("/api/highlights/leagues")
459
+ def api_highlights_leagues():return JSONResponse(_cached("hl_leagues",scrape_all_league_highlights,ttl=_cache_ttl))
460
  @app.get("/api/highlights/{league}")
461
  def api_highlights_league(league:str):
462
+ if league not in HL_LEAGUES:return JSONResponse({"error":"league not found"})
463
  return JSONResponse(_cached(f"hl_{league}",lambda:scrape_highlights_by_league(league),ttl=_cache_ttl))
464
+ @app.get("/api/highlights_config")
465
+ def api_highlights_config():return JSONResponse(HL_LEAGUES)
466
  @app.get("/api/video_url")
467
+ def api_video_url(url:str=Query(...)):
468
  if "youtube.com" in url or "youtu.be" in url:
469
  m=re.search(r'(?:v=|shorts/|youtu\.be/)([a-zA-Z0-9_-]{11})',url)
470
+ if m:vid=m.group(1);return JSONResponse({"src":f"https://www.youtube.com/embed/{vid}?autoplay=1&rel=0&enablejsapi=1","poster":f"https://i.ytimg.com/vi/{vid}/hqdefault.jpg","type":"youtube"})
471
  if "xemlaibongda.top" in url:
472
  v=extract_xemlaibongda_video(url)
473
  if v:
474
+ if v["type"]=="hls":v["src"]="/api/proxy/m3u8?url="+quote(v["src"],safe="")
 
475
  return JSONResponse(v)
476
+ if "bongdaplus.vn" in url:
477
+ try:
478
+ m=re.search(r'-(\d{6,})\.html',url)
479
+ if m:
480
+ r=requests.get(f"{BASE_BDP}/video-embed/{m.group(1)}.html",headers=HEADERS,timeout=10);r.encoding="utf-8"
481
+ soup=BeautifulSoup(r.text,"lxml");video=soup.select_one("video#videoPlayer")
482
+ if video:
483
+ source=video.find("source");src=source.get("src","") if source else "";poster=video.get("poster","")
484
+ if src:return JSONResponse({"src":"/api/proxy/video?url="+quote(src,safe=""),"poster":poster,"type":"video"})
485
+ except:pass
486
  return JSONResponse({"error":"not found"})
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
487
  @app.get("/api/bdp_videos")
488
  def api_bdp_videos():
489
  def _f():
490
  try:
491
+ soup=_get(f"{BASE_BDP}/video");arts=[];seen=set()
492
  for a in soup.find_all("a",href=True):
493
  href=a.get("href","")
494
  if"/video/" not in href or href in("/video/","/video/ban-thang-dep","/video/highlight"):continue
495
+ if not href.startswith("http"):href=BASE_BDP+href
496
+ if href in seen:continue
497
  title=re.sub(r'^\d{2}:\d{2}','',a.get_text(strip=True)).strip()
498
+ if not title or len(title)<5:continue
499
  img_tag=a.find("img") or(a.parent.find("img") if a.parent else None)
500
  img=(img_tag.get("data-src") or img_tag.get("src","")) if img_tag else ""
501
+ seen.add(href);arts.append({"title":title,"link":href,"img":img,"source":"bdp"})
502
  return arts[:20]
503
+ except:return[]
504
  return JSONResponse(_cached("bdp_videos",_f))
 
505
  # ===== NEWS =====
 
 
506
  def scrape_vne(cat_url):
507
  try:
508
+ soup=_get(cat_url);arts=[]
509
  for it in soup.select("article.item-news")[:15]:
510
  a=it.select_one("h2.title-news a") or it.select_one("h3.title-news a")
511
+ if not a:continue
512
+ t=a.get("title","") or a.get_text(strip=True);lk=a.get("href","")
513
+ if not t or not lk:continue
514
+ im=it.find("img");img=(im.get("data-src") or im.get("src","")) if im else ""
515
+ if img and'blank'in img:
516
  src=it.find("source")
517
+ if src:img=src.get("srcset","").split(",")[0].strip().split(" ")[0]
518
  arts.append({"title":t,"link":lk,"img":img,"source":"vne"})
519
  return arts
520
+ except:return[]
521
+ def scrape_vne_article(url):
522
+ try:
523
+ soup=_get(url);h1=soup.select_one("h1.title-detail");desc=soup.select_one("p.description")
524
+ og=soup.find("meta",property="og:image");og_img=og.get("content","") if og else ""
525
+ cd=soup.select_one("article.fck_detail");body=[]
526
+ if cd:
527
+ for ch in cd.children:
528
+ if not hasattr(ch,'name') or not ch.name:continue
529
+ if ch.name=="p":t=ch.get_text(strip=True);(body.append({"type":"p","text":t}) if t else None)
530
+ elif ch.name=="figure":
531
+ im=ch.find("img")
532
+ if im:s=im.get("data-src") or im.get("src","");body.append({"type":"img","src":s})
533
+ elif ch.name in("h2","h3"):body.append({"type":"heading","text":ch.get_text(strip=True)})
534
+ return{"title":h1.get_text(strip=True) if h1 else "","summary":desc.get_text(strip=True) if desc else "","og_image":og_img,"body":body,"source":"vne","url":url}
535
+ except:return None
536
+ def _scrape_dantri_homepage(cat_filter=None):
537
+ try:
538
+ soup=_get("https://dantri.com.vn/");arts=[];seen=set()
539
+ for a in soup.find_all("a",href=True):
540
+ href=a.get("href","");title=a.get("title","") or a.get_text(strip=True)
541
+ if not title or len(title)<15 or"javascript:" in href:continue
542
+ if not href.startswith("http"):href="https://dantri.com.vn"+href
543
+ if href in seen or not href.endswith(".htm"):continue
544
+ if cat_filter and f"/{cat_filter}/" not in href:continue
545
+ img_tag=a.find("img")
546
+ if not img_tag and a.parent:img_tag=a.parent.find("img")
547
+ img_src=""
548
+ if img_tag:img_src=img_tag.get("data-src","") or img_tag.get("src","")
549
+ if not img_src or "cdn" not in img_src:continue
550
+ proxied_img="/api/proxy/img?url="+quote(img_src,safe="")
551
+ seen.add(href);arts.append({"title":title,"link":href,"img":proxied_img,"source":"dantri"})
552
+ if len(arts)>=15:break
553
+ return arts
554
+ except:return[]
555
+ def scrape_dantri_hot():return _scrape_dantri_homepage()
556
+ def scrape_dantri_congnghe():
557
+ try:
558
+ soup=_get("https://dantri.com.vn/");arts=[];seen=set()
559
+ for a in soup.find_all("a",href=True):
560
+ href=a.get("href","");title=a.get("title","") or a.get_text(strip=True)
561
+ if not title or len(title)<15 or"javascript:" in href:continue
562
+ if not href.startswith("http"):href="https://dantri.com.vn"+href
563
+ if href in seen or not href.endswith(".htm"):continue
564
+ if"/cong-nghe/" not in href:continue
565
+ img_tag=a.find("img")
566
+ if not img_tag and a.parent:img_tag=a.parent.find("img")
567
+ img_src=""
568
+ if img_tag:img_src=img_tag.get("data-src","") or img_tag.get("src","")
569
+ if img_src and "cdn" in img_src:img_src="/api/proxy/img?url="+quote(img_src,safe="")
570
+ else:img_src=""
571
+ seen.add(href);arts.append({"title":title,"link":href,"img":img_src,"source":"dantri"})
572
+ if len(arts)>=15:break
573
+ return arts
574
+ except:return[]
575
  def scrape_genk_ai():
576
+ """Scrape AI articles from genk.vn - readable in-app"""
577
  try:
578
  r=requests.get("https://genk.vn/ai.chn",headers=HEADERS,timeout=15)
579
+ if r.status_code!=200:return[]
580
+ r.encoding="utf-8";soup=BeautifulSoup(r.text,"lxml")
581
+ articles=[];seen=set()
582
  for a in soup.find_all("a",href=True):
583
  href=a.get("href","")
584
+ if not href.endswith(".chn") or href=="/ai.chn":continue
585
+ if href.startswith("/"):href="https://genk.vn"+href
586
+ if href in seen or "genk.vn" not in href:continue
587
  title=a.get("title","") or a.get_text(strip=True)
588
+ if not title or len(title)<20:continue
589
+ container=a.parent;img_src=""
590
  for _ in range(6):
591
+ if container is None:break
592
  for img in container.find_all("img"):
593
  s=img.get("data-src","") or img.get("src","")
594
+ if s and "mediacdn" in s and "avatar" not in s and "logo" not in s:
595
+ img_src=s;break
596
+ if img_src:break
597
+ container=container.parent
598
  seen.add(href)
599
  if not img_src:
600
  try:
601
+ og_r=requests.get(href,headers=HEADERS,timeout=8);og_r.encoding="utf-8"
602
+ og_soup=BeautifulSoup(og_r.text,"lxml");og_tag=og_soup.find("meta",property="og:image")
603
+ if og_tag:img_src=og_tag.get("content","")
604
+ except:pass
605
  articles.append({"title":title,"link":href,"img":img_src,"source":"genk"})
606
+ if len(articles)>=30:break
607
  return articles
608
+ except:return[]
609
+
610
+ def scrape_dantri_article(url):
611
+ try:
612
+ r=requests.get(url,headers=HEADERS,timeout=15);r.encoding="utf-8";soup=BeautifulSoup(r.text,"lxml")
613
+ for tag in soup.find_all(["script","style","nav","footer","aside"]):tag.decompose()
614
+ h1=soup.find("h1");og=soup.find("meta",property="og:image");og_img=og.get("content","") if og else ""
615
+ if og_img and "cdnphoto.dantri" in og_img:og_img="/api/proxy/img?url="+quote(og_img,safe="")
616
+ content=soup.select_one("main") or soup.select_one("div.singular-content") or soup.select_one("article");body=[]
617
+ if content:
618
+ for el in content.find_all(["p","h2","h3","figure","img"],recursive=True):
619
+ if el.name=="p":t=el.get_text(strip=True);(body.append({"type":"p","text":t}) if t and len(t)>15 else None)
620
+ elif el.name in("h2","h3"):t=el.get_text(strip=True);(body.append({"type":"heading","text":t}) if t else None)
621
+ elif el.name in("figure","img"):
622
+ im=el if el.name=="img" else el.find("img")
623
+ if im:
624
+ s=im.get("data-src") or im.get("src","")
625
+ if s and"base64" not in s:
626
+ if "cdnphoto.dantri" in s:s="/api/proxy/img?url="+quote(s,safe="")
627
+ body.append({"type":"img","src":s})
628
+ desc="";sapo=soup.select_one("h2.singular-sapo") or soup.select_one("h2[class*=sapo]")
629
+ if not sapo:
630
+ og_desc=soup.find("meta",property="og:description")
631
+ if og_desc:desc=og_desc.get("content","")
632
+ else:desc=sapo.get_text(strip=True)
633
+ return{"title":h1.get_text(strip=True) if h1 else "","summary":desc,"og_image":og_img,"body":body,"source":"dantri","url":url}
634
+ except:return None
635
+ def scrape_bbc_vietnamese():
636
+ try:
637
+ r=requests.get("https://www.bbc.com/vietnamese",headers={"User-Agent":"Mozilla/5.0","Accept-Language":"en-GB"},timeout=15);r.encoding="utf-8"
638
+ soup=BeautifulSoup(r.text,"lxml");arts=[];seen=set()
639
+ for a in soup.select("a[href*='/vietnamese/']"):
640
+ href=a.get("href","")
641
+ if not href or href=="/vietnamese" or href.count("/")<3:continue
642
+ if not href.startswith("http"):href="https://www.bbc.com"+href
643
+ if href in seen:continue
644
+ title=a.get_text(strip=True)
645
+ if not title or len(title)<15 or any(x in title.lower() for x in["đăng nhập","trang chủ","bbc news"]):continue
646
+ img="";container=a.parent
647
+ for _ in range(3):
648
+ if container:
649
+ im=container.find("img")
650
+ if im:img=im.get("src","") or im.get("data-src","");break
651
+ container=container.parent
652
+ seen.add(href);arts.append({"title":title,"link":href,"img":img,"source":"bbc"})
653
+ if len(arts)>=15:break
654
+ return arts
655
+ except:return[]
656
+ def scrape_bbc_article(url):
657
+ try:
658
+ r=requests.get(url,headers={"User-Agent":"Mozilla/5.0","Accept-Language":"en-GB"},timeout=15);r.encoding="utf-8"
659
+ soup=BeautifulSoup(r.text,"lxml");h1=soup.find("h1")
660
+ og=soup.find("meta",property="og:image");og_img=og.get("content","") if og else ""
661
+ body=[]
662
+ for p in soup.select("[data-component='text-block'] p, article p, main p"):
663
+ t=p.get_text(strip=True)
664
+ if t and len(t)>20:body.append({"type":"p","text":t})
665
+ return{"title":h1.get_text(strip=True) if h1 else "","summary":"","og_image":og_img,"body":body,"source":"bbc","url":url}
666
+ except:return None
667
+
668
+ def scrape_ttvh_worldcup():
669
+ """Scrape all World Cup 2026 articles from The Thao Van Hoa RSS."""
670
+ try:
671
+ r=requests.get("https://thethaovanhoa.vn/rss/world-cup-2026.rss",headers=HEADERS,timeout=15);r.encoding="utf-8"
672
+ soup=BeautifulSoup(r.text,"xml");arts=[];seen=set()
673
+ for it in soup.find_all("item"):
674
+ title=(it.find("title").get_text(strip=True) if it.find("title") else "")
675
+ link=(it.find("link").get_text(strip=True) if it.find("link") else "")
676
+ desc=(it.find("description").get_text(" ",strip=True) if it.find("description") else "")
677
+ img="";ds=BeautifulSoup(desc,"lxml");im=ds.find("img")
678
+ if im:img=im.get("src","") or im.get("data-src","")
679
+ if title and link and link not in seen:
680
+ seen.add(link);arts.append({"title":title,"link":link,"img":img,"source":"ttvh"})
681
+ if arts:return arts
682
+ except:pass
683
+ try:
684
+ soup=_get("https://thethaovanhoa.vn/world-cup-2026.htm");arts=[];seen=set()
685
+ for a in soup.find_all("a",href=True):
686
+ href=a.get("href","")
687
+ if not href.startswith("http"):href="https://thethaovanhoa.vn"+href
688
+ if href in seen or "thethaovanhoa.vn" not in href:continue
689
+ if not re.search(r"/[^/]+-\d{8,}\.htm",href):continue
690
+ title=a.get("title","") or a.get_text(" ",strip=True)
691
+ img=None;p=a
692
+ for _ in range(5):
693
+ if p is None:break
694
+ img=p.find("img")
695
+ if img:break
696
+ p=p.parent
697
+ img_src=""
698
+ if img:
699
+ img_src=img.get("data-src","") or img.get("src","") or img.get("data-original","") or img.get("data-thumb","")
700
+ if len(title)<15:title=img.get("alt","") or img.get("title","") or title
701
+ if not title or len(title)<15:continue
702
+ seen.add(href);arts.append({"title":title,"link":href,"img":img_src,"source":"ttvh"})
703
+ if len(arts)>=24:break
704
+ return arts
705
+ except:return[]
706
 
707
+ def scrape_ttvh_article(url):
708
+ try:
709
+ soup=_get(url);h1=soup.find("h1");og=soup.find("meta",property="og:image");og_img=og.get("content","") if og else ""
710
+ og_title=soup.find("meta",property="og:title");fallback_title=og_title.get("content","") if og_title else ""
711
+ desc_el=soup.find("meta",property="og:description");desc=desc_el.get("content","") if desc_el else ""
712
+ cd=soup.select_one(".detail-content") or soup.select_one(".content-detail") or soup.select_one("article") or soup.select_one("main")
713
+ body=[]
714
+ if cd:
715
+ for el in cd.find_all(["p","h2","h3","figure","img"],recursive=True):
716
+ if el.name=="p":
717
+ t=el.get_text(strip=True)
718
+ if t and len(t)>20 and "Theo dõi" not in t:body.append({"type":"p","text":t})
719
+ elif el.name in ("h2","h3"):
720
+ t=el.get_text(strip=True)
721
+ if t:body.append({"type":"heading","text":t})
722
+ elif el.name in ("figure","img"):
723
+ im=el if el.name=="img" else el.find("img")
724
+ if im:
725
+ src=im.get("data-src") or im.get("src","") or im.get("data-original","")
726
+ if src and "base64" not in src:body.append({"type":"img","src":src})
727
+ if not body and desc:body=[{"type":"p","text":desc}]
728
+ return {"title":h1.get_text(strip=True) if h1 else fallback_title,"summary":desc,"og_image":og_img,"body":body,"source":"ttvh","url":url}
729
+ except:return None
730
+
731
+ VNE_CATS={"thoi-su":("https://vnexpress.net/thoi-su","Thời Sự"),"the-gioi":("https://vnexpress.net/the-gioi","Thế Giới"),"kinh-doanh":("https://vnexpress.net/kinh-doanh","Kinh Doanh"),"the-thao":("https://vnexpress.net/the-thao","Thể Thao"),"giai-tri":("https://vnexpress.net/giai-tri","Giải Trí"),"suc-khoe":("https://vnexpress.net/suc-khoe","Sức Khỏe"),"phap-luat":("https://vnexpress.net/phap-luat","Pháp Luật"),"giao-duc":("https://vnexpress.net/giao-duc","Giáo Dục"),"du-lich":("https://vnexpress.net/du-lich","Du Lịch"),"doi-song":("https://vnexpress.net/doi-song","Đời Sống")}
732
  @app.get("/api/homepage")
733
  def api_homepage():
734
  def _f():
735
  articles=[]
736
  with ThreadPoolExecutor(12) as ex:
737
  futs={ex.submit(scrape_vne,VNE_CATS[k][0]):VNE_CATS[k][1] for k in["thoi-su","the-gioi","kinh-doanh","the-thao","giai-tri","phap-luat","giao-duc","du-lich","doi-song"]}
738
+ futs[ex.submit(scrape_bbc_vietnamese)]="BBC"
739
  for f in as_completed(futs):
740
  try:
741
+ for a in f.result():a["group"]=futs[f];articles.append(a)
742
+ except:pass
743
  return articles
744
  return JSONResponse(_cached("homepage",_f))
 
745
  @app.get("/api/category/{cat_id}")
746
  def api_category(cat_id:str):
747
  def _f():
748
+ if cat_id=="bbc":return scrape_bbc_vietnamese()
749
+ if cat_id=="cong-nghe":return scrape_genk_ai()
750
+ if cat_id in VNE_CATS:arts=scrape_vne(VNE_CATS[cat_id][0]);[a.update({"group":VNE_CATS[cat_id][1]}) for a in arts];return arts
751
+ return[]
 
 
752
  return JSONResponse(_cached(f"cat_{cat_id}",_f))
 
753
  @app.get("/api/categories")
754
  def api_categories():
755
+ cats=[{"id":"bbc","name":"BBC Tiếng Việt","source":"bbc"},{"id":"cong-nghe","name":"Công Nghệ","source":"genk"}]
756
+ for k,(u,n) in VNE_CATS.items():cats.append({"id":k,"name":n,"source":"vne"})
757
  return JSONResponse(cats)
758
+ @app.get("/api/dantri_hot")
759
+ def api_dantri_hot():return JSONResponse(_cached("dantri_hot",scrape_dantri_hot))
760
+ @app.get("/api/genk_ai")
761
+ def api_genk_ai():return JSONResponse(_cached("genk_ai",scrape_genk_ai,ttl=_cache_ttl))
762
+ @app.get("/api/worldcup2026")
763
+ def api_worldcup2026():return JSONResponse(_cached("ttvh_worldcup",scrape_ttvh_worldcup,ttl=_cache_ttl))
764
+ def scrape_genk_article(url):
765
  try:
766
+ r=requests.get(url,headers=HEADERS,timeout=15);r.encoding="utf-8";soup=BeautifulSoup(r.text,"lxml")
767
+ h1=soup.find("h1");og=soup.find("meta",property="og:image");og_img=og.get("content","") if og else ""
768
+ og_title=soup.find("meta",property="og:title");fallback_title=og_title.get("content","") if og_title else ""
769
+ desc_el=soup.find("meta",property="og:description");desc=desc_el.get("content","") if desc_el else ""
770
+ cd=soup.select_one(".knc-content");body=[]
771
+ if cd:
772
+ for el in cd.find_all(["p","h2","h3","figure","img"],recursive=True):
773
+ if el.name=="p":t=el.get_text(strip=True);(body.append({"type":"p","text":t}) if t and len(t)>15 else None)
774
+ elif el.name in("h2","h3"):t=el.get_text(strip=True);(body.append({"type":"heading","text":t}) if t else None)
775
+ elif el.name in("figure","img"):
776
+ im=el if el.name=="img" else el.find("img")
777
+ if im:s=im.get("data-src") or im.get("src","");(body.append({"type":"img","src":s}) if s and"base64" not in s else None)
778
+ return{"title":h1.get_text(strip=True) if h1 else "","summary":desc,"og_image":og_img,"body":body,"source":"genk","url":url}
779
+ except:return None
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
780
 
781
  @app.get("/api/article")
782
  def api_article(url:str=Query(...)):
783
+ if"vnexpress.net" in url:data=scrape_vne_article(url)
784
+ elif"bbc.com" in url:data=scrape_bbc_article(url)
785
+ elif"dantri.com.vn" in url:data=scrape_dantri_article(url)
786
+ elif"genk.vn" in url:data=scrape_genk_article(url)
787
+ elif"thethaovanhoa.vn" in url:data=scrape_ttvh_article(url)
788
+ else:data=None
789
+ return JSONResponse(data if data else{"error":"not supported"})
790
+ def _web_context(topic):
791
+ """Collect real web/news context for a topic."""
792
+ bits=[]
793
+ try:
794
+ rss="https://news.google.com/rss/search?q="+quote(topic)+"&hl=vi&gl=VN&ceid=VN:vi"
795
+ r=requests.get(rss,headers=HEADERS,timeout=12);r.encoding="utf-8"
796
+ soup=BeautifulSoup(r.text,"xml")
797
+ for it in soup.find_all("item")[:8]:
798
+ title=it.find("title").get_text(" ",strip=True) if it.find("title") else ""
799
+ src=it.find("source").get_text(" ",strip=True) if it.find("source") else ""
800
+ if title:bits.append((title+(" — "+src if src else ""))[:280])
801
+ except:pass
802
+ if bits:return "\n".join(bits)
803
  try:
804
+ r=requests.get("https://html.duckduckgo.com/html/?q="+quote(topic),headers=HEADERS,timeout=12);r.encoding="utf-8"
805
+ soup=BeautifulSoup(r.text,"lxml")
806
+ for res in soup.select(".result")[:6]:
807
+ t=res.select_one(".result__title");sn=res.select_one(".result__snippet")
808
+ line=((t.get_text(" ",strip=True) if t else "")+" — "+(sn.get_text(" ",strip=True) if sn else "")).strip(" —")
809
+ if line:bits.append(line[:280])
810
+ except:pass
811
+ return "\n".join(bits)
812
+
813
+ def _jina_read(url):
814
+ try:
815
+ ju="https://r.jina.ai/http://"+url
816
+ r=requests.get(ju,headers=HEADERS,timeout=25);r.encoding="utf-8"
817
+ if r.status_code!=200 or not r.text:return None
818
+ lines=[x.rstrip() for x in r.text.splitlines()]
819
+ title="";img="";body=[];summary=""
820
+ for ln in lines[:40]:
821
+ if ln.startswith("Title:"):title=ln.replace("Title:","",1).strip()
822
+ elif ln.startswith("Image:"):img=ln.replace("Image:","",1).strip()
823
+ elif ln.startswith("Description:"):summary=ln.replace("Description:","",1).strip()
824
+ for ln in lines:
825
+ t=ln.strip()
826
+ if not t or t.startswith(("Title:","URL Source:","Published Time:","Markdown Content:","Image:","Description:")):continue
827
+ if len(t)>40:body.append({"type":"p","text":t})
828
+ if not body and summary:body=[{"type":"p","text":summary}]
829
+ return {"title":title or url,"summary":summary,"og_image":img,"body":body[:80],"source":"jina","url":url}
830
+ except:return None
831
+
832
+ def _scrape_generic_article(url):
833
+ try:
834
+ hdr={**HEADERS,"Accept":"text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8"}
835
+ r=requests.get(url,headers=hdr,timeout=15);r.encoding="utf-8"
836
+ ct=r.headers.get("content-type","").lower()
837
+ if r.status_code>=400 or "text/html" not in ct:
838
+ jr=_jina_read(url)
839
+ if jr:return jr
840
+ soup=BeautifulSoup(r.text,"lxml")
841
+ for tag in soup.find_all(["script","style","nav","footer","aside","form"]):tag.decompose()
842
+ h1=soup.find("h1")
843
+ ogt=soup.find("meta",property="og:title");title=h1.get_text(strip=True) if h1 else (ogt.get("content","") if ogt else "")
844
+ ogd=soup.find("meta",property="og:description");desc=ogd.get("content","") if ogd else ""
845
+ ogi=soup.find("meta",property="og:image");img=ogi.get("content","") if ogi else ""
846
+ main=soup.find("article") or soup.find("main") or soup.body
847
+ body=[]
848
+ if main:
849
+ for el in main.find_all(["p","h2","h3","figure","img"],recursive=True):
850
+ if el.name=="p":
851
+ t=el.get_text(" ",strip=True)
852
+ if t and len(t)>35:body.append({"type":"p","text":t})
853
+ elif el.name in ("h2","h3"):
854
+ t=el.get_text(" ",strip=True)
855
+ if t:body.append({"type":"heading","text":t})
856
+ elif el.name in ("figure","img"):
857
+ im=el if el.name=="img" else el.find("img")
858
+ if im:
859
+ src=im.get("data-src") or im.get("src","") or im.get("data-original","")
860
+ if src and "base64" not in src:body.append({"type":"img","src":src})
861
+ if not body:
862
+ jr=_jina_read(url)
863
+ if jr and jr.get("body"):return jr
864
+ if not body and desc:body=[{"type":"p","text":desc}]
865
+ return {"title":title or url,"summary":desc,"og_image":img,"body":body,"source":"generic","url":url}
866
+ except:
867
+ return _jina_read(url)
868
+
869
+ def _article_by_url(url):
870
+ if "vnexpress.net" in url:return scrape_vne_article(url)
871
+ if "bbc.com" in url:return scrape_bbc_article(url)
872
+ if "dantri.com.vn" in url:return scrape_dantri_article(url)
873
+ if "genk.vn" in url:return scrape_genk_article(url)
874
+ if "thethaovanhoa.vn" in url:return scrape_ttvh_article(url)
875
+ return _scrape_generic_article(url)
876
+
877
+ def _call_qwen(prompt, max_tokens=1800):
878
+ """Try Qwen2.5-VL via HF router; return None if unavailable."""
879
+ try:
880
+ token=os.environ.get("HF_TOKEN") or os.environ.get("HUGGINGFACEHUB_API_TOKEN") or os.environ.get("VAISTUDIO")
881
+ if not token:return None
882
+ headers={"Authorization":"Bearer "+token,"Content-Type":"application/json"}
883
+ payload={"model":"Qwen/Qwen2.5-VL-7B-Instruct","messages":[{"role":"user","content":prompt}],"max_tokens":max_tokens,"temperature":0.7}
884
+ r=requests.post("https://router.huggingface.co/v1/chat/completions",headers=headers,json=payload,timeout=75)
885
+ if r.status_code>=300:return None
886
+ j=r.json();return j.get("choices",[{}])[0].get("message",{}).get("content")
887
+ except:return None
888
+
889
+ def _collect_article_text(data, limit=28000):
890
+ title=(data or {}).get("title","");summary=(data or {}).get("summary","")
891
+ parts=[]
892
+ if summary:parts.append(summary)
893
+ for b in (data or {}).get("body",[]):
894
+ if b.get("type")=="heading":parts.append("## "+b.get("text","") )
895
+ elif b.get("type")=="p":parts.append(b.get("text","") )
896
+ text="\n".join([p.strip() for p in parts if p and p.strip()])
897
+ return title,text[:limit]
898
+
899
+ def _ai_rewrite_article(data,tone="tu-nhien"):
900
+ title,text=_collect_article_text(data)
901
+ prompt=("Bạn là biên tập viên báo điện tử tiếng Việt. Hãy viết lại bài dưới đây bằng ngôn ngữ tự nhiên, mạch lạc, không cắt khúc, không bỏ ý quan trọng. "
902
+ "Giữ đúng sự thật, không bịa, không thêm thông tin ngoài bài. Văn phong: "+tone+". "
903
+ "Đầu ra gồm: tiêu đề hấp dẫn, đoạn sapo 2-3 câu, các đoạn nội dung ngắn dễ đọc, và 3 gạch đầu dòng điểm chính.\n\n"
904
+ "TIÊU ĐỀ GỐC: "+title+"\n\nNỘI DUNG GỐC:\n"+text)
905
+ out=_call_qwen(prompt,2200)
906
+ if out and len(out)>300:return out.strip()
907
+ paras=[p.strip() for p in text.split("\n") if len(p.strip())>30]
908
+ body="\n\n".join(paras[:18])
909
+ bullets="\n".join(["• "+p[:220]+("..." if len(p)>220 else "") for p in paras[:5]])
910
+ return ("Bản tin AI viết lại: "+title+"\n\n"+
911
+ (paras[0] if paras else "")+"\n\n"+body+"\n\nĐiểm chính:\n"+bullets).strip()
912
+
913
+ def _image_for_topic(topic):
914
+ return "https://image.pollinations.ai/prompt/"+quote("editorial illustration, Vietnamese news, "+topic,safe="")+"?width=1024&height=576&nologo=true"
915
+
916
+ def _topic_articles(topic,limit=5):
917
+ items=[];seen=set()
918
+ try:
919
+ rss="https://news.google.com/rss/search?q="+quote(topic)+"&hl=vi&gl=VN&ceid=VN:vi"
920
+ r=requests.get(rss,headers=HEADERS,timeout=12);r.encoding="utf-8"
921
+ soup=BeautifulSoup(r.text,"xml")
922
+ for it in soup.find_all("item")[:limit*3]:
923
+ title=it.find("title").get_text(" ",strip=True) if it.find("title") else ""
924
+ link=it.find("link").get_text(strip=True) if it.find("link") else ""
925
+ src=it.find("source").get_text(" ",strip=True) if it.find("source") else ""
926
+ if not title or not link or link in seen:continue
927
+ seen.add(link);items.append({"title":title,"link":link,"source":src})
928
+ if len(items)>=limit:break
929
+ except:pass
930
+ return items
931
+
932
+ def _topic_article_context(topic):
933
+ """Filter readable article sources by topic, then summarize actual article bodies."""
934
+ raw_keys=[k.lower() for k in re.findall(r"[\wÀ-ỹ]+",topic) if len(k)>2]
935
+ stop={"trong","năm","the","and","của","cho","với","một","các","những","hiện","nay"}
936
+ keys=[k for k in raw_keys if k not in stop]
937
+ candidates=[];seen=set()
938
+ def add_items(items):
939
+ for a in items or []:
940
+ link=a.get("link","");title=a.get("title","")
941
+ if not link or link in seen:continue
942
+ seen.add(link);candidates.append(a)
943
+ try:add_items(scrape_genk_ai())
944
+ except:pass
945
+ try:add_items(scrape_dantri_congnghe())
946
+ except:pass
947
+ if not candidates:
948
+ try:items=_scrape_dantri_homepage();add_items(items)
949
+ except:pass
950
+ candidates=[a for a in candidates if any(k in a.get("title","").lower() for k in keys)]
951
+ if not candidates:candidates=[a for a in (items or []) if any(k in a.get("title","").lower() for k in keys)]
952
+ texts=[]
953
+ for a in candidates[:8]:
954
+ try:
955
+ data=_article_by_url(a["link"])
956
+ t=data.get("title","") if data else ""
957
+ b="\n".join(x.get("text","") for x in (data or {}).get("body",[]) if x.get("type")=="p")
958
+ if t and b:texts.append(f"[{t}]\n{b[:1200]}")
959
+ except:pass
960
+ return "\n\n".join(texts)
961
+
962
+ @app.get("/api/topic_post")
963
+ async def api_topic_post(request:Request):
964
+ try:data=await request.json()
965
+ except:data={}
966
+ topic=str(data.get("topic","")).strip()
967
+ if not topic:return JSONResponse({"error":"missing topic"},status_code=400)
968
+ context=_topic_article_context(topic)+"\n\n"+_web_context(topic)
piped_client.py DELETED
@@ -1,258 +0,0 @@
1
- """
2
- YouTube Shorts Scraper using Piped API
3
- Piped is a privacy-friendly YouTube proxy that works without JS
4
- """
5
- import requests
6
- import json
7
- import time
8
- import threading
9
-
10
- _cache = {}
11
- _lock = threading.Lock()
12
- CACHE_TTL = 900 # 15 min
13
-
14
- # Piped API instances (public)
15
- PIPED_INSTANCES = [
16
- "https://pipedapi.kavin.rocks",
17
- "https://pipedapi.adminforge.de",
18
- "https://api.piped.projectsegfau.lt",
19
- ]
20
-
21
- UA = {"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36"}
22
-
23
- def _cached(key):
24
- with _lock:
25
- if key in _cache and time.time() - _cache[key]['t'] < CACHE_TTL:
26
- return _cache[key]['d']
27
- return None
28
-
29
- def _set_cache(key, data):
30
- with _lock:
31
- _cache[key] = {'t': time.time(), 'd': data}
32
-
33
- def _piped_request(path, params=None):
34
- """Try multiple Piped instances"""
35
- last_err = None
36
- for base in PIPED_INSTANCES:
37
- try:
38
- url = f"{base}{path}"
39
- r = requests.get(url, params=params, headers=UA, timeout=15)
40
- if r.status_code == 200:
41
- return r.json()
42
- except Exception as e:
43
- last_err = e
44
- continue
45
- raise Exception(f"All Piped instances failed: {last_err}")
46
-
47
- def get_channel_videos(channel_id, max_videos=200):
48
- """Get all videos from a channel using Piped API with pagination"""
49
- cached = _cached(f'ch_vids_{channel_id}')
50
- if cached is not None:
51
- return cached
52
-
53
- all_videos = []
54
- page = None
55
-
56
- while len(all_videos) < max_videos:
57
- try:
58
- if page:
59
- data = _piped_request(f"/channels/{channel_id}/videos", {"nextpage": page})
60
- else:
61
- data = _piped_request(f"/channels/{channel_id}/videos")
62
-
63
- videos = data.get('relatedStreams', [])
64
- if not videos:
65
- break
66
-
67
- all_videos.extend(videos)
68
-
69
- # Check for next page
70
- next_page = data.get('nextpage')
71
- if not next_page or next_page == page:
72
- break
73
- page = next_page
74
-
75
- # Small delay to be polite
76
- time.sleep(0.3)
77
-
78
- if len(all_videos) >= max_videos:
79
- break
80
-
81
- except Exception as e:
82
- print(f"Piped pagination error: {e}")
83
- break
84
-
85
- result = all_videos[:max_videos]
86
- _set_cache(f'ch_vids_{channel_id}', result)
87
- return result
88
-
89
- def get_vtvnambo_shorts_piped(max_count=50):
90
- """Get shorts from VTV Nam Bộ using Piped API"""
91
- # VTV Nam Bộ channel ID
92
- channel_id = "UCJ0btJV8qh7J7R2aXb9GmGA"
93
-
94
- try:
95
- videos = get_channel_videos(channel_id, 200)
96
-
97
- shorts = []
98
- for v in videos:
99
- title = v.get('title', '')
100
- vid = v.get('url', '').replace('/watch?v=', '')
101
- if not vid:
102
- continue
103
-
104
- # Filter for shorts: title has #shorts, or duration <= 60s
105
- duration = v.get('duration', 0)
106
- is_short = (
107
- '#shorts' in title.lower() or
108
- '#short' in title.lower() or
109
- (duration > 0 and duration <= 60)
110
- )
111
-
112
- if is_short:
113
- shorts.append({
114
- 'id': vid,
115
- 'title': title,
116
- 'img': f"https://i.ytimg.com/vi/{vid}/hqdefault.jpg",
117
- 'channel': 'vtvnambo',
118
- })
119
-
120
- if shorts:
121
- return shorts[:max_count]
122
- except Exception as e:
123
- print(f"Piped shorts error: {e}")
124
-
125
- return []
126
-
127
- def get_vtvnambo_shorts_rss(max_count=50):
128
- """Get shorts from YouTube RSS feed"""
129
- cached = _cached('vtvnambo_rss')
130
- if cached is not None:
131
- return cached
132
-
133
- from xml.etree import ElementTree as ET
134
-
135
- # First get channel ID from page
136
- channel_id = None
137
- try:
138
- r = requests.get("https://www.youtube.com/@vtvnambo", headers=UA, timeout=15)
139
- if r.status_code == 200:
140
- m = re.search(r'"channelId":"(UC[^"]+)"', r.text)
141
- if m:
142
- channel_id = m.group(1)
143
- except:
144
- pass
145
-
146
- if not channel_id:
147
- channel_id = "UCJ0btJV8qh7J7R2aXb9GmGA" # fallback
148
-
149
- try:
150
- url = f"https://www.youtube.com/feeds/videos.xml?channel_id={channel_id}"
151
- r = requests.get(url, headers=UA, timeout=15)
152
- if r.status_code != 200:
153
- return []
154
-
155
- root = ET.fromstring(r.text)
156
- ns = {'atom': 'http://www.w3.org/2005/Atom', 'yt': 'http://www.youtube.com/xml/schemas/2015'}
157
-
158
- shorts = []
159
- for entry in root.findall('atom:entry', ns)[:max_count * 2]:
160
- title_el = entry.find('atom:title', ns)
161
- title = title_el.text if title_el is not None and title_el.text else ''
162
-
163
- vid_el = entry.find('yt:videoId', ns)
164
- vid = vid_el.text if vid_el is not None else ''
165
- if not vid:
166
- continue
167
-
168
- is_short = '#shorts' in title.lower() or '#short' in title.lower()
169
- link_el = entry.find('atom:link', ns)
170
- link = link_el.get('href', '') if link_el is not None else ''
171
- if '/shorts/' in link:
172
- is_short = True
173
-
174
- if is_short:
175
- shorts.append({
176
- 'id': vid,
177
- 'title': title,
178
- 'img': f"https://i.ytimg.com/vi/{vid}/hqdefault.jpg",
179
- 'channel': 'vtvnambo',
180
- })
181
-
182
- _set_cache('vtvnambo_rss', shorts[:max_count])
183
- return shorts[:max_count]
184
- except Exception as e:
185
- print(f"RSS error: {e}")
186
-
187
- return []
188
-
189
- def get_vtvnambo_shorts(max_count=50):
190
- """Get all shorts from VTV Nam Bộ. Tries Piped API first, then RSS."""
191
- cached = _cached('vtvnambo_shorts_v3')
192
- if cached is not None:
193
- return cached
194
-
195
- all_shorts = []
196
- seen_ids = set()
197
-
198
- # Method 1: Piped API (most reliable)
199
- try:
200
- piped_shorts = get_vtvnambo_shorts_piped(max_count)
201
- for s in piped_shorts:
202
- if s['id'] not in seen_ids:
203
- seen_ids.add(s['id'])
204
- all_shorts.append(s)
205
- print(f"Piped API found {len(piped_shorts)} shorts")
206
- except Exception as e:
207
- print(f"Piped method failed: {e}")
208
-
209
- # Method 2: RSS feed
210
- if len(all_shorts) < 3:
211
- try:
212
- rss_shorts = get_vtvnambo_shorts_rss(max_count)
213
- for s in rss_shorts:
214
- if s['id'] not in seen_ids:
215
- seen_ids.add(s['id'])
216
- all_shorts.append(s)
217
- print(f"RSS found {len(rss_shorts)} shorts")
218
- except Exception as e:
219
- print(f"RSS method failed: {e}")
220
-
221
- result = all_shorts[:max_count]
222
- _set_cache('vtvnambo_shorts_v3', result)
223
- return result
224
-
225
- def get_wc_related_shorts(max_count=30):
226
- """Get World Cup / football related shorts."""
227
- all_shorts = get_vtvnambo_shorts(max_count * 3)
228
-
229
- wc_kws = [
230
- 'world cup', 'wc 2026', 'worldcup', 'fifa', 'bóng đá',
231
- 'trận đấu', 'đội tuyển', 'tuyển', 'vòng loại',
232
- 'khoảnh khắc', 'highlights', 'bàn thắng', 'goal',
233
- 'kết quả', 'tỉ số', 'việt nam', 'vn',
234
- 'ngoại hạng', 'premier league', 'champions league',
235
- 'laliga', 'serie a', 'bundesliga', 'ligue 1',
236
- 'copa', 'europa', 'c1', 'c2',
237
- 'messi', 'ronaldo', 'neymar', 'mbappe', 'haaland',
238
- 'v-league', 'vleague', 'bóng đá việt',
239
- 'đội bóng', 'hlv', 'huấn luyện viên',
240
- 'chuyển nhượng', 'transfer',
241
- 'asian cup', 'aff cup', 'sea games',
242
- 'olympic', 'u23', 'u20', 'u17',
243
- ]
244
-
245
- wc_shorts = []
246
- for s in all_shorts:
247
- tl = s.get('title', '').lower()
248
- if any(k in tl for k in wc_kws):
249
- wc_shorts.append(s)
250
-
251
- if not wc_shorts:
252
- wc_shorts = all_shorts
253
-
254
- return wc_shorts[:max_count]
255
-
256
- import re
257
- # Alias for backward compatibility
258
- get_vtvnamo_shorts = get_vtvnambo_shorts
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
rewrite_fix_v2.js DELETED
@@ -1,2 +0,0 @@
1
- // No-op - all functionality built into app_v2.js
2
- (function(){})();
 
 
 
runtime.txt DELETED
@@ -1 +0,0 @@
1
- python-3.12.x
 
 
shorts_cache.py DELETED
@@ -1,86 +0,0 @@
1
- """
2
- VNEWS Shorts Runtime Cache - External Updater Module
3
- GitHub Actions fetches YouTube shorts via yt-dlp -> POST to /api/shorts/update
4
- Space saves to RAM cache + persistent file if /data available
5
- """
6
- import os
7
- import json
8
- import time
9
- import threading
10
-
11
- # Runtime cache (RAM)
12
- _shorts_runtime_cache = None
13
- _shorts_cache_ts = 0
14
- _shorts_cache_lock = threading.Lock()
15
-
16
- # Secret for authenticating update requests
17
- SHORTS_UPDATE_SECRET = os.environ.get("SHORTS_UPDATE_SECRET", "vnews-shorts-2026")
18
-
19
- # Paths
20
- SHORTS_CACHE_FILE = "/data/shorts_runtime_cache.json" if os.path.isdir("/data") else "/app/shorts_runtime_cache.json"
21
-
22
-
23
- def get_runtime_cache():
24
- """Get cached shorts (from RAM or file fallback)"""
25
- global _shorts_runtime_cache, _shorts_cache_ts
26
- with _shorts_cache_lock:
27
- if _shorts_runtime_cache is not None:
28
- age = time.time() - _shorts_cache_ts
29
- if age < 7200: # 2h fresh
30
- return _shorts_runtime_cache
31
-
32
- # Try file fallback
33
- try:
34
- if os.path.exists(SHORTS_CACHE_FILE):
35
- with open(SHORTS_CACHE_FILE, "r", encoding="utf-8") as f:
36
- data = json.load(f)
37
- age = time.time() - data.get("ts", 0)
38
- if age < 86400: # 24h stale limit
39
- items = data.get("items", [])
40
- with _shorts_cache_lock:
41
- _shorts_runtime_cache = items
42
- _shorts_cache_ts = data.get("ts", time.time())
43
- return items
44
- except Exception as e:
45
- print(f"[cache] read error: {e}")
46
-
47
- return None
48
-
49
-
50
- def set_runtime_cache(items):
51
- """Update runtime cache from external data"""
52
- global _shorts_runtime_cache, _shorts_cache_ts
53
- ts = time.time()
54
- with _shorts_cache_lock:
55
- _shorts_runtime_cache = items
56
- _shorts_cache_ts = ts
57
-
58
- # Also write to file (persistent if /data mounted)
59
- try:
60
- os.makedirs(os.path.dirname(SHORTS_CACHE_FILE), exist_ok=True)
61
- payload = {"items": items, "ts": ts, "count": len(items)}
62
- with open(SHORTS_CACHE_FILE, "w", encoding="utf-8") as f:
63
- json.dump(payload, f, ensure_ascii=False, indent=2)
64
- print(f"[cache] saved {len(items)} shorts to {SHORTS_CACHE_FILE}")
65
- except Exception as e:
66
- print(f"[cache] write skipped: {e}")
67
-
68
- return len(items)
69
-
70
-
71
- def get_cache_status():
72
- """Return status dict for the cache"""
73
- cache = None
74
- with _shorts_cache_lock:
75
- if _shorts_runtime_cache is not None:
76
- cache = _shorts_runtime_cache
77
- age = int(time.time() - _shorts_cache_ts)
78
- else:
79
- age = -1
80
- return {
81
- "cached": cache is not None,
82
- "count": len(cache) if cache else 0,
83
- "age_seconds": age,
84
- "has_persistent": os.path.isdir("/data"),
85
- "cache_file_exists": os.path.exists(SHORTS_CACHE_FILE),
86
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
shorts_rss_proxy.py DELETED
@@ -1,114 +0,0 @@
1
- """
2
- YouTube RSS Proxy - Fetches YouTube channel RSS feeds server-side
3
- Avoids CORS issues when client tries to fetch YouTube directly
4
- """
5
- import requests as req
6
- from fastapi import Query
7
- from fastapi.responses import Response
8
-
9
- HEADERS = {"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36"}
10
-
11
- YOUTUBE_CHANNELS = {
12
- "baodantri7941": "UC_x5TKhOgd6GhYvv5z4I3jg",
13
- "baosuckhoedoisongboyte": "UCBsY5fXTQLkF_JnH9kLkL4g",
14
- }
15
-
16
- def setup_rss_proxy(app):
17
- """Add RSS proxy endpoints to the FastAPI app"""
18
-
19
- @app.get("/api/proxy/rss")
20
- def proxy_rss(url: str = Query(...)):
21
- """Proxy YouTube RSS feed to avoid CORS"""
22
- try:
23
- r = req.get(url, headers=HEADERS, timeout=15)
24
- if r.status_code == 200:
25
- return Response(
26
- content=r.content,
27
- media_type="application/xml",
28
- headers={"Access-Control-Allow-Origin": "*"}
29
- )
30
- return Response(status_code=r.status_code)
31
- except Exception as e:
32
- return Response(status_code=502, content=str(e))
33
-
34
- @app.get("/api/shorts/rss")
35
- def shorts_via_rss():
36
- """Get shorts from YouTube RSS feeds server-side"""
37
- import xml.etree.ElementTree as ET
38
- import html as html_lib
39
- import re
40
-
41
- shorts = []
42
- seen = set()
43
-
44
- for handle, channel_id in YOUTUBE_CHANNELS.items():
45
- try:
46
- rss_url = f"https://www.youtube.com/feeds/videos.xml?channel_id={channel_id}"
47
- r = req.get(rss_url, headers=HEADERS, timeout=15)
48
- if r.status_code != 200:
49
- continue
50
-
51
- root = ET.fromstring(r.text)
52
- ns = {
53
- 'atom': 'http://www.w3.org/2005/Atom',
54
- 'yt': 'http://www.youtube.com/xml/schemas/2015',
55
- 'media': 'http://search.yahoo.com/mrss/'
56
- }
57
-
58
- for entry in root.findall('atom:entry', ns)[:30]:
59
- title_el = entry.find('atom:title', ns)
60
- title = html_lib.unescape(title_el.text) if title_el is not None and title_el.text else ''
61
-
62
- link_el = entry.find('atom:link', ns)
63
- link = link_el.get('href', '') if link_el is not None else ''
64
-
65
- vid_el = entry.find('yt:videoId', ns)
66
- vid = vid_el.text if vid_el is not None else ''
67
-
68
- if not vid:
69
- m = re.search(r'(?:v=|shorts/)([A-Za-z0-9_-]{11})', link)
70
- if m:
71
- vid = m.group(1)
72
-
73
- if not vid or vid in seen:
74
- continue
75
-
76
- # Check if it's a short
77
- is_short = '#shorts' in title.lower() or '#short' in title.lower() or '/shorts/' in link
78
-
79
- if not is_short:
80
- desc_el = entry.find('media:description', ns)
81
- if desc_el is not None and desc_el.text:
82
- if '#shorts' in desc_el.text.lower():
83
- is_short = True
84
-
85
- if not is_short:
86
- continue
87
-
88
- seen.add(vid)
89
-
90
- # Get thumbnail
91
- thumb = f"https://i.ytimg.com/vi/{vid}/hqdefault.jpg"
92
- media_group = entry.find('media:group', ns)
93
- if media_group is not None:
94
- thumb_el = media_group.find('media:thumbnail', ns)
95
- if thumb_el is not None:
96
- thumb = thumb_el.get('url', thumb)
97
-
98
- shorts.append({
99
- 'id': vid,
100
- 'title': title.replace('#shorts', '').replace('#short', '').strip()[:120],
101
- 'img': thumb,
102
- 'link': f'https://www.youtube.com/shorts/{vid}',
103
- 'channel': handle,
104
- 'source': 'yt'
105
- })
106
-
107
- if len(shorts) >= 40:
108
- break
109
-
110
- except Exception as e:
111
- print(f"RSS error for {handle}: {e}")
112
- continue
113
-
114
- return {"shorts": shorts, "count": len(shorts)}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
static/app_v2.js CHANGED
@@ -1,17 +1,6 @@
1
- /**
2
- * VNEWS Frontend v2 - Shorts Dantri/SKDS removed, VTV Digital CDN
3
- * v2.4 - Fixed: prependWallPost detached-element bug, makeShortVideo UI update, slide viewer for rewrite posts
4
- */
5
- function _proxyImg(url){
6
- if(!url || typeof url !== 'string') return '';
7
- if(url.startsWith('http') && !url.includes(location.host)){
8
- return '/api/proxy/img?url='+encodeURIComponent(url);
9
- }
10
- return url;
11
- }
12
-
13
- var _ttsSelections = {};
14
 
 
15
  function _fetchWithTimeout(url, ms){
16
  return new Promise((resolve,reject)=>{
17
  const ctrl=new AbortController();
@@ -28,6 +17,7 @@ async function loadHome(){
28
  const homeEl = document.getElementById('view-home');
29
  if(!homeEl) return;
30
 
 
31
  homeEl.innerHTML =
32
  '<div id="home-featured-area"></div>'
33
  +'<div class="ai-compose"><div class="ai-compose-title">🤖 AI viết bài</div><div class="ai-compose-row"><input id="topic-input" placeholder="Nhập chủ đề..."><button onclick="searchTopic()">Tìm nguồn</button></div><div class="ai-compose-row"><input id="url-input" placeholder="Dán URL bài viết..."><button class="secondary" onclick="rewriteUrl()">Rewrite</button></div><div id="hot-topics" class="hot-topic-row"></div></div>'
@@ -38,31 +28,42 @@ async function loadHome(){
38
 
39
  const afterEl = homeEl.querySelector('#home-after-wc');
40
 
 
41
  loadLivescore('today');
42
  loadHotTopics();
43
 
44
- const [featuredData, wallData, hlLeagues, wcData] = await Promise.allSettled([
 
45
  _fetchWithTimeout('/api/livescore/featured', 5000),
 
46
  _fetchWithTimeout('/api/wall', 5000),
47
  _fetchWithTimeout('/api/highlights/leagues', 10000),
 
48
  _fetchWithTimeout('/api/wc2026', 8000),
49
  ]).then(results => results.map(r => r.status === 'fulfilled' ? r.value : null));
50
 
 
51
  if(featuredData && featuredData.home){
52
  const sc=featuredData.status==='live'?'':'upcoming';
53
  const st=featuredData.status==='live'?`🔴 ${featuredData.minute||'LIVE'}`:`⏰ ${featuredData.time}`;
54
  const area=document.getElementById('home-featured-area');
55
- if(area) area.innerHTML=`<div class="featured-match" onclick="openMatch('${featuredData.event_id}')"><div class="fm-league">${featuredData.league}</div><div class="fm-teams"><div class="fm-team"><img src="${_proxyImg(featuredData.home_logo)}" onerror="this.style.display='none'"><span>${featuredData.home}</span></div><div class="fm-score">${featuredData.score||'VS'}</div><div class="fm-team"><img src="${_proxyImg(featuredData.away_logo)}" onerror="this.style.display='none'"><span>${featuredData.away}</span></div></div><div class="fm-status ${sc}">${st}</div></div>`;
56
  }
57
 
 
 
58
  _wallPosts = (wallData && wallData.posts) || [];
59
  _hlLeagueData = hlLeagues || {};
60
  _wc2026Data = wcData;
61
 
 
62
  if(wcData) switchWCTab('news');
63
 
 
 
64
  _renderWallIn(afterEl);
65
  _renderHLIn(afterEl);
 
66
  }
67
 
68
  function _renderSlidesIn(key, label, emoji, vids, afterEl){
@@ -73,9 +74,9 @@ function _renderSlidesIn(key, label, emoji, vids, afterEl){
73
  const isHL = key==='world-cup'||key==='premier-league'||key==='champions-league'||key==='la-liga'||key==='serie-a'||key==='bundesliga'||key==='friendly';
74
  vids.slice(0,isHL?8:12).forEach((a,i)=>{
75
  if(isHL){
76
- h+=`<div class="slider-item" onclick="openHighlightFeed('${key}',${i})"><div class="slider-thumb">${a.img?`<img src="${_proxyImg(a.img)}" loading="lazy">`:''}<div class="card-play">▶</div></div><div class="slider-title">${esc(a.title)}</div></div>`;
77
  } else {
78
- h+=`<div class="slider-item" onclick="readArticle('${esc(a.link)}')"><div class="slider-thumb">${a.img?`<img src="${_proxyImg(a.img)}" loading="lazy">`:''}</div><div class="slider-title">${esc(a.title)}</div></div>`;
79
  }
80
  });
81
  h+='</div>';
@@ -83,9 +84,33 @@ function _renderSlidesIn(key, label, emoji, vids, afterEl){
83
  afterEl.parentNode.insertBefore(wrap, afterEl);
84
  }
85
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
86
  function _renderWallIn(afterEl){
87
  if(!_wallPosts||!_wallPosts.length||!afterEl) return;
88
  const posts=_wallPosts;
 
 
 
 
 
 
 
 
 
89
  const wrap=document.createElement('div');
90
  wrap.className='slider-wrap';wrap.id='ai-wall-wrap';
91
  let h='<div class="slider-header"><span class="slider-label">🧱 Tường AI</span></div><div class="slider-track" id="ai-wall-track">';
@@ -106,247 +131,69 @@ function _renderHLIn(afterEl){
106
 
107
  // === WALL POST HELPERS ===
108
  function makeWallItem(p,i){
109
- var hasVideo = p.video && p.video.length > 0;
110
- var thumbContent = p.img
111
- ? '<img src="/api/proxy/img?url='+encodeURIComponent(p.img)+'" loading="lazy" onerror="this.style.display=\'none\'">'
112
- : (hasVideo ? '<video src="'+esc(p.video)+'" muted></video>' : '');
113
- var videoBadge = hasVideo ? '<div class="wall-video-badge">🎬</div>' : '';
114
- var vid = p.id||i;
115
- var lang = p.language || detectLanguage(p.title + ' ' + (p.text||''));
116
- var curVoice = p.voice || getAutoVoice(lang);
117
- var curEmotion = p.emotion || detectEmotion(p.title + ' ' + (p.text||''));
118
- var selKey = 'inline-' + vid;
119
- if(!_ttsSelections[selKey]) _ttsSelections[selKey] = {voice: curVoice, emotion: curEmotion};
120
- var voiceOpts = '';
121
- VOICE_LIST.forEach(function(v){
122
- voiceOpts += '<option value="'+v.id+'"'+(v.id===_ttsSelections[selKey].voice?' selected':'')+'>'+v.label+'</option>';
123
- });
124
- var emotOpts = '';
125
- EMOTION_LIST.forEach(function(e){
126
- emotOpts += '<option value="'+e.id+'"'+(e.id===_ttsSelections[selKey].emotion?' selected':'')+'>'+e.label+'</option>';
127
- });
128
- var spd = p.short_speed || '1.2';
129
- var voiceBar = '<div class="wall-tts-bar" style="margin:6px 0 4px;display:grid;grid-template-columns:1fr auto auto;gap:3px">'
130
- +'<select class="wvs" data-selkey="'+selKey+'" style="background:#1a1a1a;border:1px solid #333;color:#ccc;padding:3px;border-radius:6px;font-size:9px;min-width:0" onchange="_ttsSelections[this.dataset.selkey].voice=this.value">'
131
- +voiceOpts+'</select>'
132
- +'<select class="wes" data-selkey="'+selKey+'" style="background:#1a1a1a;border:1px solid #333;color:#ccc;padding:3px;border-radius:6px;font-size:9px;min-width:0" onchange="_ttsSelections[this.dataset.selkey].emotion=this.value">'
133
- +emotOpts+'</select>'
134
- +'<select class="wss" data-selkey="'+selKey+'" style="background:#1a1a1a;border:1px solid #333;color:#ccc;padding:3px;border-radius:6px;font-size:9px;min-width:0" onchange="_ttsSelections[this.dataset.selkey].speed=parseFloat(this.value)">'
135
- +'<option value="0.85"'+(spd==='0.85'?' selected':'')+'>0.85x</option>'
136
- +'<option value="1.0"'+(spd==='1.0'?' selected':'')+'>1.0x</option>'
137
- +'<option value="1.2"'+(spd==='1.2'?' selected':'')+'>1.2x</option>'
138
- +'<option value="1.35"'+(spd==='1.35'?' selected':'')+'>1.35x</option>'
139
- +'</select>'
140
- +'</div>';
141
- var makeBtn = hasVideo
142
- ? '<button class="wall-btn-video" onclick="event.stopPropagation();openShortAIFeed('+i+')">▶ Xem Short</button>'
143
- : '<button class="wall-btn-make" onclick="event.stopPropagation();makeShortVideo(\''+esc(vid)+'\',this,_ttsSelections[\'inline-'+esc(vid)+'\'].voice,parseFloat(document.querySelector(\'.wss[data-selkey=inline-'+esc(vid)+']\').value)||1.2,_ttsSelections[\'inline-'+esc(vid)+'\'].emotion)">🎬 Tạo Short</button>';
144
- return '<div class="wall-item" id="wall-item-'+esc(vid)+'"><div class="wall-thumb">'+thumbContent+videoBadge+'</div><div class="wall-title">'+esc(p.title)+'</div><div class="wall-text">'+esc((p.text||'').slice(0,180))+'</div>'+voiceBar+'<div class="wall-actions"><button class="primary" onclick="readWallPost('+i+')">Xem</button>'+makeBtn+'</div></div>';
145
  }
146
 
147
- async function makeShortVideo(postId, btn, voice, speed, emotion){
148
  if(!postId)return;
149
  const origText = btn ? btn.textContent : '🎬 Tạo Video';
150
  if(btn){btn.disabled=true;btn.textContent='⏳ Đang tạo...';}
151
  toast('⏳ Đang tạo video shorts...');
152
- if(!voice || !emotion){
153
- const selKey = 'inline-'+postId;
154
- const container = document.querySelector('.tts-selector[data-post-id="'+postId+'"]');
155
- if(container){
156
- if(_ttsSelections[selKey]){
157
- voice = voice || _ttsSelections[selKey].voice;
158
- emotion = emotion || _ttsSelections[selKey].emotion;
159
- }
160
- if(!voice){
161
- const selectedVoiceBtn = container.querySelector('.tts-voice-btn.selected') || container.querySelector('.tts-voice-btn[style*="5cb87a"]') || container.querySelector('.tts-voice-btn');
162
- voice = selectedVoiceBtn ? selectedVoiceBtn.dataset.voice : 'vi-VN-HoaiMyNeural';
163
- }
164
- if(!emotion){
165
- const selectedEmotionBtn = container.querySelector('.tts-emotion-btn.selected') || container.querySelector('.tts-emotion-btn[style*="5cb87a"]') || container.querySelector('.tts-emotion-btn');
166
- emotion = selectedEmotionBtn ? selectedEmotionBtn.dataset.emotion : 'neutral';
167
- }
168
- const speedSelect = container.querySelector('.tts-speed');
169
- speed = speedSelect ? parseFloat(speedSelect.value) || 1.2 : (speed || 1.2);
170
- } else {
171
- voice = voice || 'vi-VN-HoaiMyNeural';
172
- emotion = emotion || 'neutral';
173
- speed = speed || 1.2;
174
- }
175
- }
176
  try{
177
- const r = await fetch('/api/ai/short/' + encodeURIComponent(postId), {method:'POST',headers:{'Content-Type':'application/json'},body: JSON.stringify({voice:voice, emotion:emotion, speed:speed})});
 
 
 
 
 
178
  const j = await r.json();
179
  if(!r.ok || j.error) throw new Error(j.error||'Lỗi tạo video');
180
  toast('✅ Đã tạo video shorts!');
181
  const p = _wallPosts.find(x => String(x.id) === String(postId));
182
  if(p){
183
  p.video = j.video;
184
- p.voice = j.voice;
185
- p.emotion = j.emotion;
186
  const itemId = 'wall-item-'+postId;
187
  const el = document.getElementById(itemId);
188
  if(el){
189
  const idx = _wallPosts.indexOf(p);
190
- el.insertAdjacentHTML('afterend', makeWallItem(p, idx));
191
- el.remove();
 
192
  }
193
  }
 
194
  }catch(e){
195
  toast('❌ '+e.message);
196
  if(btn){btn.disabled=false;btn.textContent=origText;}
197
  }
198
  }
199
 
200
- var VOICE_LIST = [
201
- {id:'vi-VN-HoaiMyNeural', label:'🎙️ Hoài My (VI)', lang:'vi'},
202
- {id:'vi-VN-NamMinhNeural', label:'🎙️ Nam Minh (VI)', lang:'vi'},
203
- {id:'en-US-AndrewMultilingualNeural', label:'🎙️ Andrew (EN)', lang:'en'},
204
- {id:'en-AU-WilliamMultilingualNeural', label:'🎙️ William (EN)', lang:'en'},
205
- {id:'pt-BR-ThalitaMultilingualNeural', label:'🎙️ Thalita (PT)', lang:'pt'},
206
- {id:'fr-FR-VivienneMultilingualNeural', label:'🎙️ Vivienne (FR)', lang:'fr'},
207
- {id:'fr-FR-RemyMultilingualNeural', label:'🎙️ Rémy (FR)', lang:'fr'},
208
- {id:'de-DE-SeraphinaMultilingualNeural', label:'🎙️ Seraphina (DE)', lang:'de'},
209
- {id:'de-DE-FlorianMultilingualNeural', label:'🎙️ Florian (DE)', lang:'de'},
210
- {id:'ko-KR-HyunsuMultilingualNeural', label:'🎙️ Hyunsu (KO)', lang:'ko'},
211
- {id:'it-IT-GiuseppeMultilingualNeural', label:'🎙️ Giuseppe (IT)', lang:'it'},
212
- ];
213
- var EMOTION_LIST = [
214
- {id:'neutral', label:'😐 Trung tính'},
215
- {id:'happy', label:'😊 Vui vẻ'},
216
- {id:'excited', label:'🔥 Hào hứng'},
217
- {id:'sad', label:'😢 Buồn'},
218
- {id:'humorous', label:'😂 Hài hước'},
219
- {id:'serious', label:'⚠️ Nghiêm túc'},
220
- {id:'urgent', label:'🚨 Khẩn cấp'},
221
- {id:'warm', label:'💖 Ấm áp'},
222
- ];
223
-
224
- document.addEventListener('click',function(e){
225
- var btn = e.target.closest('.tts-voice-btn');
226
- if(btn){
227
- var container = btn.closest('.tts-selector');
228
- if(container){
229
- var selKey = 'inline-'+container.dataset.postId;
230
- if(!_ttsSelections[selKey]) _ttsSelections[selKey]={voice:btn.dataset.voice,emotion:'neutral'};
231
- var allBtns = container.querySelectorAll('.tts-voice-btn');
232
- for(var i=0;i<allBtns.length;i++){allBtns[i].style.borderColor='#333';allBtns[i].style.background='#222';allBtns[i].classList.remove('selected');}
233
- btn.style.borderColor='#5cb87a';btn.style.background='#1a2a1f';btn.classList.add('selected');
234
- _ttsSelections[selKey].voice = btn.dataset.voice;
235
- }
236
  return;
237
  }
238
- var ebtn = e.target.closest('.tts-emotion-btn');
239
- if(ebtn){
240
- var container = ebtn.closest('.tts-selector');
241
- if(container){
242
- var selKey = 'inline-'+container.dataset.postId;
243
- if(!_ttsSelections[selKey]) _ttsSelections[selKey]={voice:'vi-VN-HoaiMyNeural',emotion:ebtn.dataset.emotion};
244
- var allBtns = container.querySelectorAll('.tts-emotion-btn');
245
- for(var i=0;i<allBtns.length;i++){allBtns[i].style.borderColor='#333';allBtns[i].style.background='#222';allBtns[i].classList.remove('selected');}
246
- ebtn.style.borderColor='#5cb87a';ebtn.style.background='#1a2a1f';ebtn.classList.add('selected');
247
- _ttsSelections[selKey].emotion = ebtn.dataset.emotion;
248
  }
249
- return;
250
  }
251
- var cbtn = e.target.closest('.tts-create-btn');
252
- if(cbtn){
253
- var container = cbtn.closest('.tts-selector');
254
- if(container){
255
- var selKey = 'inline-'+container.dataset.postId;
256
- var selVoice = _ttsSelections[selKey] ? _ttsSelections[selKey].voice : 'vi-VN-HoaiMyNeural';
257
- var selEmotion = _ttsSelections[selKey] ? _ttsSelections[selKey].emotion : 'neutral';
258
- var speedSel = container.querySelector('.tts-speed');
259
- var speed = speedSel ? parseFloat(speedSel.value)||1.2 : 1.2;
260
- window.makeShortVideo(container.dataset.postId, cbtn, selVoice, speed, selEmotion);
261
- }
262
- return;
263
- }
264
- });
265
- function detectLanguage(text){
266
- if(!text) return 'vi';
267
- var t=text.toLowerCase(), chars=new Set(t);
268
- var vnChars='đăâêôơưàảãạáằẳẵặắầẩẫậấèẻẽẹéềễểệếìỉĩịíòỏõọóồổỗộốờởỡợớùủũụúừửữựứỳỷỹỵý';
269
- var vnCount=0; for(var c of vnChars){if(chars.has(c)) vnCount++;}
270
- if(vnCount>=2) return 'vi';
271
- if(chars.has('ñ')||chars.has('¿')||chars.has('¡')) return 'es';
272
- if(chars.has('ã')||chars.has('õ')) return 'pt';
273
- var words=t.split(/\s+/);
274
- var enWords=['the','is','at','which','on','and','or','but','this','that','with','from','have','been'];
275
- var enCount=words.filter(function(w){return enWords.indexOf(w)>=0;}).length;
276
- if(enCount>=2) return 'en';
277
- return 'vi';
278
- }
279
- function detectEmotion(text){
280
- if(!text) return 'neutral';
281
- var t=text.toLowerCase();
282
- var kws={
283
- happy:['vui','hạnh phúc','tuyệt','thành công','chiến thắng','feliz','maravilloso','happy','joy','wonderful','great','amazing','love','excellent'],
284
- excited:['hào hứng','phấn khích','đột phá','kỷ lục','đỉnh cao','emocionante','increíble','excited','thrilling','unbelievable','awesome','breakthrough'],
285
- sad:['buồn','đau','mất','thảm họa','khủng hoảng','triste','terrible','sad','unhappy','tragic','painful','death'],
286
- humorous:['hài hước','buồn cười','haha','đùa','engraçado','gracioso','funny','hilarious','joke','lol'],
287
- serious:['nghiêm trọng','kh���n cấp','quan trọng','lo ngại','sério','crítico','serious','critical','urgent','severe','crisis'],
288
- urgent:['khẩn cấp','báo động','ngay lập tức','urgent','breaking','alert','emergency'],
289
- warm:['ấm áp','tình cảm','yêu thương','warm','love','heart','touching']
290
- };
291
- var bestScore=0, bestEmotion='neutral';
292
- for(var em in kws){var score=0; for(var kw of kws[em]){if(t.indexOf(kw)>=0) score++;} if(score>bestScore){bestScore=score;bestEmotion=em;}}
293
- return bestEmotion;
294
- }
295
- function getAutoVoice(lang){var map={vi:'vi-VN-HoaiMyNeural',pt:'pt-BR-ThalitaMultilingualNeural',en:'en-US-AndrewMultilingualNeural',fr:'fr-FR-VivienneMultilingualNeural',de:'de-DE-SeraphinaMultilingualNeural',ko:'ko-KR-HyunsuMultilingualNeural',it:'it-IT-GiuseppeMultilingualNeural'};return map[lang]||'vi-VN-HoaiMyNeural';}
296
- function buildVoiceEmotionSelector(post){
297
- var lang=post.language||detectLanguage(post.title+' '+(post.text||''));
298
- var _oldVoiceMap = {'hoaimy':'vi-VN-HoaiMyNeural','namminh':'vi-VN-NamMinhNeural','andrew':'en-US-AndrewMultilingualNeural','jenny':'en-US-AndrewMultilingualNeural','thalita':'pt-BR-ThalitaMultilingualNeural','pt_thalita':'pt-BR-ThalitaMultilingualNeural','vivienne':'fr-FR-VivienneMultilingualNeural','remy':'fr-FR-RemyMultilingualNeural','seraphina':'de-DE-SeraphinaMultilingualNeural','florian':'de-DE-FlorianMultilingualNeural','sunhee':'ko-KR-HyunsuMultilingualNeural','hyunsu':'ko-KR-HyunsuMultilingualNeural','giuseppe':'it-IT-GiuseppeMultilingualNeural','ela':'en-US-AndrewMultilingualNeural','denise':'fr-FR-VivienneMultilingualNeural','katja':'de-DE-SeraphinaMultilingualNeural','nanami':'en-US-AndrewMultilingualNeural','xiaochen':'en-US-AndrewMultilingualNeural','es_carlos':'en-US-AndrewMultilingualNeural','pt_francisco':'pt-BR-ThalitaMultilingualNeural'};
299
- var _postVoice = post.voice ? (_oldVoiceMap[post.voice] || post.voice) : '';
300
- var autoVoice= _postVoice || getAutoVoice(lang);
301
- var autoEmotion=post.emotion||detectEmotion(post.title+' '+(post.text||''));
302
- var selKey = 'inline-'+post.id;
303
- if(!_ttsSelections[selKey]){_ttsSelections[selKey] = {voice: autoVoice, emotion: autoEmotion};}
304
- var h='<div class="tts-selector" data-post-id="'+post.id+'" style="margin-top:10px;padding:10px;background:#1a1a1a;border:1px solid #2a2a2a;border-radius:10px">';
305
- h+='<div style="font-size:11px;color:#888;margin-bottom:6px">🎙️ Giọng đọc (ngôn ngữ: '+lang.toUpperCase()+'):</div><div style="display:flex;flex-wrap:wrap;gap:4px;margin-bottom:8px">';
306
- VOICE_LIST.forEach(function(v){var sel=v.id===_ttsSelections[selKey].voice?'border-color:#5cb87a;background:#1a2a1f':'border-color:#333;background:#222';h+='<button class="tts-voice-btn '+(v.id===_ttsSelections[selKey].voice?'selected':'')+'" data-voice="'+v.id+'" data-post-id="'+post.id+'" style="'+sel+';border:1px solid;color:#ccc;padding:4px 8px;border-radius:10px;font-size:10px;cursor:pointer">'+v.label+'</button>';});
307
- h+='</div><div style="font-size:11px;color:#888;margin-bottom:6px">😊 Cảm xúc:</div><div style="display:flex;flex-wrap:wrap;gap:4px;margin-bottom:8px">';
308
- EMOTION_LIST.forEach(function(e){var sel=e.id===_ttsSelections[selKey].emotion?'border-color:#5cb87a;background:#1a2a1f':'border-color:#333;background:#222';h+='<button class="tts-emotion-btn '+(e.id===_ttsSelections[selKey].emotion?'selected':'')+'" data-emotion="'+e.id+'" data-post-id="'+post.id+'" style="'+sel+';border:1px solid;color:#ccc;padding:4px 8px;border-radius:10px;font-size:10px;cursor:pointer">'+e.label+'</button>';});
309
- h+='</div><div style="display:flex;align-items:center;gap:6px;margin-bottom:8px"><span style="font-size:11px;color:#888">⚡ Tốc độ:</span>';
310
- h+='<select class="tts-speed" style="background:#222;border:1px solid #333;color:#ccc;padding:3px 8px;border-radius:8px;font-size:10px"><option value="0.85">0.85x Chậm</option><option value="1.0">1.0x Bình thường</option><option value="1.2" selected>1.2x Nhanh</option><option value="1.35">1.35x Rất nhanh</option></select></div>';
311
- h+='<button class="tts-create-btn" style="width:100%;background:#2d8659;border:0;color:#fff;padding:8px;border-radius:10px;font-size:11px;font-weight:700;cursor:pointer">🎬 Tạo Short AI</button></div>';
312
- return h;
313
  }
314
- window.showVoiceEmotionSelector=function(postId,title,text){
315
- var overlay=document.createElement('div');
316
- overlay.style.cssText='position:fixed;inset:0;background:rgba(0,0,0,.85);z-index:99999;display:flex;align-items:center;justify-content:center;padding:16px';
317
- var box=document.createElement('div');box.style.cssText='background:#1a1a1a;border:2px solid #2d8659;border-radius:16px;padding:20px;max-width:400px;width:100%;max-height:80vh;overflow-y:auto';
318
- var lang=detectLanguage(title+' '+text);var autoEmotion=detectEmotion(title+' '+text);
319
- var h='<h3 style="color:#5cb87a;margin-bottom:12px;font-size:16px">🎬 Tạo Short AI (ngôn ngữ: '+lang.toUpperCase()+')</h3>';
320
- h+='<div style="margin-bottom:12px"><div style="color:#aaa;font-size:11px;margin-bottom:6px">🎙️ Chọn giọng đọc:</div>';
321
- VOICE_LIST.forEach(function(v){var sel=v.id===getAutoVoice(lang)?'border-color:#5cb87a;background:#1a2a1f':'border-color:#333;background:#222';h+='<button class="ve-voice-btn" data-voice="'+v.id+'" style="display:inline-block;'+sel+';border:1px solid;color:#ccc;padding:5px 10px;border-radius:12px;font-size:10px;margin:2px;cursor:pointer">'+v.label+'</button>';});
322
- h+='</div><div style="margin-bottom:12px"><div style="color:#aaa;font-size:11px;margin-bottom:6px">😊 Chọn cảm xúc:</div>';
323
- EMOTION_LIST.forEach(function(e){var sel=e.id===autoEmotion?'border-color:#5cb87a;background:#1a2a1f':'border-color:#333;background:#222';h+='<button class="ve-emotion-btn" data-emotion="'+e.id+'" style="display:inline-block;'+sel+';border:1px solid;color:#ccc;padding:5px 10px;border-radius:12px;font-size:10px;margin:2px;cursor:pointer">'+e.label+'</button>';});
324
- h+='</div><div style="margin-bottom:12px"><div style="color:#aaa;font-size:11px;margin-bottom:6px">⚡ Tốc độ:</div>';
325
- h+='<select id="ve-speed" style="background:#222;border:1px solid #333;color:#ccc;padding:6px 12px;border-radius:10px;font-size:11px"><option value="0.85">0.85x Chậm</option><option value="1.0">1.0x Bình thường</option><option value="1.2" selected>1.2x Nhanh</option><option value="1.35">1.35x Rất nhanh</option></select></div>';
326
- h+='<div style="display:flex;gap:8px"><button id="ve-create-btn" style="flex:1;background:#2d8659;border:0;color:#fff;padding:10px;border-radius:12px;font-size:12px;font-weight:700;cursor:pointer">🎬 Tạo Short</button>';
327
- h+='<button id="ve-cancel-btn" style="background:#333;border:0;color:#ccc;padding:10px 16px;border-radius:12px;font-size:11px;cursor:pointer">✕</button></div>';
328
- h+='<div id="ve-status" style="color:#888;font-size:10px;margin-top:8px;display:none"></div>';
329
- box.innerHTML=h;overlay.appendChild(box);document.body.appendChild(overlay);
330
- var selectedVoice=getAutoVoice(lang),selectedEmotion=autoEmotion;
331
- box.querySelectorAll('.ve-voice-btn').forEach(function(btn){btn.addEventListener('click',function(){box.querySelectorAll('.ve-voice-btn').forEach(function(b){b.style.borderColor='#333';b.style.background='#222';});this.style.borderColor='#5cb87a';this.style.background='#1a2a1f';selectedVoice=this.dataset.voice;});});
332
- box.querySelectorAll('.ve-emotion-btn').forEach(function(btn){btn.addEventListener('click',function(){box.querySelectorAll('.ve-emotion-btn').forEach(function(b){b.style.borderColor='#333';b.style.background='#222';});this.style.borderColor='#5cb87a';this.style.background='#1a2a1f';selectedEmotion=this.dataset.emotion;});});
333
- box.querySelector('#ve-cancel-btn').addEventListener('click',function(){overlay.remove();});
334
- box.querySelector('#ve-create-btn').addEventListener('click',async function(){
335
- this.disabled=true;this.textContent='⏳ Đang tạo...';
336
- box.querySelector('#ve-status').style.display='block';box.querySelector('#ve-status').textContent='Đang tạo video shorts...';
337
- try{
338
- var speed=parseFloat(box.querySelector('#ve-speed').value)||1.2;
339
- if(!_ttsSelections["inline-"+postId]) _ttsSelections["inline-"+postId]={voice:"vi-VN-HoaiMyNeural",emotion:"neutral"};
340
- _ttsSelections["inline-"+postId].voice=selectedVoice;_ttsSelections["inline-"+postId].emotion=selectedEmotion;
341
- var r=await fetch('/api/ai/short/'+encodeURIComponent(postId),{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({voice:selectedVoice,emotion:selectedEmotion,speed:speed})});
342
- var j=await r.json();
343
- if(!r.ok||j.error) throw new Error(j.error||'Lỗi tạo video');
344
- toast('✅ Đã tạo Short AI!');overlay.remove();
345
- var p=_wallPosts.find(function(x){return String(x.id)===String(postId);});
346
- if(p){p.video=j.video;p.voice=j.voice;p.emotion=j.emotion;}
347
- }catch(e){this.disabled=false;this.textContent='🎬 Tạo Short';box.querySelector('#ve-status').textContent='❌ '+e.message;}
348
- });
349
- };
350
 
351
  function prependWallPost(post){
352
  _wallPosts.unshift(post);
@@ -357,7 +204,8 @@ function prependWallPost(post){
357
  if(homeEl){
358
  let insertBefore=homeEl.querySelector('.slider-wrap');
359
  const newWrap=document.createElement('div');
360
- newWrap.className='slider-wrap';newWrap.id='ai-wall-wrap';
 
361
  newWrap.innerHTML=`<div class="slider-header"><span class="slider-label">🧱 Tường AI</span></div><div class="slider-track" id="ai-wall-track">${makeWallItem(post,0)}</div>`;
362
  if(insertBefore) homeEl.insertBefore(newWrap,insertBefore);
363
  else homeEl.appendChild(newWrap);
@@ -366,35 +214,68 @@ function prependWallPost(post){
366
  }
367
  return;
368
  }
369
- track.insertAdjacentHTML('afterbegin', makeWallItem(post, 0));
 
 
 
 
 
 
 
 
 
 
 
 
370
  track.scrollTo({left:0,behavior:'smooth'});
 
371
  }
372
 
 
 
373
  let _wallPosts=[];
374
  let _currentView='home';
375
  let _currentEventId=null;
376
  let _currentMatchUrl=null;
 
377
  let _htPage=0,_htTopic='';
378
- async function loadHotTopics(){const j=await fetch('/api/hot_topics').then(r=>r.json()).catch(()=>({topics:[]}));const el=document.getElementById('hot-topics');if(!el)return;el.innerHTML=(j.topics||[]).slice(0,18).map(t=>{const topicText=t.topic||t.label.replace(/^#/,'');return`<button class="hot-chip" onclick="searchTopic('${topicText.replace(/'/g,"\\'")}')">${esc(t.label)}</button>`;}).join('');}
379
  function searchTopic(topic){if(!topic){topic=document.getElementById('topic-input')?.value.trim();if(!topic){alert('Nhập chủ đề');return;}}document.getElementById('topic-input').value='';_htTopic=topic;_htPage=0;showHashtagSources(topic,0);}
380
- async function showHashtagSources(topic,page){const box=document.getElementById('hashtag-box');if(!box)return;if(page===0)box.innerHTML=`<div class="hashtag-sources"><h3>🔍 ${esc(topic)}</h3><div class="hashtag-loading"><div class="hashtag-spinner"></div>Đang tìm...</div></div>`;try{const r=await fetch(`/api/hashtag/sources?topic=${encodeURIComponent(topic)}&page=${page}`);const j=await r.json();const sources=j.sources||[];if(!sources.length&&page===0){box.innerHTML=`<div class="hashtag-sources"><h3>🔍 ${esc(topic)}</h3><div style="color:#888;padding:8px">Không tìm được bài viết liên quan</div></div>`;return;}let h='';if(page===0)h=`<div class="hashtag-sources"><h3>🔍 ${esc(topic)} <span style="font-size:10px;color:#888">(${j.total} bài từ 8 nguồn)</span></h3><div id="ht-list">`;sources.forEach((s,i)=>{const idx=page*8+i;h+=`<div class="hashtag-src-item" onclick="readArticle('${esc(s.url)}')"><div class="hashtag-src-img" id="ht-img-${idx}"></div><div class="hashtag-src-text"><div class="hashtag-src-title">${esc(s.title)}</div><div class="hashtag-src-via">${esc(s.via||'')}</div></div></div>`;});if(page===0){h+=`</div><button class="hashtag-rewrite-btn" onclick="rewriteHashtag('${esc(topic).replace(/'/g,"\\'")}')">🤖 Rewrite AI tổng hợp & đăng tường</button>`;if(j.has_more)h+=`<button class="hashtag-load-more" id="ht-more" onclick="loadMoreHashtag()">Tải thêm ▼</button>`;h+=`</div>`;box.innerHTML=h;}else{document.getElementById('ht-list')?.insertAdjacentHTML('beforeend',h);const btn=document.getElementById('ht-more');if(btn){if(!j.has_more)btn.remove();else{btn.disabled=false;btn.textContent='Tải thêm ▼';}}}sources.forEach((s,i)=>{const idx=page*8+i;if(!s.url)return;const ctrl=new AbortController();setTimeout(()=>ctrl.abort(),4000);fetch('/api/article?url='+encodeURIComponent(s.url),{signal:ctrl.signal}).then(r=>r.json()).then(d=>{if(d&&(d.og_image||d.img)){const el=document.getElementById('ht-img-'+idx);if(el)el.innerHTML=`<img src="${_proxyImg(esc(d.og_image||d.img))}" onerror="this.style.display='none'">`;}}).catch(()=>{});});}catch(e){box.innerHTML=`<div class="hashtag-sources"><h3>🔍 ${esc(topic)}</h3><div style="color:#e74c3c;padding:8px">Lỗi: ${esc(e.message)}</div></div>`;}}
381
  function loadMoreHashtag(){_htPage++;const btn=document.getElementById('ht-more');if(btn){btn.disabled=true;btn.textContent='Đang tải...';}showHashtagSources(_htTopic,_htPage);}
382
  async function rewriteHashtag(topic){const btn=event?.target;if(btn){btn.disabled=true;btn.textContent='Đang tổng hợp...';}try{const r=await fetch('/api/topic_post',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({topic})});const j=await r.json();if(!r.ok||j.error)throw new Error(j.error||'Lỗi');toast('✅ Đã đăng Tường AI!');if(btn)btn.textContent='✅ Đăng thành công!';if(j.post)prependWallPost(j.post);}catch(e){toast('❌ '+e.message);if(btn){btn.disabled=false;btn.textContent='🤖 Rewrite AI';}}}
383
  async function loadLivescore(tab){document.querySelectorAll('.ls-tab').forEach(t=>t.classList.remove('active'));document.querySelector(`.ls-tab[data-tab="${tab}"]`)?.classList.add('active');const el=document.getElementById('ls-content');if(!el)return;el.innerHTML='<div class="loading">Đang tải...</div>';let ep='/api/livescore/'+tab;if(tab.startsWith('bxh_'))ep='/api/livescore/standings/'+tab.replace('bxh_','');try{const r=await fetch(ep);const d=await r.json();el.innerHTML=d.html&&d.html.length>50?d.html:'<div class="loading">Không có dữ liệu</div>';bindMatchClicks(el);}catch(e){el.innerHTML='<div class="loading">Lỗi</div>';}}
384
- function bindMatchClicks(el){el.querySelectorAll('.match-detail').forEach(md=>{md.style.cursor='pointer';md.addEventListener('click',function(e){const statusA=this.querySelector('.status a');const teamA=this.querySelector('.teams a[href*="/tran-dau/"]');const a = statusA || teamA;if(a){e.preventDefault();e.stopPropagation();const href=a.getAttribute('href')||'';const m=href.match(/\/tran-dau\/(\d+)\//);if(m){const fullUrl=href.startsWith('http')?href:'https://bongda.com.vn'+href;openMatch(m[1],fullUrl);}}});});el.querySelectorAll('a').forEach(a=>{a.addEventListener('click',e=>{e.preventDefault();e.stopPropagation()});});}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
385
  function openMatch(id,url){if(!id)return;_currentEventId=id;if(url)_currentMatchUrl=url;document.getElementById('match-overlay').classList.add('active');document.body.style.overflow='hidden';loadMatchTab('detail')}
386
  function closeMatch(){document.getElementById('match-overlay').classList.remove('active');document.body.style.overflow=''}
387
  async function loadMatchTab(tab){document.querySelectorAll('.mo-tab').forEach(t=>t.classList.remove('active'));document.querySelectorAll('.mo-tab').forEach(t=>{if((tab==='comm'&&t.textContent==='Diễn biến')||(tab==='stats'&&t.textContent==='Thống kê')||(tab==='detail'&&t.textContent.includes('Chi tiết')))t.classList.add('active')});const el=document.getElementById('mo-body');if(!el)return;el.innerHTML='<div class="loading">Đang tải...</div>';try{let apiUrl;if(tab==='stats')apiUrl=`/api/match/${_currentEventId}/stats`;else if(tab==='comm')apiUrl=`/api/match/${_currentEventId}/commentaries`;else{apiUrl=`/api/match/${_currentEventId}/detail`;if(_currentMatchUrl)apiUrl+='?url='+encodeURIComponent(_currentMatchUrl)}const r=await fetch(apiUrl);if(!r.ok){el.innerHTML='<div class="loading">Lỗi máy chủ ('+r.status+')</div>';return}const d=await r.json();if(d.error){el.innerHTML='<div class="loading">'+esc(d.error)+'</div>';return}if(tab==='detail'&&typeof renderMatchDetail==='function'){renderMatchDetail(el,d);return}el.innerHTML=d.html||'<div class="loading">Không có dữ liệu</div>'}catch(e){el.innerHTML='<div class="loading">Lỗi</div>'}}
388
- function esc(s){return String(s||'').replace(/[&<>"']/g,m=>({'&':'&amp;','<':'&lt;','>':'&gt;','"':'&quot;',"'":'&#39;'}[m]))}
389
- function showView(id){document.querySelectorAll('.view').forEach(v=>v.classList.remove('active'));document.getElementById(id)?.classList.add('active')}
390
- function switchCat(id){document.querySelectorAll('.cat').forEach(c=>c.classList.remove('active'));document.querySelector(`[data-cat="${id}"]`)?.classList.add('active');document.querySelectorAll('.view').forEach(v=>v.classList.remove('active'));document.querySelectorAll('video').forEach(v=>{v.pause();if(v._hls){v._hls.destroy();v._hls=null}});document.querySelectorAll('iframe[data-yt-src]').forEach(f=>{f.src=''});if(id==='home')document.getElementById('view-home').classList.add('active');else if(id==='news-all'){document.getElementById('view-cat').classList.add('active');loadNewsTab()}else{document.getElementById('view-cat').classList.add('active');loadCat(id)}}
391
- function toast(msg){let t=document.getElementById('progress-toast');if(t){t.textContent=msg;t.style.display='block';setTimeout(()=>{t.style.display='none'},3500)}}
392
- function doShare(title,url,img){const shareUrl=SPACE+'/s?url='+encodeURIComponent(url)+'&title='+encodeURIComponent(title)+'&img='+encodeURIComponent(img||'');if(navigator.share)navigator.share({title,url:shareUrl}).catch(()=>{});else navigator.clipboard.writeText(shareUrl).then(()=>toast('Đã sao chép!')).catch(()=>{})}
393
  async function doInteract(videoId,type){try{const r=await fetch('/api/v2/interact',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({id:videoId,type})});return await r.json();}catch(e){return null;}}
394
  async function getInteractions(videoId){try{return await fetch('/api/v2/interactions?id='+encodeURIComponent(videoId)).then(r=>r.json());}catch(e){return{views:0,likes:0,comments:0};}}
395
  async function getComments(videoId){try{const j=await fetch('/api/v2/comments?id='+encodeURIComponent(videoId)).then(r=>r.json());return j.comments||[];}catch(e){return[];}}
396
  async function postComment(videoId,text){try{const j=await fetch('/api/v2/comment',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({id:videoId,text})}).then(r=>r.json());return j.comments||[];}catch(e){return[];}}
397
- function buildTikTokSlide(opts){return`<div class="tiktok-slide" data-vid="${esc(opts.videoId)}">${opts.vtag}<div class="tiktok-bottom"><span class="badge ${opts.badgeClass||'badge-fpt'}">${opts.badge||''}</span><p class="tiktok-title">${esc(opts.title)}</p></div><div class="tiktok-right"><button class="tiktok-right-btn" onclick="event.stopPropagation();doView('${esc(opts.videoId)}',this)"><div class="icon">👁</div><div class="count" id="vc-${opts.idx}">0</div></button><button class="tiktok-right-btn" onclick="event.stopPropagation();doLike('${esc(opts.videoId)}',this)"><div class="icon">❤️</div><div class="count" id="lc-${opts.idx}">0</div></button><button class="tiktok-right-btn" onclick="event.stopPropagation();toggleComments('${esc(opts.videoId)}',${opts.idx})"><div class="icon">💬</div><div class="count" id="cc-${opts.idx}">0</div></button><button class="tiktok-right-btn" onclick="event.stopPropagation();doShare('${esc(opts.title)}','${esc(opts.shareUrl||'')}','')"><div class="icon">📤</div></button><button class="tiktok-right-btn" onclick="event.stopPropagation();toggle169View('${esc(opts.videoId)}','169-toggle-${opts.idx}')"><div class="icon" id="169-toggle-${opts.idx}">🖥️</div></button>${opts.extraBtn||''}</div><span class="tiktok-counter">${opts.idx+1}/${opts.total}</span><div class="inline-comments" id="cmt-inline-${opts.idx}" style="display:none"></div></div>`;}
398
  async function doView(videoId,btn){const j=await doInteract(videoId,'view');if(j){const c=btn.querySelector('.count');if(c)c.textContent=fmtNum(j.views);}}
399
  async function doLike(videoId,btn){const j=await doInteract(videoId,'like');if(j){const c=btn.querySelector('.count');if(c)c.textContent=fmtNum(j.likes);}}
400
  function fmtNum(n){if(!n)return'0';if(n>=1000000)return(n/1000000).toFixed(1)+'M';if(n>=1000)return(n/1000).toFixed(1)+'K';return String(n);}
@@ -402,24 +283,91 @@ async function loadCounters(videoIds){for(let i=0;i<videoIds.length;i++){const i
402
  async function toggleComments(videoId,idx){const panel=document.getElementById('cmt-inline-'+idx);if(!panel)return;if(panel.style.display!=='none'){panel.style.display='none';return;}panel.style.display='block';panel.innerHTML='<div style="padding:8px;color:#888;font-size:11px">Đang tải...</div>';const cmts=await getComments(videoId);renderInlineComments(panel,videoId,idx,cmts);}
403
  function renderInlineComments(panel,videoId,idx,cmts){let h='<div class="inline-cmt-header"><span>💬 Bình luận</span><button onclick="document.getElementById(\'cmt-inline-'+idx+'\').style.display=\'none\'">✕</button></div><div class="inline-cmt-list">';if(cmts.length){cmts.slice(-30).forEach(c=>{h+=`<div class="inline-cmt-item"><span class="inline-cmt-time">${c.time||''}</span>${esc(c.text)}</div>`;});}else{h+='<div style="color:#777;font-size:11px;padding:4px">Chưa có bình luận</div>';}h+=`</div><div class="inline-cmt-input"><input id="cmt-input-${idx}" placeholder="Viết bình luận..." onkeydown="if(event.key==='Enter')submitInlineCmt('${esc(videoId)}',${idx})"><button onclick="submitInlineCmt('${esc(videoId)}',${idx})">Gửi</button></div>`;panel.innerHTML=h;const list=panel.querySelector('.inline-cmt-list');if(list)list.scrollTop=list.scrollHeight;}
404
  async function submitInlineCmt(videoId,idx){const inp=document.getElementById('cmt-input-'+idx);if(!inp)return;const text=inp.value.trim();if(!text)return;inp.value='';inp.disabled=true;const cmts=await postComment(videoId,text);inp.disabled=false;const panel=document.getElementById('cmt-inline-'+idx);if(panel)renderInlineComments(panel,videoId,idx,cmts);const cc=document.getElementById('cc-'+idx);if(cc)cc.textContent=fmtNum(cmts.length);}
405
- function toggle169View(videoId,iconId){const slide=videoId?document.querySelector(`.tiktok-slide[data-vid="${videoId}"]`):null;if(slide){slide.classList.toggle('ratio-wide');const iconEl=iconId?document.getElementById(iconId):null;if(iconEl)iconEl.textContent=slide.classList.contains('ratio-wide')?'📺':'🖥️';else{const btn=slide.querySelector('.tiktok-right-btn .icon');if(btn)btn.textContent=slide.classList.contains('ratio-wide')?'📺':'🖥️';}return}document.querySelectorAll('.tiktok-slide.ratio-wide').forEach(s=>s.classList.remove('ratio-wide'));document.querySelectorAll('.tiktok-slide').forEach(s=>s.classList.add('ratio-wide'));document.querySelectorAll('.tiktok-right-btn .icon').forEach(b=>{if(b.textContent==='🖥️')b.textContent='📺';})}
406
  function initTikTokFeed(){const feed=document.getElementById('tiktok-feed');if(!feed)return;const slides=feed.querySelectorAll('.tiktok-slide');let cur=-1;function act(i){if(i===cur)return;slides.forEach((sl,idx)=>{const v=sl.querySelector('video');const fr=sl.querySelector('iframe');if(idx===i){if(v&&v.dataset.hls&&!v._hls&&typeof Hls!=='undefined'&&Hls.isSupported()){const hls=new Hls();hls.loadSource(v.dataset.hls);hls.attachMedia(v);hls.on(Hls.Events.MANIFEST_PARSED,()=>v.play().catch(()=>{}));v._hls=hls}else if(v)v.play().catch(()=>{});if(fr&&!fr.src&&fr.dataset.ytSrc)fr.src=fr.dataset.ytSrc;const vid=sl.dataset.vid;if(vid&&!sl._viewed){sl._viewed=true;doInteract(vid,'view').then(j=>{if(j){const vc=document.getElementById('vc-'+idx);if(vc)vc.textContent=fmtNum(j.views);}});}}else{if(v){v.pause();if(v._hls){v._hls.destroy();v._hls=null}}if(fr&&fr.src)fr.src=''}});cur=i}let sT;feed.addEventListener('scroll',()=>{clearTimeout(sT);sT=setTimeout(()=>{const rect=feed.getBoundingClientRect(),ctr=rect.top+rect.height/2;let best=-1,bestD=1e9;slides.forEach((sl,i)=>{const d=Math.abs(sl.getBoundingClientRect().top+sl.getBoundingClientRect().height/2-ctr);if(d<bestD){bestD=d;best=i}});if(best>=0)act(best)},150)});setTimeout(()=>act(0),400);slides.forEach(sl=>{const v=sl.querySelector('video');if(v)v.addEventListener('click',e=>{e.preventDefault();v.paused?v.play().catch(()=>{}):v.pause()})});const ids=[...slides].map(sl=>sl.dataset.vid||'');loadCounters(ids)}
407
- async function openHighlightFeed(league,idx,forceUrl){showView('view-tiktok');const el=document.getElementById('view-tiktok');el.innerHTML='<div class="loading">Đang tải...</div>';let articles=(_hlLeagueData||{})[league]||[];if(!articles.length){try{articles=await fetch('/api/highlights/'+league).then(r=>r.json())}catch(e){articles=[]}}if(!articles.length){el.innerHTML='<div class="loading">Không có video</div>';return}const vids=[];const results=await Promise.all(articles.map(async(a,i)=>{try{const r=await fetch('/api/video_url?url='+encodeURIComponent(a.link)+'&img='+encodeURIComponent(a.img||''));const v=await r.json();if(v&&v.src){return{_idx:i,title:a.title||v.title||'',link:a.link||'',img:a.img||v.poster||'',src:v.src,type:v.type||'',poster:v.poster||a.img||''}}return null}catch(e){}return null}));results.forEach(r=>{if(r)vids.push(r)});vids.sort((a,b)=>a._idx-b._idx);if(!vids.length){el.innerHTML='<div class="loading">Không tìm thấy video</div>';return}let ti=vids.findIndex(v=>v._idx===idx);if(ti<0)ti=0;const ordered=ti>0?[...vids.slice(ti),...vids.slice(0,ti)]:vids;let h=`<button class="back-btn" onclick="switchCat('home')">← Highlight</button><div class="tiktok-container"><div class="tiktok-feed" id="tiktok-feed">`;ordered.forEach((v,i)=>{const isYT=v.type==='youtube',isHLS=!isYT&&v.src?.includes('.m3u8'),poster=v.poster?` poster="${esc(v.poster)}"`:'';const vtag=isYT?`<iframe data-yt-src="${v.src}" allowfullscreen allow="accelerometer;autoplay;clipboard-write;encrypted-media;gyroscope;picture-in-picture"></iframe>`:isHLS?`<video playsinline preload="none"${poster} data-hls="${v.src}" loop controls></video>`:`<video playsinline preload="none"${poster} loop controls><source src="${v.src}" type="video/mp4"></video>`;const videoId='hl-'+league+'-'+v._idx;h+=buildTikTokSlide({vtag,title:v.title,badge:'HL',badgeClass:'badge-fpt',videoId,idx:i,total:ordered.length,shareUrl:v.link||'',extraBtn:`<button class="tiktok-right-btn" onclick="event.stopPropagation();openHighlightFeed('${league}',${(i+1)%ordered.length})"><div class="icon">⏭️</div></button>`})});h+='</div></div>';el.innerHTML=h;setTimeout(()=>initTikTokFeed(),200);}
408
- async function openYTShortsFeed(idx){showView('view-tiktok');const el=document.getElementById('view-tiktok');el.innerHTML='<div class="loading">Đã xóa Shorts Dân trí/SKĐS</div>';}
409
- async function openShortAIFeed(idx){showView('view-tiktok');const el=document.getElementById('view-tiktok');if(!_wallPosts||!_wallPosts.length){el.innerHTML='<div class="loading">Không có Short AI</div>';return}const aiPosts=_wallPosts.filter(p=>p.video);if(!aiPosts.length||idx>=aiPosts.length){el.innerHTML='<div class="loading">Không có Short AI</div>';return}const ordered=aiPosts.slice(idx).concat(aiPosts.slice(0,idx));let h=`<button class="back-btn" onclick="switchCat('home')">← Tường AI</button><div class="tiktok-container"><div class="tiktok-feed" id="tiktok-feed">`;ordered.forEach((p,i)=>{const baseIdx=_wallPosts.indexOf(p);const vtag=p.video?`<video playsinline preload="none" src="${esc(p.video)}" loop controls></video>`:'';h+=buildTikTokSlide({vtag,title:p.title,badge:'Short AI',badgeClass:'badge-ai',videoId:p.id||'ai-'+i,idx:i,total:ordered.length,shareUrl:'',extraBtn:`<button class="tiktok-right-btn" onclick="event.stopPropagation();showVoiceEmotionSelector('${esc(p.id)}','${esc(p.title)}','${esc((p.text||'').slice(0,200))}')"><div class="icon">🎤</div></button>`})});h+='</div></div>';el.innerHTML=h;setTimeout(()=>initTikTokFeed(),200);}
410
- function makeShortVideo(postId, btn, voice, speed, emotion){
411
- if(!postId)return;
412
- if(!voice||!emotion){voice='vi-VN-HoaiMyNeural';emotion='neutral';}
413
- toast(' Đang tạo...');
414
- fetch('/api/ai/short/'+encodeURIComponent(postId),{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({voice,emotion,speed:speed||1.2})}).then(r=>r.json()).then(j=>{if(j.error)throw new Error(j.error);toast('✅ OK!');}).catch(e=>toast(''+e.message));
415
- }
416
- function readWallPost(idx){const p=_wallPosts&&_wallPosts[idx];if(!p)return;if(p.slides&&p.slides.length){readSlidePost(idx);return}readArticle(p.url||'','','',p.title,p.text);}
417
- /** Show rewrite slide viewer - vertical slides with text+image */
418
- function readSlidePost(idx){const p=_wallPosts[idx];if(!p||!p.slides)return;showView('view-article');const el=document.getElementById('view-article');let h=`<button class="back-btn" onclick="switchCat('home')">← Tường AI</button><div class="slide-viewer" style="padding:12px;max-width:600px;margin:0 auto">`;p.slides.forEach((s,i)=>{h+=`<div class="slide-card" style="background:#1a1a1a;border:1px solid #2a2a2a;border-radius:12px;padding:16px;margin-bottom:12px"><div class="slide-num" style="color:#5cb87a;font-size:12px;font-weight:700;margin-bottom:6px">Slide ${s.index||i+1}/${p.slides.length}</div>${s.image?`<img src="${_proxyImg(s.image)}" style="width:100%;max-height:300px;object-fit:cover;border-radius:8px;margin-bottom:8px" loading="lazy" onerror="this.style.display=\'none\'">`:''}<p style="color:#ddd;font-size:14px;line-height:1.6;margin:0">${esc(s.text)}</p></div>`;});h+=`</div>`;el.innerHTML=h;}
419
- function readNewsTab(tab){loadNewsTab();}
420
- function loadNewsTab(){const el=document.getElementById('view-cat');if(!el)return;el.innerHTML='<div class="loading">Đang tải tin tức...</div>';fetch('/api/homepage').then(r=>r.json()).then(articles=>{if(!articles||!articles.length){el.innerHTML='<div class="loading">Không có tin</div>';return}let h='<div class="grid">';articles.forEach(a=>{const src=a.source||'vne';const badge=a.group||a.source||'';h+=`<div class="card" onclick="readArticle('${esc(a.link)}')"><div class="card-img">${a.img?`<img src="${_proxyImg(a.img)}" loading="lazy" onerror="this.style.display=\'none\'">`:''}</div><div class="card-body"><span class="badge badge-${src}">${esc(badge)}</span><div class="card-title">${esc(a.title)}</div></div></div>`;});h+='</div>';el.innerHTML=h;}).catch(()=>{el.innerHTML='<div class="loading">Lỗi tải</div>';});}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
421
 
422
- function readArticle(url,title,img,presetTitle,presetText){showView('view-article');const el=document.getElementById('view-article');el.innerHTML='<div class="loading">Đang tải...</div><button class="back-btn" onclick="switchCat(\'home\')">← Quay lại</button>';if(presetTitle){el.innerHTML=`<button class="back-btn" onclick="switchCat('home')">← Quay lại</button><div class="article-view"><h1 class="article-title">${esc(presetTitle)}</h1>${presetText?`<div class="article-summary">${esc(presetText)}</div>`:''}</div>`;return;}if(!url)return;fetch('/api/article?url='+encodeURIComponent(url)).then(r=>r.json()).then(d=>{let h=`<button class="back-btn" onclick="switchCat('home')">← Quay lại</button><div class="article-view">`;if(d.title)h+=`<h1 class="article-title">${esc(d.title)}</h1>`;if(d.summary)h+=`<div class="article-summary">${esc(d.summary)}</div>`;if(d.body)d.body.forEach(b=>{if(b.type==='p')h+=`<p class="article-p">${esc(b.text)}</p>`;else if(b.type==='heading')h+=`<h2 class="article-h2">${esc(b.text)}</h2>`;else if(b.type==='img'&&b.src)h+=`<img class="article-img" src="${_proxyImg(b.src)}" loading="lazy" onerror="this.style.display=\'none\'">`;});h+=`</div><div class="article-actions"><button onclick="doShare('${esc(d.title||'')}','${esc(url)}','${esc(d.og_image||'')}')">📤 Chia sẻ</button><button class="primary" onclick="rewriteSlide('${esc(url)}')">🤖 Slide Rewrite AI</button></div>`;el.innerHTML=h;}).catch(()=>{el.innerHTML=`<button class="back-btn" onclick="switchCat('home')">← Quay lại</button><div class="article-view"><p>Không thể tải bài viết</p></div>`;});}
423
- async function rewriteSlide(url){if(!url)return;const btn=document.querySelector('.article-actions .primary')||event?.target;if(btn){btn.disabled=true;btn.textContent='⏳ Đang tạo slides...';}toast('⏳ Đang tạo slide rewrite...');try{const r=await fetch('/api/rewrite_slide',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({url})});const j=await r.json();if(!r.ok||j.error)throw new Error(j.error||'Lỗi');toast('✅ Đã tạo slide! Xem trong Tường AI.');if(btn)btn.textContent='✅ Hoàn tất';if(j.post)prependWallPost(j.post);}catch(e){toast('❌ '+e.message);if(btn){btn.disabled=false;btn.textContent='🤖 Slide Rewrite AI';}}}
424
- function rewriteUrl(){const url=document.getElementById('url-input')?.value.trim();if(!url){alert('Nhập URL');return;}readArticle(url);}
425
- function loadCat(id){const el=document.getElementById('view-cat');el.innerHTML='<div class="loading">Đang tải...</div>';fetch('/api/category/'+id).then(r=>r.json()).then(articles=>{if(!articles||!articles.length){el.innerHTML='<div class="loading">Không có bài viết</div>';return}let h='<div class="grid">';articles.forEach(a=>{h+=`<div class="card" onclick="readArticle('${esc(a.link)}')"><div class="card-img">${a.img?`<img src="${_proxyImg(a.img)}" loading="lazy" onerror="this.style.display=\'none\'">`:''}</div><div class="card-body"><span class="badge badge-${a.source||'vne'}">${esc(a.source||'')}</span><div class="card-title">${esc(a.title)}</div></div></div>`;});h+='</div>';el.innerHTML=h;}).catch(()=>{el.innerHTML='<div class="loading">Lỗi tải</div>';});}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ // === VNEWS Frontend v2 - Optimized for speed ===
 
 
 
 
 
 
 
 
 
 
 
 
2
 
3
+ // === LOAD HOME - Fast: immediate shell + parallel fetch ===
4
  function _fetchWithTimeout(url, ms){
5
  return new Promise((resolve,reject)=>{
6
  const ctrl=new AbortController();
 
17
  const homeEl = document.getElementById('view-home');
18
  if(!homeEl) return;
19
 
20
+ // Build shell IMMEDIATELY — no skeleton, no delay
21
  homeEl.innerHTML =
22
  '<div id="home-featured-area"></div>'
23
  +'<div class="ai-compose"><div class="ai-compose-title">🤖 AI viết bài</div><div class="ai-compose-row"><input id="topic-input" placeholder="Nhập chủ đề..."><button onclick="searchTopic()">Tìm nguồn</button></div><div class="ai-compose-row"><input id="url-input" placeholder="Dán URL bài viết..."><button class="secondary" onclick="rewriteUrl()">Rewrite</button></div><div id="hot-topics" class="hot-topic-row"></div></div>'
 
28
 
29
  const afterEl = homeEl.querySelector('#home-after-wc');
30
 
31
+ // Start critical loads immediately
32
  loadLivescore('today');
33
  loadHotTopics();
34
 
35
+ // Fetch all data in parallel with shorter timeouts
36
+ const [featuredData, shortsData, wallData, hlLeagues, aiData, wcData] = await Promise.allSettled([
37
  _fetchWithTimeout('/api/livescore/featured', 5000),
38
+ _fetchWithTimeout('/api/shorts', 8000),
39
  _fetchWithTimeout('/api/wall', 5000),
40
  _fetchWithTimeout('/api/highlights/leagues', 10000),
41
+ _fetchWithTimeout('/api/genk_ai', 8000),
42
  _fetchWithTimeout('/api/wc2026', 8000),
43
  ]).then(results => results.map(r => r.status === 'fulfilled' ? r.value : null));
44
 
45
+ // Render featured match
46
  if(featuredData && featuredData.home){
47
  const sc=featuredData.status==='live'?'':'upcoming';
48
  const st=featuredData.status==='live'?`🔴 ${featuredData.minute||'LIVE'}`:`⏰ ${featuredData.time}`;
49
  const area=document.getElementById('home-featured-area');
50
+ if(area) area.innerHTML=`<div class="featured-match" onclick="openMatch('${featuredData.event_id}')"><div class="fm-league">${featuredData.league}</div><div class="fm-teams"><div class="fm-team"><img src="${featuredData.home_logo}" onerror="this.style.display='none'"><span>${featuredData.home}</span></div><div class="fm-score">${featuredData.score||'VS'}</div><div class="fm-team"><img src="${featuredData.away_logo}" onerror="this.style.display='none'"><span>${featuredData.away}</span></div></div><div class="fm-status ${sc}">${st}</div></div>`;
51
  }
52
 
53
+ // Store globally
54
+ _shortsData = shortsData || [];
55
  _wallPosts = (wallData && wallData.posts) || [];
56
  _hlLeagueData = hlLeagues || {};
57
  _wc2026Data = wcData;
58
 
59
+ // Render WC if data arrived
60
  if(wcData) switchWCTab('news');
61
 
62
+ // Render sections into the after-wc area
63
+ _renderShortsIn(afterEl);
64
  _renderWallIn(afterEl);
65
  _renderHLIn(afterEl);
66
+ if(aiData && aiData.length) _renderSlidesIn('ai-articles','Ứng dụng AI','🤖',aiData,afterEl);
67
  }
68
 
69
  function _renderSlidesIn(key, label, emoji, vids, afterEl){
 
74
  const isHL = key==='world-cup'||key==='premier-league'||key==='champions-league'||key==='la-liga'||key==='serie-a'||key==='bundesliga'||key==='friendly';
75
  vids.slice(0,isHL?8:12).forEach((a,i)=>{
76
  if(isHL){
77
+ h+=`<div class="slider-item" onclick="openHighlightFeed('${key}',${i})"><div class="slider-thumb">${a.img?`<img src="${a.img}" loading="lazy">`:''}<div class="card-play">▶</div></div><div class="slider-title">${esc(a.title)}</div></div>`;
78
  } else {
79
+ h+=`<div class="slider-item" onclick="readArticle('${esc(a.link)}')"><div class="slider-thumb">${a.img?`<img src="${a.img}" loading="lazy">`:''}</div><div class="slider-title">${esc(a.title)}</div></div>`;
80
  }
81
  });
82
  h+='</div>';
 
84
  afterEl.parentNode.insertBefore(wrap, afterEl);
85
  }
86
 
87
+ function _renderShortsIn(afterEl){
88
+ if(!_shortsData||!_shortsData.length||!afterEl) return;
89
+ const mixed=interleaveShorts(_shortsData);
90
+ if(!mixed.length) return;
91
+ const wrap=document.createElement('div');
92
+ wrap.className='slider-wrap';
93
+ let h=`<div class="slider-header"><span class="slider-label">📱 Shorts Dân trí & SKĐS</span><span class="slider-note">Mới nhất · xen kẽ</span></div><div class="slider-track">`;
94
+ mixed.slice(0,30).forEach((a,i)=>{
95
+ const badge=a.channel==='baosuckhoedoisongboyte'?'SKĐS':'Dân trí';
96
+ h+=`<div class="slider-item shorts-item" onclick="openYTShortsFeed(${i})"><div class="slider-thumb shorts-thumb">${a.img?`<img src="${a.img}" loading="lazy">`:''}<div class="card-play">▶</div></div><div class="slider-title"><span style="color:#f0c040;font-size:8px">${badge}</span> ${esc(a.title)}</div></div>`;
97
+ });
98
+ h+='</div>';wrap.innerHTML=h;
99
+ afterEl.parentNode.insertBefore(wrap,afterEl);
100
+ }
101
+
102
  function _renderWallIn(afterEl){
103
  if(!_wallPosts||!_wallPosts.length||!afterEl) return;
104
  const posts=_wallPosts;
105
+ const aiShorts=posts.filter(p=>p.video);
106
+ if(aiShorts.length){
107
+ const wrap=document.createElement('div');
108
+ wrap.className='slider-wrap';
109
+ let h='<div class="slider-header"><span class="slider-label">🎬 Short AI</span></div><div class="slider-track">';
110
+ aiShorts.slice(0,20).forEach((p,i)=>{h+=`<div class="slider-item shorts-item" onclick="openShortAIFeed(${i})"><div class="slider-thumb shorts-thumb"><video src="${p.video}" muted preload="metadata"></video><div class="card-play">▶</div></div><div class="slider-title">${esc(p.title)}</div></div>`;});
111
+ h+='</div>';wrap.innerHTML=h;
112
+ afterEl.parentNode.insertBefore(wrap,afterEl);
113
+ }
114
  const wrap=document.createElement('div');
115
  wrap.className='slider-wrap';wrap.id='ai-wall-wrap';
116
  let h='<div class="slider-header"><span class="slider-label">🧱 Tường AI</span></div><div class="slider-track" id="ai-wall-track">';
 
131
 
132
  // === WALL POST HELPERS ===
133
  function makeWallItem(p,i){
134
+ const hasVideo = p.video && p.video.length > 0;
135
+ const thumbContent = p.img
136
+ ? `<img src="${esc(p.img)}" loading="lazy" onerror="this.style.display='none'">`
137
+ : (hasVideo ? `<video src="${esc(p.video)}" muted></video>` : '');
138
+ const videoBadge = hasVideo ? `<div class="wall-video-badge">🎬</div>` : '';
139
+ const videoBtn = hasVideo
140
+ ? `<button class="wall-btn-video" onclick="event.stopPropagation();openShortAIFeed(${i})">▶ Xem Short</button>`
141
+ : `<button class="wall-btn-make" onclick="event.stopPropagation();makeShortVideo('${esc(p.id||i)}',this)">🎬 Tạo Video</button>`;
142
+ return `<div class="wall-item" id="wall-item-${esc(p.id||i)}"><div class="wall-thumb">${thumbContent}${videoBadge}</div><div class="wall-title">${esc(p.title)}</div><div class="wall-text">${esc((p.text||'').slice(0,180))}</div><div class="wall-actions"><button class="primary" onclick="readWallPost(${i})">Xem</button>${videoBtn}</div></div>`;
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
143
  }
144
 
145
+ async function makeShortVideo(postId, btn, voice, speed){
146
  if(!postId)return;
147
  const origText = btn ? btn.textContent : '🎬 Tạo Video';
148
  if(btn){btn.disabled=true;btn.textContent='⏳ Đang tạo...';}
149
  toast('⏳ Đang tạo video shorts...');
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
150
  try{
151
+ let url = '/api/ai/short/'+encodeURIComponent(postId);
152
+ const params = [];
153
+ if(voice) params.push('voice='+encodeURIComponent(voice));
154
+ if(speed) params.push('speed='+encodeURIComponent(speed));
155
+ if(params.length) url += '?' + params.join('&');
156
+ const r = await fetch(url, {method:'POST'});
157
  const j = await r.json();
158
  if(!r.ok || j.error) throw new Error(j.error||'Lỗi tạo video');
159
  toast('✅ Đã tạo video shorts!');
160
  const p = _wallPosts.find(x => String(x.id) === String(postId));
161
  if(p){
162
  p.video = j.video;
 
 
163
  const itemId = 'wall-item-'+postId;
164
  const el = document.getElementById(itemId);
165
  if(el){
166
  const idx = _wallPosts.indexOf(p);
167
+ el.outerHTML = makeWallItem(p, idx);
168
+ const newEl = document.getElementById(itemId);
169
+ if(newEl) newEl.className = 'wall-item wall-item-new';
170
  }
171
  }
172
+ refreshShortAISlider();
173
  }catch(e){
174
  toast('❌ '+e.message);
175
  if(btn){btn.disabled=false;btn.textContent=origText;}
176
  }
177
  }
178
 
179
+ function refreshShortAISlider(){
180
+ const aiShorts = _wallPosts.filter(p=>p.video);
181
+ let shortAISection = document.getElementById('short-ai-section');
182
+ if(aiShorts.length === 0){
183
+ if(shortAISection) shortAISection.remove();
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
184
  return;
185
  }
186
+ if(shortAISection){
187
+ const track = shortAISection.querySelector('.slider-track');
188
+ if(track){
189
+ let h = '';
190
+ aiShorts.slice(0,20).forEach((p,i)=>{
191
+ h+=`<div class="slider-item shorts-item" onclick="openShortAIFeed(${i})"><div class="slider-thumb shorts-thumb"><video src="${esc(p.video)}" muted preload="metadata"></video><div class="card-play">▶</div></div><div class="slider-title">${esc(p.title)}</div></div>`;
192
+ });
193
+ track.innerHTML = h;
 
 
194
  }
 
195
  }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
196
  }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
197
 
198
  function prependWallPost(post){
199
  _wallPosts.unshift(post);
 
204
  if(homeEl){
205
  let insertBefore=homeEl.querySelector('.slider-wrap');
206
  const newWrap=document.createElement('div');
207
+ newWrap.className='slider-wrap';
208
+ newWrap.id='ai-wall-wrap';
209
  newWrap.innerHTML=`<div class="slider-header"><span class="slider-label">🧱 Tường AI</span></div><div class="slider-track" id="ai-wall-track">${makeWallItem(post,0)}</div>`;
210
  if(insertBefore) homeEl.insertBefore(newWrap,insertBefore);
211
  else homeEl.appendChild(newWrap);
 
214
  }
215
  return;
216
  }
217
+ const div=document.createElement('div');
218
+ div.className='wall-item wall-item-new';
219
+ div.id='wall-item-'+(post.id||'new-'+Date.now());
220
+ const hasVideo = post.video && post.video.length > 0;
221
+ const thumbContent = post.img
222
+ ? `<img src="${esc(post.img)}" loading="lazy" onerror="this.style.display='none'">`
223
+ : (hasVideo ? `<video src="${esc(post.video)}" muted></video>` : '');
224
+ const videoBadge = hasVideo ? `<div class="wall-video-badge">🎬</div>` : '';
225
+ const videoBtn = hasVideo
226
+ ? `<button class="wall-btn-video" onclick="event.stopPropagation();openShortAIFeed(0)">▶ Xem Short</button>`
227
+ : `<button class="wall-btn-make" onclick="event.stopPropagation();makeShortVideo('${esc(post.id)}',this)">🎬 Tạo Video</button>`;
228
+ div.innerHTML=`<div class="wall-thumb">${thumbContent}${videoBadge}</div><div class="wall-title">${esc(post.title)}</div><div class="wall-text">${esc((post.text||'').slice(0,180))}</div><div class="wall-actions"><button class="primary" onclick="readWallPost(0)">Xem</button>${videoBtn}</div>`;
229
+ track.prepend(div);
230
  track.scrollTo({left:0,behavior:'smooth'});
231
+ if(hasVideo) refreshShortAISlider();
232
  }
233
 
234
+ // === REST OF FUNCTIONS ===
235
+ let _shortsData=[];
236
  let _wallPosts=[];
237
  let _currentView='home';
238
  let _currentEventId=null;
239
  let _currentMatchUrl=null;
240
+ function interleaveShorts(shorts){const dt=shorts.filter(s=>s.channel==='baodantri7941');const sk=shorts.filter(s=>s.channel==='baosuckhoedoisongboyte');const result=[];let i=0,j=0;while(i<dt.length||j<sk.length){if(i<dt.length)result.push(dt[i++]);if(j<sk.length)result.push(sk[j++]);}return result;}
241
  let _htPage=0,_htTopic='';
242
+ async function loadHotTopics(){const j=await fetch('/api/hot_topics').then(r=>r.json()).catch(()=>({topics:[]}));const el=document.getElementById('hot-topics');if(!el)return;el.innerHTML=(j.topics||[]).slice(0,18).map(t=>{const topicText=t.topic||t.label.replace(/^#/,'');return`<button class="hot-chip" onclick="searchTopic('${topicText.replace(/'/g,"\\'")}')">${esc(t.label)}</button>`;}).join('');if(j.topics&&j.topics[0]){const firstTopic=j.topics[0].topic||j.topics[0].label.replace(/^#/,'');setTimeout(()=>searchTopic(firstTopic),800);}}
243
  function searchTopic(topic){if(!topic){topic=document.getElementById('topic-input')?.value.trim();if(!topic){alert('Nhập chủ đề');return;}}document.getElementById('topic-input').value='';_htTopic=topic;_htPage=0;showHashtagSources(topic,0);}
244
+ async function showHashtagSources(topic,page){const box=document.getElementById('hashtag-box');if(!box)return;if(page===0)box.innerHTML=`<div class="hashtag-sources"><h3>🔍 ${esc(topic)}</h3><div class="hashtag-loading"><div class="hashtag-spinner"></div>Đang tìm...</div></div>`;try{const r=await fetch(`/api/hashtag/sources?topic=${encodeURIComponent(topic)}&page=${page}`);const j=await r.json();const sources=j.sources||[];if(!sources.length&&page===0){box.innerHTML=`<div class="hashtag-sources"><h3>🔍 ${esc(topic)}</h3><div style="color:#888;padding:8px">Không tìm được bài viết liên quan</div></div>`;return;}let h='';if(page===0)h=`<div class="hashtag-sources"><h3>🔍 ${esc(topic)} <span style="font-size:10px;color:#888">(${j.total} bài từ 8 nguồn)</span></h3><div id="ht-list">`;sources.forEach((s,i)=>{const idx=page*8+i;h+=`<div class="hashtag-src-item" onclick="readArticle('${esc(s.url)}')"><div class="hashtag-src-img" id="ht-img-${idx}"></div><div class="hashtag-src-text"><div class="hashtag-src-title">${esc(s.title)}</div><div class="hashtag-src-via">${esc(s.via||'')}</div></div></div>`;});if(page===0){h+=`</div><button class="hashtag-rewrite-btn" onclick="rewriteHashtag('${esc(topic).replace(/'/g,"\\'")}')">🤖 Rewrite AI tổng hợp & đăng tường</button>`;if(j.has_more)h+=`<button class="hashtag-load-more" id="ht-more" onclick="loadMoreHashtag()">Tải thêm ▼</button>`;h+=`</div>`;box.innerHTML=h;}else{document.getElementById('ht-list')?.insertAdjacentHTML('beforeend',h);const btn=document.getElementById('ht-more');if(btn){if(!j.has_more)btn.remove();else{btn.disabled=false;btn.textContent='Tải thêm ▼';}}}sources.forEach((s,i)=>{const idx=page*8+i;if(!s.url)return;fetch('/api/article?url='+encodeURIComponent(s.url)).then(r=>r.json()).then(d=>{if(d&&(d.og_image||d.img)){const el=document.getElementById('ht-img-'+idx);if(el)el.innerHTML=`<img src="${esc(d.og_image||d.img)}" onerror="this.style.display='none'">`;}}).catch(()=>{});});}catch(e){box.innerHTML=`<div class="hashtag-sources"><h3>🔍 ${esc(topic)}</h3><div style="color:#e74c3c;padding:8px">Lỗi: ${esc(e.message)}</div></div>`;}}
245
  function loadMoreHashtag(){_htPage++;const btn=document.getElementById('ht-more');if(btn){btn.disabled=true;btn.textContent='Đang tải...';}showHashtagSources(_htTopic,_htPage);}
246
  async function rewriteHashtag(topic){const btn=event?.target;if(btn){btn.disabled=true;btn.textContent='Đang tổng hợp...';}try{const r=await fetch('/api/topic_post',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({topic})});const j=await r.json();if(!r.ok||j.error)throw new Error(j.error||'Lỗi');toast('✅ Đã đăng Tường AI!');if(btn)btn.textContent='✅ Đăng thành công!';if(j.post)prependWallPost(j.post);}catch(e){toast('❌ '+e.message);if(btn){btn.disabled=false;btn.textContent='🤖 Rewrite AI';}}}
247
  async function loadLivescore(tab){document.querySelectorAll('.ls-tab').forEach(t=>t.classList.remove('active'));document.querySelector(`.ls-tab[data-tab="${tab}"]`)?.classList.add('active');const el=document.getElementById('ls-content');if(!el)return;el.innerHTML='<div class="loading">Đang tải...</div>';let ep='/api/livescore/'+tab;if(tab.startsWith('bxh_'))ep='/api/livescore/standings/'+tab.replace('bxh_','');try{const r=await fetch(ep);const d=await r.json();el.innerHTML=d.html&&d.html.length>50?d.html:'<div class="loading">Không có dữ liệu</div>';bindMatchClicks(el);}catch(e){el.innerHTML='<div class="loading">Lỗi</div>';}}
248
+ function bindMatchClicks(el){
249
+ el.querySelectorAll('.match-detail').forEach(md=>{
250
+ md.style.cursor='pointer';
251
+ md.addEventListener('click',function(e){
252
+ const statusA=this.querySelector('.status a');
253
+ const teamA=this.querySelector('.teams a[href*="/tran-dau/"]');
254
+ const a = statusA || teamA;
255
+ if(a){
256
+ e.preventDefault();
257
+ e.stopPropagation();
258
+ const href=a.getAttribute('href')||'';
259
+ const m=href.match(/\/tran-dau\/(\d+)\//);
260
+ if(m){
261
+ const fullUrl=href.startsWith('http')?href:'https://bongda.com.vn'+href;
262
+ openMatch(m[1],fullUrl);
263
+ }
264
+ }
265
+ });
266
+ });
267
+ el.querySelectorAll('a').forEach(a=>{
268
+ a.addEventListener('click',e=>{e.preventDefault();e.stopPropagation()});
269
+ });
270
+ }
271
  function openMatch(id,url){if(!id)return;_currentEventId=id;if(url)_currentMatchUrl=url;document.getElementById('match-overlay').classList.add('active');document.body.style.overflow='hidden';loadMatchTab('detail')}
272
  function closeMatch(){document.getElementById('match-overlay').classList.remove('active');document.body.style.overflow=''}
273
  async function loadMatchTab(tab){document.querySelectorAll('.mo-tab').forEach(t=>t.classList.remove('active'));document.querySelectorAll('.mo-tab').forEach(t=>{if((tab==='comm'&&t.textContent==='Diễn biến')||(tab==='stats'&&t.textContent==='Thống kê')||(tab==='detail'&&t.textContent.includes('Chi tiết')))t.classList.add('active')});const el=document.getElementById('mo-body');if(!el)return;el.innerHTML='<div class="loading">Đang tải...</div>';try{let apiUrl;if(tab==='stats')apiUrl=`/api/match/${_currentEventId}/stats`;else if(tab==='comm')apiUrl=`/api/match/${_currentEventId}/commentaries`;else{apiUrl=`/api/match/${_currentEventId}/detail`;if(_currentMatchUrl)apiUrl+='?url='+encodeURIComponent(_currentMatchUrl)}const r=await fetch(apiUrl);if(!r.ok){el.innerHTML='<div class="loading">Lỗi máy chủ ('+r.status+')</div>';return}const d=await r.json();if(d.error){el.innerHTML='<div class="loading">'+esc(d.error)+'</div>';return}if(tab==='detail'&&typeof renderMatchDetail==='function'){renderMatchDetail(el,d);return}el.innerHTML=d.html||'<div class="loading">Không có dữ liệu</div>'}catch(e){el.innerHTML='<div class="loading">Lỗi</div>'}}
 
 
 
 
 
274
  async function doInteract(videoId,type){try{const r=await fetch('/api/v2/interact',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({id:videoId,type})});return await r.json();}catch(e){return null;}}
275
  async function getInteractions(videoId){try{return await fetch('/api/v2/interactions?id='+encodeURIComponent(videoId)).then(r=>r.json());}catch(e){return{views:0,likes:0,comments:0};}}
276
  async function getComments(videoId){try{const j=await fetch('/api/v2/comments?id='+encodeURIComponent(videoId)).then(r=>r.json());return j.comments||[];}catch(e){return[];}}
277
  async function postComment(videoId,text){try{const j=await fetch('/api/v2/comment',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({id:videoId,text})}).then(r=>r.json());return j.comments||[];}catch(e){return[];}}
278
+ function buildTikTokSlide(opts){return`<div class="tiktok-slide" data-vid="${esc(opts.videoId)}">${opts.vtag}<div class="tiktok-bottom"><span class="badge ${opts.badgeClass||'badge-fpt'}">${opts.badge||''}</span><p class="tiktok-title">${esc(opts.title)}</p></div><div class="tiktok-right"><button class="tiktok-right-btn" onclick="event.stopPropagation();doView('${esc(opts.videoId)}',this)"><div class="icon">👁</div><div class="count" id="vc-${opts.idx}">0</div></button><button class="tiktok-right-btn" onclick="event.stopPropagation();doLike('${esc(opts.videoId)}',this)"><div class="icon">❤️</div><div class="count" id="lc-${opts.idx}">0</div></button><button class="tiktok-right-btn" onclick="event.stopPropagation();toggleComments('${esc(opts.videoId)}',${opts.idx})"><div class="icon">💬</div><div class="count" id="cc-${opts.idx}">0</div></button><button class="tiktok-right-btn" onclick="event.stopPropagation();doShare('${esc(opts.title)}','${esc(opts.shareUrl||'')}','')"><div class="icon">📤</div></button>${opts.extraBtn||''}</div><span class="tiktok-counter">${opts.idx+1}/${opts.total}</span><div class="inline-comments" id="cmt-inline-${opts.idx}" style="display:none"></div></div>`;}
279
  async function doView(videoId,btn){const j=await doInteract(videoId,'view');if(j){const c=btn.querySelector('.count');if(c)c.textContent=fmtNum(j.views);}}
280
  async function doLike(videoId,btn){const j=await doInteract(videoId,'like');if(j){const c=btn.querySelector('.count');if(c)c.textContent=fmtNum(j.likes);}}
281
  function fmtNum(n){if(!n)return'0';if(n>=1000000)return(n/1000000).toFixed(1)+'M';if(n>=1000)return(n/1000).toFixed(1)+'K';return String(n);}
 
283
  async function toggleComments(videoId,idx){const panel=document.getElementById('cmt-inline-'+idx);if(!panel)return;if(panel.style.display!=='none'){panel.style.display='none';return;}panel.style.display='block';panel.innerHTML='<div style="padding:8px;color:#888;font-size:11px">Đang tải...</div>';const cmts=await getComments(videoId);renderInlineComments(panel,videoId,idx,cmts);}
284
  function renderInlineComments(panel,videoId,idx,cmts){let h='<div class="inline-cmt-header"><span>💬 Bình luận</span><button onclick="document.getElementById(\'cmt-inline-'+idx+'\').style.display=\'none\'">✕</button></div><div class="inline-cmt-list">';if(cmts.length){cmts.slice(-30).forEach(c=>{h+=`<div class="inline-cmt-item"><span class="inline-cmt-time">${c.time||''}</span>${esc(c.text)}</div>`;});}else{h+='<div style="color:#777;font-size:11px;padding:4px">Chưa có bình luận</div>';}h+=`</div><div class="inline-cmt-input"><input id="cmt-input-${idx}" placeholder="Viết bình luận..." onkeydown="if(event.key==='Enter')submitInlineCmt('${esc(videoId)}',${idx})"><button onclick="submitInlineCmt('${esc(videoId)}',${idx})">Gửi</button></div>`;panel.innerHTML=h;const list=panel.querySelector('.inline-cmt-list');if(list)list.scrollTop=list.scrollHeight;}
285
  async function submitInlineCmt(videoId,idx){const inp=document.getElementById('cmt-input-'+idx);if(!inp)return;const text=inp.value.trim();if(!text)return;inp.value='';inp.disabled=true;const cmts=await postComment(videoId,text);inp.disabled=false;const panel=document.getElementById('cmt-inline-'+idx);if(panel)renderInlineComments(panel,videoId,idx,cmts);const cc=document.getElementById('cc-'+idx);if(cc)cc.textContent=fmtNum(cmts.length);}
 
286
  function initTikTokFeed(){const feed=document.getElementById('tiktok-feed');if(!feed)return;const slides=feed.querySelectorAll('.tiktok-slide');let cur=-1;function act(i){if(i===cur)return;slides.forEach((sl,idx)=>{const v=sl.querySelector('video');const fr=sl.querySelector('iframe');if(idx===i){if(v&&v.dataset.hls&&!v._hls&&typeof Hls!=='undefined'&&Hls.isSupported()){const hls=new Hls();hls.loadSource(v.dataset.hls);hls.attachMedia(v);hls.on(Hls.Events.MANIFEST_PARSED,()=>v.play().catch(()=>{}));v._hls=hls}else if(v)v.play().catch(()=>{});if(fr&&!fr.src&&fr.dataset.ytSrc)fr.src=fr.dataset.ytSrc;const vid=sl.dataset.vid;if(vid&&!sl._viewed){sl._viewed=true;doInteract(vid,'view').then(j=>{if(j){const vc=document.getElementById('vc-'+idx);if(vc)vc.textContent=fmtNum(j.views);}});}}else{if(v){v.pause();if(v._hls){v._hls.destroy();v._hls=null}}if(fr&&fr.src)fr.src=''}});cur=i}let sT;feed.addEventListener('scroll',()=>{clearTimeout(sT);sT=setTimeout(()=>{const rect=feed.getBoundingClientRect(),ctr=rect.top+rect.height/2;let best=-1,bestD=1e9;slides.forEach((sl,i)=>{const d=Math.abs(sl.getBoundingClientRect().top+sl.getBoundingClientRect().height/2-ctr);if(d<bestD){bestD=d;best=i}});if(best>=0)act(best)},150)});setTimeout(()=>act(0),400);slides.forEach(sl=>{const v=sl.querySelector('video');if(v)v.addEventListener('click',e=>{e.preventDefault();v.paused?v.play().catch(()=>{}):v.pause()})});const ids=[...slides].map(sl=>sl.dataset.vid||'');loadCounters(ids)}
287
+ async function openHighlightFeed(league,idx,forceUrl){showView('view-tiktok');const el=document.getElementById('view-tiktok');el.innerHTML='<div class="loading">Đang tải...</div>';let articles=(_hlLeagueData||{})[league]||[];if(!articles.length){try{articles=await fetch('/api/highlights/'+league).then(r=>r.json())}catch(e){articles=[]}}if(!articles.length){el.innerHTML='<div class="loading">Không có video</div>';return}const vids=[];const results=await Promise.all(articles.map(async(a,i)=>{try{const r=await fetch('/api/video_url?url='+encodeURIComponent(a.link));const v=await r.json();if(v&&v.src)return{...a,...v,_idx:i}}catch(e){}return null}));results.forEach(r=>{if(r)vids.push(r)});vids.sort((a,b)=>a._idx-b._idx);if(!vids.length){el.innerHTML='<div class="loading">Không tìm thấy video</div>';return}let ti=vids.findIndex(v=>v._idx===idx);if(ti<0)ti=0;const ordered=ti>0?[...vids.slice(ti),...vids.slice(0,ti)]:vids;let h=`<button class="back-btn" onclick="switchCat('home')">← Highlight</button><div class="tiktok-container"><div class="tiktok-feed" id="tiktok-feed">`;ordered.forEach((v,i)=>{const isYT=v.type==='youtube',isHLS=!isYT&&v.src?.includes('.m3u8'),poster=v.poster?` poster="${v.poster}"`:'';const vtag=isYT?`<iframe data-yt-src="${v.src}" allowfullscreen allow="accelerometer;autoplay;clipboard-write;encrypted-media;gyroscope;picture-in-picture"></iframe>`:isHLS?`<video playsinline preload="none"${poster} data-hls="${v.src}" loop controls></video>`:`<video playsinline preload="none"${poster} loop controls><source src="${v.src}" type="video/mp4"></video>`;const videoId='hl-'+league+'-'+(v.id||v._idx);h+=buildTikTokSlide({vtag,title:v.title,badge:'HL',badgeClass:'badge-fpt',videoId,idx:i,total:ordered.length,shareUrl:v.link||'',extraBtn:`<button class="tiktok-right-btn" onclick="event.stopPropagation();this.closest('.tiktok-slide').classList.toggle('ratio-wide')"><div class="icon"></div><div class="count">16:9</div></button>`});});h+='</div></div>';el.innerHTML=h;initTikTokFeed();}
288
+ async function openYTShortsFeed(startIdx){showView('view-tiktok');const el=document.getElementById('view-tiktok');el.innerHTML='<div class="loading">Đang tải...</div>';const arts=_shortsData.length?_shortsData:await fetch('/api/shorts').then(r=>r.json()).catch(()=>[]);if(!arts.length){el.innerHTML='<div class="loading">Không có shorts</div>';return}const ordered=startIdx>0?[...arts.slice(startIdx),...arts.slice(0,startIdx)]:arts;let h=`<button class="back-btn" onclick="switchCat('home')">← Shorts</button><div class="tiktok-container"><div class="tiktok-feed" id="tiktok-feed">`;ordered.forEach((v,i)=>{const id=v.id||'';const src=`https://www.youtube.com/embed/${id}?autoplay=1&rel=0&playsinline=1`;const vtag=`<iframe data-yt-src="${src}" allowfullscreen allow="accelerometer;autoplay;clipboard-write;encrypted-media;gyroscope;picture-in-picture"></iframe>`;const badge=v.channel==='baosuckhoedoisongboyte'?'SKĐS':'Dân trí';const videoId='yt-'+id;h+=buildTikTokSlide({vtag,title:v.title,badge,badgeClass:'badge-fpt',videoId,idx:i,total:ordered.length,shareUrl:'https://youtube.com/watch?v='+id});});h+='</div></div>';el.innerHTML=h;initTikTokFeed();}
289
+ async function openShortAIFeed(startIdx){showView('view-tiktok');const el=document.getElementById('view-tiktok');el.innerHTML='<div class="loading">Đang tải...</div>';const wall=(await fetch('/api/wall').then(r=>r.json()).catch(()=>({posts:[]}))).posts||[];const vids=wall.filter(p=>p.video);if(!vids.length){el.innerHTML='<div class="loading">Chưa có Short AI</div>';return}const ordered=startIdx>0?[...vids.slice(startIdx),...vids.slice(0,startIdx)]:vids;let h=`<button class="back-btn" onclick="switchCat('home')">← Short AI</button><div class="tiktok-container"><div class="tiktok-feed" id="tiktok-feed">`;ordered.forEach((p,i)=>{const vtag=`<video src="${p.video}" playsinline loop controls></video>`;const videoId='ai-'+(p.id||i);h+=buildTikTokSlide({vtag,title:p.title,badge:'AI',badgeClass:'badge-ai',videoId,idx:i,total:ordered.length,shareUrl:SPACE});});h+='</div></div>';el.innerHTML=h;initTikTokFeed();}
290
+ async function readArticle(url){showView('view-article');const el=document.getElementById('view-article');el.innerHTML='<div class="loading">Đang tải...</div>';try{const r=await fetch('/api/article?url='+encodeURIComponent(url));const data=await r.json();if(data&&!data.error&&data.body&&data.body.length){_currentArticle={url,data};let h=`<button class="back-btn" onclick="switchCat('home')">← Quay lại</button><div class="article-view"><h1 class="article-title">${esc(data.title)}</h1>`;if(data.summary)h+=`<div class="article-summary">${esc(data.summary)}</div>`;const seen={};data.body.forEach(b=>{if(b.type==='p')h+=`<p class="article-p">${b.text}</p>`;else if(b.type==='img'&&b.src&&!seen[b.src]){seen[b.src]=1;h+=`<img class="article-img" src="${esc(b.src)}" onerror="this.style.display='none'">`}else if(b.type==='heading')h+=`<h2 class="article-h2">${esc(b.text)}</h2>`});h+=`<div class="article-actions"><button class="primary" onclick="rewriteArticle()">🤖 Rewrite AI đăng tường</button><button onclick="doShare('${esc(data.title)}','${esc(url)}','${esc(data.og_image||'')}')">📤</button><button onclick="window.open('${esc(url)}','_blank')">🔗 Gốc</button></div><div class="article-ai-ask"><h3 style="font-size:14px;color:#5cb87a">🤖 Hỏi AI</h3><textarea id="ask-q" placeholder="Hỏi về bài viết..."></textarea><button onclick="askAI()">Hỏi</button><div id="ask-a" class="article-ai-answer"></div></div></div>`;el.innerHTML=h;window.scrollTo(0,0);return;}}catch(e){}el.innerHTML=`<button class="back-btn" onclick="switchCat('home')">← Quay lại</button><div class="loading"><p>Không đọc được.</p><a href="${esc(url)}" target="_blank" style="color:#5cb87a">Mở gốc →</a></div>`;}
291
+ async function rewriteArticle(){const url=_currentArticle?.url;if(!url)return;toast('⏳ Đang rewrite...');try{const r=await fetch('/api/rewrite_share',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({url,context:document.querySelector('.article-view')?.innerText?.slice(0,14000)||''})});const j=await r.json();if(!r.ok||j.error)throw new Error(j.error);toast('✅ Đã đăng Tường AI!');if(j.post)prependWallPost(j.post);}catch(e){toast('❌ '+e.message)}}
292
+ async function rewriteUrl(){const url=document.getElementById('url-input')?.value.trim();if(!url)return alert('Dán URL');toast('⏳ Đang rewrite...');try{const r=await fetch('/api/url_wall',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({url})});const j=await r.json();if(!r.ok||j.error)throw new Error(j.error);toast('✅ Đã đăng!');document.getElementById('url-input').value='';if(j.post)prependWallPost(j.post);}catch(e){toast('❌ '+e.message)}}
293
+ async function askAI(){const q=document.getElementById('ask-q')?.value.trim();if(!q)return alert('Nhập câu hỏi');const a=document.getElementById('ask-a');a.textContent='Đang hỏi...';try{const r=await fetch('/api/article/ask',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({url:_currentArticle?.url||'',question:q,context:document.querySelector('.article-view')?.innerText?.slice(0,12000)||''})});const j=await r.json();a.textContent=j.answer||'Không trả lời được';}catch(e){a.textContent='Lỗi: '+e.message}}
294
+ async function readWallPost(i){const p=_wallPosts[i];if(!p)return;showView('view-article');
295
+ const images = p.images || [];
296
+ let imgGallery = '';
297
+ if(images.length > 0){
298
+ imgGallery = '<div class="article-image-gallery">';
299
+ images.forEach((imgUrl, imgIdx) => {
300
+ if(imgIdx === 0){
301
+ imgGallery += `<img class="article-img article-hero-img" src="${esc(imgUrl)}" onerror="this.style.display='none" loading="eager">`;
302
+ } else {
303
+ if(imgIdx === 1) imgGallery += '<div class="gallery-thumbs">';
304
+ imgGallery += `<div class="gallery-thumb"><img src="${esc(imgUrl)}" onerror="this.parentElement.style.display='none'" loading="lazy"></div>`;
305
+ }
306
+ });
307
+ if(images.length > 1) imgGallery += '</div>';
308
+ imgGallery += '</div>';
309
+ }
310
+ const hasVideo = p.video && p.video.length > 0;
311
+ const voiceOptions = [
312
+ {id:'hoaimy', label:'🎙️ Nữ — Hoài My'},
313
+ {id:'namminh', label:'🎙️ Nam — Nam Minh'},
314
+ ];
315
+ let voiceSelector = '';
316
+ if(!hasVideo){
317
+ voiceSelector = `<div class="tts-selector"><div class="tts-selector-label">🎙️ Chọn giọng đọc:</div><div class="tts-voice-btns">`;
318
+ voiceOptions.forEach(v=>{
319
+ voiceSelector += `<button class="tts-voice-btn" onclick="document.querySelectorAll('.tts-voice-btn').forEach(b=>b.classList.remove('active'));this.classList.add('active');document.getElementById('selected-voice').value='${v.id}'">${v.label}</button>`;
320
+ });
321
+ voiceSelector += `</div><div class="tts-speed-row"><span>Tốc độ:</span><select id="selected-speed"><option value="1.0">1.0x — Bình thường</option><option value="1.2" selected>1.2x — Nhanh</option><option value="1.5">1.5x — Rất nhanh</option><option value="0.8">0.8x — Chậm</option></select></div>`;
322
+ voiceSelector += `<input type="hidden" id="selected-voice" value="hoaimy"></div>`;
323
+ }
324
+ document.getElementById('view-article').innerHTML=`<button class="back-btn" onclick="switchCat('home')">← Quay lại</button><div class="article-view"><span class="badge badge-ai">AI</span><h1 class="article-title">${esc(p.title)}</h1>${imgGallery}<p class="article-p" style="white-space:pre-wrap">${esc(p.text)}</p>${hasVideo?`<video class="article-img" src="${esc(p.video)}" controls playsinline style="max-height:400px"></video>`:''}<div class="article-actions">${hasVideo?`<button onclick="openShortAIFeed(${i})">🎬 Xem Short</button>${voiceSelector}<button class="primary" onclick="makeShortVideo('${esc(p.id)}',this,document.getElementById('selected-voice')?.value,parseFloat(document.getElementById('selected-speed')?.value)||1.2)">🔄 Tạo lại Short</button>`:`${voiceSelector}<button class="primary" onclick="makeShortVideo('${esc(p.id)}',this,document.getElementById('selected-voice')?.value,parseFloat(document.getElementById('selected-speed')?.value)||1.2)">🎬 Tạo Video Shorts</button>`}<button onclick="doShare('${esc(p.title)}','${SPACE}','${esc(p.img||'')}')">📤</button></div></div>`;
325
+ const firstVoiceBtn = document.querySelector('.tts-voice-btn');
326
+ if(firstVoiceBtn) firstVoiceBtn.classList.add('active');
327
+ window.scrollTo(0,0)}
328
+ async function loadNewsTab(){const el=document.getElementById('view-cat');el.innerHTML='<div class="loading">Đang tải...</div>';try{const r=await fetch('/api/homepage');const news=await r.json();if(!news.length){el.innerHTML='<div class="loading">Không có tin</div>';return}const groups={};news.forEach(a=>{if(!groups[a.group])groups[a.group]=[];groups[a.group].push(a)});let h='';for(const[g,arts]of Object.entries(groups)){h+=`<div class="section-title">${g}</div><div class="grid">`;arts.slice(0,6).forEach(a=>{h+=`<div class="card" onclick="readArticle('${esc(a.link)}')"><div class="card-img">${a.img?`<img src="${a.img}" loading="lazy">`:''}</div><div class="card-body"><span class="badge badge-vne">${esc(a.source||'VnE')}</span><div class="card-title">${esc(a.title)}</div></div></div>`});h+='</div>'}el.innerHTML=h}catch(e){el.innerHTML='<div class="loading">Lỗi</div>'}}
329
+ async function loadCat(id){const el=document.getElementById('view-cat');el.innerHTML='<div class="loading">Đang tải...</div>';const arts=await fetch('/api/category/'+id).then(r=>r.json()).catch(()=>[]);if(!arts.length){el.innerHTML='<div class="loading">Không có tin</div>';return}let h='<div class="grid">';arts.forEach(a=>{h+=`<div class="card" onclick="readArticle('${esc(a.link)}')"><div class="card-img">${a.img?`<img src="${a.img}" loading="lazy">`:''}</div><div class="card-body"><span class="badge badge-vne">${esc(a.source||'')}</span><div class="card-title">${esc(a.title)}</div></div></div>`});h+='</div>';el.innerHTML=h}
330
+ fetch('/api/storage_status').then(r=>r.json()).then(j=>{if(!j.persistent){const home=document.getElementById('view-home');if(home){const w=document.createElement('div');w.className='storage-warn';w.innerHTML='⚠️ Persistent Storage chưa bật.';home.prepend(w)}}}).catch(()=>{});
331
+
332
+ (function(){
333
+ try{
334
+ var hash = window.location.hash;
335
+ if(hash && hash.length > 1){
336
+ var articleUrl = decodeURIComponent(hash.substring(1));
337
+ if(articleUrl.startsWith('http')){
338
+ history.replaceState(null, '', window.location.pathname);
339
+ setTimeout(function(){
340
+ if(typeof readArticle==='function') readArticle(articleUrl);
341
+ }, 1500);
342
+ }
343
+ }
344
+ }catch(e){}
345
+ })();
346
 
347
+ (function(){
348
+ try{
349
+ const pa=localStorage.getItem('pending_article');
350
+ const pv=localStorage.getItem('pending_video');
351
+ if(pa){
352
+ localStorage.removeItem('pending_article');
353
+ setTimeout(()=>{
354
+ if(typeof readArticle==='function') readArticle(pa);
355
+ },1500);
356
+ }
357
+ if(pv){
358
+ localStorage.removeItem('pending_video');
359
+ try{
360
+ const v=JSON.parse(pv);
361
+ if(v&&v.url) setTimeout(()=>{window.open(v.url,'_blank')},1500);
362
+ }catch(e){}
363
+ }
364
+ }catch(e){}
365
+ })();
366
+
367
+ if (document.readyState === 'loading') {
368
+ document.addEventListener('DOMContentLoaded', function() {
369
+ if (typeof loadHome === 'function') loadHome();
370
+ });
371
+ } else {
372
+ if (typeof loadHome === 'function') loadHome();
373
+ }
static/app_v2_shorts_fix.js DELETED
@@ -1,2 +0,0 @@
1
- // No-op - all functionality built into app_v2.js
2
- (function(){})();
 
 
 
static/fm_fix.css CHANGED
@@ -31,76 +31,6 @@
31
  @keyframes ht-spin{to{transform:rotate(360deg)}}
32
  .wall-img-count{position:absolute;bottom:4px;left:4px;background:rgba(0,0,0,.7);color:#fff;font-size:9px;padding:1px 5px;border-radius:4px}
33
  #progress-toast{position:fixed!important;bottom:70px!important;left:50%!important;transform:translateX(-50%)!important;background:#2d8659!important;color:#fff!important;padding:10px 20px!important;border-radius:20px!important;font-size:12px!important;z-index:99998!important;box-shadow:0 4px 12px rgba(0,0,0,.4)!important;display:none;white-space:nowrap!important}
34
- /* === ARTICLE / REWRITE VIEW FIX === *//* === WORLD CUP 2026 SECTION === */
35
- .wc-section{margin:8px 4px!important;background:#1a1a1a!important;border:1px solid #2a2a2a!important;border-radius:10px!important;overflow:hidden!important}
36
- .wc-header{padding:10px 12px!important;background:linear-gradient(135deg,#0d1b2b,#1a3a2a 50%,#0b4b8b)!important;display:flex!important;align-items:center!important;justify-content:space-between!important}
37
- .wc-title{font-size:14px!important;font-weight:800!important;color:#fff!important}
38
- .wc-tabs{display:flex!important;gap:4px!important;padding:8px 10px!important;background:#1a1a1a!important;overflow-x:auto!important;scrollbar-width:none!important;border-bottom:1px solid #2a2a2a!important}
39
- .wc-tabs::-webkit-scrollbar{display:none!important}
40
- .wc-tab{padding:6px 12px!important;background:#222!important;border:1px solid #333!important;border-radius:12px!important;color:#999!important;font-size:10px!important;white-space:nowrap!important;cursor:pointer!important;flex-shrink:0!important}
41
- .wc-tab.active{background:#0b6bcb!important;border-color:#0b6bcb!important;color:#fff!important;font-weight:700!important}
42
- .wc-tab-content{padding:10px!important;max-height:500px!important;overflow-y:auto!important}
43
- .wc-news-list{display:flex!important;flex-direction:column!important;gap:8px!important}
44
- .wc-news-item{display:flex!important;gap:10px!important;padding:8px!important;background:#202020!important;border-radius:8px!important;cursor:pointer!important}
45
- .wc-news-item:hover{background:#2a2a2a!important}
46
- .wc-news-img{flex:0 0 90px!important;aspect-ratio:16/9!important;background:#333!important;border-radius:6px!important;overflow:hidden!important}
47
- .wc-news-img img{width:100%!important;height:100%!important;object-fit:cover!important}
48
- .wc-news-text{flex:1!important;min-width:0!important;display:flex!important;flex-direction:column!important;justify-content:space-between!important}
49
- .wc-news-title{font-size:12px!important;font-weight:700!important;color:#eee!important;display:-webkit-box!important;-webkit-line-clamp:2!important;-webkit-box-orient:vertical!important;overflow:hidden!important}
50
- .wc-news-source{font-size:10px!important;color:#888!important;margin-top:4px!important}
51
-
52
- /* World Cup Fixtures */
53
- .wc-fixtures-list{display:flex!important;flex-direction:column!important;gap:6px!important}
54
- .wc-match-item{padding:10px!important;background:#202020!important;border-radius:8px!important;border-left:3px solid #333!important}
55
- .wc-match-item:nth-child(odd){background:#1e1e1e!important}
56
- .wc-match-date{font-size:10px!important;color:#888!important;margin-bottom:4px!important}
57
- .wc-match-teams{display:flex!important;align-items:center!important;justify-content:center!important;gap:12px!important}
58
- .wc-match-teams .wc-team{flex:1!important;font-size:12px!important;color:#ddd!important;text-align:center!important}
59
- .wc-match-teams .wc-score{font-size:16px!important;font-weight:900!important;color:#f0c040!important;min-width:60px!important;text-align:center!important}
60
- .wc-score.wc-live{color:#e74c3c!important}
61
- .wc-score.wc-finished{color:#888!important}
62
- .wc-score.wc-upcoming{color:#5cb87a!important}
63
- .wc-match-location{font-size:9px!important;color:#777!important;text-align:center!important;margin-top:4px!important}
64
-
65
- /* World Cup Standings & Stats Tables */
66
- .wc-standings-table,.wc-stats-table{font-size:11px!important;color:#ccc!important;overflow-x:auto!important;width:100%!important}
67
- .wc-standings-table table,.wc-stats-table table{width:100%!important;border-collapse:collapse!important;table-layout:fixed!important}
68
- .wc-standings-table th,.wc-stats-table th{background:#222!important;color:#999!important;padding:6px 4px!important;font-size:10px!important;border-bottom:1px solid #333!important;text-align:center!important}
69
- .wc-standings-table th:first-child,.wc-stats-table th:first-child{text-align:left!important}
70
- .wc-standings-table td,.wc-stats-table td{padding:5px 4px!important;border-bottom:1px solid #1a1a1a!important;text-align:center!important;font-size:11px!important}
71
- .wc-standings-table td:first-child,.wc-stats-table td:first-child{text-align:left!important}
72
- .wc-standings-table .pts,.wc-stats-table .pts{font-weight:800!important;color:#f0c040!important}
73
- .wc-standings-table .team-name,.wc-stats-table .team-name{display:flex!important;align-items:center!important;gap:4px!important}
74
- .wc-standings-table .team-name img,.wc-stats-table .team-name img{width:16px!important;height:16px!important;object-fit:contain!important}
75
-
76
- /* World Cup Highlights */
77
- .wc-highlights-grid{display:grid!important;grid-template-columns:repeat(2,1fr)!important;gap:8px!important}/* World Cup Stats */
78
- .wc-stats{padding:10px!important}
79
- .wc-stat-group{margin-bottom:16px!important;padding:10px!important;background:#202020!important;border-radius:8px!important}
80
- .wc-stat-group h4{font-size:12px!important;color:#5cb87a!important;margin-bottom:8px!important;font-weight:700!important}
81
- .wc-stat-group .wc-empty{color:#888!important;font-size:11px!important;font-style:italic!important}
82
- .wc-stat-group table{width:100%!important;border-collapse:collapse!important}
83
- .wc-stat-group th{background:#2a2a2a!important;color:#999!important;padding:6px 4px!important;font-size:10px!important;text-align:left!important;border-bottom:1px solid #333!important}
84
- .wc-stat-group td{padding:5px 4px!important;border-bottom:1px solid #1a1a1a!important;font-size:11px!important;color:#ccc!important}
85
-
86
- /* World Cup Group Standings */
87
- .wc-group{margin-bottom:16px!important}
88
- .wc-group h4{font-size:12px!important;color:#f0c040!important;margin-bottom:6px!important;padding:6px 8px!important;background:#222!important;border-radius:6px!important;font-weight:700!important}
89
- .wc-table{width:100%!important;border-collapse:collapse!important;font-size:11px!important}
90
- .wc-table th{background:#2a2a2a!important;color:#999!important;padding:6px 4px!important;font-size:10px!important;text-align:center!important;border-bottom:1px solid #333!important}
91
- .wc-table th:first-child{text-align:left!important}
92
- .wc-table td{padding:5px 4px!important;border-bottom:1px solid #1a1a1a!important;text-align:center!important;color:#ccc!important}
93
- .wc-table td:first-child{text-align:left!important;font-weight:600!important}
94
- .wc-table .pts{font-weight:800!important;color:#f0c040!important}
95
-
96
- /* World Cup Highlights */
97
- .wc-highlights-grid{display:grid!important;grid-template-columns:repeat(2,1fr)!important;gap:8px!important}
98
- .wc-hl-item{cursor:pointer!important}
99
- .wc-hl-thumb{position:relative!important;aspect-ratio:16/9!important;background:#333!important;border-radius:6px!important;overflow:hidden!important}
100
- .wc-hl-thumb img{width:100%!important;height:100%!important;object-fit:cover!important}
101
- .wc-hl-thumb .card-play{position:absolute!important;left:50%!important;top:50%!important;transform:translate(-50%,-50%)!important;width:30px!important;height:30px!important;border-radius:50%!important;background:rgba(0,0,0,.55)!important;display:flex!important;align-items:center!important;justify-content:center!important;color:#fff!important;font-size:12px!important}
102
- .wc-hl-title{font-size:10px!important;color:#ccc!important;margin-top:4px!important;line-height:1.2!important;display:-webkit-box!important;-webkit-line-clamp:2!important;-webkit-box-orient:vertical!important;overflow:hidden!important}
103
-
104
  /* === ARTICLE / REWRITE VIEW FIX === */
105
  .article-view{padding:12px 8px 40px!important;max-width:760px!important;margin:0 auto!important}
106
  .article-title{font-size:18px!important;font-weight:800!important;line-height:1.3!important;margin-bottom:8px!important;color:#fff!important}
 
31
  @keyframes ht-spin{to{transform:rotate(360deg)}}
32
  .wall-img-count{position:absolute;bottom:4px;left:4px;background:rgba(0,0,0,.7);color:#fff;font-size:9px;padding:1px 5px;border-radius:4px}
33
  #progress-toast{position:fixed!important;bottom:70px!important;left:50%!important;transform:translateX(-50%)!important;background:#2d8659!important;color:#fff!important;padding:10px 20px!important;border-radius:20px!important;font-size:12px!important;z-index:99998!important;box-shadow:0 4px 12px rgba(0,0,0,.4)!important;display:none;white-space:nowrap!important}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
34
  /* === ARTICLE / REWRITE VIEW FIX === */
35
  .article-view{padding:12px 8px 40px!important;max-width:760px!important;margin:0 auto!important}
36
  .article-title{font-size:18px!important;font-weight:800!important;line-height:1.3!important;margin-bottom:8px!important;color:#fff!important}
static/index_v2.html CHANGED
@@ -10,12 +10,10 @@
10
  <link rel="canonical" href="https://bep40-vnews.hf.space">
11
  <link rel="stylesheet" href="/static/wc2026.css">
12
  <script src="https://cdn.jsdelivr.net/npm/hls.js@1/dist/hls.min.js"></script>
13
- <style>
14
- *{box-sizing:border-box;margin:0;padding:0}body{background:#111;color:#eee;font-family:-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,sans-serif;overflow-x:hidden}.header{background:linear-gradient(135deg,#0d1117,#1a3a2a 50%,#8b7500);padding:12px;text-align:center}.header h1{font-size:18px;color:#fff}.header p{font-size:10px;color:#aaa}.cats{display:flex;overflow-x:auto;background:#1a1a1a;border-bottom:1px solid #333;padding:0 4px;position:sticky;top:0;z-index:50;scrollbar-width:none}.cats::-webkit-scrollbar{display:none}.cat{padding:9px 11px;color:#888;font-size:11px;white-space:nowrap;border-bottom:2px solid transparent;cursor:pointer;flex-shrink:0}.cat.active{color:#5cb87a;border-bottom-color:#5cb87a;font-weight:700}.view{display:none}.view.active{display:block}.loading{text-align:center;padding:30px;color:#777;font-size:12px}.slider-wrap{margin:6px 4px;background:#1a1a1a;border:1px solid #2a2a2a;border-radius:8px;overflow:hidden}.slider-header{padding:7px 10px;display:flex;align-items:center;justify-content:space-between}.slider-label{color:#f0c040;font-size:13px;font-weight:800}.slider-note{font-size:10px;color:#777}.slider-track{display:flex;overflow-x:auto;gap:8px;padding:4px 10px 10px;scrollbar-width:none}.slider-track::-webkit-scrollbar{display:none}.slider-item{flex:0 0 160px;cursor:pointer}.slider-thumb{position:relative;width:100%;aspect-ratio:16/9;border-radius:6px;overflow:hidden;background:#333}.slider-thumb img,.slider-thumb video{width:100%;height:100%;object-fit:cover}.slider-title{font-size:10px;color:#ccc;margin-top:3px;line-height:1.2;display:-webkit-box;-webkit-line-clamp:2;-webkit-box-orient:vertical;overflow:hidden}.card-play{position:absolute;left:50%;top:50%;transform:translate(-50%,-50%);width:30px;height:30px;border-radius:50%;background:rgba(0,0,0,.55);display:flex;align-items:center;justify-content:center;color:#fff;font-size:12px}.grid{display:grid;grid-template-columns:repeat(2,1fr);gap:6px;padding:6px 4px}@media(min-width:650px){.grid{grid-template-columns:repeat(3,1fr)}}.card{background:#1a1a1a;border:1px solid #222;border-radius:8px;overflow:hidden;cursor:pointer}.card-img{position:relative;aspect-ratio:16/9;background:#333}.card-img img{width:100%;height:100%;object-fit:cover}.card-body{padding:6px 8px}.card-title{font-size:11px;line-height:1.35;color:#eee;display:-webkit-box;-webkit-line-clamp:2;-webkit-box-orient:vertical;overflow:hidden}.badge{font-size:8px;padding:1px 5px;border-radius:3px;font-weight:700;display:inline-block;margin-bottom:2px;color:#fff}.badge-vne{background:#c0392b}.badge-genk{background:#6a1b9a}.badge-ai{background:#2d8659}.badge-wc{background:#0b6bcb}.section-title{font-size:13px;font-weight:800;color:#5cb87a;margin:8px 0 4px;padding-left:8px;border-left:3px solid #5cb87a}.back-btn{background:#111;color:#fff;border:none;padding:10px;font-size:12px;width:100%;position:sticky;top:0;z-index:60;cursor:pointer}.article-view{padding:12px 8px 40px;max-width:760px;margin:0 auto}.article-title{font-size:18px;font-weight:800;line-height:1.3;margin-bottom:8px}.article-summary{background:#1a2a1f;border-left:3px solid #2d8659;padding:10px;margin-bottom:14px;color:#ccc;font-size:13px}.article-p{font-size:14px;line-height:1.7;color:#ccc;margin-bottom:10px}.article-img{width:100%;border-radius:6px;margin:10px 0}.article-h2{font-size:16px;margin:16px 0 8px;color:#eee}.article-actions{display:flex;gap:8px;flex-wrap:wrap;border-top:1px solid #333;margin-top:16px;padding-top:10px}.article-actions button{background:#1a1a1a;border:1px solid #333;color:#ccc;padding:7px 12px;border-radius:14px;font-size:11px;cursor:pointer}.article-actions button.primary{background:#2d8659;border-color:#2d8659;color:#fff}.article-ai-ask{margin-top:12px;background:#141414;border:1px solid #2a2a2a;border-radius:10px;padding:10px}.article-ai-ask textarea{width:100%;min-height:60px;background:#222;border:1px solid #444;color:#eee;border-radius:10px;padding:9px;font-size:12px}.article-ai-ask button{background:#2d8659;border:0;color:#fff;border-radius:10px;padding:8px 12px;margin-top:6px;font-size:11px;cursor:pointer}.article-ai-answer{white-space:pre-wrap;color:#ccc;font-size:13px;line-height:1.55;margin-top:8px}.tiktok-container{width:100%;height:80vh;max-height:680px;min-height:400px;background:#000}.tiktok-feed{height:100%;overflow-y:scroll;scroll-snap-type:y mandatory;scrollbar-width:none}.tiktok-feed::-webkit-scrollbar{display:none}.tiktok-slide{height:80vh;max-height:680px;min-height:400px;scroll-snap-align:start;position:relative;background:#000;display:flex;align-items:center;justify-content:center}.tiktok-slide video,.tiktok-slide iframe{width:100%;height:100%;object-fit:cover;border:none}.tiktok-slide.ratio-wide video,.tiktok-slide.ratio-wide iframe{object-fit:contain}.tiktok-bottom{position:absolute;bottom:0;left:0;right:60px;padding:12px 10px 16px;background:linear-gradient(transparent,rgba(0,0,0,.85));z-index:3}.tiktok-title{font-size:12px;color:#fff}.tiktok-counter{position:absolute;top:8px;left:8px;background:rgba(0,0,0,.5);font-size:9px;padding:2px 7px;border-radius:8px;color:#fff;z-index:4}.tiktok-right{position:absolute;right:8px;bottom:100px;display:flex;flex-direction:column;align-items:center;gap:14px;z-index:5}.tiktok-right-btn{display:flex;flex-direction:column;align-items:center;gap:2px;background:none;border:0;color:#fff;cursor:pointer;font-size:10px}.tiktok-right-btn .icon{width:42px;height:42px;border-radius:50%;background:rgba(255,255,255,.12);display:flex;align-items:center;justify-content:center;font-size:20px}.tiktok-right-btn .count{font-size:10px;color:#ddd}.inline-comments{position:absolute;bottom:0;left:0;right:0;max-height:50%;background:rgba(18,18,18,.95);border-radius:14px 14px 0 0;z-index:10;overflow:hidden;display:flex;flex-direction:column}.inline-cmt-header{display:flex;justify-content:space-between;align-items:center;padding:8px 12px;border-bottom:1px solid #333;color:#5cb87a;font-size:12px;font-weight:700}.inline-cmt-header button{background:none;border:0;color:#fff;font-size:16px;cursor:pointer}.inline-cmt-list{flex:1;overflow-y:auto;padding:6px 10px;max-height:180px}.inline-cmt-item{background:#222;border-radius:8px;padding:6px 8px;margin:4px 0;color:#ccc;font-size:11px;line-height:1.3}.inline-cmt-time{font-size:9px;color:#777;margin-right:6px}.inline-cmt-input{display:flex;gap:6px;padding:8px 10px;border-top:1px solid #333}.inline-cmt-input input{flex:1;background:#222;border:1px solid #444;color:#eee;border-radius:16px;padding:7px 12px;font-size:11px}.inline-cmt-input button{background:#2d8659;border:0;color:#fff;border-radius:16px;padding:7px 12px;font-size:11px;cursor:pointer}.wc2026-section{margin:6px 4px;background:linear-gradient(135deg,#0d1117,#1a1a3a);border:1px solid #1a3a5a;border-radius:10px;overflow:hidden}.wc-header{display:flex;align-items:center;justify-content:space-between;padding:10px 12px;background:linear-gradient(90deg,#0b2e4a,#1a3a5a)}.wc-header h2{font-size:15px;color:#fff;margin:0}.wc-live-badge{font-size:10px;color:#e74c3c;font-weight:700;animation:wc-pulse 1.5s infinite}@keyframes wc-pulse{0%,100%{opacity:1}50%{opacity:.4}}.wc-tabs{display:flex;gap:4px;padding:8px 10px;overflow-x:auto;scrollbar-width:none}.wc-tabs::-webkit-scrollbar{display:none}.wc-tab{padding:5px 10px;background:#1a2a3a;border:1px solid #2a3a4a;border-radius:12px;color:#8ab4d8;font-size:10px;cursor:pointer;white-space:nowrap;flex-shrink:0}.wc-tab.active{background:#0b6bcb;border-color:#0b6bcb;color:#fff;font-weight:700}.wc-content{padding:8px 10px;max-height:500px;overflow-y:auto}.wc-news-grid{display:flex;flex-direction:column;gap:8px}.wc-news-item{display:flex;gap:8px;padding:8px;background:#1a2030;border-radius:8px;cursor:pointer}.wc-news-item:active{opacity:.8}.wc-news-img{flex:0 0 70px;aspect-ratio:16/9;border-radius:6px;overflow:hidden;background:#222}.wc-news-img img{width:100%;height:100%;object-fit:cover}.wc-news-text{flex:1;min-width:0}.wc-news-title{font-size:11px;font-weight:700;color:#eee;line-height:1.3;display:-webkit-box;-webkit-line-clamp:2;-webkit-box-orient:vertical;overflow:hidden}.wc-news-via{font-size:9px;color:#6a9fca;margin-top:2px}.ls-section{margin:6px 4px;background:#1a1a1a;border:1px solid #2a2a2a;border-radius:8px;overflow:hidden}.ls-header{padding:7px 10px;display:flex;align-items:center;justify-content:space-between}.ls-header h3{color:#f0c040;font-size:13px;font-weight:800}.ls-tabs{display:flex;gap:4px;padding:0 10px 8px;overflow-x:auto;scrollbar-width:none}.ls-tabs::-webkit-scrollbar{display:none}.ls-tab{padding:4px 10px;background:#222;border:1px solid #333;border-radius:12px;color:#999;font-size:10px;white-space:nowrap;cursor:pointer;flex-shrink:0}.ls-tab.active{background:#2d8659;border-color:#2d8659;color:#fff;font-weight:700}.ls-content{max-height:420px;overflow-y:auto;padding:0 6px 8px;font-size:12px;color:#ddd}.ls-content ul{list-style:none;padding:0;margin:0}.ls-content .title-content{display:flex;gap:6px;align-items:center;background:#222;border-radius:4px;margin:4px 0;padding:5px 8px}.ls-content .title-content img{width:18px;height:18px}.ls-content .title-content strong{font-size:11px;color:#ccc}.ls-content .match-detail{padding:6px;border-bottom:1px solid #262626;cursor:pointer}.ls-content .match-detail:hover{background:#1a2a1f}.ls-content .match{display:flex;flex-wrap:wrap;align-items:center;gap:4px}.ls-content .datetime{width:100%;font-size:9px;color:#888}.ls-content .teams{display:flex;width:100%;align-items:center;gap:4px}.ls-content .team{flex:1;display:flex;align-items:center;gap:4px;min-width:0;text-decoration:none}.ls-content .team .name{font-size:11px;color:#ddd;white-space:nowrap;overflow:hidden;text-overflow:ellipsis}.ls-content .team .logo img{width:18px;height:18px}.ls-content .home-team{justify-content:flex-end;text-align:right}.ls-content .status{flex:0 0 54px;text-align:center}.ls-content .status a{color:#fff;text-decoration:none;font-weight:800;font-size:12px}.ls-content .status .label{font-size:8px;color:#888;display:block}.ls-content .status .label.live{color:#e74c3c}.ls-content .info,.ls-content .btns{display:none}.ls-content table{width:100%;border-collapse:collapse;font-size:11px;color:#ccc}.ls-content table th{background:#222;color:#999;padding:5px 4px;font-size:10px;border-bottom:1px solid #333}.ls-content table td{padding:4px 3px;border-bottom:1px solid #1a1a1a}.ls-content table .team-name{display:flex;align-items:center;gap:4px}.ls-content table .team-name img{width:16px;height:16px}.ls-content table .pts{font-weight:800;color:#f0c040}.match-overlay{position:fixed;inset:0;background:#111;z-index:9999;display:none;flex-direction:column;overflow:auto}.match-overlay.active{display:flex}.mo-header{padding:10px;background:#1a1a1a;display:flex;justify-content:space-between;align-items:center;position:sticky;top:0;z-index:1}.mo-header h3{font-size:13px;color:#eee}.mo-close{background:none;border:0;color:#fff;font-size:22px;cursor:pointer}.mo-tabs{display:flex;gap:4px;padding:8px 10px;background:#1a1a1a;overflow-x:auto}.mo-tab{padding:5px 12px;background:#222;border:1px solid #333;border-radius:10px;color:#999;font-size:10px;cursor:pointer;white-space:nowrap}.mo-tab.active{background:#2d8659;color:#fff}.mo-body{padding:8px;overflow-x:auto;font-size:12px;color:#ddd}.mo-body ul{list-style:none;padding:0;margin:0}.mo-body li{padding:5px 0;border-bottom:1px solid #222}.featured-match{margin:6px 4px;background:linear-gradient(135deg,#1a2a1f,#0d1117);border:1px solid #2d8659;border-radius:10px;padding:12px;cursor:pointer}.fm-league{text-align:center;color:#5cb87a;font-size:9px;font-weight:700;text-transform:uppercase}.fm-teams{display:flex;align-items:center;justify-content:center;gap:10px;margin-top:6px}.fm-team{flex:1;display:flex;flex-direction:column;align-items:center;gap:4px}.fm-team img{width:32px;height:32px;object-fit:contain}.fm-team span{font-size:10px;color:#ccc;text-align:center}.fm-score{font-size:22px;font-weight:900;min-width:60px;text-align:center;color:#fff}.fm-status{text-align:center;margin-top:6px;font-size:9px;color:#e74c3c;font-weight:700}.fm-status.upcoming{color:#f0c040}.ai-compose{margin:6px 4px;background:#141414;border:1px solid #2a2a2a;border-radius:10px;padding:10px}.ai-compose-title{font-size:13px;font-weight:800;color:#5cb87a;margin-bottom:8px}.ai-compose-row{display:flex;gap:6px;margin-top:6px}.ai-compose input{flex:1;background:#222;border:1px solid #333;color:#eee;border-radius:18px;padding:9px 12px;font-size:12px;min-width:0}.ai-compose button{background:#2d8659;border:0;color:#fff;border-radius:18px;padding:9px 12px;font-size:11px;font-weight:700;cursor:pointer;white-space:nowrap}.ai-compose button.secondary{background:#333}.hot-topic-row{display:flex;gap:6px;overflow-x:auto;padding:4px 0;scrollbar-width:none}.hot-topic-row::-webkit-scrollbar{display:none}.hot-chip{flex:0 0 auto;background:#222;border:1px solid #333;color:#ddd;border-radius:16px;padding:5px 10px;font-size:11px;cursor:pointer;white-space:nowrap}.hot-chip:active{transform:scale(.96)}.hashtag-sources{margin:8px 4px;background:#1a1a1a;border:1px solid #2a2a2a;border-radius:10px;padding:10px}.hashtag-sources h3{font-size:13px;color:#5cb87a;margin-bottom:8px}.hashtag-src-item{display:flex;gap:8px;padding:8px;background:#202020;border-radius:8px;margin:6px 0;cursor:pointer}.hashtag-src-item:active{opacity:.8}.hashtag-src-img{flex:0 0 80px;aspect-ratio:16/9;background:#333;border-radius:6px;overflow:hidden}.hashtag-src-img img{width:100%;height:100%;object-fit:cover}.hashtag-src-text{flex:1;min-width:0}.hashtag-src-title{font-size:12px;font-weight:700;color:#eee;display:-webkit-box;-webkit-line-clamp:2;-webkit-box-orient:vertical;overflow:hidden}.hashtag-src-via{font-size:10px;color:#888;margin-top:2px}.hashtag-rewrite-btn{width:100%;margin-top:8px;background:#2d8659;border:0;color:#fff;padding:9px;border-radius:10px;font-size:12px;font-weight:700;cursor:pointer}.hashtag-load-more{width:100%;margin-top:8px;background:#222;border:1px solid #333;color:#ccc;padding:9px;border-radius:10px;font-size:12px;cursor:pointer}.hashtag-loading{display:flex;align-items:center;gap:8px;padding:12px;color:#888;font-size:12px}.hashtag-spinner{width:16px;height:16px;border:2px solid #333;border-top-color:#5cb87a;border-radius:50%;animation:ht-spin .8s linear infinite}@keyframes ht-spin{to{transform:rotate(360deg)}}.wall-item{flex:0 0 260px;background:#141414;border:1px solid #2b2b2b;border-radius:10px;padding:8px}.wall-item-new{animation:wall-flash 1.8s ease-out}@keyframes wall-flash{0%{border-color:#f0c040;box-shadow:0 0 18px rgba(240,192,64,.35)}30%{border-color:#f0c040;box-shadow:0 0 12px rgba(240,192,64,.2)}100%{border-color:#2b2b2b;box-shadow:none}}.wall-thumb{width:100%;aspect-ratio:16/9;border-radius:8px;background:#222;overflow:hidden;margin-bottom:6px;position:relative}.wall-thumb img{width:100%;height:100%;object-fit:cover}.wall-video-badge{position:absolute;top:4px;right:4px;background:rgba(45,134,89,.9);color:#fff;font-size:10px;padding:2px 6px;border-radius:6px;font-weight:700}.wall-title{font-size:12px;color:#5cb87a;font-weight:800;line-height:1.3;margin-bottom:4px;display:-webkit-box;-webkit-line-clamp:2;-webkit-box-orient:vertical;overflow:hidden}.wall-text{font-size:11px;color:#bbb;line-height:1.4;white-space:pre-wrap;display:-webkit-box;-webkit-line-clamp:4;-webkit-box-orient:vertical;overflow:hidden}.wall-actions{display:flex;gap:6px;margin-top:8px}.wall-actions button{flex:1;border:1px solid #333;background:#222;color:#ddd;border-radius:14px;padding:6px 8px;font-size:10px;cursor:pointer}.wall-actions button.primary{background:#2d8659;border-color:#2d8659;color:#fff}#progress-toast{position:fixed;bottom:70px;left:50%;transform:translateX(-50%);background:#2d8659;color:#fff;padding:10px 20px;border-radius:20px;font-size:12px;z-index:99998;box-shadow:0 4px 12px rgba(0,0,0,.4);display:none;white-space:nowrap}.storage-warn{background:#332200;border:1px solid #664400;color:#ffcc00;padding:8px 12px;border-radius:8px;font-size:11px;margin:6px 4px}
15
- </style>
16
  </head>
17
  <body>
18
- <div class="header"><h1>📰 VNEWS</h1><p>Tin tức · Bóng đá LIVE · Highlight · AI · World Cup 2026</p></div>
19
  <div class="cats" id="cat-bar"></div>
20
  <div id="view-home" class="view active"><div class="loading">Đang tải...</div></div>
21
  <div id="view-cat" class="view"></div>
@@ -32,19 +30,18 @@
32
  var _cats=[],_hlLeagueData={},_currentArticle=null;window._currentEventId='';
33
  function esc(s){return String(s||'').replace(/[&<>"']/g,m=>({'&':'&amp;','<':'&lt;','>':'&gt;','"':'&quot;',"'":'&#39;'}[m]))}
34
  function showView(id){document.querySelectorAll('.view').forEach(v=>v.classList.remove('active'));document.getElementById(id)?.classList.add('active')}
35
- function switchCat(id){document.querySelectorAll('.cat').forEach(c=>c.classList.remove('active'));document.querySelector('[data-cat="'+id+'"]')?.classList.add('active');document.querySelectorAll('.view').forEach(v=>v.classList.remove('active'));document.querySelectorAll('video').forEach(v=>{v.pause();if(v._hls){v._hls.destroy();v._hls=null}});document.querySelectorAll('iframe[data-yt-src]').forEach(f=>{f.src=''});if(id==='home')document.getElementById('view-home').classList.add('active');else if(id==='news-all'){document.getElementById('view-cat').classList.add('active');loadNewsTab()}else{document.getElementById('view-cat').classList.add('active');loadCat(id)}}
36
  function toast(msg){let t=document.getElementById('progress-toast');if(t){t.textContent=msg;t.style.display='block';setTimeout(()=>{t.style.display='none'},3500)}}
37
  function doShare(title,url,img){const shareUrl=SPACE+'/s?url='+encodeURIComponent(url)+'&title='+encodeURIComponent(title)+'&img='+encodeURIComponent(img||'');if(navigator.share)navigator.share({title,url:shareUrl}).catch(()=>{});else navigator.clipboard.writeText(shareUrl).then(()=>alert('Đã sao chép!')).catch(()=>{})}
38
- async function init(){_cats=await fetch('/api/categories').then(r=>r.json()).catch(()=>[]);let bar='<div class="cat active" data-cat="home">🏠</div><div class="cat" data-cat="news-all">📰 Tin tức</div>';_cats.forEach(c=>{bar+='<div class="cat" data-cat="'+c.id+'">'+c.name+'</div>'});document.getElementById('cat-bar').innerHTML=bar;document.querySelectorAll('.cat').forEach(t=>{t.onclick=()=>switchCat(t.dataset.cat)});}
39
  var SPACE=location.origin;
40
  </script>
41
- <script src="/static/app_v2.js?v=20260701"></script>
42
  <script src="/static/yt_live.js"></script>
43
- <script src="/static/vtv_init.js"></script>
44
- <script src="/static/hot_multi.js?v=1"></script>
45
  <script src="/static/wc2026_v2.js"></script>
46
  <script src="/static/live_mode.js"></script>
47
  <script src="/static/match_detail_v6.js"></script>
48
- <script>init();loadHome();</script>
49
  </body>
50
  </html>
 
10
  <link rel="canonical" href="https://bep40-vnews.hf.space">
11
  <link rel="stylesheet" href="/static/wc2026.css">
12
  <script src="https://cdn.jsdelivr.net/npm/hls.js@1/dist/hls.min.js"></script>
13
+ <style>*{box-sizing:border-box;margin:0;padding:0}body{background:#111;color:#eee;font-family:-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,sans-serif;overflow-x:hidden}.header{background:linear-gradient(135deg,#0d1117,#1a3a2a 50%,#8b7500);padding:12px;text-align:center}.header h1{font-size:18px;color:#fff}.header p{font-size:10px;color:#aaa}.cats{display:flex;overflow-x:auto;background:#1a1a1a;border-bottom:1px solid #333;padding:0 4px;position:sticky;top:0;z-index:50;scrollbar-width:none}.cats::-webkit-scrollbar{display:none}.cat{padding:9px 11px;color:#888;font-size:11px;white-space:nowrap;border-bottom:2px solid transparent;cursor:pointer;flex-shrink:0}.cat.active{color:#5cb87a;border-bottom-color:#5cb87a;font-weight:700}.view{display:none}.view.active{display:block}.loading{text-align:center;padding:30px;color:#777;font-size:12px}.slider-wrap{margin:6px 4px;background:#1a1a1a;border:1px solid #2a2a2a;border-radius:8px;overflow:hidden}.slider-header{padding:7px 10px;display:flex;align-items:center;justify-content:space-between}.slider-label{color:#f0c040;font-size:13px;font-weight:800}.slider-note{font-size:10px;color:#777}.slider-track{display:flex;overflow-x:auto;gap:8px;padding:4px 10px 10px;scrollbar-width:none}.slider-track::-webkit-scrollbar{display:none}.slider-item{flex:0 0 160px;cursor:pointer}.shorts-item{flex:0 0 110px!important}.slider-thumb{position:relative;width:100%;aspect-ratio:16/9;border-radius:6px;overflow:hidden;background:#333}.shorts-thumb{aspect-ratio:3/4!important;border-radius:8px!important}.slider-thumb img,.slider-thumb video{width:100%;height:100%;object-fit:cover}.slider-title{font-size:10px;color:#ccc;margin-top:3px;line-height:1.2;display:-webkit-box;-webkit-line-clamp:2;-webkit-box-orient:vertical;overflow:hidden}.card-play{position:absolute;left:50%;top:50%;transform:translate(-50%,-50%);width:30px;height:30px;border-radius:50%;background:rgba(0,0,0,.55);display:flex;align-items:center;justify-content:center;color:#fff;font-size:12px}.grid{display:grid;grid-template-columns:repeat(2,1fr);gap:6px;padding:6px 4px}@media(min-width:650px){.grid{grid-template-columns:repeat(3,1fr)}}.card{background:#1a1a1a;border:1px solid #222;border-radius:8px;overflow:hidden;cursor:pointer}.card-img{position:relative;aspect-ratio:16/9;background:#333}.card-img img{width:100%;height:100%;object-fit:cover}.card-body{padding:6px 8px}.card-title{font-size:11px;line-height:1.35;color:#eee;display:-webkit-box;-webkit-line-clamp:2;-webkit-box-orient:vertical;overflow:hidden}.badge{font-size:8px;padding:1px 5px;border-radius:3px;font-weight:700;display:inline-block;margin-bottom:2px;color:#fff}.badge-vne{background:#c0392b}.badge-bbc{background:#b80000}.badge-dt{background:#1565c0}.badge-genk{background:#6a1b9a}.badge-fpt{background:#f26522}.badge-ai{background:#2d8659}.badge-wc{background:#0b6bcb}.section-title{font-size:13px;font-weight:800;color:#5cb87a;margin:8px 0 4px;padding-left:8px;border-left:3px solid #5cb87a}.back-btn{background:#111;color:#fff;border:none;padding:10px;font-size:12px;width:100%;position:sticky;top:0;z-index:60;cursor:pointer}.article-view{padding:12px 8px 40px;max-width:760px;margin:0 auto}.article-title{font-size:18px;font-weight:800;line-height:1.3;margin-bottom:8px}.article-summary{background:#1a2a1f;border-left:3px solid #2d8659;padding:10px;margin-bottom:14px;color:#ccc;font-size:13px}.article-p{font-size:14px;line-height:1.7;color:#ccc;margin-bottom:10px}.article-img{width:100%;border-radius:6px;margin:10px 0}.article-h2{font-size:16px;margin:16px 0 8px;color:#eee}.article-actions{display:flex;gap:8px;flex-wrap:wrap;border-top:1px solid #333;margin-top:16px;padding-top:10px}.article-actions button{background:#1a1a1a;border:1px solid #333;color:#ccc;padding:7px 12px;border-radius:14px;font-size:11px;cursor:pointer}.article-actions button.primary{background:#2d8659;border-color:#2d8659;color:#fff}.article-ai-ask{margin-top:12px;background:#141414;border:1px solid #2a2a2a;border-radius:10px;padding:10px}.article-ai-ask textarea{width:100%;min-height:60px;background:#222;border:1px solid #444;color:#eee;border-radius:10px;padding:9px;font-size:12px}.article-ai-ask button{background:#2d8659;border:0;color:#fff;border-radius:10px;padding:8px 12px;margin-top:6px;font-size:11px;cursor:pointer}.article-ai-answer{white-space:pre-wrap;color:#ccc;font-size:13px;line-height:1.55;margin-top:8px}.tiktok-container{width:100%;height:80vh;max-height:680px;min-height:400px;background:#000}.tiktok-feed{height:100%;overflow-y:scroll;scroll-snap-type:y mandatory;scrollbar-width:none}.tiktok-feed::-webkit-scrollbar{display:none}.tiktok-slide{height:80vh;max-height:680px;min-height:400px;scroll-snap-align:start;position:relative;background:#000;display:flex;align-items:center;justify-content:center}.tiktok-slide video,.tiktok-slide iframe{width:100%;height:100%;object-fit:cover;border:none}.tiktok-slide.ratio-wide video,.tiktok-slide.ratio-wide iframe{object-fit:contain}.tiktok-bottom{position:absolute;bottom:0;left:0;right:60px;padding:12px 10px 16px;background:linear-gradient(transparent,rgba(0,0,0,.85));z-index:3}.tiktok-title{font-size:12px;color:#fff}.tiktok-counter{position:absolute;top:8px;left:8px;background:rgba(0,0,0,.5);font-size:9px;padding:2px 7px;border-radius:8px;color:#fff;z-index:4}.tiktok-right{position:absolute;right:8px;bottom:100px;display:flex;flex-direction:column;align-items:center;gap:14px;z-index:5}.tiktok-right-btn{display:flex;flex-direction:column;align-items:center;gap:2px;background:none;border:0;color:#fff;cursor:pointer;font-size:10px}.tiktok-right-btn .icon{width:42px;height:42px;border-radius:50%;background:rgba(255,255,255,.12);display:flex;align-items:center;justify-content:center;font-size:20px}.tiktok-right-btn .count{font-size:10px;color:#ddd}.inline-comments{position:absolute;bottom:0;left:0;right:0;max-height:50%;background:rgba(18,18,18,.95);border-radius:14px 14px 0 0;z-index:10;overflow:hidden;display:flex;flex-direction:column}.inline-cmt-header{display:flex;justify-content:space-between;align-items:center;padding:8px 12px;border-bottom:1px solid #333;color:#5cb87a;font-size:12px;font-weight:700}.inline-cmt-header button{background:none;border:0;color:#fff;font-size:16px;cursor:pointer}.inline-cmt-list{flex:1;overflow-y:auto;padding:6px 10px;max-height:180px}.inline-cmt-item{background:#222;border-radius:8px;padding:6px 8px;margin:4px 0;color:#ccc;font-size:11px;line-height:1.3}.inline-cmt-time{font-size:9px;color:#777;margin-right:6px}.inline-cmt-input{display:flex;gap:6px;padding:8px 10px;border-top:1px solid #333}.inline-cmt-input input{flex:1;background:#222;border:1px solid #444;color:#eee;border-radius:16px;padding:7px 12px;font-size:11px}.inline-cmt-input button{background:#2d8659;border:0;color:#fff;border-radius:16px;padding:7px 12px;font-size:11px;cursor:pointer}.wc2026-section{margin:6px 4px;background:linear-gradient(135deg,#0d1117,#1a1a3a);border:1px solid #1a3a5a;border-radius:10px;overflow:hidden}.wc-header{display:flex;align-items:center;justify-content:space-between;padding:10px 12px;background:linear-gradient(90deg,#0b2e4a,#1a3a5a)}.wc-header h2{font-size:15px;color:#fff;margin:0}.wc-live-badge{font-size:10px;color:#e74c3c;font-weight:700;animation:wc-pulse 1.5s infinite}@keyframes wc-pulse{0%,100%{opacity:1}50%{opacity:.4}}.wc-tabs{display:flex;gap:4px;padding:8px 10px;overflow-x:auto;scrollbar-width:none}.wc-tabs::-webkit-scrollbar{display:none}.wc-tab{padding:5px 10px;background:#1a2a3a;border:1px solid #2a3a4a;border-radius:12px;color:#8ab4d8;font-size:10px;cursor:pointer;white-space:nowrap;flex-shrink:0}.wc-tab.active{background:#0b6bcb;border-color:#0b6bcb;color:#fff;font-weight:700}.wc-content{padding:8px 10px;max-height:500px;overflow-y:auto}.wc-news-grid{display:flex;flex-direction:column;gap:8px}.wc-news-item{display:flex;gap:8px;padding:8px;background:#1a2030;border-radius:8px;cursor:pointer}.wc-news-item:active{opacity:.8}.wc-news-img{flex:0 0 70px;aspect-ratio:16/9;border-radius:6px;overflow:hidden;background:#222}.wc-news-img img{width:100%;height:100%;object-fit:cover}.wc-news-text{flex:1;min-width:0}.wc-news-title{font-size:11px;font-weight:700;color:#eee;line-height:1.3;display:-webkit-box;-webkit-line-clamp:2;-webkit-box-orient:vertical;overflow:hidden}.wc-news-via{font-size:9px;color:#6a9fca;margin-top:2px}.ls-section{margin:6px 4px;background:#1a1a1a;border:1px solid #2a2a2a;border-radius:8px;overflow:hidden}.ls-header{padding:7px 10px;display:flex;align-items:center;justify-content:space-between}.ls-header h3{color:#f0c040;font-size:13px;font-weight:800}.ls-tabs{display:flex;gap:4px;padding:0 10px 8px;overflow-x:auto;scrollbar-width:none}.ls-tabs::-webkit-scrollbar{display:none}.ls-tab{padding:4px 10px;background:#222;border:1px solid #333;border-radius:12px;color:#999;font-size:10px;white-space:nowrap;cursor:pointer;flex-shrink:0}.ls-tab.active{background:#2d8659;border-color:#2d8659;color:#fff;font-weight:700}.ls-content{max-height:420px;overflow-y:auto;padding:0 6px 8px;font-size:12px;color:#ddd}.ls-content ul{list-style:none;padding:0;margin:0}.ls-content .title-content{display:flex;gap:6px;align-items:center;background:#222;border-radius:4px;margin:4px 0;padding:5px 8px}.ls-content .title-content img{width:18px;height:18px}.ls-content .title-content strong{font-size:11px;color:#ccc}.ls-content .match-detail{padding:6px;border-bottom:1px solid #262626;cursor:pointer}.ls-content .match-detail:hover{background:#1a2a1f}.ls-content .match{display:flex;flex-wrap:wrap;align-items:center;gap:4px}.ls-content .datetime{width:100%;font-size:9px;color:#888}.ls-content .teams{display:flex;width:100%;align-items:center;gap:4px}.ls-content .team{flex:1;display:flex;align-items:center;gap:4px;min-width:0;text-decoration:none}.ls-content .team .name{font-size:11px;color:#ddd;white-space:nowrap;overflow:hidden;text-overflow:ellipsis}.ls-content .team .logo img{width:18px;height:18px}.ls-content .home-team{justify-content:flex-end;text-align:right}.ls-content .status{flex:0 0 54px;text-align:center}.ls-content .status a{color:#fff;text-decoration:none;font-weight:800;font-size:12px}.ls-content .status .label{font-size:8px;color:#888;display:block}.ls-content .status .label.live{color:#e74c3c}.ls-content .info,.ls-content .btns{display:none}.ls-content table{width:100%;border-collapse:collapse;font-size:11px;color:#ccc}.ls-content table th{background:#222;color:#999;padding:5px 4px;font-size:10px;border-bottom:1px solid #333}.ls-content table td{padding:4px 3px;border-bottom:1px solid #1a1a1a}.ls-content table .team-name{display:flex;align-items:center;gap:4px}.ls-content table .team-name img{width:16px;height:16px}.ls-content table .pts{font-weight:800;color:#f0c040}.match-overlay{position:fixed;inset:0;background:#111;z-index:9999;display:none;flex-direction:column;overflow:auto}.match-overlay.active{display:flex}.mo-header{padding:10px;background:#1a1a1a;display:flex;justify-content:space-between;align-items:center;position:sticky;top:0;z-index:1}.mo-header h3{font-size:13px;color:#eee}.mo-close{background:none;border:0;color:#fff;font-size:22px;cursor:pointer}.mo-tabs{display:flex;gap:4px;padding:8px 10px;background:#1a1a1a;overflow-x:auto}.mo-tab{padding:5px 12px;background:#222;border:1px solid #333;border-radius:10px;color:#999;font-size:10px;cursor:pointer;white-space:nowrap}.mo-tab.active{background:#2d8659;color:#fff}.mo-body{padding:8px;overflow-x:auto;font-size:12px;color:#ddd}.mo-body ul{list-style:none;padding:0;margin:0}.mo-body li{padding:5px 0;border-bottom:1px solid #222}.featured-match{margin:6px 4px;background:linear-gradient(135deg,#1a2a1f,#0d1117);border:1px solid #2d8659;border-radius:10px;padding:12px;cursor:pointer}.fm-league{text-align:center;color:#5cb87a;font-size:9px;font-weight:700;text-transform:uppercase}.fm-teams{display:flex;align-items:center;justify-content:center;gap:10px;margin-top:6px}.fm-team{flex:1;display:flex;flex-direction:column;align-items:center;gap:4px}.fm-team img{width:32px;height:32px;object-fit:contain}.fm-team span{font-size:10px;color:#ccc;text-align:center}.fm-score{font-size:22px;font-weight:900;min-width:60px;text-align:center;color:#fff}.fm-status{text-align:center;margin-top:6px;font-size:9px;color:#e74c3c;font-weight:700}.fm-status.upcoming{color:#f0c040}.ai-compose{margin:6px 4px;background:#141414;border:1px solid #2a2a2a;border-radius:10px;padding:10px}.ai-compose-title{font-size:13px;font-weight:800;color:#5cb87a;margin-bottom:8px}.ai-compose-row{display:flex;gap:6px;margin-top:6px}.ai-compose input{flex:1;background:#222;border:1px solid #333;color:#eee;border-radius:18px;padding:9px 12px;font-size:12px;min-width:0}.ai-compose button{background:#2d8659;border:0;color:#fff;border-radius:18px;padding:9px 12px;font-size:11px;font-weight:700;cursor:pointer;white-space:nowrap}.ai-compose button.secondary{background:#333}.hot-topic-row{display:flex;gap:6px;overflow-x:auto;padding:4px 0;scrollbar-width:none}.hot-topic-row::-webkit-scrollbar{display:none}.hot-chip{flex:0 0 auto;background:#222;border:1px solid #333;color:#ddd;border-radius:16px;padding:5px 10px;font-size:11px;cursor:pointer;white-space:nowrap}.hot-chip:active{transform:scale(.96)}.hashtag-sources{margin:8px 4px;background:#1a1a1a;border:1px solid #2a2a2a;border-radius:10px;padding:10px}.hashtag-sources h3{font-size:13px;color:#5cb87a;margin-bottom:8px}.hashtag-src-item{display:flex;gap:8px;padding:8px;background:#202020;border-radius:8px;margin:6px 0;cursor:pointer}.hashtag-src-item:active{opacity:.8}.hashtag-src-img{flex:0 0 80px;aspect-ratio:16/9;background:#333;border-radius:6px;overflow:hidden}.hashtag-src-img img{width:100%;height:100%;object-fit:cover}.hashtag-src-text{flex:1;min-width:0}.hashtag-src-title{font-size:12px;font-weight:700;color:#eee;display:-webkit-box;-webkit-line-clamp:2;-webkit-box-orient:vertical;overflow:hidden}.hashtag-src-via{font-size:10px;color:#888;margin-top:2px}.hashtag-rewrite-btn{width:100%;margin-top:8px;background:#2d8659;border:0;color:#fff;padding:9px;border-radius:10px;font-size:12px;font-weight:700;cursor:pointer}.hashtag-load-more{width:100%;margin-top:8px;background:#222;border:1px solid #333;color:#ccc;padding:9px;border-radius:10px;font-size:12px;cursor:pointer}.hashtag-loading{display:flex;align-items:center;gap:8px;padding:12px;color:#888;font-size:12px}.hashtag-spinner{width:16px;height:16px;border:2px solid #333;border-top-color:#5cb87a;border-radius:50%;animation:ht-spin .8s linear infinite}@keyframes ht-spin{to{transform:rotate(360deg)}}.wall-item{flex:0 0 260px;background:#141414;border:1px solid #2b2b2b;border-radius:10px;padding:8px}.wall-item-new{animation:wall-flash 1.8s ease-out}@keyframes wall-flash{0%{border-color:#f0c040;box-shadow:0 0 18px rgba(240,192,64,.35)}30%{border-color:#f0c040;box-shadow:0 0 12px rgba(240,192,64,.2)}100%{border-color:#2b2b2b;box-shadow:none}}.wall-thumb{width:100%;aspect-ratio:16/9;border-radius:8px;background:#222;overflow:hidden;margin-bottom:6px;position:relative}.wall-thumb img{width:100%;height:100%;object-fit:cover}.wall-video-badge{position:absolute;top:4px;right:4px;background:rgba(45,134,89,.9);color:#fff;font-size:10px;padding:2px 6px;border-radius:6px;font-weight:700}.wall-title{font-size:12px;color:#5cb87a;font-weight:800;line-height:1.3;margin-bottom:4px;display:-webkit-box;-webkit-line-clamp:2;-webkit-box-orient:vertical;overflow:hidden}.wall-text{font-size:11px;color:#bbb;line-height:1.4;white-space:pre-wrap;display:-webkit-box;-webkit-line-clamp:4;-webkit-box-orient:vertical;overflow:hidden}.wall-actions{display:flex;gap:6px;margin-top:8px}.wall-actions button{flex:1;border:1px solid #333;background:#222;color:#ddd;border-radius:14px;padding:6px 8px;font-size:10px;cursor:pointer}.wall-actions button.primary{background:#2d8659;border-color:#2d8659;color:#fff}#progress-toast{position:fixed;bottom:70px;left:50%;transform:translateX(-50%);background:#2d8659;color:#fff;padding:10px 20px;border-radius:20px;font-size:12px;z-index:99998;box-shadow:0 4px 12px rgba(0,0,0,.4);display:none;white-space:nowrap}.storage-warn{background:#332200;border:1px solid #664400;color:#ffcc00;padding:8px 12px;border-radius:8px;font-size:11px;margin:6px 4px}</style>
 
 
14
  </head>
15
  <body>
16
+ <div class="header"><h1>📰 VNEWS</h1><p>Tin tức · Bóng đá LIVE · Video · AI · World Cup 2026</p></div>
17
  <div class="cats" id="cat-bar"></div>
18
  <div id="view-home" class="view active"><div class="loading">Đang tải...</div></div>
19
  <div id="view-cat" class="view"></div>
 
30
  var _cats=[],_hlLeagueData={},_currentArticle=null;window._currentEventId='';
31
  function esc(s){return String(s||'').replace(/[&<>"']/g,m=>({'&':'&amp;','<':'&lt;','>':'&gt;','"':'&quot;',"'":'&#39;'}[m]))}
32
  function showView(id){document.querySelectorAll('.view').forEach(v=>v.classList.remove('active'));document.getElementById(id)?.classList.add('active')}
33
+ function switchCat(id){document.querySelectorAll('.cat').forEach(c=>c.classList.remove('active'));document.querySelector(`[data-cat="${id}"]`)?.classList.add('active');document.querySelectorAll('.view').forEach(v=>v.classList.remove('active'));document.querySelectorAll('video').forEach(v=>{v.pause();if(v._hls){v._hls.destroy();v._hls=null}});document.querySelectorAll('iframe[data-yt-src]').forEach(f=>{f.src=''});if(id==='home')document.getElementById('view-home').classList.add('active');else if(id==='news-all'){document.getElementById('view-cat').classList.add('active');loadNewsTab()}else{document.getElementById('view-cat').classList.add('active');loadCat(id)}}
34
  function toast(msg){let t=document.getElementById('progress-toast');if(t){t.textContent=msg;t.style.display='block';setTimeout(()=>{t.style.display='none'},3500)}}
35
  function doShare(title,url,img){const shareUrl=SPACE+'/s?url='+encodeURIComponent(url)+'&title='+encodeURIComponent(title)+'&img='+encodeURIComponent(img||'');if(navigator.share)navigator.share({title,url:shareUrl}).catch(()=>{});else navigator.clipboard.writeText(shareUrl).then(()=>alert('Đã sao chép!')).catch(()=>{})}
36
+ async function init(){_cats=await fetch('/api/categories').then(r=>r.json()).catch(()=>[]);let bar='<div class="cat active" data-cat="home">🏠</div><div class="cat" data-cat="news-all">📰 Tin tức</div>';_cats.forEach(c=>{bar+=`<div class="cat" data-cat="${c.id}">${c.name}</div>`});document.getElementById('cat-bar').innerHTML=bar;document.querySelectorAll('.cat').forEach(t=>{t.onclick=()=>switchCat(t.dataset.cat)});}
37
  var SPACE=location.origin;
38
  </script>
39
+ <script src="/static/app_v2.js"></script>
40
  <script src="/static/yt_live.js"></script>
41
+ <script src="/static/hot_multi.js"></script>
 
42
  <script src="/static/wc2026_v2.js"></script>
43
  <script src="/static/live_mode.js"></script>
44
  <script src="/static/match_detail_v6.js"></script>
45
+ <script>init();</script>
46
  </body>
47
  </html>
static/restart_trigger.txt ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ Rebuild triggered at 2026-06-13 08:22 UTC
2
+ Changes:
3
+ - app_v2.js: homepage instant shell, shorter timeouts, lazy loading
4
+ - wc2026_v2.js: WC highlights via backend proxy (CORS fix)
5
+ - yt_live.js: miniplayer auto-close when returning to home
6
+ - app_v2_entry.py: /api/proxy/xlb endpoint for xemlaibongda.top
static/rewrite_fix.js ADDED
@@ -0,0 +1,90 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ // Fix rewriteArticle - call correct endpoint
2
+ // This file patches the rewriteArticle function to use /api/rewrite_slide instead of /api/rewrite_share
3
+
4
+ (function(){
5
+ // Override rewriteArticle to call /api/rewrite_slide
6
+ const origRewrite = window.rewriteArticle;
7
+ window.rewriteArticle = async function(){
8
+ const url = _currentArticle?.url;
9
+ if(!url) return;
10
+ toast('⏳ Đang tạo slide tóm tắt...');
11
+ try {
12
+ const r = await fetch('/api/rewrite_slide', {
13
+ method: 'POST',
14
+ headers: {'Content-Type': 'application/json'},
15
+ body: JSON.stringify({url, context: document.querySelector('.article-view')?.innerText?.slice(0,14000) || ''})
16
+ });
17
+ const j = await r.json();
18
+ if (!r.ok || j.error) throw new Error(j.error);
19
+ toast('✅ Đã đăng Tường AI!');
20
+ if (j.post) prependWallPost(j.post);
21
+ // Navigate to the new post on Tường AI (home). Slide overlay (if any) stays on top.
22
+ if (j.post && typeof goToWallPost === 'function') goToWallPost(j.post.id);
23
+ // Show slides preview
24
+ if (j.slides && j.slides.length) {
25
+ showSlidePreview(j.slides, j.post?.title || '');
26
+ }
27
+ } catch(e) {
28
+ // Fallback: try /api/rewrite_share (old endpoint from ai_ext)
29
+ try {
30
+ const r2 = await fetch('/api/rewrite_share', {
31
+ method: 'POST',
32
+ headers: {'Content-Type': 'application/json'},
33
+ body: JSON.stringify({url, context: document.querySelector('.article-view')?.innerText?.slice(0,14000) || ''})
34
+ });
35
+ const j2 = await r2.json();
36
+ if (r2.ok && !j2.error) {
37
+ toast('✅ Đã đăng Tường AI!');
38
+ if (j2.post) prependWallPost(j2.post);
39
+ if (j2.post && typeof goToWallPost === 'function') goToWallPost(j2.post.id);
40
+ return;
41
+ }
42
+ } catch(e2) {}
43
+ toast('❌ ' + e.message);
44
+ }
45
+ };
46
+
47
+ // Show slides as fullscreen overlay
48
+ window.showSlidePreview = function(slides, title) {
49
+ if (!slides || !slides.length) return;
50
+ const overlay = document.createElement('div');
51
+ overlay.id = 'slide-preview';
52
+ overlay.style.cssText = 'position:fixed;inset:0;background:#000;z-index:99999;display:flex;flex-direction:column;overflow:hidden';
53
+
54
+ let currentSlide = 0;
55
+ function renderSlide(idx) {
56
+ const s = slides[idx];
57
+ overlay.innerHTML = `
58
+ <div style="position:absolute;top:10px;left:10px;right:10px;display:flex;justify-content:space-between;align-items:center;z-index:2">
59
+ <button onclick="document.getElementById('slide-preview').remove()" style="background:rgba(0,0,0,.6);border:0;color:#fff;padding:8px 14px;border-radius:20px;font-size:12px;cursor:pointer">✕ Đóng</button>
60
+ <span style="color:#fff;font-size:11px;background:rgba(0,0,0,.6);padding:4px 10px;border-radius:10px">${idx+1}/${slides.length}</span>
61
+ </div>
62
+ <div style="flex:1;display:flex;align-items:center;justify-content:center;padding:20px">
63
+ ${s.image ? `<img src="${esc(s.image)}" style="max-width:100%;max-height:60vh;border-radius:10px;object-fit:contain" onerror="this.style.display='none'">` : ''}
64
+ </div>
65
+ <div style="padding:16px 20px;background:linear-gradient(transparent,rgba(0,0,0,.9));min-height:100px">
66
+ <p style="color:#fff;font-size:14px;line-height:1.6">${esc(s.text)}</p>
67
+ </div>
68
+ <div style="display:flex;gap:10px;padding:10px 20px 20px;justify-content:center">
69
+ <button onclick="prevSlide()" style="background:#333;border:0;color:#fff;padding:10px 20px;border-radius:20px;font-size:12px;cursor:pointer" ${idx===0?'disabled style="opacity:.3"':''}>← Trước</button>
70
+ <button onclick="nextSlide()" style="background:#2d8659;border:0;color:#fff;padding:10px 20px;border-radius:20px;font-size:12px;cursor:pointer" ${idx===slides.length-1?'disabled style="opacity:.3"':''}>Tiếp →</button>
71
+ </div>
72
+ `;
73
+ }
74
+
75
+ window.nextSlide = function() { if (currentSlide < slides.length - 1) { currentSlide++; renderSlide(currentSlide); } };
76
+ window.prevSlide = function() { if (currentSlide > 0) { currentSlide--; renderSlide(currentSlide); } };
77
+
78
+ renderSlide(0);
79
+ document.body.appendChild(overlay);
80
+
81
+ // Swipe support
82
+ let startX = 0;
83
+ overlay.addEventListener('touchstart', e => { startX = e.touches[0].clientX; });
84
+ overlay.addEventListener('touchend', e => {
85
+ const diff = e.changedTouches[0].clientX - startX;
86
+ if (diff < -50) nextSlide();
87
+ else if (diff > 50) prevSlide();
88
+ });
89
+ };
90
+ })();
static/shorts_fresh.js DELETED
@@ -1,5 +0,0 @@
1
- /**
2
- * VNEWS Shorts — REMOVED per user request
3
- * Dantri/SKDS slides removed from homepage entirely
4
- */
5
- (function(){'use strict';console.log('[Shorts] Disabled - Dantri/SKDS removed');})();
 
 
 
 
 
 
static/vtv_init.js DELETED
@@ -1,397 +0,0 @@
1
- // ===== VNEWS VTV Player v8 — VTV6 proxy, EPG multi-source, FPTPlay fallback =====
2
- (function() {
3
- if (window._vtvInitLoaded) return;
4
- window._vtvInitLoaded = true;
5
-
6
- // VTV6 first (ưu tiên hàng đầu)
7
- var CHANNELS = [
8
- {id:'vtv6',name:'VTV6',badge:'Thanh Niên'},{id:'vtv1',name:'VTV1',badge:'Tin tức'},
9
- {id:'vtv2',name:'VTV2',badge:'Khoa học'},{id:'vtv3',name:'VTV3',badge:'Giải trí'},
10
- {id:'vtv4',name:'VTV4',badge:'Quốc Tế'},{id:'vtv5',name:'VTV5',badge:'Miền Nam'},
11
- {id:'vtv7',name:'VTV7',badge:'Giáo Dục'},{id:'vtv8',name:'VTV8',badge:'Miền Trung'},
12
- {id:'vtv9',name:'VTV9',badge:'Miền Bắc'},{id:'vtv10',name:'VTV10',badge:'Cần Thơ'},
13
- {id:'vtvprime',name:'VTVPrime',badge:'Prime'},
14
- ];
15
-
16
- var _hlsInst = null;
17
- var _currentCh = null;
18
- var _epgTimer = null;
19
- var _epgData = [];
20
- var _timeTimer = null;
21
-
22
- // ===== CSS =====
23
- (function injectCSS() {
24
- if (document.getElementById('vtv-pro-css')) return;
25
- var s = document.createElement('style');
26
- s.id = 'vtv-pro-css';
27
- s.textContent = [
28
- '#vtv-player-section{display:block!important;margin:8px 0 10px;background:linear-gradient(180deg,#0a0e17,#111622);border:1px solid rgba(0,150,255,.15);border-radius:14px;overflow:hidden;font-family:-apple-system,BlinkMacSystemFont,"Segoe UI",Roboto,sans-serif}',
29
- '#vtv-player-section .vtv-pro-head{display:flex;align-items:center;gap:10px;padding:10px 14px;background:linear-gradient(90deg,#0a1628,rgba(0,102,204,.08))}',
30
- '#vtv-player-section .vtv-pro-logo{width:22px;height:22px;border-radius:50%;background:linear-gradient(135deg,#0066cc,#00ccff);display:flex;align-items:center;justify-content:center;font-size:11px;color:#fff;font-weight:800;flex-shrink:0}',
31
- '#vtv-player-section .vtv-pro-title{font-size:13px;font-weight:700;color:#e8eaed;letter-spacing:.3px}',
32
- '#vtv-player-section .vtv-pro-live{font-size:9px;font-weight:700;color:#00cc88;background:rgba(0,204,136,.12);padding:2px 8px;border-radius:10px;display:flex;align-items:center;gap:4px;margin-left:auto}',
33
- '#vtv-player-section .vtv-pro-live-dot{width:5px;height:5px;border-radius:50%;background:#00cc88;animation:vtv-pro-pulse 1.2s infinite}',
34
- '@keyframes vtv-pro-pulse{0%,100%{opacity:1}50%{opacity:.3}}',
35
- '#vtv-player-section .vtv-pro-tabs{display:flex;flex-wrap:wrap;gap:2px;padding:6px 10px 8px;overflow-x:auto;scrollbar-width:none}',
36
- '#vtv-player-section .vtv-pro-tabs::-webkit-scrollbar{display:none}',
37
- '#vtv-player-section .vtv-pro-tab{padding:5px 10px;background:rgba(255,255,255,.04);border:1px solid rgba(255,255,255,.06);border-radius:8px;color:#9aa0a6;font-size:10px;font-weight:500;cursor:pointer;white-space:nowrap;flex-shrink:0;transition:all .2s ease}',
38
- '#vtv-player-section .vtv-pro-tab:hover{background:rgba(0,102,204,.1);border-color:rgba(0,102,204,.3);color:#e8eaed}',
39
- '#vtv-player-section .vtv-pro-tab.on{background:rgba(0,102,204,.2);border-color:#0066cc;color:#fff;font-weight:600}',
40
- '#vtv-player-section .vtv-pro-tab .b{font-size:7px;opacity:.5;display:block;margin-top:1px}',
41
- '#vtv-player-section .vtv-pro-frame{position:relative;width:100%;aspect-ratio:16/9;background:#000;min-height:200px;border-top:1px solid rgba(255,255,255,.04)}',
42
- '#vtv-player-section .vtv-pro-frame video{position:absolute;inset:0;width:100%!important;height:100%!important;object-fit:contain;background:#000}',
43
- '#vtv-player-section .vtv-pro-load{display:flex;flex-direction:column;align-items:center;justify-content:center;height:100%;min-height:200px;gap:12px;color:#9aa0a6;font-size:12px}',
44
- '#vtv-player-section .vtv-pro-spinner{width:28px;height:28px;border:2px solid rgba(255,255,255,.06);border-top-color:#0066cc;border-radius:50%;animation:vtv-pro-spin .7s linear infinite}',
45
- '@keyframes vtv-pro-spin{to{transform:rotate(360deg)}}',
46
- '#vtv-player-section .vtv-pro-err{display:flex;flex-direction:column;align-items:center;justify-content:center;height:100%;min-height:200px;gap:10px;color:#9aa0a6;font-size:12px;text-align:center;padding:20px}',
47
- '#vtv-player-section .vtv-pro-err .icon{font-size:28px;opacity:.5}',
48
- '#vtv-player-section .vtv-pro-err .msg{color:#9aa0a6}',
49
- '#vtv-player-section .vtv-pro-err button{background:rgba(0,102,204,.15);border:1px solid rgba(0,102,204,.3);color:#8ab4f8;padding:7px 16px;border-radius:8px;font-size:11px;cursor:pointer;transition:all .15s}',
50
- '#vtv-player-section .vtv-pro-err button:hover{background:rgba(0,102,204,.25)}',
51
- '#vtv-player-section .vtv-pro-controls{display:flex;align-items:center;gap:6px;padding:6px 12px;background:rgba(0,0,0,.3);border-top:1px solid rgba(255,255,255,.04)}',
52
- '#vtv-player-section .vtv-pro-btn{background:rgba(255,255,255,.06);border:1px solid rgba(255,255,255,.08);color:#9aa0a6;font-size:10px;padding:4px 10px;border-radius:6px;cursor:pointer;transition:all .15s;display:flex;align-items:center;gap:4px}',
53
- '#vtv-player-section .vtv-pro-btn:hover{background:rgba(0,102,204,.12);color:#e8eaed}',
54
- '#vtv-player-section .vtv-pro-btn.active{background:rgba(0,102,204,.2);border-color:#0066cc;color:#8ab4f8}',
55
- '#vtv-player-section .vtv-pro-epg{border-top:1px solid rgba(255,255,255,.04);background:rgba(0,0,0,.15)}',
56
- '#vtv-player-section .vtv-pro-epg-hdr{display:flex;align-items:center;justify-content:space-between;padding:8px 12px 4px}',
57
- '#vtv-player-section .vtv-pro-epg-title{font-size:10px;font-weight:600;color:#9aa0a6;letter-spacing:.5px;text-transform:uppercase}',
58
- '#vtv-player-section .vtv-pro-epg-time{font-size:9px;color:#5f6368;font-variant-numeric:tabular-nums}',
59
- '#vtv-player-section .vtv-pro-epg-list{padding:2px 8px 10px;display:flex;flex-direction:column;gap:2px;max-height:200px;overflow-y:auto;scrollbar-width:thin}',
60
- '#vtv-player-section .vtv-pro-epg-list::-webkit-scrollbar{width:3px}',
61
- '#vtv-player-section .vtv-pro-epg-list::-webkit-scrollbar-thumb{background:rgba(255,255,255,.08);border-radius:3px}',
62
- '#vtv-player-section .vtv-pro-epg-row{display:flex;align-items:center;gap:8px;padding:5px 8px;border-radius:6px;transition:all .2s;border-left:2px solid transparent;font-size:11px}',
63
- '#vtv-player-section .vtv-pro-epg-row:hover{background:rgba(255,255,255,.03)}',
64
- '#vtv-player-section .vtv-pro-epg-row.now{border-left-color:#00cc88;background:rgba(0,204,136,.06)}',
65
- '#vtv-player-section .vtv-pro-epg-row.passed{opacity:.35}',
66
- '#vtv-player-section .vtv-pro-epg-row .t{font-size:10px;color:#5f6368;min-width:38px;font-variant-numeric:tabular-nums}',
67
- '#vtv-player-section .vtv-pro-epg-row.now .t{color:#00cc88;font-weight:600}',
68
- '#vtv-player-section .vtv-pro-epg-row .n{color:#e8eaed;line-height:1.3}',
69
- '#vtv-player-section .vtv-pro-epg-row.now .n{color:#fff;font-weight:500}',
70
- '#vtv-player-section .vtv-pro-epg-row .bar{flex:0 0 2px;height:12px;border-radius:1px;background:rgba(255,255,255,.08)}',
71
- '#vtv-player-section .vtv-pro-epg-row.now .bar{background:#00cc88}',
72
- '#vtv-player-section .vtv-pro-epg-empty{color:#5f6368;font-size:11px;padding:12px 14px;text-align:center}',
73
- '#vtv-player-section .vtv-pro-epg-load{color:#5f6368;font-size:11px;padding:12px 14px;text-align:center;display:flex;align-items:center;justify-content:center;gap:6px}',
74
- ].join('');
75
- document.head.appendChild(s);
76
- })();
77
-
78
- // ===== URL handler =====
79
- // VTV6: vtvdigital CDN uses signed tokens + CORS blocks. Proxy through server.
80
- // Other channels: play directly from CDN (no proxy — FPTPlay blocks server).
81
- function proxyUrl(url, chId) {
82
- if (!url) return url;
83
- if (url.indexOf(location.origin) === 0) return url;
84
- if (url.indexOf('/api/proxy/') === 0) return url;
85
- // VTV6: use server proxy endpoint (handles CORS + token rewriting)
86
- if (chId === 'vtv6' || url.indexOf('vtvdigital') !== -1) {
87
- return '/api/proxy/vtv6/master.m3u8';
88
- }
89
- // FPTPlay: play directly in browser (CDN blocks server)
90
- if (url.indexOf('fptplay') !== -1) {
91
- return url;
92
- }
93
- // Other: use generic proxy
94
- return '/api/proxy/m3u8/vtv?url=' + encodeURIComponent(url);
95
- }
96
-
97
- function formatTime(d) {
98
- var h = d.getUTCHours(), m = d.getUTCMinutes();
99
- return (h < 10 ? '0' : '') + h + ':' + (m < 10 ? '0' : '') + m;
100
- }
101
-
102
- // GMT+0 (UTC)
103
- function getVNTime() {
104
- return new Date();
105
- }
106
-
107
- // ===== EPG =====
108
- function loadEPG(chId) {
109
- var epgEl = document.getElementById('vtv-pro-epg-body');
110
- if (!epgEl) return;
111
- epgEl.innerHTML = '<div class="vtv-pro-epg-load"><div class="vtv-pro-spinner" style="width:12px;height:12px;border-width:1.5px;flex-shrink:0"></div>Đang tải lịch...</div>';
112
- var xhr = new XMLHttpRequest();
113
- xhr.open('GET', '/api/vtv/epg/' + chId, true);
114
- xhr.timeout = 15000;
115
- xhr.onload = function() {
116
- try {
117
- var data = JSON.parse(xhr.responseText);
118
- _epgData = data.programs || [];
119
- if (_epgData.length === 0) {
120
- epgEl.innerHTML = '<div class="vtv-pro-epg-empty">Không có lịch phát sóng</div>';
121
- } else {
122
- renderEPG();
123
- }
124
- } catch(e) {
125
- epgEl.innerHTML = '<div class="vtv-pro-epg-empty">Không có lịch phát sóng</div>';
126
- }
127
- };
128
- xhr.onerror = xhr.ontimeout = function() {
129
- epgEl.innerHTML = '<div class="vtv-pro-epg-empty">Không tải được lịch phát sóng</div>';
130
- };
131
- xhr.send();
132
- }
133
-
134
- function renderEPG() {
135
- var epgEl = document.getElementById('vtv-pro-epg-body');
136
- if (!epgEl) return;
137
- if (!_epgData || !_epgData.length) {
138
- epgEl.innerHTML = '<div class="vtv-pro-epg-empty">Chưa có lịch phát sóng</div>';
139
- return;
140
- }
141
- var now = new Date();
142
- var nowStr = formatTime(now);
143
-
144
- _epgData.sort(function(a, b) { return a.time.localeCompare(b.time); });
145
-
146
- var foundNow = false;
147
- for (var i = 0; i < _epgData.length; i++) {
148
- _epgData[i].now = false;
149
- var p = _epgData[i];
150
- var next = _epgData[i + 1];
151
- if (p.time <= nowStr && (!next || next.time > nowStr)) {
152
- p.now = true;
153
- p.end_time = next ? next.time : '';
154
- foundNow = true;
155
- }
156
- }
157
-
158
- if (!foundNow) {
159
- for (var i = 0; i < _epgData.length; i++) {
160
- if (_epgData[i].time > nowStr) {
161
- _epgData[i].now = true;
162
- _epgData[i].end_time = (_epgData[i+1] || {}).time || '';
163
- break;
164
- }
165
- }
166
- }
167
-
168
- var html = '';
169
- for (var i = 0; i < _epgData.length; i++) {
170
- var p = _epgData[i];
171
- var cls = 'vtv-pro-epg-row';
172
- if (p.now) cls += ' now';
173
- else if (p.time < nowStr) cls += ' passed';
174
- var extra = '';
175
- if (p.end_time) {
176
- extra = ' → ' + p.end_time;
177
- } else if (i < _epgData.length - 1) {
178
- extra = ' → ' + _epgData[i + 1].time;
179
- }
180
- html += '<div class="' + cls + '">' +
181
- '<span class="t">' + p.time + extra + '</span>' +
182
- '<span class="bar"></span>' +
183
- '<span class="n">' + p.title + '</span>' +
184
- '</div>';
185
- }
186
- epgEl.innerHTML = html;
187
-
188
- var nowEl = epgEl.querySelector('.now');
189
- if (nowEl) {
190
- nowEl.scrollIntoView({ block: 'center', behavior: 'smooth' });
191
- }
192
- }
193
-
194
- function startEpgRefresh() {
195
- if (_epgTimer) clearInterval(_epgTimer);
196
- _epgTimer = setInterval(function() {
197
- if (_currentCh) {
198
- loadEPG(_currentCh);
199
- }
200
- }, 30000);
201
- }
202
-
203
- function updateClock() {
204
- var timeEl = document.getElementById('vtv-pro-time');
205
- if (timeEl) {
206
- var now = new Date();
207
- timeEl.textContent = formatTime(now) + ' GMT+0';
208
- }
209
- }
210
-
211
- // ===== Switch channel =====
212
- window._vtvProSwitch = function(chId) {
213
- if (_hlsInst) { try { _hlsInst.destroy(); } catch(e){} _hlsInst = null; }
214
- _currentCh = chId;
215
-
216
- var tabs = document.querySelectorAll('#vtv-player-section .vtv-pro-tab');
217
- for (var i = 0; i < tabs.length; i++) {
218
- tabs[i].className = 'vtv-pro-tab' + (tabs[i].getAttribute('data-ch') === chId ? ' on' : '');
219
- }
220
-
221
- updateClock();
222
- if (_timeTimer) clearInterval(_timeTimer);
223
- _timeTimer = setInterval(updateClock, 60000);
224
-
225
- var frame = document.getElementById('vtv-pro-frame');
226
- if (!frame) return;
227
- frame.innerHTML = '<div class="vtv-pro-load"><div class="vtv-pro-spinner"></div><span>Đang kết nối ' + chId.toUpperCase() + '...</span></div>';
228
-
229
- loadEPG(chId);
230
- startEpgRefresh();
231
-
232
- var xhr = new XMLHttpRequest();
233
- xhr.open('GET', '/api/vtv/stream/' + chId, true);
234
- xhr.timeout = 30000;
235
- xhr.onload = function() {
236
- try {
237
- var data = JSON.parse(xhr.responseText);
238
- var url = data.stream_url;
239
- if (!url) {
240
- frame.innerHTML = '<div class="vtv-pro-err"><div class="icon">📡</div><div class="msg">Kênh đang tạm ngừng phát sóng</div><button onclick="_vtvProSwitch(\''+chId+'\')">Thử lại</button></div>';
241
- return;
242
- }
243
- // VTV6: proxy through server, with FPTPlay fallback
244
- if (chId === 'vtv6') {
245
- playVTV6(url, frame);
246
- } else {
247
- url = proxyUrl(url, chId);
248
- playStream(url, frame);
249
- }
250
- } catch(e) {
251
- frame.innerHTML = '<div class="vtv-pro-err"><div class="icon">⚠️</div><div class="msg">Lỗi tải dữ liệu</div><button onclick="_vtvProSwitch(\''+chId+'\')">Thử lại</button></div>';
252
- }
253
- };
254
- xhr.onerror = xhr.ontimeout = function() {
255
- frame.innerHTML = '<div class="vtv-pro-err"><div class="icon">🔌</div><div class="msg">Mất kết nối máy chủ</div><button onclick="_vtvProSwitch(\''+chId+'\')">Thử lại</button></div>';
256
- };
257
- xhr.send();
258
- };
259
-
260
- // VTV6: try server proxy first, fallback to FPTPlay direct
261
- function playVTV6(vtvdigitalUrl, frame) {
262
- // Try proxy first (handles CORS + token rewriting)
263
- var proxyUrl = '/api/proxy/vtv6/master.m3u8';
264
-
265
- // Check if proxy works by doing a HEAD request
266
- var check = new XMLHttpRequest();
267
- check.open('GET', proxyUrl, true);
268
- check.timeout = 10000;
269
- check.onload = function() {
270
- if (check.status === 200) {
271
- // Proxy works! Use it
272
- playStream(proxyUrl, frame);
273
- } else {
274
- // Proxy failed (403 or 502) — try FPTPlay direct
275
- var fallback = 'https://live.fptplay53.net/fnxhd1/vtv6_vhls.smil/chunklist_b5000000.m3u8';
276
- playStream(fallback, frame);
277
- }
278
- };
279
- check.onerror = function() {
280
- // Proxy unreachable — try FPTPlay direct
281
- var fallback = 'https://live.fptplay53.net/fnxhd1/vtv6_vhls.smil/chunklist_b5000000.m3u8';
282
- playStream(fallback, frame);
283
- };
284
- check.send();
285
- }
286
-
287
- function playStream(url, frame) {
288
- var video = document.createElement('video');
289
- video.id = 'vtv-pro-video';
290
- video.style.cssText = 'width:100%;height:100%;object-fit:contain';
291
- video.setAttribute('controls', '');
292
- video.setAttribute('autoplay', '');
293
- video.setAttribute('playsinline', '');
294
- frame.innerHTML = '';
295
- frame.appendChild(video);
296
-
297
- if (typeof Hls !== 'undefined' && Hls.isSupported()) {
298
- var hls = new Hls({ debug: false, enableWorker: true, maxBufferLength: 30 });
299
- hls.loadSource(url);
300
- hls.attachMedia(video);
301
- hls.on(Hls.Events.MANIFEST_PARSED, function() {
302
- video.play().catch(function(){});
303
- });
304
- hls.on(Hls.Events.ERROR, function(ev, data) {
305
- if (data.fatal) {
306
- hls.destroy();
307
- _hlsInst = null;
308
- // If VTV6 proxy failed, try direct FPTPlay
309
- if (_currentCh === 'vtv6' && url.indexOf('/api/proxy/') === 0) {
310
- var fallback = 'https://live.fptplay53.net/fnxhd1/vtv6_vhls.smil/chunklist_b5000000.m3u8';
311
- playStream(fallback, frame);
312
- return;
313
- }
314
- frame.innerHTML = '<div class="vtv-pro-err"><div class="icon">📹</div><div class="msg">Lỗi phát video</div><button onclick="_vtvProSwitch(\''+_currentCh+'\')">Thử lại</button></div>';
315
- }
316
- });
317
- _hlsInst = hls;
318
- } else if (video.canPlayType('application/vnd.apple.mpegurl')) {
319
- video.src = url;
320
- } else {
321
- frame.innerHTML = '<div class="vtv-pro-err"><div class="msg">Trình duyệt không hỗ trợ HLS</div></div>';
322
- }
323
- }
324
-
325
- // ===== HTML builder =====
326
- function buildHTML() {
327
- var tabs = '';
328
- for (var i = 0; i < CHANNELS.length; i++) {
329
- var ch = CHANNELS[i];
330
- tabs += '<span class="vtv-pro-tab" data-ch="'+ch.id+'" onclick="_vtvProSwitch(\''+ch.id+'\')">' +
331
- ch.name + '<span class="b">'+ch.badge+'</span></span>';
332
- }
333
- return '<div id="vtv-player-section">' +
334
- '<div class="vtv-pro-head">' +
335
- '<div class="vtv-pro-logo">V</div>' +
336
- '<span class="vtv-pro-title">VTV Player</span>' +
337
- '<span class="vtv-pro-live"><span class="vtv-pro-live-dot"></span>TRỰC TIẾP</span>' +
338
- '</div>' +
339
- '<div class="vtv-pro-tabs">'+tabs+'</div>' +
340
- '<div class="vtv-pro-frame" id="vtv-pro-frame">' +
341
- '<div class="vtv-pro-load"><div class="vtv-pro-spinner"></div><span>Chọn kênh để xem trực tiếp</span></div>' +
342
- '</div>' +
343
- '<div class="vtv-pro-controls">' +
344
- '<span id="vtv-pro-time" class="vtv-pro-btn" style="background:none;border:none;font-size:10px;color:#5f6368;margin-right:auto"></span>' +
345
- '</div>' +
346
- '<div class="vtv-pro-epg">' +
347
- '<div class="vtv-pro-epg-hdr">' +
348
- '<span class="vtv-pro-epg-title">Lịch phát sóng</span>' +
349
- '</div>' +
350
- '<div class="vtv-pro-epg-list" id="vtv-pro-epg-body">' +
351
- '<div class="vtv-pro-epg-empty">Chọn kênh để xem lịch phát sóng</div>' +
352
- '</div>' +
353
- '</div>' +
354
- '</div>';
355
- }
356
-
357
- function inject() {
358
- var homeEl = document.getElementById('view-home');
359
- if (!homeEl || document.getElementById('vtv-player-section')) return;
360
-
361
- var featured = document.getElementById('home-featured-area') || homeEl.querySelector('.featured-match, .fm-section, .slider-wrap');
362
- if (featured && featured.parentNode) {
363
- featured.insertAdjacentHTML('afterend', buildHTML());
364
- } else if (homeEl.firstChild) {
365
- homeEl.insertAdjacentHTML('afterbegin', buildHTML());
366
- }
367
-
368
- // Auto-load VTV6 as default
369
- setTimeout(function() {
370
- if (window._vtvProSwitch) {
371
- window._vtvProSwitch('vtv6');
372
- }
373
- }, 500);
374
- }
375
-
376
- // Hook into loadHome
377
- var orig = window.loadHome;
378
- if (typeof orig === 'function') {
379
- window.loadHome = function() {
380
- var r = orig.apply(this, arguments);
381
- if (r && typeof r.then === 'function') {
382
- return r.then(function(v) { setTimeout(inject, 2000); return v; });
383
- } else {
384
- setTimeout(inject, 2000);
385
- return r;
386
- }
387
- };
388
- } else {
389
- (function waitAndInject() {
390
- if (document.getElementById('view-home') && !document.getElementById('vtv-player-section')) {
391
- inject();
392
- } else if (!document.getElementById('vtv-player-section')) {
393
- setTimeout(waitAndInject, 1000);
394
- }
395
- })();
396
- }
397
- })();
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
static/vtv_init_loader.html DELETED
@@ -1,2 +0,0 @@
1
- <script src="/static/yt_live.js"></script>
2
- <script src="/static/vtv_init.js"></script>
 
 
 
static/wc2026_v2.js CHANGED
@@ -1,6 +1,5 @@
1
- // === WORLD CUP 2026 - Highlights + Shorts ===
2
  let _wc2026Data=null;let _wcRefreshInterval=null;
3
- let _wcShortsData=[]; // YouTube shorts: VTV Nam Bộ + WC moments
4
  const WC_TEAMS={'Mexico':'🇲🇽 Mexico','Colombia':'🇨🇴 Colombia','USA':'🇺🇸 Mỹ','United States':'🇺🇸 Mỹ','Brazil':'🇧🇷 Brazil','Argentina':'🇦🇷 Argentina','Germany':'🇩🇪 Đức','France':'🇫🇷 Pháp','Spain':'🇪🇸 Tây Ban Nha','England':'🏴󠁧󠁢󠁥󠁮󠁧󠁿 Anh','Portugal':'🇵🇹 Bồ Đào Nha','Netherlands':'🇳🇱 Hà Lan','Italy':'🇮🇹 Ý','Belgium':'🇧🇪 Bỉ','Croatia':'🇭🇷 Croatia','Uruguay':'🇺🇾 Uruguay','Japan':'🇯🇵 Nhật Bản','South Korea':'🇰🇷 Hàn Quốc','Korea Republic':'🇰🇷 Hàn Quốc','Australia':'🇦🇺 Úc','Saudi Arabia':'🇸🇦 Ả Rập Xê Út','Iran':'🇮🇷 Iran','Qatar':'🇶🇦 Qatar','Canada':'🇨🇦 Canada','Morocco':'🇲🇦 Morocco','Senegal':'🇸🇳 Senegal','Ecuador':'🇪🇨 Ecuador','Serbia':'🇷🇸 Serbia','Switzerland':'🇨🇭 Thụy Sĩ','Denmark':'🇩🇰 Đan Mạch','Poland':'🇵🇱 Ba Lan','Turkey':'🇹🇷 Thổ Nhĩ Kỳ','Türkiye':'🇹🇷 Thổ Nhĩ Kỳ','Ukraine':'🇺🇦 Ukraine','Egypt':'🇪🇬 Ai Cập','Nigeria':'🇳🇬 Nigeria','Cameroon':'🇨🇲 Cameroon','Ghana':'🇬🇭 Ghana','Tunisia':'🇹🇳 Tunisia','Algeria':'🇩🇿 Algeria','South Africa':'🇿🇦 Nam Phi','Indonesia':'🇮🇩 Indonesia','Vietnam':'🇻🇳 Việt Nam','China PR':'🇨🇳 Trung Quốc','New Zealand':'🇳🇿 New Zealand','Panama':'🇵🇦 Panama','Costa Rica':'🇨🇷 Costa Rica','Jamaica':'🇯🇲 Jamaica','Honduras':'🇭🇳 Honduras','Paraguay':'🇵🇾 Paraguay','Bolivia':'🇧🇴 Bolivia','Chile':'🇨🇱 Chile','Peru':'🇵🇪 Peru','Venezuela':'🇻🇪 Venezuela','Norway':'🇳🇴 Na Uy','Sweden':'🇸🇪 Thụy Điển','Austria':'🇦🇹 Áo','Scotland':'🏴󠁧󠁢󠁳󠁣󠁴󠁿 Scotland','Wales':'🏴󠁧󠁢󠁷󠁬󠁳󠁿 Xứ Wales','Ireland':'🇮🇪 Ireland','Romania':'🇷🇴 Romania','Hungary':'🇭🇺 Hungary','Greece':'🇬🇷 Hy Lạp','Uzbekistan':'🇺🇿 Uzbekistan','Iraq':'🇮🇶 Iraq','TBD':'🏳️ TBD'};
5
  function _t(n){return WC_TEAMS[n]||WC_TEAMS[(n||'').trim()]||'🏴 '+n;}
6
 
@@ -29,19 +28,6 @@ const WC_CSS=`<style>
29
  .wc-bxh table{width:100%!important;border-collapse:collapse!important;font-size:11px!important;display:table!important}
30
  .wc-bxh th{background:#0b1a2a!important;color:#6a9fca!important;padding:5px 3px!important;font-size:9px!important;text-align:center!important}
31
  .wc-bxh td{padding:4px 3px!important;border-bottom:1px solid #0d1a2a!important;font-size:10px!important;color:#ddd!important}
32
- /* WC Shorts slider */
33
- .wc-shorts-slider{margin:8px 0}
34
- .wc-shorts-track{display:flex;overflow-x:auto;gap:8px;padding:4px 0 10px;scrollbar-width:none}
35
- .wc-shorts-track::-webkit-scrollbar{display:none}
36
- .wc-shorts-item{flex:0 0 130px;cursor:pointer;border-radius:8px;overflow:hidden;background:#0a1520;border:1px solid #1a3a5a;transition:border-color .2s}
37
- .wc-shorts-item:hover{border-color:#f0c040}
38
- .wc-shorts-thumb{position:relative;width:100%;aspect-ratio:9/16;background:#1a2a3a;overflow:hidden}
39
- .wc-shorts-thumb img{width:100%;height:100%;object-fit:cover}
40
- .wc-shorts-thumb .card-play{position:absolute;left:50%;top:50%;transform:translate(-50%,-50%);width:32px;height:32px;border-radius:50%;background:rgba(0,0,0,.55);display:flex;align-items:center;justify-content:center;color:#fff;font-size:14px}
41
- .wc-shorts-info{padding:4px 6px}
42
- .wc-shorts-badge{font-size:8px;font-weight:700;padding:1px 4px;border-radius:3px;display:inline-block;margin-bottom:2px}
43
- .wc-shorts-title{font-size:10px;color:#ccc;line-height:1.2;display:-webkit-box;-webkit-line-clamp:2;-webkit-box-orient:vertical;overflow:hidden}
44
- .wc-shorts-viewall{flex:0 0 100px;display:flex;align-items:center;justify-content:center;background:linear-gradient(135deg,#0b2e4a,#1a3a5a);border-radius:8px;cursor:pointer;border:1px dashed #2a5a8a;color:#8ab4d8;font-size:11px;font-weight:700;text-align:center;padding:8px}
45
  </style>`;
46
 
47
  function switchWCTab(tab){
@@ -68,6 +54,7 @@ function renderWCNews(el){
68
  h+=`<div class="wc-news-item" onclick="readArticle('${esc(a.link)}')"><div class="wc-news-img" id="wcimg-${i}">${a.img?`<img src="${esc(a.img)}" loading="lazy" onerror="this.parentElement.innerHTML=''">`:''}</div><div class="wc-news-text">${badge}<div class="wc-news-title">${esc(a.title)}</div><div class="wc-news-via">${esc(a.source||'')}</div></div></div>`;
69
  });
70
  h+='</div>';el.innerHTML=h;
 
71
  combined.slice(0,20).forEach((a,i)=>{if(a.img)return;fetch('/api/article?url='+encodeURIComponent(a.link)).then(r=>r.json()).then(d=>{if(d&&(d.og_image||d.img)){const x=document.getElementById('wcimg-'+i);if(x)x.innerHTML=`<img src="${esc(d.og_image||d.img)}" loading="lazy" onerror="this.parentElement.innerHTML=''">`;}}).catch(()=>{});});
72
  }
73
 
@@ -121,27 +108,31 @@ function renderWCStats(el){
121
  else el.innerHTML='<div class="loading">Thống kê cập nhật khi giải bắt đầu<br><small style="color:#6a9fca">Vua phá lưới · Kiến tạo · Thẻ phạt</small></div>';
122
  }
123
 
124
- // === HIGHLIGHTS - WC match highlights from xemlaibongda.top ===
125
  async function renderWCHighlights(el){
126
  el.innerHTML='<div class="loading">Đang tải highlight World Cup...</div>';
127
  try{
128
- const wcRes = await fetch('/api/proxy/xlb?path=the-gioi/world-cup', {signal: AbortSignal.timeout(15000)}).then(r=>r.json()).catch(()=>({videos:[]}));
129
- let wcVids = (wcRes.videos||[]);
130
- // Deduplicate by link
131
- const seen = new Map();
132
- wcVids.forEach((v,i)=>{if(!seen.has(v.link)) seen.set(v.link, {...v, _li: seen.size});});
133
- wcVids = Array.from(seen.values());
134
-
135
- if(!wcVids.length){
136
- el.innerHTML='<div class="loading">Chưa có highlight World Cup<br><small style="color:#6a9fca">Highlight sẽ xuất hiện khi giải diễn ra</small></div>';
 
 
 
137
  return;
138
  }
139
 
140
  let h='<div style="display:grid;grid-template-columns:repeat(2,1fr);gap:6px;max-height:450px;overflow-y:auto;padding:4px">';
141
- wcVids.forEach((v,i)=>{
142
- h+=`<div style="cursor:pointer;border-radius:6px;overflow:hidden;background:#0a1520" onclick="openHighlightFeed('world-cup',${v._li},'${esc(v.link)}')">`;
 
143
  h+=`<div style="position:relative;aspect-ratio:16/9;background:#1a2a3a">${v.img?`<img src="${esc(v.img)}" loading="lazy" style="width:100%;height:100%;object-fit:cover" onerror="this.style.display='none'">`:''}<div class="card-play">▶</div></div>`;
144
- h+=`<div style="padding:4px 6px"><span style="font-size:8px;color:#f0c040">🌍 World Cup</span><div style="font-size:10px;color:#ccc;line-height:1.2;display:-webkit-box;-webkit-line-clamp:2;-webkit-box-orient:vertical;overflow:hidden">${esc(v.title)}</div></div>`;
145
  h+='</div>';
146
  });
147
  h+='</div>';el.innerHTML=h;
@@ -150,159 +141,6 @@ async function renderWCHighlights(el){
150
  }
151
  }
152
 
153
- // === WC SHORTS SLIDE — shown inside WC section ===
154
- function renderWCShortsSlide(container){
155
- if(!_wcShortsData || !_wcShortsData.length) return;
156
- const wrap = document.createElement('div');
157
- wrap.className = 'slider-wrap wc-shorts-slider';
158
- let h = '<div class="slider-header"><span class="slider-label">📱 Shorts World Cup</span><span style="font-size:10px;color:#8ab4d8;cursor:pointer" onclick="openWCShortsFeed(0)">Xem tất cả →</span></div>';
159
- h += '<div class="wc-shorts-track">';
160
- _wcShortsData.slice(0,10).forEach((s,i)=>{
161
- const badge = s.channel==='vtvnambo' ? 'VTV Nam Bộ' : '⚽ WC';
162
- const badgeColor = s.channel==='vtvnambo' ? '#0b6bcb' : '#f26522';
163
- h += `<div class="wc-shorts-item" onclick="openWCShortsFeed(${i})">`;
164
- h += `<div class="wc-shorts-thumb">${s.img?`<img src="${esc(s.img)}" loading="lazy" onerror="this.style.display='none'">`:''}<div class="card-play">▶</div></div>`;
165
- h += `<div class="wc-shorts-info"><span class="wc-shorts-badge" style="background:${badgeColor};color:#fff">${badge}</span><div class="wc-shorts-title">${esc(s.title)}</div></div>`;
166
- h += '</div>';
167
- });
168
- // "View all" card
169
- h += `<div class="wc-shorts-viewall" onclick="openWCShortsFeed(0)">📱 Xem tất cả<br><small style="font-size:9px;color:#6a9fca">${_wcShortsData.length} shorts</small></div>`;
170
- h += '</div>';
171
- wrap.innerHTML = h;
172
- container.appendChild(wrap);
173
- }
174
-
175
- // === WC SHORTS TIKTOK FEED — full-screen vertical scroll, continuous autoplay ===
176
- async function openWCShortsFeed(startIdx){
177
- showView('view-tiktok');
178
- const el = document.getElementById('view-tiktok');
179
- el.innerHTML = '<div class="loading">Đang tải shorts...</div>';
180
-
181
- // Fetch shorts if not loaded
182
- if(!_wcShortsData || !_wcShortsData.length){
183
- try{
184
- const r = await fetch('/api/shorts?channel=vtvnambo&count=30');
185
- _wcShortsData = await r.json() || [];
186
- }catch(e){
187
- _wcShortsData = [];
188
- }
189
- }
190
-
191
- if(!_wcShortsData.length){
192
- el.innerHTML = '<div class="loading">Chưa có shorts</div>';
193
- return;
194
- }
195
-
196
- // Rotate so startIdx is first
197
- const ordered = startIdx > 0 ? [..._wcShortsData.slice(startIdx), ..._wcShortsData.slice(0,startIdx)] : _wcShortsData;
198
-
199
- let h = `<button class="back-btn" onclick="switchCat('home')">← World Cup Shorts</button>`;
200
- h += '<div class="tiktok-container"><div class="tiktok-feed" id="wc-shorts-feed">';
201
-
202
- ordered.forEach((s,i)=>{
203
- const id = s.id || '';
204
- const src = `https://www.youtube.com/embed/${id}?autoplay=0&rel=0&playsinline=1&mute=1`;
205
- const vtag = `<iframe data-yt-src="${src}" allowfullscreen allow="accelerometer;autoplay;clipboard-write;encrypted-media;gyroscope;picture-in-picture"></iframe>`;
206
- const badge = s.channel==='vtvnambo' ? 'VTV Nam Bộ' : '⚽ WC';
207
- const badgeClass = s.channel==='vtvnambo' ? 'badge-wc' : 'badge-fpt';
208
- const videoId = 'wcshort-' + id;
209
- h += buildTikTokSlide({vtag,title:s.title,badge,badgeClass,videoId,idx:i,total:ordered.length,shareUrl:'https://youtube.com/shorts/'+id});
210
- });
211
-
212
- h += '</div></div>';
213
- el.innerHTML = h;
214
-
215
- // Init continuous autoplay feed
216
- initWCShortsFeed();
217
- }
218
-
219
- function initWCShortsFeed(){
220
- const feed = document.getElementById('wc-shorts-feed');
221
- if(!feed) return;
222
- const slides = feed.querySelectorAll('.tiktok-slide');
223
- let cur = -1;
224
- let isManualScroll = false;
225
- let manualTimer = null;
226
-
227
- function activate(i){
228
- if(i === cur) return;
229
- slides.forEach((sl,idx)=>{
230
- const fr = sl.querySelector('iframe');
231
- if(idx === i){
232
- // Load and play
233
- if(fr && !fr.src && fr.dataset.ytSrc) fr.src = fr.dataset.ytSrc;
234
- // For continuous feel: set autoplay=1 on activate
235
- if(fr && fr.src && !fr.src.includes('autoplay=1')){
236
- fr.src = fr.src.replace('autoplay=0','autoplay=1');
237
- }
238
- const vid = sl.dataset.vid;
239
- if(vid && !sl._viewed){ sl._viewed = true; doInteract(vid,'view'); }
240
- } else {
241
- // Pause: reload with autoplay=0 to stop
242
- if(fr && fr.src){
243
- fr.src = fr.src.replace('autoplay=1','autoplay=0');
244
- }
245
- }
246
- });
247
- cur = i;
248
- }
249
-
250
- // Scroll-based activation with snap
251
- let scrollTimer;
252
- feed.addEventListener('scroll', ()=>{
253
- isManualScroll = true;
254
- clearTimeout(manualTimer);
255
- manualTimer = setTimeout(()=>{ isManualScroll = false; }, 2000);
256
- clearTimeout(scrollTimer);
257
- scrollTimer = setTimeout(()=>{
258
- const rect = feed.getBoundingClientRect();
259
- const ctr = rect.top + rect.height / 2;
260
- let best = -1, bestD = 1e9;
261
- slides.forEach((sl,i)=>{
262
- const d = Math.abs(sl.getBoundingClientRect().top + sl.getBoundingClientRect().height/2 - ctr);
263
- if(d < bestD){ bestD = d; best = i; }
264
- });
265
- if(best >= 0) activate(best);
266
- }, 100);
267
- });
268
-
269
- // Auto-advance every 8 seconds for continuous playback
270
- let autoTimer = setInterval(()=>{
271
- if(isManualScroll) return;
272
- if(cur < slides.length - 1){
273
- const nextSl = slides[cur + 1];
274
- if(nextSl){
275
- nextSl.scrollIntoView({behavior:'smooth', block:'start'});
276
- // activate will be called by scroll event
277
- }
278
- } else {
279
- // Loop back to first
280
- slides[0].scrollIntoView({behavior:'smooth', block:'start'});
281
- }
282
- }, 8000);
283
-
284
- // Store timer for cleanup
285
- feed._autoTimer = autoTimer;
286
-
287
- // Activate first after short delay
288
- setTimeout(() => {
289
- if(slides.length > 0){
290
- slides[0].scrollIntoView({behavior:'start'});
291
- activate(0);
292
- }
293
- }, 300);
294
-
295
- // Cleanup on leave
296
- const origSwitchCat = window.switchCat;
297
- if(origSwitchCat && !origSwitchCat.__wcCleaned){
298
- window.switchCat = function(id){
299
- if(feed._autoTimer) clearInterval(feed._autoTimer);
300
- return origSwitchCat.apply(this, arguments);
301
- };
302
- window.switchCat.__wcCleaned = true;
303
- }
304
- }
305
-
306
  // === LIVE AUTO-REFRESH (90s) ===
307
  function startWCLiveRefresh(){
308
  if(_wcRefreshInterval)clearInterval(_wcRefreshInterval);
@@ -314,4 +152,4 @@ function startWCLiveRefresh(){
314
  }catch(e){}
315
  },90000);
316
  }
317
- setTimeout(startWCLiveRefresh,5000);
 
1
+ // === WORLD CUP 2026 - FIXED: proxy-based highlights + reliable data ===
2
  let _wc2026Data=null;let _wcRefreshInterval=null;
 
3
  const WC_TEAMS={'Mexico':'🇲🇽 Mexico','Colombia':'🇨🇴 Colombia','USA':'🇺🇸 Mỹ','United States':'🇺🇸 Mỹ','Brazil':'🇧🇷 Brazil','Argentina':'🇦🇷 Argentina','Germany':'🇩🇪 Đức','France':'🇫🇷 Pháp','Spain':'🇪🇸 Tây Ban Nha','England':'🏴󠁧󠁢󠁥󠁮󠁧󠁿 Anh','Portugal':'🇵🇹 Bồ Đào Nha','Netherlands':'🇳🇱 Hà Lan','Italy':'🇮🇹 Ý','Belgium':'🇧🇪 Bỉ','Croatia':'🇭🇷 Croatia','Uruguay':'🇺🇾 Uruguay','Japan':'🇯🇵 Nhật Bản','South Korea':'🇰🇷 Hàn Quốc','Korea Republic':'🇰🇷 Hàn Quốc','Australia':'🇦🇺 Úc','Saudi Arabia':'🇸🇦 Ả Rập Xê Út','Iran':'🇮🇷 Iran','Qatar':'🇶🇦 Qatar','Canada':'🇨🇦 Canada','Morocco':'🇲🇦 Morocco','Senegal':'🇸🇳 Senegal','Ecuador':'🇪🇨 Ecuador','Serbia':'🇷🇸 Serbia','Switzerland':'🇨🇭 Thụy Sĩ','Denmark':'🇩🇰 Đan Mạch','Poland':'🇵🇱 Ba Lan','Turkey':'🇹🇷 Thổ Nhĩ Kỳ','Türkiye':'🇹🇷 Thổ Nhĩ Kỳ','Ukraine':'🇺🇦 Ukraine','Egypt':'🇪🇬 Ai Cập','Nigeria':'🇳🇬 Nigeria','Cameroon':'🇨🇲 Cameroon','Ghana':'🇬🇭 Ghana','Tunisia':'🇹🇳 Tunisia','Algeria':'🇩🇿 Algeria','South Africa':'🇿🇦 Nam Phi','Indonesia':'🇮🇩 Indonesia','Vietnam':'🇻🇳 Việt Nam','China PR':'🇨🇳 Trung Quốc','New Zealand':'🇳🇿 New Zealand','Panama':'🇵🇦 Panama','Costa Rica':'🇨🇷 Costa Rica','Jamaica':'🇯🇲 Jamaica','Honduras':'🇭🇳 Honduras','Paraguay':'🇵🇾 Paraguay','Bolivia':'🇧🇴 Bolivia','Chile':'🇨🇱 Chile','Peru':'🇵🇪 Peru','Venezuela':'🇻🇪 Venezuela','Norway':'🇳🇴 Na Uy','Sweden':'🇸🇪 Thụy Điển','Austria':'🇦🇹 Áo','Scotland':'🏴󠁧󠁢󠁳󠁣󠁴󠁿 Scotland','Wales':'🏴󠁧󠁢󠁷󠁬󠁳󠁿 Xứ Wales','Ireland':'🇮🇪 Ireland','Romania':'🇷🇴 Romania','Hungary':'🇭🇺 Hungary','Greece':'🇬🇷 Hy Lạp','Uzbekistan':'🇺🇿 Uzbekistan','Iraq':'🇮🇶 Iraq','TBD':'🏳️ TBD'};
4
  function _t(n){return WC_TEAMS[n]||WC_TEAMS[(n||'').trim()]||'🏴 '+n;}
5
 
 
28
  .wc-bxh table{width:100%!important;border-collapse:collapse!important;font-size:11px!important;display:table!important}
29
  .wc-bxh th{background:#0b1a2a!important;color:#6a9fca!important;padding:5px 3px!important;font-size:9px!important;text-align:center!important}
30
  .wc-bxh td{padding:4px 3px!important;border-bottom:1px solid #0d1a2a!important;font-size:10px!important;color:#ddd!important}
 
 
 
 
 
 
 
 
 
 
 
 
 
31
  </style>`;
32
 
33
  function switchWCTab(tab){
 
54
  h+=`<div class="wc-news-item" onclick="readArticle('${esc(a.link)}')"><div class="wc-news-img" id="wcimg-${i}">${a.img?`<img src="${esc(a.img)}" loading="lazy" onerror="this.parentElement.innerHTML=''">`:''}</div><div class="wc-news-text">${badge}<div class="wc-news-title">${esc(a.title)}</div><div class="wc-news-via">${esc(a.source||'')}</div></div></div>`;
55
  });
56
  h+='</div>';el.innerHTML=h;
57
+ // Fetch missing images
58
  combined.slice(0,20).forEach((a,i)=>{if(a.img)return;fetch('/api/article?url='+encodeURIComponent(a.link)).then(r=>r.json()).then(d=>{if(d&&(d.og_image||d.img)){const x=document.getElementById('wcimg-'+i);if(x)x.innerHTML=`<img src="${esc(d.og_image||d.img)}" loading="lazy" onerror="this.parentElement.innerHTML=''">`;}}).catch(()=>{});});
59
  }
60
 
 
108
  else el.innerHTML='<div class="loading">Thống kê cập nhật khi giải bắt đầu<br><small style="color:#6a9fca">Vua phá lưới · Kiến tạo · Thẻ phạt</small></div>';
109
  }
110
 
111
+ // === HIGHLIGHTS - Uses backend proxy to avoid CORS ===
112
  async function renderWCHighlights(el){
113
  el.innerHTML='<div class="loading">Đang tải highlight World Cup...</div>';
114
  try{
115
+ // Use backend proxy endpoint (avoids CORS)
116
+ const [wcRes, frRes] = await Promise.all([
117
+ fetch('/api/proxy/xlb?path=the-gioi/world-cup', {signal: AbortSignal.timeout(15000)}).then(r=>r.json()).catch(()=>({videos:[]})),
118
+ fetch('/api/proxy/xlb?path=giai-khac/friendly', {signal: AbortSignal.timeout(15000)}).then(r=>r.json()).catch(()=>({videos:[]}))
119
+ ]);
120
+
121
+ const wcVids = (wcRes.videos||[]).map((v,i)=>({...v,_league:'WC',_li:i}));
122
+ const frVids = (frRes.videos||[]).map((v,i)=>({...v,_league:'Friendly',_li:i}));
123
+ const all = [...wcVids, ...frVids];
124
+
125
+ if(!all.length){
126
+ el.innerHTML='<div class="loading">Chưa có highlight World Cup / Giao hữu</div>';
127
  return;
128
  }
129
 
130
  let h='<div style="display:grid;grid-template-columns:repeat(2,1fr);gap:6px;max-height:450px;overflow-y:auto;padding:4px">';
131
+ all.forEach((v,i)=>{
132
+ const badge=v._league==='WC'?'🌍 WC':'🤝 Friendly';
133
+ h+=`<div style="cursor:pointer;border-radius:6px;overflow:hidden;background:#0a1520" onclick="openHighlightFeed('${v._league==='WC'?'world-cup':'friendly'}',${v._li},'${esc(v.link)}')">`;
134
  h+=`<div style="position:relative;aspect-ratio:16/9;background:#1a2a3a">${v.img?`<img src="${esc(v.img)}" loading="lazy" style="width:100%;height:100%;object-fit:cover" onerror="this.style.display='none'">`:''}<div class="card-play">▶</div></div>`;
135
+ h+=`<div style="padding:4px 6px"><span style="font-size:8px;color:#f0c040">${badge}</span><div style="font-size:10px;color:#ccc;line-height:1.2;display:-webkit-box;-webkit-line-clamp:2;-webkit-box-orient:vertical;overflow:hidden">${esc(v.title)}</div></div>`;
136
  h+='</div>';
137
  });
138
  h+='</div>';el.innerHTML=h;
 
141
  }
142
  }
143
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
144
  // === LIVE AUTO-REFRESH (90s) ===
145
  function startWCLiveRefresh(){
146
  if(_wcRefreshInterval)clearInterval(_wcRefreshInterval);
 
152
  }catch(e){}
153
  },90000);
154
  }
155
+ setTimeout(startWCLiveRefresh,5000);
static/wc_shorts_inject.js DELETED
@@ -1,76 +0,0 @@
1
- /**
2
- * WC Shorts Inject — Chạy SAU app_v2.js
3
- * Tự tạo anchor DIV rồi render WC Shorts slide TRÊN World Cup section
4
- */
5
- (function(){
6
- 'use strict';
7
-
8
- function inject(){
9
- // Tìm WC section
10
- const wcSection = document.getElementById('wc2026-live-section');
11
- const homeEl = document.getElementById('view-home');
12
- if(!homeEl) return;
13
-
14
- // Đã inject?
15
- if(document.getElementById('wc-shorts-slide-section')) return;
16
-
17
- // Đảm bảo renderWCSSlideSection đã load
18
- if(typeof renderWCSSlideSection !== 'function'){
19
- // Chờ rồi thử lại
20
- setTimeout(inject, 500);
21
- return;
22
- }
23
-
24
- // Tạo container rồi render vào
25
- const div = document.createElement('div');
26
- div.id = 'wc-shorts-slide-container-tmp';
27
- div.style.display = 'none';
28
-
29
- // Insert TRƯỚC WC section, hoặc vào đầu home
30
- if(wcSection && wcSection.parentNode){
31
- wcSection.parentNode.insertBefore(div, wcSection);
32
- } else {
33
- homeEl.insertBefore(div, homeEl.firstChild);
34
- }
35
-
36
- // Render slide section vào div
37
- renderWCSSlideSection(div);
38
-
39
- // Bỏ wrapper thừa, di chuyển section ra đúng vị trí
40
- const section = div.querySelector('.wc-shorts-slide-section');
41
- if(section){
42
- section.id = 'wc-shorts-slide-section';
43
- if(wcSection && wcSection.parentNode){
44
- wcSection.parentNode.insertBefore(section, wcSection);
45
- }
46
- div.remove();
47
- }
48
- }
49
-
50
- // Chạy khi DOM sẵn sàng
51
- if(document.readyState === 'loading'){
52
- document.addEventListener('DOMContentLoaded', ()=> setTimeout(inject, 800));
53
- } else {
54
- setTimeout(inject, 800);
55
- }
56
-
57
- // Cũng chạy sau khi loadHome xong
58
- const origLoadHome = window.loadHome;
59
- if(origLoadHome){
60
- window.loadHome = function(){
61
- const result = origLoadHome.apply(this, arguments);
62
- setTimeout(inject, 200);
63
- setTimeout(inject, 800);
64
- setTimeout(inject, 2000);
65
- return result;
66
- };
67
- }
68
-
69
- // Fallback: retry nhiều lần
70
- let retries = 0;
71
- const retryIv = setInterval(()=>{
72
- retries++;
73
- inject();
74
- if(retries > 20) clearInterval(retryIv);
75
- }, 1000);
76
- })();
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
static/yt_live.js CHANGED
@@ -1,62 +1,50 @@
1
- // === VNEWS — VTV LIVE + Inline Recorder v9 ===
2
- // Features: PiP, mini-player, INLINE RECORDER, VTVGO POPUP players for each channel
3
- // FIX v10: Correct VTVGO popup URLs for all channels + World Cup 2026 TS stream links
4
- // FIX v11: Allow clicking off tabs (always clickable), better stream status handling
5
 
6
  (function(){
7
  if(window._ytLiveLoaded) return;
8
  window._ytLiveLoaded = true;
9
 
10
  const CHANNELS = [
11
- {id:'vtv1',name:'VTV1',badge:'Tin tức'},{id:'vtv2',name:'VTV2',badge:'Khoa học'},
12
- {id:'vtv3',name:'VTV3',badge:'Giải trí'},{id:'vtv4',name:'VTV4',badge:'Quốc tế'},
13
- {id:'vtv5',name:'VTV5',badge:'Miền Nam'},{id:'vtv6',name:'VTV6',badge:'Thanh niên'},
14
- {id:'vtv7',name:'VTV7',badge:'Giáo dục'},{id:'vtv8',name:'VTV8',badge:'Miền Trung'},
15
- {id:'vtv9',name:'VTV9',badge:'Miền Bắc'},{id:'vtv10',name:'VTV10',badge:'Cần Thơ'},
16
- {id:'vtvprime',name:'VTVPrime',badge:'Prime'},
 
 
 
 
 
17
  ];
18
-
19
- // VTVGO popup URLs for each channel (FIXED v10 - verified correct URLs)
20
- const VTV_POPUP_URLS = {
21
- vtv1: 'https://package.vtvgo.vn/channel/vtv1-1,1.html',
22
- vtv2: 'https://package.vtvgo.vn/channel/vtv2-1,2.html',
23
- vtv3: 'https://package.vtvgo.vn/channel/vtv3-1,3.html',
24
- vtv4: 'https://package.vtvgo.vn/channel/vtv4-1,4.html',
25
- vtv5: 'https://package.vtvgo.vn/channel/vtv5-1,5.html',
26
- vtv6: 'https://package.vtvgo.vn/channel/vtv6-1,13.html',
27
- vtv7: 'https://package.vtvgo.vn/channel/vtv7-1,27.html',
28
- vtv8: 'https://package.vtvgo.vn/channel/vtv8-1,36.html',
29
- vtv9: 'https://package.vtvgo.vn/channel/vtv9-1,39.html',
30
- vtv10: 'https://package.vtvgo.vn/channel/vtv10-1,6.html',
31
- };
32
-
33
- const DEFAULT_CHANNEL = 'vtv3';
34
- const NEEDS_PROXY = /fptplay\.net|vtvgo\.vn/;
35
  const STREAMS = {};
36
- let _currentCh = null, _hls = null, _loading = false, _blockInserted = false;
37
- let _streamsLoaded = false, _pipActive = false, _miniActive = false, _vtvPinned = false, _vtvPopupActive = false;
38
-
39
- // ===== RECORDER STATE =====
40
- const _rec = {
41
- active: false, startTime: null, endTime: null,
42
- isRecording: false, recorder: null, chunks: [], blob: null,
43
- ratio: 'original', _dragMarker: null, _recTimer: null,
44
- };
45
 
46
  // ===== STYLES =====
47
  const style = document.createElement('style');
48
  style.textContent = `
49
- .vtv-wrap{position:relative;margin:6px 4px;background:#111;border:1px solid #0066cc;border-radius:10px;overflow:hidden}
50
  .vtv-head{display:flex;align-items:center;gap:8px;padding:8px 10px;background:linear-gradient(90deg,#003366,#1a1a1a)}
51
  .vtv-title{font-size:13px;font-weight:800;color:#00ccff}
52
  .vtv-badge{font-size:10px;font-weight:800;color:#00ccff;animation:vtvp 1.3s infinite}
53
  @keyframes vtvp{0%,100%{opacity:1}50%{opacity:.3}}
54
- .vtv-tabs{display:flex;flex-wrap:wrap;gap:3px;padding:6px 8px;overflow-x:auto;scrollbar-width:none;background:#0d1a2a}
55
  .vtv-tabs::-webkit-scrollbar{display:none}
56
  .vtv-tab{padding:4px 8px;background:#1a2a3a;border:1px solid #2a3a4a;border-radius:10px;color:#8ab4d8;font-size:9px;cursor:pointer;white-space:nowrap;flex-shrink:0;transition:all .2s}
57
  .vtv-tab:hover{background:#0b4a7a;color:#fff}
58
  .vtv-tab.on{background:#0066cc;border-color:#00ccff;color:#fff;font-weight:700}
59
- .vtv-tab.off{opacity:.35}
60
  .vtv-frame{position:relative;width:100%;aspect-ratio:16/9;background:#000;min-height:180px}
61
  .vtv-frame video{position:absolute;inset:0;width:100%;height:100%;object-fit:contain}
62
  .vtv-err{display:flex;align-items:center;justify-content:center;height:180px;color:#888;font-size:12px;text-align:center;padding:20px;flex-direction:column;gap:8px}
@@ -85,65 +73,8 @@
85
  .vtv-epg-empty{color:#666;font-size:9px;padding:4px}
86
  .vtv-epg-loading{color:#00ccff;font-size:9px;padding:4px;display:flex;align-items:center;gap:6px}
87
  .vtv-epg-sp{width:10px;height:10px;border:1px solid #333;border-top-color:#00ccff;border-radius:50%;animation:vtvspin .8s linear infinite}
88
- .vtv-pin-btn{position:absolute;top:6px;right:8px;z-index:5;background:rgba(0,0,0,.5);border:1px solid #2a3a4a;color:#8ab4d8;font-size:9px;padding:2px 6px;border-radius:4px;cursor:pointer;font-weight:700;display:flex;align-items:center;gap:3px}
89
- .vtv-pin-btn:hover{background:#0b4a7a;color:#fff}
90
- .vtv-pin-btn.pinned{background:#0066cc;border-color:#00ccff;color:#fff}
91
- .vtv-wrap.vtv-sticky{position:sticky;top:0;z-index:48;transition:all .25s ease;box-shadow:0 4px 24px rgba(0,102,204,.35)}
92
- .vtv-wrap.vtv-sticky .vtv-epg{display:none}
93
- .vtv-wrap.vtv-sticky .vtv-controls{display:none}
94
- .vtv-wrap.vtv-sticky .vtv-tabs{padding:3px 8px}
95
- .vtv-wrap.vtv-sticky .vtv-tab{padding:3px 7px;font-size:8px}
96
- .vtv-wrap.vtv-sticky .vtv-frame{max-height:140px;min-height:100px}
97
- .vtv-wrap.vtv-sticky .vtv-frame video{max-height:140px}
98
- .vtv-wrap.vtv-sticky .vtv-load{height:100px}
99
- .vtv-wrap.vtv-sticky .vtv-err{height:100px}
100
- .vtv-wrap.vtv-sticky .vtv-head{padding:5px 10px}
101
- .vtv-wrap.vtv-sticky .vtv-title{font-size:11px}
102
- .vtv-wrap.vtv-sticky .vtv-badge{font-size:8px}
103
- .vtv-wrap.vtv-sticky .vtv-pin-btn{top:4px;right:6px}
104
- .vtv-rec-btn{background:#660000;border:1px solid #990000;color:#ff6666;font-size:9px;padding:3px 8px;border-radius:6px;cursor:pointer;transition:all .2s;display:flex;align-items:center;gap:4px;font-weight:700}
105
- .vtv-rec-btn:hover{background:#990000;color:#fff}
106
- .vtv-rec-btn.recording{background:#cc0000;border-color:#ff0000;color:#fff;animation:vtv-rec-pulse 1s infinite}
107
- @keyframes vtv-rec-pulse{0%,100%{opacity:1}50%{opacity:.5}}
108
- .vtv-rec-btn svg{width:12px;height:12px;fill:currentColor}
109
- .vtv-inline-rec{display:none;padding:8px 10px;background:#0a0a1a;border-top:1px solid #1a1a3a}
110
- .vtv-inline-rec.show{display:block}
111
- .vtv-inline-rec .rec-bar{position:relative;height:32px;background:#000;border-radius:4px;overflow:hidden;cursor:pointer;border:1px solid #2a2a4a;margin-bottom:6px;user-select:none}
112
- .vtv-inline-rec .rec-bar .rec-progress{position:absolute;top:0;bottom:0;background:rgba(155,89,182,.3);left:0;width:0%;pointer-events:none}
113
- .vtv-inline-rec .rec-bar .rec-marker{position:absolute;top:0;bottom:0;width:4px;z-index:2;border-radius:2px}
114
- .vtv-inline-rec .rec-bar .rec-marker.s{background:#2ecc71}
115
- .vtv-inline-rec .rec-bar .rec-marker.e{background:#e74c3c}
116
- .vtv-inline-rec .rec-time{display:flex;justify-content:space-between;font-size:9px;color:#888;margin-bottom:6px}
117
- .vtv-inline-rec .rec-controls{display:flex;gap:4px}
118
- .vtv-inline-rec .rec-controls button{flex:1;padding:6px;border:none;border-radius:6px;font-size:10px;font-weight:700;cursor:pointer}
119
- .vtv-inline-rec .rec-controls .rec-set-start{background:#1a3a1a;border:1px solid #2d6a2d;color:#5cb87a}
120
- .vtv-inline-rec .rec-controls .rec-set-end{background:#3a1a1a;border:1px solid #6a2d2d;color:#e74c3c}
121
- .vtv-inline-rec .rec-controls .rec-go{background:#1a0a3a;border:1px solid #3a2a6a;color:#9b59b6}
122
- .vtv-inline-rec .rec-controls .rec-reset{background:#222;border:1px solid #333;color:#888}
123
- .vtv-inline-rec .rec-hint{font-size:9px;color:#666;text-align:center;margin-top:4px}
124
- .vtv-inline-rec .rec-status{font-size:10px;color:#888;text-align:center;padding:4px;margin-top:4px;background:#111;border-radius:4px}
125
- .vtv-inline-rec .rec-status.ok{color:#2ecc71}
126
- .vtv-inline-rec .rec-status.err{color:#e74c3c}
127
- .vtv-inline-rec .rec-status.recording{color:#e74c3c;animation:vtv-rec-pulse 1s infinite}
128
- .vtv-rec-panel{display:none;padding:10px;background:#0d0d20;border-top:1px solid #2a1a4a}
129
- .vtv-rec-panel.show{display:block}
130
- .vtv-rec-panel .rec-panel-title{font-size:11px;font-weight:700;color:#9b59b6;margin-bottom:8px}
131
- .vtv-rec-panel .rec-preview-wrap{margin-bottom:8px;text-align:center}
132
- .vtv-rec-panel .rec-preview-wrap video{max-width:100%;max-height:180px;border-radius:6px;background:#000}
133
- .vtv-rec-panel .rec-ratio-row{display:flex;gap:4px;margin-bottom:8px}
134
- .vtv-rec-panel .rec-ratio-row button{flex:1;padding:6px;background:#1a1a2e;border:1px solid #2a2a4a;border-radius:6px;color:#888;font-size:10px;cursor:pointer}
135
- .vtv-rec-panel .rec-ratio-row button.active{border-color:#9b59b6;color:#9b59b6;background:#2a1a4a}
136
- .vtv-rec-panel .rec-actions{display:flex;gap:4px}
137
- .vtv-rec-panel .rec-actions button{flex:1;padding:8px;border:none;border-radius:6px;font-size:11px;font-weight:700;cursor:pointer}
138
- .vtv-rec-panel .rec-actions .rec-download{background:#2d8659;color:#fff}
139
- .vtv-rec-panel .rec-actions .rec-share{background:#9b59b6;color:#fff}
140
- .vtv-rec-panel .rec-proc{text-align:center;padding:12px;color:#9b59b6;font-size:12px;display:none}
141
- .vtv-rec-panel .rec-title-input{width:100%;background:#1a1a2e;border:1px solid #2a2a4a;border-radius:6px;padding:8px;color:#ccc;font-size:11px;margin-bottom:8px;box-sizing:border-box}
142
- .vtv-rec-panel .rec-title-input:focus{border-color:#9b59b6;outline:none}
143
- .vtv-rec-panel .rec-ai-title-row{display:flex;gap:4px;margin-bottom:8px}
144
- .vtv-rec-panel .rec-ai-title-row button{flex:1;padding:6px;background:#1a1a2e;border:1px solid #2a2a4a;border-radius:6px;color:#888;font-size:10px;cursor:pointer}
145
- .vtv-rec-panel .rec-ai-title-row button:hover{border-color:#9b59b6;color:#9b59b6}
146
- .vtv-rec-panel .rec-ai-title-row button:disabled{opacity:.4;cursor:not-allowed}
147
  .vtv-mini{position:fixed;top:0;left:0;right:0;z-index:99990;background:#000;border-bottom:2px solid #0066cc;box-shadow:0 4px 20px rgba(0,102,204,.4);display:none;transition:transform .3s ease}
148
  .vtv-mini.show{display:block}
149
  .vtv-mini.hidden{transform:translateY(-88%)}
@@ -159,17 +90,335 @@
159
  .vtv-mini-btn.x:hover{background:#900;color:#fff}
160
  .vtv-mini-peek{position:absolute;bottom:-18px;right:10px;background:#0066cc;color:#fff;font-size:9px;padding:2px 8px;border-radius:0 0 6px 6px;cursor:pointer;display:none}
161
  .vtv-mini.hidden .vtv-mini-peek{display:block}
162
- .vtv-popup{position:fixed;top:0;left:0;right:0;margin:8px auto 0;background:rgba(0,0,0,.95);z-index:99999;display:none;flex-direction:column;max-height:calc(100vh - 16px);max-width:800px;border-radius:12px;overflow:hidden}
163
- .vtv-popup.show{display:flex}
164
- .vtv-popup:has(#vtv-popup-hdr[style*="none"]) .vtv-popup-frame{height:100%}
165
- .vtv-popup-header{display:flex;align-items:center;justify-content:space-between;padding:8px 12px;background:#111;border-bottom:1px solid #333}
166
- .vtv-popup-title{font-size:14px;font-weight:700;color:#00ccff}
167
- .vtv-popup-close{background:#333;border:none;color:#fff;font-size:18px;width:32px;height:32px;border-radius:6px;cursor:pointer}
168
- .vtv-popup-frame{flex:1;position:relative;background:#000;min-height:300px}
169
- .vtv-popup-frame iframe{position:absolute;inset:0;width:100%;height:100%;border:none}
170
- .vtv-popup-frame iframe{position:absolute;inset:0;width:100%;height:100%;border:none}
171
- .vtv-popup-loader{display:flex;align-items:center;justify-content:center;height:280px;color:#00ccff;font-size:12px;flex-direction:column;gap:8px}
172
- .vtv-popup-btn{background:#1a2a3a;border:1px solid #2a3a4a;color:#8ab4d8;font-size:9px;padding:3px 8px;border-radius:6px;cursor:pointer;margin-left:4px}
173
- .vtv-popup-btn:hover{background:#0b4a7a;color:#fff}
174
  `;
175
- document.head.appendChild(style);
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ // === VNEWS — VTV1-VTV10 + VTVPrime LIVE CHANNELS ===
2
+ // Default channel: VTV6
3
+ // Features: PiP button, floating mini-player when reading articles/videos
 
4
 
5
  (function(){
6
  if(window._ytLiveLoaded) return;
7
  window._ytLiveLoaded = true;
8
 
9
  const CHANNELS = [
10
+ {id:'vtv1', name:'VTV1', badge:'Tin tức'},
11
+ {id:'vtv2', name:'VTV2', badge:'Khoa học'},
12
+ {id:'vtv3', name:'VTV3', badge:'Giải trí'},
13
+ {id:'vtv4', name:'VTV4', badge:'Quốc tế'},
14
+ {id:'vtv5', name:'VTV5', badge:'Miền Nam'},
15
+ {id:'vtv6', name:'VTV6', badge:'Thanh niên'},
16
+ {id:'vtv7', name:'VTV7', badge:'Giáo dục'},
17
+ {id:'vtv8', name:'VTV8', badge:'Miền Trung'},
18
+ {id:'vtv9', name:'VTV9', badge:'Miền Bắc'},
19
+ {id:'vtv10', name:'VTV10', badge:'VTV Cần Thơ'},
20
+ {id:'vtvprime', name:'VTVPrime', badge:'Prime'},
21
  ];
22
+ const DEFAULT_CHANNEL = 'vtv6';
23
+ const NEEDS_PROXY = /fptplay\.net/;
24
+
 
 
 
 
 
 
 
 
 
 
 
 
 
 
25
  const STREAMS = {};
26
+ let _currentCh = null;
27
+ let _hls = null;
28
+ let _loading = false;
29
+ let _blockInserted = false;
30
+ let _streamsLoaded = false;
31
+ let _pipActive = false;
32
+ let _miniActive = false;
 
 
33
 
34
  // ===== STYLES =====
35
  const style = document.createElement('style');
36
  style.textContent = `
37
+ .vtv-wrap{margin:6px 4px;background:#111;border:1px solid #0066cc;border-radius:10px;overflow:hidden}
38
  .vtv-head{display:flex;align-items:center;gap:8px;padding:8px 10px;background:linear-gradient(90deg,#003366,#1a1a1a)}
39
  .vtv-title{font-size:13px;font-weight:800;color:#00ccff}
40
  .vtv-badge{font-size:10px;font-weight:800;color:#00ccff;animation:vtvp 1.3s infinite}
41
  @keyframes vtvp{0%,100%{opacity:1}50%{opacity:.3}}
42
+ .vtv-tabs{display:flex;gap:3px;padding:6px 8px;overflow-x:auto;scrollbar-width:none;background:#0d1a2a}
43
  .vtv-tabs::-webkit-scrollbar{display:none}
44
  .vtv-tab{padding:4px 8px;background:#1a2a3a;border:1px solid #2a3a4a;border-radius:10px;color:#8ab4d8;font-size:9px;cursor:pointer;white-space:nowrap;flex-shrink:0;transition:all .2s}
45
  .vtv-tab:hover{background:#0b4a7a;color:#fff}
46
  .vtv-tab.on{background:#0066cc;border-color:#00ccff;color:#fff;font-weight:700}
47
+ .vtv-tab.off{opacity:.35;pointer-events:none}
48
  .vtv-frame{position:relative;width:100%;aspect-ratio:16/9;background:#000;min-height:180px}
49
  .vtv-frame video{position:absolute;inset:0;width:100%;height:100%;object-fit:contain}
50
  .vtv-err{display:flex;align-items:center;justify-content:center;height:180px;color:#888;font-size:12px;text-align:center;padding:20px;flex-direction:column;gap:8px}
 
73
  .vtv-epg-empty{color:#666;font-size:9px;padding:4px}
74
  .vtv-epg-loading{color:#00ccff;font-size:9px;padding:4px;display:flex;align-items:center;gap:6px}
75
  .vtv-epg-sp{width:10px;height:10px;border:1px solid #333;border-top-color:#00ccff;border-radius:50%;animation:vtvspin .8s linear infinite}
76
+
77
+ /* ===== MINI PLAYER ===== */
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
78
  .vtv-mini{position:fixed;top:0;left:0;right:0;z-index:99990;background:#000;border-bottom:2px solid #0066cc;box-shadow:0 4px 20px rgba(0,102,204,.4);display:none;transition:transform .3s ease}
79
  .vtv-mini.show{display:block}
80
  .vtv-mini.hidden{transform:translateY(-88%)}
 
90
  .vtv-mini-btn.x:hover{background:#900;color:#fff}
91
  .vtv-mini-peek{position:absolute;bottom:-18px;right:10px;background:#0066cc;color:#fff;font-size:9px;padding:2px 8px;border-radius:0 0 6px 6px;cursor:pointer;display:none}
92
  .vtv-mini.hidden .vtv-mini-peek{display:block}
 
 
 
 
 
 
 
 
 
 
 
 
93
  `;
94
+ document.head.appendChild(style);
95
+
96
+ // ===== MINI PLAYER =====
97
+ function createMiniPlayer(){
98
+ if(document.getElementById('vtv-mini')) return;
99
+ const mini = document.createElement('div');
100
+ mini.className = 'vtv-mini';
101
+ mini.id = 'vtv-mini';
102
+ mini.innerHTML = `
103
+ <div class="vtv-mini-frame"><video id="vtv-mini-vid" playsinline muted preload="auto"></video></div>
104
+ <div class="vtv-mini-bar">
105
+ <div style="display:flex;align-items:center;flex:1;min-width:0">
106
+ <span class="vtv-mini-ch" id="vtv-mini-ch">VTV</span>
107
+ <span class="vtv-mini-epg" id="vtv-mini-epg"></span>
108
+ </div>
109
+ <div class="vtv-mini-btns">
110
+ <button class="vtv-mini-btn" id="vtv-mini-pip" title="PiP">📺</button>
111
+ <button class="vtv-mini-btn" id="vtv-mini-expand" title="Mở rộng">⬆️</button>
112
+ <button class="vtv-mini-btn" id="vtv-mini-hide" title="Thu nhỏ">⬇️</button>
113
+ <button class="vtv-mini-btn x" id="vtv-mini-close" title="Đóng">✕</button>
114
+ </div>
115
+ </div>
116
+ <div class="vtv-mini-peek" id="vtv-mini-peek">📺 Xem</div>`;
117
+ document.body.appendChild(mini);
118
+ document.getElementById('vtv-mini-pip').onclick = toggleMiniPiP;
119
+ document.getElementById('vtv-mini-expand').onclick = expandFromMini;
120
+ document.getElementById('vtv-mini-hide').onclick = ()=>{ mini.classList.add('hidden'); };
121
+ document.getElementById('vtv-mini-close').onclick = closeMiniPlayer;
122
+ document.getElementById('vtv-mini-peek').onclick = ()=>{ mini.classList.remove('hidden'); };
123
+ }
124
+
125
+ function closeMiniPlayer(){
126
+ const mini = document.getElementById('vtv-mini');
127
+ if(mini){
128
+ const v = document.getElementById('vtv-mini-vid');
129
+ if(v){ v.pause(); v.src = ''; }
130
+ mini.remove();
131
+ }
132
+ _miniActive = false;
133
+ }
134
+
135
+ function expandFromMini(){
136
+ // Close mini player and go home — user wants to see full player
137
+ closeMiniPlayer();
138
+ if(typeof switchCat === 'function') switchCat('home');
139
+ }
140
+
141
+ function activateMiniPlayer(){
142
+ if(!_currentCh) return;
143
+ createMiniPlayer();
144
+ _miniActive = true;
145
+ const mini = document.getElementById('vtv-mini');
146
+ const miniVid = document.getElementById('vtv-mini-vid');
147
+ const mainVid = document.getElementById('vtv-player');
148
+
149
+ if(_hls){
150
+ const hls = new Hls({enableWorker:true,lowLatencyMode:true,startLevel:-1,capLevelToPlayerSize:true,maxBufferLength:10});
151
+ hls.loadSource(STREAMS[_currentCh][0]);
152
+ hls.attachMedia(miniVid);
153
+ miniVid.play().catch(()=>{});
154
+ } else if(mainVid && mainVid.src){
155
+ miniVid.src = mainVid.src;
156
+ miniVid.play().catch(()=>{});
157
+ }
158
+
159
+ const ch = CHANNELS.find(c=>c.id===_currentCh);
160
+ if(ch) document.getElementById('vtv-mini-ch').textContent = ch.name;
161
+ updateMiniEpg();
162
+ mini.classList.add('show');
163
+ mini.classList.remove('hidden');
164
+ }
165
+
166
+ function updateMiniEpg(){
167
+ if(!_currentCh) return;
168
+ fetch('/api/vtv/epg/'+_currentCh).then(r=>r.json()).then(d=>{
169
+ const progs = d.programs||[];
170
+ const now = progs.find(p=>p.now);
171
+ document.getElementById('vtv-mini-epg').textContent = now ? now.time+' '+now.title : (progs[0] ? progs[0].time+' '+progs[0].title : '');
172
+ }).catch(()=>{});
173
+ }
174
+
175
+ async function toggleMiniPiP(){
176
+ const v = document.getElementById('vtv-mini-vid');
177
+ if(!v) return;
178
+ try{
179
+ if(document.pictureInPictureElement === v){
180
+ await document.exitPictureInPicture();
181
+ } else {
182
+ await v.requestPictureInPicture();
183
+ }
184
+ } catch(e){ alert('Không hỗ trợ PiP'); }
185
+ }
186
+
187
+ // ===== MAIN PLAYER =====
188
+ async function loadAllStreams(){
189
+ if(_loading) return;
190
+ _loading = true;
191
+ const el = document.getElementById('vtv-load');
192
+ if(el) el.innerHTML = '<div class="vtv-spinner"></div>Đang tải kênh...';
193
+ try{
194
+ const r = await fetch('/api/vtv/streams', {signal: AbortSignal.timeout(10000)});
195
+ if(r.ok){
196
+ const data = await r.json();
197
+ CHANNELS.forEach(ch=>{
198
+ const info = data[ch.id];
199
+ if(info && info.stream_url){
200
+ let url = info.stream_url;
201
+ if(NEEDS_PROXY.test(url)) url = '/api/proxy/m3u8/vtv?url='+encodeURIComponent(url);
202
+ STREAMS[ch.id] = [url];
203
+ } else {
204
+ STREAMS[ch.id] = [];
205
+ }
206
+ });
207
+ }
208
+ } catch(e){ console.warn('VTV API error:', e); }
209
+ CHANNELS.forEach(ch=>{
210
+ const tab = document.getElementById('vtvt-'+ch.id);
211
+ if(tab){
212
+ if(STREAMS[ch.id]&&STREAMS[ch.id].length>0){ tab.classList.remove('off'); tab.textContent=ch.name; }
213
+ else { tab.style.opacity='0.35'; tab.textContent=ch.name+' ✕'; }
214
+ }
215
+ });
216
+ _loading = false;
217
+ _streamsLoaded = true;
218
+ }
219
+
220
+ function buildBlock(){
221
+ const w = document.createElement('div');
222
+ w.className = 'vtv-wrap';
223
+ w.id = 'vtv-block';
224
+ let tabs = '';
225
+ CHANNELS.forEach(ch=>{
226
+ tabs += '<button class="vtv-tab off" id="vtvt-'+ch.id+'" onclick="window._vtvPlay(\''+ch.id+'\')">'+ch.name+'</button>';
227
+ });
228
+ w.innerHTML =
229
+ '<div class="vtv-head"><span class="vtv-title">📺 VTV Trực Tuyến</span><span class="vtv-badge">● LIVE</span></div>' +
230
+ '<div class="vtv-tabs">'+tabs+'</div>' +
231
+ '<div class="vtv-frame">' +
232
+ '<div class="vtv-load" id="vtv-load"><div class="vtv-spinner"></div>Đang tải kênh...</div>' +
233
+ '<video id="vtv-player" playsinline muted controls preload="auto" style="display:none"></video>' +
234
+ '<div class="vtv-err" id="vtv-err" style="display:none"><span id="vtv-err-msg">Không thể tải kênh</span><button onclick="window._vtvRetry()">Thử lại</button></div>' +
235
+ '</div>' +
236
+ '<div class="vtv-controls">' +
237
+ '<button class="vtv-pip-btn" id="vtv-pip-btn" onclick="window._vtvTogglePiP()">' +
238
+ '<svg viewBox="0 0 24 24"><rect x="2" y="4" width="20" height="16" rx="2" fill="none" stroke="currentColor" stroke-width="2"/><rect x="12" y="10" width="8" height="6" rx="1" fill="currentColor"/></svg>PiP' +
239
+ '</button>' +
240
+ '<span style="flex:1"></span>' +
241
+ '<span style="font-size:9px;color:#666" id="vtv-status"></span>' +
242
+ '</div>' +
243
+ '<div class="vtv-epg" id="vtv-epg">' +
244
+ '<div class="vtv-epg-header"><span class="vtv-epg-title">📋 Lịch phát sóng</span><button class="vtv-epg-toggle" id="vtv-epg-toggle" onclick="window._vtvToggleEpg()">Ẩn</button></div>' +
245
+ '<div class="vtv-epg-list" id="vtv-epg-list"><div class="vtv-epg-loading"><div class="vtv-epg-sp"></div>Đang tải...</div></div>' +
246
+ '</div>';
247
+ return w;
248
+ }
249
+
250
+ async function loadEpg(chId){
251
+ const el = document.getElementById('vtv-epg-list');
252
+ if(!el) return;
253
+ el.innerHTML = '<div class="vtv-epg-loading"><div class="vtv-epg-sp"></div>Đang tải lịch...</div>';
254
+ try{
255
+ const r = await fetch('/api/vtv/epg/'+chId, {signal: AbortSignal.timeout(10000)});
256
+ if(!r.ok) throw new Error('fail');
257
+ const data = await r.json();
258
+ const progs = data.programs||[];
259
+ if(!progs.length){ el.innerHTML='<div class="vtv-epg-empty">Chưa có lịch phát sóng</div>'; return; }
260
+ let h = '';
261
+ progs.forEach(p=>{
262
+ const nw = p.now ? ' now' : '';
263
+ h += '<div class="vtv-epg-item'+nw+'"><span class="t">'+(p.time||'')+(p.end_time?'-'+p.end_time:'')+'</span> <span class="n">'+(p.title||'')+'</span></div>';
264
+ });
265
+ el.innerHTML = h;
266
+ const nowItem = el.querySelector('.vtv-epg-item.now');
267
+ if(nowItem) nowItem.scrollIntoView({behavior:'smooth',inline:'center',block:'nearest'});
268
+ } catch(e){
269
+ el.innerHTML = '<div class="vtv-epg-empty">Không tải được lịch</div>';
270
+ }
271
+ }
272
+
273
+ window._vtvToggleEpg = function(){
274
+ const l = document.getElementById('vtv-epg-list'), b = document.getElementById('vtv-epg-toggle');
275
+ if(!l||!b) return;
276
+ if(l.style.display==='none'){ l.style.display='flex'; b.textContent='Ẩn'; }
277
+ else { l.style.display='none'; b.textContent='Hiện'; }
278
+ };
279
+
280
+ window._vtvTogglePiP = async function(){
281
+ const v = document.getElementById('vtv-player');
282
+ const btn = document.getElementById('vtv-pip-btn');
283
+ if(!v) return;
284
+ try{
285
+ if(document.pictureInPictureElement === v){
286
+ await document.exitPictureInPicture();
287
+ _pipActive = false;
288
+ if(btn) btn.classList.remove('on');
289
+ } else {
290
+ await v.requestPictureInPicture();
291
+ _pipActive = true;
292
+ if(btn) btn.classList.add('on');
293
+ }
294
+ } catch(e){ alert('Không hỗ trợ PiP'); }
295
+ };
296
+
297
+ function pinBlock(){
298
+ const h = document.getElementById('view-home');
299
+ if(!h||document.getElementById('vtv-block')) return;
300
+ h.insertBefore(buildBlock(), h.firstChild);
301
+ _blockInserted = true;
302
+ if(!_streamsLoaded){
303
+ loadAllStreams().then(()=>{
304
+ setTimeout(()=>window._vtvPlay(DEFAULT_CHANNEL), 300);
305
+ });
306
+ }
307
+ }
308
+
309
+ window._vtvRetry = function(){ if(_currentCh) window._vtvPlay(_currentCh); };
310
+
311
+ window._vtvPlay = function(chId){
312
+ const ch = CHANNELS.find(c=>c.id===chId);
313
+ if(!ch) return;
314
+ _currentCh = chId;
315
+ document.querySelectorAll('.vtv-tab').forEach(t=>t.classList.remove('on'));
316
+ const tab = document.getElementById('vtvt-'+chId);
317
+ if(tab) tab.classList.add('on');
318
+ const video = document.getElementById('vtv-player');
319
+ const errEl = document.getElementById('vtv-err');
320
+ const loadEl = document.getElementById('vtv-load');
321
+ const errMsg = document.getElementById('vtv-err-msg');
322
+ video.style.display='none';
323
+ errEl.style.display='none';
324
+ loadEl.style.display='flex';
325
+ loadEl.innerHTML = '<div class="vtv-spinner"></div>Đang kết nối '+ch.name+'...';
326
+ if(_hls){_hls.destroy(); _hls=null;}
327
+ const urls = STREAMS[chId]||[];
328
+ if(!urls.length){
329
+ loadEl.style.display='none'; errEl.style.display='flex';
330
+ errMsg.textContent = chId==='vtvprime' ? 'VTVPrime: Kênh trả phí.' : ch.name+': Không có luồng.';
331
+ return;
332
+ }
333
+ _tryPlay(video, urls, 0, ch.name, loadEl, errEl, errMsg);
334
+ loadEpg(chId);
335
+ if(_miniActive){
336
+ document.getElementById('vtv-mini-ch').textContent = ch.name;
337
+ updateMiniEpg();
338
+ }
339
+ };
340
+
341
+ function _tryPlay(video, urls, idx, name, loadEl, errEl, errMsg){
342
+ if(idx>=urls.length){
343
+ loadEl.style.display='none'; errEl.style.display='flex';
344
+ errMsg.textContent = name+': Tất cả nguồn lỗi.';
345
+ return;
346
+ }
347
+ const src = urls[idx];
348
+ loadEl.innerHTML = '<div class="vtv-spinner"></div>Kết nối '+name+' ('+(idx+1)+'/'+urls.length+')...';
349
+ if(typeof Hls!=='undefined'&&Hls.isSupported()){
350
+ const hls = new Hls({enableWorker:true,lowLatencyMode:true,startLevel:-1,capLevelToPlayerSize:true,maxBufferLength:20});
351
+ _hls = hls;
352
+ hls.loadSource(src);
353
+ hls.attachMedia(video);
354
+ hls.on(Hls.Events.MANIFEST_PARSED,()=>{ video.play().catch(()=>{}); loadEl.style.display='none'; video.style.display='block'; });
355
+ let rec = 0;
356
+ hls.on(Hls.Events.ERROR,(ev,data)=>{
357
+ if(data.fatal){
358
+ if(data.type===Hls.ErrorTypes.NETWORK_ERROR){
359
+ rec++;
360
+ if(rec<=3) setTimeout(()=>hls.startLoad(), 2000);
361
+ else { hls.destroy(); _hls=null; _tryPlay(video,urls,idx+1,name,loadEl,errEl,errMsg); }
362
+ } else if(data.type===Hls.ErrorTypes.MEDIA_ERROR){
363
+ try{hls.recoverMediaError();}catch(e){}
364
+ } else { hls.destroy(); _hls=null; _tryPlay(video,urls,idx+1,name,loadEl,errEl,errMsg); }
365
+ }
366
+ });
367
+ } else if(video.canPlayType('application/vnd.apple.mpegurl')){
368
+ video.src = src;
369
+ video.addEventListener('loadedmetadata',()=>{ video.play().catch(()=>{}); loadEl.style.display='none'; video.style.display='block'; },{once:true});
370
+ video.addEventListener('error',()=>{_tryPlay(video,urls,idx+1,name,loadEl,errEl,errMsg);},{once:true});
371
+ } else {
372
+ loadEl.style.display='none'; errEl.style.display='flex'; errMsg.textContent='Không hỗ trợ HLS';
373
+ }
374
+ }
375
+
376
+ // ===== NAVIGATION HOOKS =====
377
+ // Close mini player when returning to home
378
+ const _origShowView = window.showView;
379
+ window.showView = function(id){
380
+ if(id==='view-home' && _miniActive){
381
+ closeMiniPlayer();
382
+ } else if(id!=='view-home' && _currentCh && STREAMS[_currentCh] && STREAMS[_currentCh].length>0 && !_miniActive){
383
+ activateMiniPlayer();
384
+ }
385
+ return _origShowView.apply(this, arguments);
386
+ };
387
+
388
+ // Override readArticle — activate mini when leaving home
389
+ const _origReadArticle = window.readArticle;
390
+ if(_origReadArticle){
391
+ window.readArticle = function(url){
392
+ if(_currentCh && STREAMS[_currentCh] && STREAMS[_currentCh].length>0 && !_miniActive) activateMiniPlayer();
393
+ return _origReadArticle.apply(this, arguments);
394
+ };
395
+ }
396
+
397
+ // Override switchCat — close mini when going to home, activate when leaving
398
+ const _origSwitchCat = window.switchCat;
399
+ if(_origSwitchCat){
400
+ window.switchCat = function(id){
401
+ if(id==='home'){
402
+ // Returning to home — close mini player
403
+ if(_miniActive) closeMiniPlayer();
404
+ } else if(_currentCh && STREAMS[_currentCh] && STREAMS[_currentCh].length>0 && !_miniActive){
405
+ activateMiniPlayer();
406
+ }
407
+ return _origSwitchCat.apply(this, arguments);
408
+ };
409
+ }
410
+
411
+ // Override loadHome to inject VTV block
412
+ const _origLoadHome = window.loadHome;
413
+ if(_origLoadHome && !_origLoadHome.__vtvWrapped){
414
+ window.loadHome = async function(){
415
+ const old = document.getElementById('vtv-block');
416
+ if(old) old.remove();
417
+ _blockInserted = false;
418
+ const r = await _origLoadHome.apply(this, arguments);
419
+ try{ pinBlock(); }catch(e){}
420
+ return r;
421
+ };
422
+ window.loadHome.__vtvWrapped = true;
423
+ }
424
+ })();
vtv_api.py CHANGED
@@ -1,334 +1,186 @@
1
- # VTV Stream Fix v4 — EPG multi-source, VTV6 proxy + fallback
2
- import re, time, threading, json, base64, urllib.parse
 
 
 
 
 
3
  from fastapi import APIRouter, Query
4
  from fastapi.responses import JSONResponse, Response
 
5
  from datetime import datetime, timedelta, timezone
6
 
7
- # GMT+0 (UTC)
8
- VN_TZ = timezone(timedelta(hours=0))
9
  router = APIRouter()
10
 
11
  UA = {
12
  "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/125.0.0.0 Safari/537.36",
13
  "Accept-Language": "vi-VN,vi;q=0.9",
14
- "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
15
  }
16
 
17
  CHANNEL_NAMES = {
18
- "vtv1":"VTV1","vtv2":"VTV2","vtv3":"VTV3","vtv4":"VTV4","vtv5":"VTV5",
19
- "vtv6":"VTV6","vtv7":"VTV7","vtv8":"VTV8","vtv9":"VTV9","vtv10":"VTV10",
20
- "vtvprime":"VTVPrime",
 
 
 
 
 
 
 
 
21
  }
22
 
23
- # ===== VTV6 PRIORITY STREAM =====
24
- VTV6_PRIORITY_URL = "https://vtvgolive-ssaimh.vtvdigital.vn/o-QNezId5ssiM7KaV4MfXQ/1781372128/manifest/vtv6tt/master.m3u8"
 
 
 
 
 
 
 
 
 
 
25
 
26
- # ===== ALL STREAMS =====
27
  FPTPLAY_URLS = {
28
- "vtv1": "https://live.fptplay53.net/fnxhd1/vtv1_vhls.smil/chunklist_b5000000.m3u8",
29
- "vtv2": "https://live.fptplay53.net/fnxhd1/vtv2_vhls.smil/chunklist_b5000000.m3u8",
30
- "vtv3": "https://live.fptplay53.net/fnxhd1/vtv3_vhls.smil/chunklist_b5000000.m3u8",
31
- "vtv4": "https://live.fptplay53.net/fnxhd1/vtv4_vhls.smil/chunklist_b5000000.m3u8",
32
- "vtv5": "https://live.fptplay53.net/fnxhd1/vtv5_vhls.smil/chunklist_b5000000.m3u8",
33
- "vtv6": "https://live.fptplay53.net/fnxhd1/vtv6_vhls.smil/chunklist_b5000000.m3u8",
34
- "vtv7": "https://live.fptplay53.net/fnxhd1/vtv7_vhls.smil/chunklist_b5000000.m3u8",
35
- "vtv8": "https://live.fptplay53.net/fnxhd1/vtv8_vhls.smil/chunklist_b5000000.m3u8",
36
- "vtv9": "https://live.fptplay53.net/fnxhd1/vtv9_vhls.smil/chunklist_b5000000.m3u8",
37
- "vtv10": "https://live.fptplay53.net/fnxhd1/vtv10_vhls.smil/chunklist_b5000000.m3u8",
38
  }
39
 
40
- # Cache
41
- _cache = {}
42
- _lock = threading.Lock()
43
- _CACHE_TTL = 300
44
 
45
- def _cached(k):
46
- with _lock:
47
- if k in _cache and time.time() - _cache[k]['t'] < _CACHE_TTL:
48
- return _cache[k]['d']
49
  return None
50
- def _set_cache(k, d):
51
- with _lock:
52
- _cache[k] = {'t': time.time(), 'd': d}
53
-
54
- # ===== FETCH STREAM =====
55
- def fetch_vtv_stream(channel_id):
56
- channel_id = channel_id.lower().strip()
57
- cached = _cached(channel_id)
58
- if cached is not None:
59
- return cached
60
-
61
- # VTV6: ưu tiên link vtvdigital
62
- if channel_id == "vtv6":
63
- result = VTV6_PRIORITY_URL
64
- _set_cache(channel_id, result)
65
- return result
66
-
67
- # Các kênh khác: FPTPlay CDN
68
- result = FPTPLAY_URLS.get(channel_id)
69
- _set_cache(channel_id, result)
70
- return result
71
-
72
-
73
- # ===== EPG (LỊCH PHÁT SÓNG) — ĐA NGUỒN =====
74
 
75
- def _fetch_epg_vtvgo(channel_slug, date_str):
76
- """Lấy EPG từ VTVGo API"""
77
- try:
78
- import requests
79
- api_url = f"https://vtvgo.vn/api/schedule?channel={channel_slug}&date={date_str}"
80
- r = requests.get(api_url, headers={"User-Agent": UA["User-Agent"]}, timeout=10)
81
- if r.status_code == 200:
82
- data = r.json()
83
- items = data if isinstance(data, list) else data.get('data', data.get('schedule', []))
84
- programs = []
85
- for item in items:
86
- if isinstance(item, dict):
87
- t = item.get('time', item.get('start_time', ''))
88
- programs.append({
89
- 'time': t[:5] if len(t) >= 5 else t,
90
- 'title': item.get('title', item.get('name', item.get('program', ''))),
91
- })
92
- if programs:
93
- return programs
94
- except:
95
- pass
96
  return None
97
 
98
- def _fetch_epg_vtvvn(channel_slug):
99
- """Lấy EPG từ vtv.vn bằng BeautifulSoup"""
 
 
100
  try:
101
- import requests
102
- from bs4 import BeautifulSoup
103
- h = {"User-Agent": UA["User-Agent"], "Accept-Language": "vi-VN,vi;q=0.9"}
104
- ch_url = f"https://vtv.vn/lich-phat-song-{channel_slug}.htm"
105
- r = requests.get(ch_url, headers=h, timeout=10)
106
- r.encoding = 'utf-8'
107
- if r.status_code != 200:
108
- return None
109
-
110
- soup = BeautifulSoup(r.text, 'lxml')
111
- programs = []
112
-
113
- # Pattern 1: JSON data in <script>
114
- for script in soup.find_all('script'):
115
- text = script.string or ''
116
- if 'time' in text.lower() and ('title' in text.lower() or 'program' in text.lower()):
117
- for m in re.finditer(r'\[.*?\]', text, re.DOTALL):
118
- try:
119
- data = json.loads(m.group(0))
120
- if isinstance(data, list) and len(data) > 0:
121
- for item in data:
122
- if isinstance(item, dict) and 'time' in item:
123
- programs.append({
124
- 'time': item.get('time', ''),
125
- 'title': item.get('title', item.get('name', item.get('program', ''))),
126
- })
127
- except:
128
- pass
129
-
130
- # Pattern 2: HTML tables
131
- for table in soup.find_all('table'):
132
- for row in table.find_all('tr'):
133
- cells = row.find_all(['td', 'th'])
134
- if len(cells) >= 2:
135
- time_text = cells[0].get_text(strip=True)
136
- title_text = cells[1].get_text(strip=True)
137
- if re.match(r'^\d{1,2}:\d{2}', time_text) and len(title_text) >= 3:
138
- programs.append({'time': time_text[:5], 'title': title_text})
139
-
140
- # Pattern 3: Text with timestamps
141
- for el in soup.find_all(['div', 'li', 'p', 'span', 'td']):
142
- text = el.get_text(strip=True)
143
- tm = re.match(r'^(\d{1,2}:\d{2})\s*[-–—:]\s*(.+)', text)
144
- if tm and len(tm.group(2)) >= 3:
145
- programs.append({'time': tm.group(1), 'title': tm.group(2).strip()})
146
-
147
- # Pattern 4: Schedule class elements
148
- for cls_pattern in [r'schedule', r'program', r'lich', r'epg', r'timeline', r'table']:
149
- for el in soup.find_all(class_=re.compile(cls_pattern, re.I)):
150
- for item in el.find_all(['li', 'div', 'p']):
151
- text = item.get_text(strip=True)
152
- tm = re.match(r'^(\d{1,2}:\d{2})\s*[-–—:]\s*(.+)', text)
153
- if tm and len(tm.group(2)) >= 3:
154
- programs.append({'time': tm.group(1), 'title': tm.group(2).strip()})
155
-
156
- if programs:
157
- return programs
158
- except:
159
- pass
160
- return None
161
-
162
- def _fetch_epg_webtv(channel_slug):
163
- """Lấy EPG từ webtv.vtv.vn"""
164
- try:
165
- import requests
166
- h = {"User-Agent": UA["User-Agent"]}
167
- # Map channel slug to webtv channel ID
168
- webtv_map = {
169
- 'vtv1': '1', 'vtv2': '2', 'vtv3': '3', 'vtv4': '4',
170
- 'vtv5': '5', 'vtv6': '6', 'vtv7': '7', 'vtv8': '8',
171
- 'vtv9': '9', 'vtv10': '10',
172
- }
173
- ch_id = webtv_map.get(channel_slug)
174
- if not ch_id:
175
- return None
176
-
177
- today = datetime.now(VN_TZ).strftime("%Y-%m-%d")
178
- # Try webtv.vtv.vn API
179
- urls = [
180
- f"https://webtv.vtv.vn/api/schedule?channel_id={ch_id}&date={today}",
181
- f"https://vtvgo.vn/api/schedule-v2?channel_id={ch_id}&date={today}",
182
- ]
183
- for url in urls:
184
- try:
185
- r = requests.get(url, headers=h, timeout=10)
186
- if r.status_code == 200:
187
- data = r.json()
188
- items = data if isinstance(data, list) else data.get('data', data.get('schedule', data.get('result', [])))
189
- programs = []
190
- for item in items:
191
- if isinstance(item, dict):
192
- t = item.get('time', item.get('start_time', item.get('start', '')))
193
- programs.append({
194
- 'time': t[:5] if len(t) >= 5 else t,
195
- 'title': item.get('title', item.get('name', item.get('program', item.get('event', '')))),
196
- })
197
- if programs:
198
- return programs
199
- except:
200
- continue
201
  except:
202
  pass
203
  return None
204
 
205
- def _fetch_epg_xmltv(channel_slug):
206
- """Lấy EPG từ nguồn XMLTV công cộng"""
 
 
207
  try:
208
- import requests
209
- h = {"User-Agent": UA["User-Agent"]}
210
- # Map channel to XMLTV channel ID
211
- xmltv_map = {
212
- 'vtv1': 'VTV1.vtv.vn',
213
- 'vtv2': 'VTV2.vtv.vn',
214
- 'vtv3': 'VTV3.vtv.vn',
215
- 'vtv4': 'VTV4.vtv.vn',
216
- 'vtv5': 'VTV5.vtv.vn',
217
- 'vtv6': 'VTV6.vtv.vn',
218
- 'vtv7': 'VTV7.vtv.vn',
219
- 'vtv8': 'VTV8.vtv.vn',
220
- 'vtv9': 'VTV9.vtv.vn',
221
- 'vtv10': 'VTV10.vtv.vn',
222
- }
223
- ch_id = xmltv_map.get(channel_slug)
224
- if not ch_id:
225
- return None
226
-
227
- today = datetime.now(VN_TZ).strftime("%Y%m%d")
228
- # Try iptv-org EPG sources
229
- epg_sources = [
230
- f"https://epg.pm/api/epg?channel={ch_id}&date={today}",
231
- f"https://epg.iptvx.one/api/epg?channel={ch_id}&date={today}",
232
- ]
233
- for url in epg_sources:
234
- try:
235
- r = requests.get(url, headers=h, timeout=10)
236
- if r.status_code == 200:
237
- data = r.json()
238
- programs = []
239
- for item in data if isinstance(data, list) else data.get('epg', data.get('programs', [])):
240
- if isinstance(item, dict):
241
- start = item.get('start', item.get('time', ''))
242
- title = item.get('title', item.get('name', ''))
243
- if start and title:
244
- programs.append({
245
- 'time': start[-8:-3] if len(start) >= 8 else start[:5],
246
- 'title': title,
247
- })
248
- if programs:
249
- return programs
250
- except:
251
- continue
252
  except:
253
  pass
254
  return None
255
 
256
-
257
- def fetch_epg(channel_id):
258
- """Lấy EPG từ đa nguồn, ưu tiên VTVGo API"""
259
  channel_id = channel_id.lower().strip()
260
- EPG_SLUG = {
261
- 'vtv1': 'vtv1', 'vtv2': 'vtv2', 'vtv3': 'vtv3', 'vtv4': 'vtv4',
262
- 'vtv5': 'vtv5', 'vtv6': 'vtv6', 'vtv7': 'vtv7', 'vtv8': 'vtv8',
263
- 'vtv9': 'vtv9', 'vtv10': 'vtv10',
 
 
264
  }
265
- epg_ch = EPG_SLUG.get(channel_id)
266
- if not epg_ch:
267
- return {"programs": [], "channel": channel_id, "date": "", "timezone": "+0"}
268
-
269
- today = datetime.now(VN_TZ).strftime("%Y-%m-%d")
270
- cache_key = f"epg_{epg_ch}_{today}"
271
- cached = _cached(cache_key)
272
  if cached is not None:
273
  return cached
274
-
275
- programs = []
276
-
277
- # Nguồn 1: VTVGo API (nhanh, đáng tin cậy nhất)
278
- programs = _fetch_epg_vtvgo(epg_ch, today)
279
-
280
- # Nguồn 2: vtv.vn scraping
281
- if not programs:
282
- programs = _fetch_epg_vtvvn(epg_ch)
283
-
284
- # Nguồn 3: webtv.vtv.vn API
285
- if not programs:
286
- programs = _fetch_epg_webtv(epg_ch)
287
-
288
- # Nguồn 4: XMLTV
289
- if not programs:
290
- programs = _fetch_epg_xmltv(epg_ch)
291
-
292
- # Nếu vẫn không có, tạo schedule mẫu
293
- if not programs:
294
- # Tạo schedule giả định theo khung giờ VTV điển hình
295
- sample = [
296
- ("00:00", "Phim truyện"),
297
- ("06:00", "Chào buổi sáng"),
298
- ("07:00", "Bản tin sáng"),
299
- ("09:00", "Chương trình VTV {ch}"),
300
- ("12:00", "Bản tin trưa"),
301
- ("13:00", "Phim truyện chiều"),
302
- ("17:00", "Bản tin chiều"),
303
- ("19:00", "Thời sự"),
304
- ("20:00", "Phim truyện tối"),
305
- ("22:00", "Bản tin đêm"),
306
- ]
307
- ch_num = channel_id.replace('vtv', '')
308
- programs = [{'time': t, 'title': tt.replace('{ch}', ch_num)} for t, tt in sample]
309
-
310
- # Deduplicate + sort
311
- seen = set()
312
- unique = []
313
- for p in programs:
314
- key = f"{p['time']}|{p['title']}"
315
- if key not in seen:
316
- seen.add(key)
317
- unique.append(p)
318
- unique.sort(key=lambda x: x['time'])
319
-
320
- result = {
321
- "programs": unique[:50],
322
- "channel": channel_id,
323
- "channel_name": CHANNEL_NAMES.get(channel_id, channel_id.upper()),
324
- "date": today,
325
- "timezone": "+0",
326
- }
327
- _set_cache(cache_key, result)
328
- return result
329
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
330
 
331
- # ===== API ENDPOINTS =====
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
332
 
333
  @router.get("/api/vtv/streams")
334
  def api_vtv_streams():
@@ -338,173 +190,276 @@ def api_vtv_streams():
338
  result[ch_id] = {"name": CHANNEL_NAMES[ch_id], "stream_url": stream_url, "status": "ok" if stream_url else "offline"}
339
  return JSONResponse(result)
340
 
341
-
342
  @router.get("/api/vtv/stream/{channel_id}")
343
  def api_vtv_stream(channel_id: str):
344
- channel_id = channel_id.lower().strip()
345
- if channel_id not in CHANNEL_NAMES:
346
- channel_id = "vtv6"
347
  stream_url = fetch_vtv_stream(channel_id)
348
  if stream_url:
349
  return JSONResponse({"stream_url": stream_url, "status": "ok"})
350
  return JSONResponse({"error": "stream not found", "status": "offline"}, status_code=404)
351
 
352
-
353
- @router.get("/api/vtv/epg/{channel_id}")
354
- def api_vtv_epg(channel_id: str):
355
- channel_id = channel_id.lower().strip()
356
- if channel_id not in CHANNEL_NAMES:
357
- channel_id = "vtv6"
358
- data = fetch_epg(channel_id)
359
- return JSONResponse(data)
360
-
361
-
362
- # ===== PROXY: VTV6 (vtvdigital → cần CORS proxy) =====
363
- # VTVdigital CDN uses signed tokens and may block CORS from HF Space.
364
- # We proxy the manifest and segments through the server.
365
- # If the server gets 403 (non-VN IP), the frontend falls back to FPTPlay.
366
-
367
- @router.get("/api/proxy/vtv6/master.m3u8")
368
- def proxy_vtv6_master():
369
- """Proxy VTV6 master playlist — crucial for CORS-broken vtvdigital CDN"""
370
- url = VTV6_PRIORITY_URL
371
  try:
372
- import requests
373
- h = {
374
- "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36",
375
- "Accept": "*/*",
376
- "Referer": "https://vtvgo.vn/",
377
- "Origin": "https://vtvgo.vn",
378
- }
379
- r = requests.get(url, headers=h, timeout=15)
380
  if r.status_code != 200:
381
- # Try with different referer
382
- h["Referer"] = "https://vtv.vn/"
383
- h["Origin"] = "https://vtv.vn/"
384
- r = requests.get(url, headers=h, timeout=15)
385
-
 
 
 
 
 
 
 
 
 
 
386
  if r.status_code != 200:
387
- return Response(
388
- content=json.dumps({"error": f"upstream {r.status_code}", "fallback": FPTPLAY_URLS.get("vtv6")}),
389
- status_code=502,
390
- media_type="application/json",
391
- headers={"Access-Control-Allow-Origin": "*", "X-Fallback-Url": FPTPLAY_URLS.get("vtv6", "")}
392
- )
393
-
394
- # Rewrite segment URLs to go through proxy
395
- base_url = url[:url.rfind('/')]
396
  content = r.text
397
- lines = content.strip().split('\n')
398
  rewritten = []
 
399
  for line in lines:
400
- stripped = line.strip()
401
- if stripped.startswith('#') or not stripped:
402
  rewritten.append(line)
403
- elif stripped.startswith('http'):
404
- rewritten.append(f"/api/proxy/seg/vtv6?url={urllib.parse.quote(stripped, safe='')}")
405
  else:
406
- seg_url = f"{base_url}/{stripped}"
407
- rewritten.append(f"/api/proxy/seg/vtv6?url={urllib.parse.quote(seg_url, safe='')}")
408
-
409
- return Response(
410
- content='\n'.join(rewritten).encode('utf-8'),
411
- media_type="application/vnd.apple.mpegurl",
412
- headers={
413
- "Access-Control-Allow-Origin": "*",
414
- "Access-Control-Allow-Methods": "GET, OPTIONS",
415
- "Access-Control-Allow-Headers": "*",
416
- "Cache-Control": "public, max-age=10",
417
- }
418
- )
419
  except Exception as e:
420
- return Response(
421
- content=json.dumps({"error": str(e), "fallback": FPTPLAY_URLS.get("vtv6")}),
422
- status_code=502,
423
- media_type="application/json",
424
- headers={"Access-Control-Allow-Origin": "*", "X-Fallback-Url": FPTPLAY_URLS.get("vtv6", "")}
425
- )
426
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
427
 
428
- @router.get("/api/proxy/seg/vtv6")
429
- def proxy_seg_vtv6(url: str = Query(...)):
 
 
 
 
 
 
430
  try:
431
- import requests
432
- h = {
433
- "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36",
434
- "Accept": "*/*",
435
- "Referer": "https://vtvgo.vn/",
436
- "Origin": "https://vtvgo.vn",
437
  }
438
- r = requests.get(url, headers=h, timeout=30)
439
  if r.status_code != 200:
440
- return Response(status_code=502, content=f"upstream error (status {r.status_code})")
441
-
442
- return Response(
443
- content=r.content,
444
- media_type="video/mp2t",
445
- headers={
446
- "Access-Control-Allow-Origin": "*",
447
- "Access-Control-Allow-Methods": "GET, OPTIONS",
448
- "Access-Control-Allow-Headers": "*",
449
- "Cache-Control": "public, max-age=3600",
450
- }
451
- )
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
452
  except Exception as e:
453
- return Response(status_code=502, content=f"proxy error: {e}")
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
454
 
455
 
456
- # ===== LEGACY PROXY (for non-VTV6 channels) =====
 
 
 
457
 
458
- @router.get("/api/proxy/m3u8/vtv")
459
- def proxy_m3u8_vtv(url: str = Query(...)):
460
- try:
461
- import requests
462
- h = {"User-Agent": UA["User-Agent"], "Accept": "*/*",
463
- "Referer": "https://sv2.xemtivitop.com/"}
464
- r = requests.get(url, headers=h, timeout=15, verify=False)
465
- if r.status_code != 200:
466
- return Response(status_code=502, content=f"upstream error (status {r.status_code})")
467
-
468
- base_url = url[:url.rfind('/')]
469
- lines = r.text.strip().split('\n')
470
- rewritten = []
471
- for line in lines:
472
- stripped = line.strip()
473
- if stripped.startswith('#') or not stripped:
474
- rewritten.append(line)
475
- elif stripped.startswith('http'):
476
- rewritten.append(f"/api/proxy/seg/vtv?url={urllib.parse.quote(stripped, safe='')}")
477
  else:
478
- seg_url = f"{base_url}/{stripped}"
479
- rewritten.append(f"/api/proxy/seg/vtv?url={urllib.parse.quote(seg_url, safe='')}")
480
-
481
- return Response(
482
- content='\n'.join(rewritten).encode('utf-8'),
483
- media_type="application/vnd.apple.mpegurl",
484
- headers={"Access-Control-Allow-Origin": "*", "Cache-Control": "public, max-age=60"}
485
- )
486
- except Exception as e:
487
- return Response(status_code=502, content=f"proxy error: {e}")
488
 
 
 
 
489
 
490
- @router.get("/api/proxy/seg/vtv")
491
- def proxy_seg_vtv(url: str = Query(...)):
492
- try:
493
- import requests
494
- h = {"User-Agent": UA["User-Agent"], "Accept": "*/*",
495
- "Referer": "https://sv2.xemtivitop.com/"}
496
- r = requests.get(url, headers=h, timeout=30, verify=False)
497
- if r.status_code != 200:
498
- return Response(status_code=502, content=f"upstream error (status {r.status_code})")
499
-
500
- data = r.content
501
- if len(data) > 188 and data[0:4] == b'\x89PNG' and data[188] == 0x47:
502
- data = data[188:]
503
-
504
- return Response(
505
- content=data,
506
- media_type="video/mp2t",
507
- headers={"Access-Control-Allow-Origin": "*", "Cache-Control": "public, max-age=3600"}
508
- )
509
- except Exception as e:
510
- return Response(status_code=502, content=f"proxy error: {e}")
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ VTV Channels API - Backend endpoints for VTV1-VTV10 + VTVPrime
3
+ Fetches stream URLs from hd.xemtv.net PHP endpoints
4
+ EPG data scraped from https://vtv.vn/lich-phat-song.htm
5
+ """
6
+ import re, time, threading
7
+ import requests
8
  from fastapi import APIRouter, Query
9
  from fastapi.responses import JSONResponse, Response
10
+ from bs4 import BeautifulSoup
11
  from datetime import datetime, timedelta, timezone
12
 
13
+ VN_TZ = timezone(timedelta(hours=7))
14
+
15
  router = APIRouter()
16
 
17
  UA = {
18
  "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/125.0.0.0 Safari/537.36",
19
  "Accept-Language": "vi-VN,vi;q=0.9",
20
+ "Referer": "https://hd.xemtv.net/",
21
+ }
22
+
23
+ XEMTV_PHP_ENDPOINTS = {
24
+ "vtv1": "https://hd.xemtv.net/kenh/vtv1.php",
25
+ "vtv2": "https://hd.xemtv.net/kenh/vtv2.php",
26
+ "vtv3": "https://hd.xemtv.net/kenh/vtv3.php",
27
+ "vtv4": "https://hd.xemtv.net/kenh/vtv4.php",
28
+ "vtv5": "https://hd.xemtv.net/kenh/vtv5.php",
29
+ "vtv6": "https://hd.xemtv.net/kenh/vtv6.php",
30
+ "vtv7": "https://hd.xemtv.net/kenh/vtv7.php",
31
+ "vtv8": "https://hd.xemtv.net/kenh/vtv8.php",
32
+ "vtv9": "https://hd.xemtv.net/kenh/vtv9.php",
33
+ "vtv10": "https://hd.xemtv.net/kenh/vtv10.php",
34
+ "vtvprime": "https://hd.xemtv.net/kenh/vtvprime.php",
35
  }
36
 
37
  CHANNEL_NAMES = {
38
+ "vtv1": "VTV1",
39
+ "vtv2": "VTV2",
40
+ "vtv3": "VTV3",
41
+ "vtv4": "VTV4",
42
+ "vtv5": "VTV5",
43
+ "vtv6": "VTV6",
44
+ "vtv7": "VTV7",
45
+ "vtv8": "VTV8",
46
+ "vtv9": "VTV9",
47
+ "vtv10": "VTV10",
48
+ "vtvprime": "VTVPrime",
49
  }
50
 
51
+ VTVGO_FAILOVER = {
52
+ "vtv1": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv1-manifest.m3u8",
53
+ "vtv2": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv2-manifest.m3u8",
54
+ "vtv3": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv3-manifest.m3u8",
55
+ "vtv4": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv4-manifest.m3u8",
56
+ "vtv5": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv5-manifest.m3u8",
57
+ "vtv7": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv7-manifest.m3u8",
58
+ "vtv8": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv8-manifest.m3u8",
59
+ "vtv9": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv9-manifest.m3u8",
60
+ }
61
+
62
+ XEMTV_ONLY_CHANNELS = {"vtv6", "vtv10"}
63
 
 
64
  FPTPLAY_URLS = {
65
+ "vtv1": "https://live.fptplay53.net/fnxch2/vtv1hd_abr.smil/chunklist.m3u8",
66
+ "vtv2": "https://live.fptplay53.net/fnxch2/vtv2hd_abr.smil/chunklist.m3u8",
67
+ "vtv3": "https://live.fptplay53.net/fnxch2/vtv3hd_abr.smil/chunklist.m3u8",
68
+ "vtv4": "https://live.fptplay53.net/fnxch2/vtv4hd_abr.smil/chunklist.m3u8",
69
+ "vtv5": "https://live-a.fptplay53.net/live/media/VTV5HD/live_hls_avc/index.m3u8",
70
+ "vtv6": "https://live-a.fptplay53.net/live/media/vtv6/live247-hls-avc/vtv6-avc1_5600000=10000-mp4a_131600=20000.m3u8",
71
+ "vtv7": "https://live.fptplay53.net/fnxhd1/vtv7hd_vhls.smil/chunklist_b5000000.m3u8",
72
+ "vtv8": "https://live.fptplay53.net/epzhd1/vtv8hd_vhls.smil/chunklist.m3u8",
73
+ "vtv9": "https://live.fptplay53.net/fnxhd1/vtv9hd_vhls.smil/chunklist.m3u8",
74
+ "vtv10": "https://live.fptplay53.net/fnxch2/vtvcantho_abr.smil/chunklist.m3u8",
75
  }
76
 
77
+ _vtv_cache = {}
78
+ _vtv_lock = threading.Lock()
79
+ _CACHE_TTL = 180
 
80
 
81
+ def _cached(key):
82
+ with _vtv_lock:
83
+ if key in _vtv_cache and time.time() - _vtv_cache[key]['t'] < _CACHE_TTL:
84
+ return _vtv_cache[key]['d']
85
  return None
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
86
 
87
+ def _set_cache(key, data):
88
+ with _vtv_lock:
89
+ _vtv_cache[key] = {'t': time.time(), 'd': data}
90
+
91
+ def extract_m3u8_from_html(html):
92
+ if not html:
93
+ return None
94
+ m = re.search(r"file\s*:\s*['\"]([^'\"]*\.m3u8[^'\"]*)['\"]", html, re.IGNORECASE)
95
+ if m:
96
+ url = m.group(1).strip()
97
+ if len(url) > 20:
98
+ return url
99
+ m = re.search(r"(https?://[^\s\"'<>\\]+\.m3u8[^\s\"'<>\\]*)", html, re.IGNORECASE)
100
+ if m:
101
+ url = m.group(1).strip()
102
+ if len(url) > 20:
103
+ return url
 
 
 
 
104
  return None
105
 
106
+ def fetch_xemtv_stream(channel_id):
107
+ php_url = XEMTV_PHP_ENDPOINTS.get(channel_id)
108
+ if not php_url:
109
+ return None
110
  try:
111
+ r = requests.get(php_url, headers=UA, timeout=15, allow_redirects=True)
112
+ if r.status_code == 200:
113
+ m3u8 = extract_m3u8_from_html(r.text)
114
+ if m3u8:
115
+ return m3u8
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
116
  except:
117
  pass
118
  return None
119
 
120
+ def fetch_fptplay_stream(channel_id):
121
+ url = FPTPLAY_URLS.get(channel_id)
122
+ if not url:
123
+ return None
124
  try:
125
+ headers = {"User-Agent": UA["User-Agent"], "Referer": "https://fptplay.vn/", "Origin": "https://fptplay.vn"}
126
+ r = requests.get(url, headers=headers, timeout=15, allow_redirects=True)
127
+ if r.status_code == 200:
128
+ return url
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
129
  except:
130
  pass
131
  return None
132
 
133
+ def fetch_vtv_stream(channel_id):
 
 
134
  channel_id = channel_id.lower().strip()
135
+ name_map = {
136
+ 'vtvct': 'vtv10', 'vtv-can-tho': 'vtv10', 'vtv can tho': 'vtv10',
137
+ 'vtv_can_tho': 'vtv10', 'cantho': 'vtv10',
138
+ 'vietnam_vtv1': 'vtv1', 'vietnam_vtv2': 'vtv2', 'vietnam_vtv3': 'vtv3',
139
+ 'vietnam_vtv4': 'vtv4', 'vietnam_vtv5': 'vtv5', 'vietnam_vtv6': 'vtv6',
140
+ 'vietnam_vtv7': 'vtv7', 'vietnam_vtv8': 'vtv8', 'vietnam_vtv9': 'vtv9',
141
  }
142
+ channel_id = name_map.get(channel_id, channel_id)
143
+ cached = _cached(channel_id)
 
 
 
 
 
144
  if cached is not None:
145
  return cached
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
146
 
147
+ if channel_id in XEMTV_ONLY_CHANNELS:
148
+ xemtv_url = fetch_xemtv_stream(channel_id)
149
+ if xemtv_url:
150
+ _set_cache(channel_id, xemtv_url)
151
+ return xemtv_url
152
+ fpt_url = FPTPLAY_URLS.get(channel_id)
153
+ if fpt_url:
154
+ _set_cache(channel_id, fpt_url)
155
+ return fpt_url
156
+ _set_cache(channel_id, None)
157
+ return None
158
+
159
+ if channel_id in VTVGO_FAILOVER:
160
+ result = VTVGO_FAILOVER[channel_id]
161
+ _set_cache(channel_id, result)
162
+ return result
163
 
164
+ if channel_id == 'vtvprime':
165
+ xemtv_url = fetch_xemtv_stream('vtvprime')
166
+ if xemtv_url:
167
+ _set_cache(channel_id, xemtv_url)
168
+ return xemtv_url
169
+ _set_cache(channel_id, None)
170
+ return None
171
+
172
+ xemtv_url = fetch_xemtv_stream(channel_id)
173
+ if xemtv_url:
174
+ _set_cache(channel_id, xemtv_url)
175
+ return xemtv_url
176
+
177
+ fpt_url = fetch_fptplay_stream(channel_id)
178
+ if fpt_url:
179
+ _set_cache(channel_id, fpt_url)
180
+ return fpt_url
181
+
182
+ _set_cache(channel_id, None)
183
+ return None
184
 
185
  @router.get("/api/vtv/streams")
186
  def api_vtv_streams():
 
190
  result[ch_id] = {"name": CHANNEL_NAMES[ch_id], "stream_url": stream_url, "status": "ok" if stream_url else "offline"}
191
  return JSONResponse(result)
192
 
 
193
  @router.get("/api/vtv/stream/{channel_id}")
194
  def api_vtv_stream(channel_id: str):
 
 
 
195
  stream_url = fetch_vtv_stream(channel_id)
196
  if stream_url:
197
  return JSONResponse({"stream_url": stream_url, "status": "ok"})
198
  return JSONResponse({"error": "stream not found", "status": "offline"}, status_code=404)
199
 
200
+ @router.get("/api/proxy/page")
201
+ def proxy_page(url: str = Query(...)):
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
202
  try:
203
+ headers = {**UA}
204
+ if "xemtv.net" in url:
205
+ headers["Referer"] = "https://hd.xemtv.net/"
206
+ r = requests.get(url, headers=headers, timeout=15, allow_redirects=True)
 
 
 
 
207
  if r.status_code != 200:
208
+ return Response(status_code=502, content="upstream error")
209
+ return Response(content=r.text.encode("utf-8"), media_type="text/html; charset=utf-8", headers={"Access-Control-Allow-Origin": "*"})
210
+ except:
211
+ return Response(status_code=502, content="proxy error")
212
+
213
+ @router.get("/api/proxy/m3u8/vtv")
214
+ def proxy_vtv_m3u8(url: str = Query(...)):
215
+ try:
216
+ headers = {"User-Agent": UA["User-Agent"], "Accept": "*/*"}
217
+ if "fptplay" in url:
218
+ headers["Referer"] = "https://fptplay.vn/"
219
+ headers["Origin"] = "https://fptplay.vn"
220
+ elif "xemtv" in url:
221
+ headers["Referer"] = "https://hd.xemtv.net/"
222
+ r = requests.get(url, headers=headers, timeout=15, allow_redirects=True)
223
  if r.status_code != 200:
224
+ return Response(status_code=502, content="upstream error")
 
 
 
 
 
 
 
 
225
  content = r.text
226
+ lines = content.split('\n')
227
  rewritten = []
228
+ base_url = url.rsplit('/', 1)[0] + '/'
229
  for line in lines:
230
+ line = line.strip()
231
+ if not line or line.startswith('#'):
232
  rewritten.append(line)
 
 
233
  else:
234
+ seg_url = line
235
+ if not seg_url.startswith('http'):
236
+ seg_url = base_url + seg_url
237
+ rewritten.append("/api/proxy/seg/vtv?url=" + requests.utils.quote(seg_url, safe=""))
238
+ return Response(content='\n'.join(rewritten).encode("utf-8"), media_type="application/vnd.apple.mpegurl", headers={"Access-Control-Allow-Origin": "*", "Cache-Control": "no-cache"})
 
 
 
 
 
 
 
 
239
  except Exception as e:
240
+ return Response(status_code=502, content="proxy error: " + str(e))
 
 
 
 
 
241
 
242
+ @router.get("/api/proxy/seg/vtv")
243
+ def proxy_vtv_segment(url: str = Query(...)):
244
+ try:
245
+ headers = {"User-Agent": UA["User-Agent"], "Accept": "*/*"}
246
+ if "fptplay" in url:
247
+ headers["Referer"] = "https://fptplay.vn/"
248
+ headers["Origin"] = "https://fptplay.vn"
249
+ r = requests.get(url, headers=headers, timeout=30, allow_redirects=True)
250
+ if r.status_code != 200:
251
+ return Response(status_code=502, content="upstream error")
252
+ data = r.content
253
+ if len(data) > 188 and data[0:4] == b'\x89PNG' and data[188] == 0x47:
254
+ data = data[188:]
255
+ return Response(content=data, media_type="video/mp2t", headers={"Access-Control-Allow-Origin": "*", "Cache-Control": "public, max-age=3600"})
256
+ except:
257
+ return Response(status_code=502, content="proxy error")
258
+
259
+
260
+ # ===== EPG from vtv.vn/lich-phat-song.htm =====
261
+
262
+ _epg_cache = {}
263
+ _epg_cache_time = 0
264
+ _EPG_CACHE_TTL = 1800 # 30 minutes
265
+
266
+ # Map vtv.vn channel IDs to our channel IDs
267
+ VTV_CHANNEL_MAP = {
268
+ "vtv1": "vtv1",
269
+ "vtv2": "vtv2",
270
+ "vtv3": "vtv3",
271
+ "vtv4": "vtv4",
272
+ "vtv5": "vtv5",
273
+ "vtv5-tay-nam-bo": "vtv5",
274
+ "vtv5-tay-nguyen": "vtv5",
275
+ "vtv7": "vtv7",
276
+ "vtv8": "vtv8",
277
+ "vtv9": "vtv9",
278
+ "vtv-can-tho": "vtv10",
279
+ }
280
 
281
+ def _fetch_epg_from_vtv():
282
+ """Fetch EPG from https://vtv.vn/lich-phat-song.htm"""
283
+ global _epg_cache, _epg_cache_time
284
+ now_ts = time.time()
285
+ if _epg_cache and now_ts - _epg_cache_time < _EPG_CACHE_TTL:
286
+ return _epg_cache
287
+
288
+ epg_data = {}
289
  try:
290
+ headers = {
291
+ "User-Agent": UA["User-Agent"],
292
+ "Accept-Language": "vi-VN,vi;q=0.9",
293
+ "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8",
294
+ "Referer": "https://vtv.vn/",
 
295
  }
296
+ r = requests.get("https://vtv.vn/lich-phat-song.htm", headers=headers, timeout=20)
297
  if r.status_code != 200:
298
+ print(f"EPG vtv.vn fetch failed: {r.status_code}")
299
+ return epg_data
300
+
301
+ r.encoding = "utf-8"
302
+ soup = BeautifulSoup(r.text, "lxml")
303
+
304
+ # Get channel order from list-channel section
305
+ channel_order = []
306
+ list_channel = soup.find(class_=re.compile(r'list-channel'))
307
+ if list_channel:
308
+ for link in list_channel.find_all('a', href=re.compile(r'truyen-hinh-truc-tuyen/([^.]+)\.htm')):
309
+ ch_id = re.search(r'truyen-hinh-truc-tuyen/([^.]+)\.htm', link.get('href', ''))
310
+ if ch_id:
311
+ channel_order.append(ch_id.group(1))
312
+
313
+ # If no list-channel found, try to find channel links elsewhere
314
+ if not channel_order:
315
+ for link in soup.find_all('a', href=re.compile(r'truyen-hinh-truc-tuyen/([^.]+)\.htm')):
316
+ ch_id = re.search(r'truyen-hinh-truc-tuyen/([^.]+)\.htm', link.get('href', ''))
317
+ if ch_id and ch_id.group(1) not in channel_order:
318
+ channel_order.append(ch_id.group(1))
319
+
320
+ # Find all <ul class="programs"> containers
321
+ prog_containers = soup.find_all('ul', class_=re.compile(r'\bprograms\b'))
322
+
323
+ print(f"EPG: {len(channel_order)} channels, {len(prog_containers)} program containers")
324
+
325
+ # Map each container to channel by order
326
+ for i, container in enumerate(prog_containers):
327
+ if i >= len(channel_order):
328
+ break
329
+ vtv_ch_id = channel_order[i]
330
+ our_ch_id = VTV_CHANNEL_MAP.get(vtv_ch_id, vtv_ch_id)
331
+
332
+ if our_ch_id not in epg_data:
333
+ epg_data[our_ch_id] = []
334
+
335
+ # Parse each <li class="program"> in this container
336
+ for li in container.find_all('li', class_=re.compile(r'\bprogram\b')):
337
+ time_span = li.find('span', class_=re.compile(r'\btime\b'))
338
+ title_span = li.find('span', class_=re.compile(r'\btitle\b'))
339
+ genre_a = li.find('a', class_=re.compile(r'\bgenre\b'))
340
+
341
+ time_str = time_span.get_text(strip=True) if time_span else ""
342
+ # Title: prefer genre (specific program name), fallback to title (category)
343
+ title = ""
344
+ if genre_a:
345
+ title = genre_a.get_text(strip=True)
346
+ if not title and title_span:
347
+ title = title_span.get_text(strip=True)
348
+
349
+ if not time_str or not title:
350
+ continue
351
+
352
+ # Parse time to datetime
353
+ start_dt = _parse_time(time_str)
354
+ if not start_dt:
355
+ continue
356
+
357
+ epg_data[our_ch_id].append({
358
+ "time": time_str[:5],
359
+ "title": title[:80],
360
+ "start_dt": start_dt,
361
+ })
362
+
363
+ # Sort by time and calculate end times
364
+ for ch_id in epg_data:
365
+ epg_data[ch_id].sort(key=lambda x: x.get("start_dt") or datetime.min)
366
+ # Remove duplicates
367
+ seen = set()
368
+ unique = []
369
+ for p in epg_data[ch_id]:
370
+ key = (p["time"], p["title"])
371
+ if key not in seen:
372
+ seen.add(key)
373
+ unique.append(p)
374
+ epg_data[ch_id] = unique
375
+
376
+ print(f"EPG parsed: {len(epg_data)} channels, {sum(len(v) for v in epg_data)} programmes")
377
+
378
  except Exception as e:
379
+ print(f"EPG vtv.vn error: {e}")
380
+
381
+ _epg_cache = epg_data
382
+ _epg_cache_time = now_ts
383
+ return epg_data
384
+
385
+
386
+ def _parse_time(time_str):
387
+ """Parse HH:MM to datetime today"""
388
+ if not time_str:
389
+ return None
390
+ time_str = time_str.strip().replace("h", ":").replace("H", ":")
391
+ m = re.search(r'(\d{1,2}):(\d{2})', time_str)
392
+ if m:
393
+ try:
394
+ hour, minute = int(m.group(1)), int(m.group(2))
395
+ now = datetime.now(VN_TZ)
396
+ return now.replace(hour=hour, minute=minute, second=0, microsecond=0)
397
+ except:
398
+ pass
399
+ return None
400
 
401
 
402
+ def _get_epg_for_channel(channel_id):
403
+ """Get EPG for a channel with current program marking."""
404
+ epg_data = _fetch_epg_from_vtv()
405
+ programmes = epg_data.get(channel_id, [])
406
 
407
+ if not programmes:
408
+ return []
409
+
410
+ now = datetime.now(VN_TZ)
411
+ result = []
412
+ for i, p in enumerate(programmes):
413
+ start_dt = p.get("start_dt")
414
+ stop_dt = None
415
+ if i + 1 < len(programmes):
416
+ stop_dt = programmes[i + 1].get("start_dt")
417
+
418
+ is_now = False
419
+ if start_dt:
420
+ if stop_dt:
421
+ is_now = start_dt <= now < stop_dt
 
 
 
 
422
  else:
423
+ is_now = start_dt <= now
 
 
 
 
 
 
 
 
 
424
 
425
+ end_time = ""
426
+ if stop_dt:
427
+ end_time = stop_dt.strftime("%H:%M")
428
 
429
+ result.append({
430
+ "time": p["time"],
431
+ "title": p["title"],
432
+ "end_time": end_time,
433
+ "now": is_now,
434
+ })
435
+
436
+ return result
437
+
438
+
439
+ @router.get("/api/vtv/epg/{channel_id}")
440
+ def api_vtv_epg(channel_id: str):
441
+ """Get EPG for a VTV channel from vtv.vn."""
442
+ channel_id = channel_id.lower().strip()
443
+ if channel_id not in CHANNEL_NAMES:
444
+ return JSONResponse({"error": "channel not found"}, status_code=404)
445
+ programs = _get_epg_for_channel(channel_id)
446
+ return JSONResponse({
447
+ "channel": channel_id,
448
+ "channel_name": CHANNEL_NAMES.get(channel_id, channel_id),
449
+ "date": datetime.now(VN_TZ).strftime("%Y-%m-%d"),
450
+ "programs": programs,
451
+ })
452
+
453
+
454
+ @router.get("/api/vtv/epg")
455
+ def api_vtv_epg_refresh():
456
+ """Force refresh EPG cache."""
457
+ global _epg_cache, _epg_cache_time
458
+ _epg_cache = {}
459
+ _epg_cache_time = 0
460
+ epg_data = _fetch_epg_from_vtv()
461
+ return JSONResponse({
462
+ "status": "refreshed",
463
+ "channels": len(epg_data),
464
+ "total_programmes": sum(len(v) for v in epg_data),
465
+ })
vtv_scraper.py CHANGED
@@ -1,11 +1,9 @@
1
  """
2
- VTV Channels Scraper - Optimized for stable streaming
3
- Fetches stream URLs from multiple CDN sources for VTV1-VTV10 + VTV Cần Thơ
 
4
  """
5
  import requests, re, time, threading
6
- from datetime import datetime, timedelta, timezone
7
-
8
- VN_TZ = timezone(timedelta(hours=17))
9
 
10
  UA = {
11
  "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/125.0.0.0 Safari/537.36",
@@ -13,9 +11,8 @@ UA = {
13
  "Referer": "https://hd.xemtv.net/",
14
  }
15
 
16
- # ===== PRIMARY CDN SOURCES (Optimized for stability) =====
17
- # Priority order: FPTPlay > VTVGo > MediaCDN
18
-
19
  XEMTV_PHP_ENDPOINTS = {
20
  "vtv1": "https://hd.xemtv.net/kenh/vtv1.php",
21
  "vtv2": "https://hd.xemtv.net/kenh/vtv2.php",
@@ -26,45 +23,42 @@ XEMTV_PHP_ENDPOINTS = {
26
  "vtv7": "https://hd.xemtv.net/kenh/vtv7.php",
27
  "vtv8": "https://hd.xemtv.net/kenh/vtv8.php",
28
  "vtv9": "https://hd.xemtv.net/kenh/vtv9.php",
29
- "vtv10": "https://hd.xemtv.net/kenh/vtv10.php",
30
  }
31
 
 
32
  CHANNEL_NAMES = {
33
- "vtv1": "VTV1", "vtv2": "VTV2", "vtv3": "VTV3", "vtv4": "VTV4",
34
- "vtv5": "VTV5", "vtv6": "VTV6", "vtv7": "VTV7", "vtv8": "VTV8",
35
- "vtv9": "VTV9", "vtv10": "VTV10",
 
 
 
 
 
 
 
36
  }
37
 
38
- # ===== RELIABLE CDN BACKUPS (verified working URLs) =====
39
- CDN_STREAMS = {
40
- # FPTPlay - Primary (most stable)
41
- "vtv1": "https://live-a.fptplay53.net/live/media/vtv1/live247-hls-avc/index.m3u8",
42
- "vtv2": "https://live-a.fptplay53.net/live/media/vtv2/live247-hls-avc/index.m3u8",
43
- "vtv3": "https://live-a.fptplay53.net/live/media/vtv3/live247-hls-avc/index.m3u8",
44
- "vtv4": "https://live-a.fptplay53.net/live/media/vtv4/live247-hls-avc/index.m3u8",
45
- "vtv5": "https://live-a.fptplay53.net/live/media/vtv5/live247-hls-avc/index.m3u8",
46
- "vtv6": "https://live-a.fptplay53.net/live/media/vtv6/live247-hls-avc/index.m3u8",
47
- "vtv7": "https://live-a.fptplay53.net/live/media/vtv7/live247-hls-avc/index.m3u8",
48
- "vtv8": "https://live-a.fptplay53.net/live/media/vtv8/live-hls-avc/index.m3u8",
49
- "vtv9": "https://live-a.fptplay53.net/live/media/vtv9/live247-hls-avc/index.m3u8",
50
- "vtv10": "https://live-a.fptplay53.net/live/media/vtv10/live247-hls-avc/index.m3u8",
51
-
52
- # VTVGo Failover
53
- "vtv1_fb": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv1-manifest.m3u8",
54
- "vtv2_fb": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv2-manifest.m3u8",
55
- "vtv3_fb": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv3-manifest.m3u8",
56
- "vtv4_fb": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv4-manifest.m3u8",
57
- "vtv5_fb": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv5-manifest.m3u8",
58
- "vtv6_fb": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv6-manifest.m3u8",
59
- "vtv7_fb": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv7-manifest.m3u8",
60
- "vtv8_fb": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv8-manifest.m3u8",
61
- "vtv9_fb": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv9-manifest.m3u8",
62
- "vtv10_fb": "https://vtvgolive-failover.vtvdigital.vn/vtvgo/vtv10-manifest.m3u8",
63
  }
64
 
65
  _vtv_cache = {}
66
  _vtv_lock = threading.Lock()
67
- _CACHE_TTL = 180
 
68
 
69
  def _cached(key):
70
  with _vtv_lock:
@@ -72,39 +66,36 @@ def _cached(key):
72
  return _vtv_cache[key]['d']
73
  return None
74
 
 
75
  def _set_cache(key, data):
76
  with _vtv_lock:
77
  _vtv_cache[key] = {'t': time.time(), 'd': data}
78
 
 
79
  def extract_m3u8_from_html(html):
80
- if not html: return None
81
- # Look for file: "..." pattern
 
 
82
  m = re.search(r"file\s*:\s*['\"]([^'\"]*\.m3u8[^'\"]*)['\"]", html, re.IGNORECASE)
83
  if m:
84
  url = m.group(1).strip()
85
- if len(url) > 20: return url
86
- # Look for direct m3u8 URL
 
87
  m = re.search(r"(https?://[^\s\"'<>\\]+\.m3u8[^\s\"'<>\\]*)", html, re.IGNORECASE)
88
  if m:
89
  url = m.group(1).strip()
90
- if len(url) > 20: return url
91
- return None
92
-
93
- def verify_cdn(url, referer="", timeout=8):
94
- """Quick verify CDN is working"""
95
- if not url: return None
96
- try:
97
- r = requests.get(url, headers={"User-Agent": UA["User-Agent"], "Referer": referer}, timeout=timeout, allow_redirects=True, verify=False)
98
- if r.status_code == 200 and '#EXTM3U' in r.text[:500]:
99
  return url
100
- except: pass
101
  return None
102
 
 
103
  def fetch_vtv_stream(channel_id):
104
- """Fetch VTV stream with priority order for maximum stability"""
105
  channel_id = channel_id.lower().strip()
106
-
107
- # Normalize channel ID
108
  name_map = {
109
  'vtvct': 'vtv10', 'vtv-can-tho': 'vtv10', 'vtv can tho': 'vtv10',
110
  'vtv_can_tho': 'vtv10', 'cantho': 'vtv10', 'cần thơ': 'vtv10',
@@ -113,35 +104,53 @@ def fetch_vtv_stream(channel_id):
113
  'vietnam_vtv7': 'vtv7', 'vietnam_vtv8': 'vtv8', 'vietnam_vtv9': 'vtv9',
114
  }
115
  channel_id = name_map.get(channel_id, channel_id)
116
-
117
- # Try direct CDN URLs first (most stable)
118
- if channel_id in CDN_STREAMS:
119
- return CDN_STREAMS[channel_id]
120
-
121
- # Try PHP endpoints as fallback
122
  php_url = XEMTV_PHP_ENDPOINTS.get(channel_id)
123
- if php_url:
124
- try:
125
- r = requests.get(php_url, headers=UA, timeout=15, allow_redirects=True, verify=False)
126
- if r.status_code == 200:
127
- m3u8 = extract_m3u8_from_html(r.text)
128
- if m3u8: return m3u8
129
- except: pass
130
-
131
- # Try failover backup
132
- fb_key = f"{channel_id}_fb"
133
- if fb_key in CDN_STREAMS:
134
- return CDN_STREAMS[fb_key]
135
-
 
 
 
 
 
 
 
 
 
 
 
136
  return None
137
 
 
138
  def get_all_vtv_streams():
 
139
  channels = []
140
- for ch_id in CHANNEL_NAMES:
141
  stream_url = fetch_vtv_stream(ch_id)
142
  channels.append({
143
  'id': ch_id,
144
  'name': CHANNEL_NAMES.get(ch_id, ch_id.upper()),
145
  'stream_url': stream_url,
146
  })
147
- return channels
 
 
 
 
 
 
1
  """
2
+ VTV Channels Scraper
3
+ Fetches stream URLs from hd.xemtv.net PHP endpoints for VTV1-VTV10 + VTV Cần Thơ
4
+ The PHP endpoints return jwplayer config with fresh stream URLs (important for VTV6 which has expiring signatures)
5
  """
6
  import requests, re, time, threading
 
 
 
7
 
8
  UA = {
9
  "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/125.0.0.0 Safari/537.36",
 
11
  "Referer": "https://hd.xemtv.net/",
12
  }
13
 
14
+ # Channel ID -> xemtv.net PHP endpoint mapping
15
+ # These PHP pages return jwplayer config with fresh stream URLs
 
16
  XEMTV_PHP_ENDPOINTS = {
17
  "vtv1": "https://hd.xemtv.net/kenh/vtv1.php",
18
  "vtv2": "https://hd.xemtv.net/kenh/vtv2.php",
 
23
  "vtv7": "https://hd.xemtv.net/kenh/vtv7.php",
24
  "vtv8": "https://hd.xemtv.net/kenh/vtv8.php",
25
  "vtv9": "https://hd.xemtv.net/kenh/vtv9.php",
26
+ "vtv10": "https://hd.xemtv.net/kenh/vtv10.php", # VTV Cần Thơ
27
  }
28
 
29
+ # Channel display names
30
  CHANNEL_NAMES = {
31
+ "vtv1": "VTV1",
32
+ "vtv2": "VTV2",
33
+ "vtv3": "VTV3",
34
+ "vtv4": "VTV4",
35
+ "vtv5": "VTV5",
36
+ "vtv6": "VTV6",
37
+ "vtv7": "VTV7",
38
+ "vtv8": "VTV8",
39
+ "vtv9": "VTV9",
40
+ "vtv10": "VTV Cần Thơ",
41
  }
42
 
43
+ # Fallback CDN streams (fptplay) used when xemtv scraping fails
44
+ CDN_FALLBACK = {
45
+ "vtv1": "https://live.fptplay53.net/fnxch2/vtv1hd_abr.smil/chunklist.m3u8",
46
+ "vtv2": "https://live.fptplay53.net/fnxch2/vtv2hd_abr.smil/chunklist.m3u8",
47
+ "vtv3": "https://live.fptplay53.net/fnxch2/vtv3hd_abr.smil/chunklist.m3u8",
48
+ "vtv4": "https://live.fptplay53.net/fnxch2/vtv4hd_abr.smil/chunklist.m3u8",
49
+ "vtv5": "https://live-a.fptplay53.net/live/media/VTV5HD/live_hls_avc/index.m3u8",
50
+ "vtv7": "https://live.fptplay53.net/fnxhd1/vtv7hd_vhls.smil/chunklist_b5000000.m3u8",
51
+ "vtv8": "https://live.fptplay53.net/epzhd1/vtv8hd_vhls.smil/chunklist.m3u8",
52
+ "vtv9": "https://live.fptplay53.net/fnxhd1/vtv9hd_vhls.smil/chunklist.m3u8",
53
+ "vtv10": "https://live.fptplay53.net/fnxch2/vtvcantho_abr.smil/chunklist.m3u8",
54
+ # VTV6 fallback: try canthotv as alternative
55
+ "vtv6": "https://live.canthotv.vn/live/tv/chunklist.m3u8",
 
 
 
 
 
 
 
 
 
 
 
 
56
  }
57
 
58
  _vtv_cache = {}
59
  _vtv_lock = threading.Lock()
60
+ _CACHE_TTL = 180 # 3 minutes — refresh frequently for VTV6 sign URLs
61
+
62
 
63
  def _cached(key):
64
  with _vtv_lock:
 
66
  return _vtv_cache[key]['d']
67
  return None
68
 
69
+
70
  def _set_cache(key, data):
71
  with _vtv_lock:
72
  _vtv_cache[key] = {'t': time.time(), 'd': data}
73
 
74
+
75
  def extract_m3u8_from_html(html):
76
+ """Extract m3u8 URL from xemtv PHP page (jwplayer config)."""
77
+ if not html:
78
+ return None
79
+ # jwplayer config: file: 'URL'
80
  m = re.search(r"file\s*:\s*['\"]([^'\"]*\.m3u8[^'\"]*)['\"]", html, re.IGNORECASE)
81
  if m:
82
  url = m.group(1).strip()
83
+ if len(url) > 20:
84
+ return url
85
+ # Generic m3u8 pattern
86
  m = re.search(r"(https?://[^\s\"'<>\\]+\.m3u8[^\s\"'<>\\]*)", html, re.IGNORECASE)
87
  if m:
88
  url = m.group(1).strip()
89
+ if len(url) > 20:
 
 
 
 
 
 
 
 
90
  return url
 
91
  return None
92
 
93
+
94
  def fetch_vtv_stream(channel_id):
95
+ """Fetch m3u8 stream URL for a VTV channel by scraping xemtv.net PHP endpoint."""
96
  channel_id = channel_id.lower().strip()
97
+
98
+ # Normalize name
99
  name_map = {
100
  'vtvct': 'vtv10', 'vtv-can-tho': 'vtv10', 'vtv can tho': 'vtv10',
101
  'vtv_can_tho': 'vtv10', 'cantho': 'vtv10', 'cần thơ': 'vtv10',
 
104
  'vietnam_vtv7': 'vtv7', 'vietnam_vtv8': 'vtv8', 'vietnam_vtv9': 'vtv9',
105
  }
106
  channel_id = name_map.get(channel_id, channel_id)
107
+
108
+ # Check cache
109
+ cached = _cached(channel_id)
110
+ if cached:
111
+ return cached
112
+
113
  php_url = XEMTV_PHP_ENDPOINTS.get(channel_id)
114
+ if not php_url:
115
+ # Fallback
116
+ fallback = CDN_FALLBACK.get(channel_id)
117
+ if fallback:
118
+ _set_cache(channel_id, fallback)
119
+ return fallback
120
+ return None
121
+
122
+ try:
123
+ r = requests.get(php_url, headers=UA, timeout=15, allow_redirects=True)
124
+ if r.status_code == 200:
125
+ m3u8 = extract_m3u8_from_html(r.text)
126
+ if m3u8:
127
+ _set_cache(channel_id, m3u8)
128
+ return m3u8
129
+ except:
130
+ pass
131
+
132
+ # Fallback to CDN
133
+ fallback = CDN_FALLBACK.get(channel_id)
134
+ if fallback:
135
+ _set_cache(channel_id, fallback)
136
+ return fallback
137
+
138
  return None
139
 
140
+
141
  def get_all_vtv_streams():
142
+ """Fetch all VTV channel streams. Returns list of {id, name, stream_url}."""
143
  channels = []
144
+ for ch_id, php_url in XEMTV_PHP_ENDPOINTS.items():
145
  stream_url = fetch_vtv_stream(ch_id)
146
  channels.append({
147
  'id': ch_id,
148
  'name': CHANNEL_NAMES.get(ch_id, ch_id.upper()),
149
  'stream_url': stream_url,
150
  })
151
+ return channels
152
+
153
+
154
+ # Legacy compatibility
155
+ XEMTV_CHANNELS = {v: k for k, v in CHANNEL_NAMES.items()}
156
+ CDN_STREAMS = {v: k for k, v in CDN_FALLBACK.items()}
vtv_shorts.py DELETED
@@ -1,140 +0,0 @@
1
- """
2
- VTV Nam Bộ YouTube Shorts Scraper
3
- """
4
- import requests
5
- import re
6
- import json
7
- import subprocess
8
- import html as html_lib
9
- import time
10
- import threading
11
- from xml.etree import ElementTree as ET
12
-
13
- _cache = {}
14
- _lock = threading.Lock()
15
- CACHE_TTL = 1800
16
-
17
- UA = {
18
- "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/125.0.0.0 Safari/537.36",
19
- "Accept-Language": "vi-VN,vi;q=0.9,en;q=0.8",
20
- }
21
-
22
- def _cached(key):
23
- with _lock:
24
- if key in _cache and time.time() - _cache[key]['t'] < CACHE_TTL:
25
- return _cache[key]['d']
26
- return None
27
-
28
- def _set_cache(key, data):
29
- with _lock:
30
- _cache[key] = {'t': time.time(), 'd': data}
31
-
32
- def get_channel_id(username):
33
- cached = _cached(f'ch_id_{username}')
34
- if cached:
35
- return cached
36
- channel_id = None
37
- try:
38
- url = f"https://www.youtube.com/@{username}"
39
- r = requests.get(url, headers=UA, timeout=15)
40
- if r.status_code == 200:
41
- m = re.search(r'<meta\s+property="og:url"\s+content="https://www.youtube.com/channel/(UC[^"]+)"', r.text)
42
- if m:
43
- channel_id = m.group(1)
44
- if not channel_id:
45
- m = re.search(r'"channelId":"(UC[^"]+)"', r.text)
46
- if m:
47
- channel_id = m.group(1)
48
- except Exception as e:
49
- print(f"Error getting channel ID for @{username}: {e}")
50
- if channel_id:
51
- _set_cache(f'ch_id_{username}', channel_id)
52
- return channel_id
53
-
54
- def scrape_via_rss(channel_id, max_count=100):
55
- shorts = []
56
- try:
57
- url = f"https://www.youtube.com/feeds/videos.xml?channel_id={channel_id}"
58
- r = requests.get(url, headers=UA, timeout=15)
59
- if r.status_code != 200:
60
- return shorts
61
- root = ET.fromstring(r.text)
62
- ns = {'atom': 'http://www.w3.org/2005/Atom', 'yt': 'http://www.youtube.com/xml/schemas/2015'}
63
- for entry in root.findall('atom:entry', ns)[:max_count]:
64
- title_el = entry.find('atom:title', ns)
65
- title = html_lib.unescape(title_el.text) if title_el is not None and title_el.text else ''
66
- vid_el = entry.find('yt:videoId', ns)
67
- vid = vid_el.text if vid_el is not None else ''
68
- if not vid:
69
- continue
70
- is_short = '#shorts' in title.lower() or '#short' in title.lower()
71
- link_el = entry.find('atom:link', ns)
72
- link = link_el.get('href', '') if link_el is not None else ''
73
- if '/shorts/' in link:
74
- is_short = True
75
- if is_short:
76
- shorts.append({'id': vid, 'title': title, 'img': f"https://i.ytimg.com/vi/{vid}/hqdefault.jpg", 'channel': 'vtvnambo'})
77
- except Exception as e:
78
- print(f"RSS scrape error: {e}")
79
- return shorts
80
-
81
- def scrape_via_ydlp(username, count=50):
82
- shorts = []
83
- try:
84
- url = f"https://www.youtube.com/@{username}/shorts"
85
- result = subprocess.run(
86
- ["yt-dlp", "--dump-json", "--flat-playlist", "--no-download", "--playlist-end", str(count), url],
87
- capture_output=True, text=True, timeout=90
88
- )
89
- if result.returncode == 0 and result.stdout.strip():
90
- seen_ids = set()
91
- for line in result.stdout.strip().split('\n'):
92
- line = line.strip()
93
- if not line:
94
- continue
95
- try:
96
- entry = json.loads(line)
97
- vid = entry.get('id', '')
98
- if not vid or vid in seen_ids:
99
- continue
100
- seen_ids.add(vid)
101
- title = entry.get('title', 'VTV Nam Bộ Short')
102
- shorts.append({'id': vid, 'title': title, 'img': f"https://i.ytimg.com/vi/{vid}/hqdefault.jpg", 'channel': 'vtvnambo'})
103
- except json.JSONDecodeError:
104
- continue
105
- except (subprocess.TimeoutExpired, FileNotFoundError) as e:
106
- print(f"yt-dlp error: {e}")
107
- return shorts
108
-
109
- def get_vtvnambo_shorts(max_count=50):
110
- cached = _cached('vtvnambo_shorts_v2')
111
- if cached is not None:
112
- return cached
113
- all_shorts = []
114
- seen_ids = set()
115
- channel_id = get_channel_id('vtvnambo')
116
- if channel_id:
117
- rss_shorts = scrape_via_rss(channel_id, max_count * 2)
118
- for s in rss_shorts:
119
- if s['id'] not in seen_ids:
120
- seen_ids.add(s['id'])
121
- all_shorts.append(s)
122
- if len(all_shorts) < 3:
123
- ydlp_shorts = scrape_via_ydlp('vtvnambo', max_count)
124
- for s in ydlp_shorts:
125
- if s['id'] not in seen_ids:
126
- seen_ids.add(s['id'])
127
- all_shorts.append(s)
128
- result = all_shorts[:max_count]
129
- _set_cache('vtvnambo_shorts_v2', result)
130
- return result
131
-
132
- get_vtvnamo_shorts = get_vtvnambo_shorts
133
-
134
- def get_wc_related_shorts(max_count=30):
135
- all_shorts = get_vtvnambo_shorts(max_count * 3)
136
- wc_kws = ['world cup', 'wc 2026', 'worldcup', 'fifa', 'bóng đá', 'trận đấu', 'đội tuyển']
137
- wc_shorts = [s for s in all_shorts if any(k in s.get('title', '').lower() for k in wc_kws)]
138
- if not wc_shorts:
139
- wc_shorts = all_shorts
140
- return wc_shorts[:max_count]
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
yt_scraper.py DELETED
@@ -1,235 +0,0 @@
1
- """
2
- YouTube Shorts Scraper using yt-dlp (already installed on Space)
3
- Runs yt-dlp as subprocess to extract video info and direct URLs
4
- """
5
- import subprocess
6
- import json
7
- import time
8
- import threading
9
- import os
10
- import re as re_mod
11
-
12
- _cache = {}
13
- _lock = threading.Lock()
14
- CACHE_TTL = 600 # 10 min cache
15
-
16
- def _cached(key):
17
- with _lock:
18
- if key in _cache and time.time() - _cache[key]['t'] < CACHE_TTL:
19
- return _cache[key]['d']
20
- return None
21
-
22
- def _set_cache(key, data):
23
- with _lock:
24
- _cache[key] = {'t': time.time(), 'd': data}
25
-
26
- def run_yt_dlp(args, timeout=120):
27
- """Run yt-dlp and return parsed JSON lines"""
28
- try:
29
- result = subprocess.run(
30
- ["yt-dlp"] + args,
31
- capture_output=True, text=True, timeout=timeout
32
- )
33
- if result.returncode != 0 and not result.stdout.strip():
34
- print(f"yt-dlp error: {result.stderr[:200]}")
35
- return []
36
- lines = result.stdout.strip().split('\n')
37
- items = []
38
- for line in lines:
39
- line = line.strip()
40
- if not line:
41
- continue
42
- try:
43
- items.append(json.loads(line))
44
- except json.JSONDecodeError:
45
- continue
46
- return items
47
- except subprocess.TimeoutExpired:
48
- print("yt-dlp timeout")
49
- return []
50
- except FileNotFoundError:
51
- print("yt-dlp not found!")
52
- return []
53
- except Exception as e:
54
- print(f"yt-dlp exception: {e}")
55
- return []
56
-
57
- def get_channel_shorts_via_playlist(channel_username, max_count=100):
58
- """Get shorts from channel's shorts page using yt-dlp"""
59
- shorts = []
60
-
61
- # Method 1: Fetch from /shorts page
62
- url = f"https://www.youtube.com/@{channel_username}/shorts"
63
- items = run_yt_dlp([
64
- "--dump-json",
65
- "--flat-playlist",
66
- "--no-download",
67
- "--playlist-end", str(max_count),
68
- "--no-check-certificates",
69
- "--user-agent", "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36",
70
- url
71
- ], timeout=90)
72
-
73
- seen_ids = set()
74
- for item in items:
75
- vid = item.get('id', '')
76
- if not vid or vid in seen_ids:
77
- continue
78
- seen_ids.add(vid)
79
-
80
- title = item.get('title', 'VTV Nam Bộ Short')
81
- duration = item.get('duration', 0) or 0
82
-
83
- # Only include actual shorts (<= 120s to be safe)
84
- if duration <= 120 or '#shorts' in title.lower() or '#short' in title.lower():
85
- shorts.append({
86
- 'id': vid,
87
- 'title': title,
88
- 'duration': duration,
89
- 'channel': channel_username,
90
- 'img': f"https://i.ytimg.com/vi/{vid}/hqdefault.jpg",
91
- })
92
-
93
- return shorts
94
-
95
- def get_channel_videos_filter_shorts(channel_username, max_count=200):
96
- """Get all videos from /videos page and filter for shorts by duration"""
97
- shorts = []
98
-
99
- url = f"https://www.youtube.com/@{channel_username}/videos"
100
- items = run_yt_dlp([
101
- "--dump-json",
102
- "--flat-playlist",
103
- "--no-download",
104
- "--playlist-end", str(max_count),
105
- "--match-filter", "duration > 0 and duration <= 120",
106
- "--no-check-certificates",
107
- "--user-agent", "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36",
108
- url
109
- ], timeout=120)
110
-
111
- seen_ids = set()
112
- for item in items:
113
- vid = item.get('id', '')
114
- if not vid or vid in seen_ids:
115
- continue
116
- seen_ids.add(vid)
117
-
118
- title = item.get('title', 'VTV Nam Bộ Short')
119
- duration = item.get('duration', 0) or 0
120
-
121
- shorts.append({
122
- 'id': vid,
123
- 'title': title,
124
- 'duration': duration,
125
- 'channel': channel_username,
126
- 'img': f"https://i.ytimg.com/vi/{vid}/hqdefault.jpg",
127
- })
128
-
129
- return shorts
130
-
131
- def get_shorts_with_direct_url(video_ids):
132
- """Get direct video download URLs for given video IDs"""
133
- results = []
134
-
135
- for vid in video_ids[:20]: # Limit to avoid timeout
136
- try:
137
- url = f"https://www.youtube.com/shorts/{vid}"
138
- items = run_yt_dlp([
139
- "--dump-json",
140
- "--no-download",
141
- "--no-check-certificates",
142
- "--format", "best[filesize<10M]/best",
143
- url
144
- ], timeout=30)
145
-
146
- if items:
147
- info = items[0]
148
- direct_url = info.get('url', '')
149
- if not direct_url:
150
- # Try to get from formats
151
- formats = info.get('formats', [])
152
- for f in formats:
153
- if f.get('vcodec') != 'none' and f.get('acodec') != 'none':
154
- direct_url = f.get('url', '')
155
- break
156
-
157
- if direct_url:
158
- results.append({
159
- 'id': vid,
160
- 'title': info.get('title', ''),
161
- 'direct_url': direct_url,
162
- 'thumbnail': info.get('thumbnail', f"https://i.ytimg.com/vi/{vid}/hqdefault.jpg"),
163
- 'duration': info.get('duration', 0),
164
- })
165
- except Exception as e:
166
- print(f"Error getting URL for {vid}: {e}")
167
-
168
- return results
169
-
170
- def get_vtvnambo_shorts(max_count=50):
171
- """Get all shorts from VTV Nam Bộ using yt-dlp"""
172
- cached = _cached('vtvnambo_shorts_yt')
173
- if cached is not None:
174
- return cached
175
-
176
- all_shorts = []
177
- seen_ids = set()
178
-
179
- # Method 1: /shorts page
180
- print(f"[yt-dlp] Fetching /shorts page...")
181
- shorts_page = get_channel_shorts_via_playlist('vtvnambo', max_count)
182
- for s in shorts_page:
183
- if s['id'] not in seen_ids:
184
- seen_ids.add(s['id'])
185
- all_shorts.append(s)
186
- print(f"[yt-dlp] /shorts page: {len(shorts_page)} shorts")
187
-
188
- # Method 2: /videos page with duration filter
189
- if len(all_shorts) < 5:
190
- print(f"[yt-dlp] Fetching /videos page with filter...")
191
- videos_filtered = get_channel_videos_filter_shorts('vtvnambo', max_count * 2)
192
- for s in videos_filtered:
193
- if s['id'] not in seen_ids:
194
- seen_ids.add(s['id'])
195
- all_shorts.append(s)
196
- print(f"[yt-dlp] /videos filter: {len(videos_filtered)} shorts")
197
-
198
- result = all_shorts[:max_count]
199
- _set_cache('vtvnambo_shorts_yt', result)
200
- print(f"[yt-dlp] Total: {len(result)} shorts from VTV Nam Bộ")
201
- return result
202
-
203
- def get_wc_related_shorts(max_count=30):
204
- """Get World Cup / football related shorts"""
205
- all_shorts = get_vtvnambo_shorts(max_count * 3)
206
-
207
- wc_kws = [
208
- 'world cup', 'wc 2026', 'worldcup', 'fifa', 'bóng đá',
209
- 'trận đấu', 'đội tuyển', 'tuyển', 'vòng loại',
210
- 'khoảnh khắc', 'highlights', 'bàn thắng', 'goal',
211
- 'kết quả', 'tỉ số', 'việt nam', 'vn',
212
- 'ngoại hạng', 'premier league', 'champions league',
213
- 'laliga', 'serie a', 'bundesliga', 'ligue 1',
214
- 'copa', 'europa', 'c1', 'c2',
215
- 'messi', 'ronaldo', 'neymar', 'mbappe', 'haaland',
216
- 'v-league', 'vleague', 'bóng đá việt',
217
- 'đội bóng', 'hlv', 'huấn luyện viên',
218
- 'chuyển nhượng', 'transfer',
219
- 'asian cup', 'aff cup', 'sea games',
220
- 'olympic', 'u23', 'u20', 'u17',
221
- ]
222
-
223
- wc_shorts = []
224
- for s in all_shorts:
225
- tl = s.get('title', '').lower()
226
- if any(k in tl for k in wc_kws):
227
- wc_shorts.append(s)
228
-
229
- if not wc_shorts:
230
- wc_shorts = all_shorts
231
-
232
- return wc_shorts[:max_count]
233
-
234
- # Aliases
235
- get_vtvnamo_shorts = get_vtvnambo_shorts
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
yt_scraper_fixed.py DELETED
@@ -1,162 +0,0 @@
1
- """
2
- YouTube Shorts Scraper using yt-dlp (already installed on Space)
3
- Optimized for fast load with long cache + fallback
4
- """
5
- import subprocess
6
- import json
7
- import time
8
- import threading
9
- import os
10
- import re as re_mod
11
-
12
- _cache = {}
13
- _lock = threading.Lock()
14
- CACHE_TTL = 1800 # 30 min cache - longer to reduce timeout issues
15
-
16
- def _get_cached(key):
17
- """Get cached data if still valid"""
18
- with _lock:
19
- if key in _cache:
20
- entry = _cache[key]
21
- if time.time() - entry['t'] < CACHE_TTL:
22
- return entry['d']
23
- return None
24
-
25
- def _set_cached(key, data):
26
- """Set cache with timestamp"""
27
- with _lock:
28
- _cache[key] = {'t': time.time(), 'd': data}
29
-
30
- def run_yt_dlp(args, timeout=45):
31
- """Run yt-dlp and return parsed JSON lines - with shorter timeout"""
32
- try:
33
- result = subprocess.run(
34
- ["yt-dlp"] + args,
35
- capture_output=True, text=True, timeout=timeout
36
- )
37
- if result.returncode != 0 and not result.stdout.strip():
38
- return []
39
- lines = result.stdout.strip().split('\n')
40
- items = []
41
- for line in lines:
42
- line = line.strip()
43
- if not line:
44
- continue
45
- try:
46
- items.append(json.loads(line))
47
- except json.JSONDecodeError:
48
- continue
49
- return items
50
- except subprocess.TimeoutExpired:
51
- print("yt-dlp timeout (this is OK - using fallback)")
52
- return []
53
- except FileNotFoundError:
54
- print("yt-dlp not found - using fallback")
55
- return []
56
- except Exception as e:
57
- print(f"yt-dlp exception: {e}")
58
- return []
59
-
60
- def get_channel_shorts_fast(channel_username, max_count=25):
61
- """Get shorts fast - prioritize /shorts page only to avoid timeout"""
62
- shorts = []
63
-
64
- url = f"https://www.youtube.com/@{channel_username}/shorts"
65
- items = run_yt_dlp([
66
- "--dump-json",
67
- "--flat-playlist",
68
- "--no-download",
69
- "--playlist-end", str(max_count),
70
- "--no-check-certificates",
71
- "--quiet", # Reduce output for speed
72
- "--no-warnings",
73
- url
74
- ], timeout=35) # Increased timeout but still reasonable
75
-
76
- seen_ids = set()
77
- for item in items:
78
- vid = item.get('id', '')
79
- if not vid or vid in seen_ids:
80
- continue
81
- seen_ids.add(vid)
82
-
83
- title = item.get('title', f'{channel_username} Short')
84
-
85
- shorts.append({
86
- 'id': vid,
87
- 'title': title,
88
- 'channel': channel_username,
89
- 'img': f"https://i.ytimg.com/vi/{vid}/hqdefault.jpg",
90
- })
91
-
92
- if len(shorts) >= max_count:
93
- break
94
-
95
- return shorts
96
-
97
- def get_dantri_shorts(max_count=25):
98
- """Get Dantri shorts - fast, no fallback needed, cache for 30min"""
99
- cached = _get_cached('dantri_shorts_yt')
100
- if cached is not None:
101
- return cached
102
-
103
- shorts = get_channel_shorts_fast('baodantri7941', max_count)
104
-
105
- if shorts:
106
- _set_cached('dantri_shorts_yt', shorts)
107
- return shorts
108
-
109
- # Fallback to static list if scrape fails
110
- return [
111
- {"id":"Lu_iCQ5YwNM","title":"Công an lập hồ sơ xử lý người phụ nữ chửi bới tát nam tài xế ô tô ở Hà Nội","channel":"baodantri7941"},
112
- {"id":"CwWvijF8BOA","title":"Chú rể Ninh Bình bật khóc nhận món quà bí mật người cha","channel":"baodantri7941"},
113
- ]
114
-
115
- def get_skds_shorts(max_count=25):
116
- """Get SKĐS shorts - fast, no fallback needed, cache for 30min"""
117
- cached = _get_cached('skds_shorts_yt')
118
- if cached is not None:
119
- return cached
120
-
121
- shorts = get_channel_shorts_fast('baosuckhoedoisongboyte', max_count)
122
-
123
- if shorts:
124
- _set_cached('skds_shorts_yt', shorts)
125
- return shorts
126
-
127
- # Fallback to static list if scrape fails
128
- return [
129
- {"id":"7Pd6vZ2Lz1M","title":"Hành động ấm lòng của người đàn ông tìm kiếm 5 học sinh tử vong","channel":"baosuckhoedoisongboyte"},
130
- {"id":"SlHLt_ZyPiE","title":"Xử phạt người đàn ông xóa số điện thoại cứu hộ trên cao tốc Bắc Nam","channel":"baosuckhoedoisongboyte"},
131
- ]
132
-
133
- def get_dantri_skds_shorts(max_count=50):
134
- """Get interleaved Dantri + SKĐS shorts - optimized with separate caching"""
135
- # Get each channel's shorts separately (allows partial fallback)
136
- dantri = get_dantri_shorts(max_count // 2 + 10)
137
- skds = get_skds_shorts(max_count // 2 + 10)
138
-
139
- # Interleave them
140
- result = []
141
- seen = set()
142
- i, j = 0, 0
143
-
144
- while (i < len(dantri) or j < len(skds)) and len(result) < max_count:
145
- if i < len(dantri):
146
- item = dantri[i]
147
- if item.get('id') not in seen:
148
- seen.add(item.get('id'))
149
- result.append(item)
150
- i += 1
151
- if j < len(skds):
152
- item = skds[j]
153
- if item.get('id') not in seen:
154
- seen.add(item.get('id'))
155
- result.append(item)
156
- j += 1
157
-
158
- return result
159
-
160
- # For backward compatibility
161
- get_vtvnambo_shorts = get_dantri_skds_shorts
162
- get_wc_related_shorts = get_dantri_skds_shorts