Spaces:
Running
Running
Revert to c726397: restore 100% to target commit
Browse files- ai_ext.py +1 -134
- app_v2_entry.py +1341 -48
ai_ext.py
CHANGED
|
@@ -197,137 +197,4 @@ def _fallback_summary_from_prompt(prompt: str, max_units: int = 6) -> str:
|
|
| 197 |
break
|
| 198 |
if chunks:
|
| 199 |
return "\n".join("• " + c for c in chunks)
|
| 200 |
-
return "• Không có đủ nội dung để tóm tắt."
|
| 201 |
-
|
| 202 |
-
|
| 203 |
-
# ===== URL scraping & article processing =====
|
| 204 |
-
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"}
|
| 205 |
-
|
| 206 |
-
try:
|
| 207 |
-
_shorts_base = "/data" if os.path.isdir("/data") else os.path.join(os.path.dirname(os.path.abspath(__file__)), "data")
|
| 208 |
-
except Exception:
|
| 209 |
-
_shorts_base = os.path.join(os.path.dirname(os.path.abspath(__file__)), "data")
|
| 210 |
-
SHORTS_DIR = os.path.join(_shorts_base, "ai_shorts")
|
| 211 |
-
os.makedirs(SHORTS_DIR, exist_ok=True)
|
| 212 |
-
|
| 213 |
-
import random as _random2
|
| 214 |
-
from datetime import datetime, timezone, timedelta
|
| 215 |
-
_VN_TZ = timezone(timedelta(hours=7))
|
| 216 |
-
|
| 217 |
-
|
| 218 |
-
def _safe_name(filename: str) -> str:
|
| 219 |
-
"""Sanitize filename."""
|
| 220 |
-
return re.sub(r"[^a-zA-Z0-9_.-]", "_", filename)[:120]
|
| 221 |
-
|
| 222 |
-
|
| 223 |
-
def pollinations_image_url(topic: str) -> str:
|
| 224 |
-
"""Generate a placeholder image URL via Pollinations."""
|
| 225 |
-
try:
|
| 226 |
-
return "https://image.pollinations.ai/prompt/" + quote("Vietnamese editorial illustration, " + topic, safe="") + "?width=1024&height=576&nologo=true"
|
| 227 |
-
except Exception:
|
| 228 |
-
return ""
|
| 229 |
-
|
| 230 |
-
|
| 231 |
-
def _download_image(url: str, fallback_title: str, out_path: str) -> str:
|
| 232 |
-
"""Download an image from URL or create a placeholder."""
|
| 233 |
-
if url:
|
| 234 |
-
try:
|
| 235 |
-
r = requests.get(url, headers=HEADERS, timeout=15)
|
| 236 |
-
if r.status_code == 200 and len(r.content) > 1200:
|
| 237 |
-
os.makedirs(os.path.dirname(out_path), exist_ok=True)
|
| 238 |
-
with open(out_path, "wb") as f:
|
| 239 |
-
f.write(r.content)
|
| 240 |
-
return out_path
|
| 241 |
-
except Exception:
|
| 242 |
-
pass
|
| 243 |
-
# Fallback: create a placeholder image
|
| 244 |
-
try:
|
| 245 |
-
from PIL import Image, ImageDraw, ImageFont
|
| 246 |
-
img = Image.new("RGB", (1080, 760), (24, 24, 24))
|
| 247 |
-
draw = ImageDraw.Draw(img)
|
| 248 |
-
try:
|
| 249 |
-
font = ImageFont.truetype("/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf", 48)
|
| 250 |
-
except Exception:
|
| 251 |
-
font = None
|
| 252 |
-
text = (fallback_title or "VNEWS")[:40]
|
| 253 |
-
try:
|
| 254 |
-
bbox = draw.textbbox((0, 0), text, font=font)
|
| 255 |
-
tw = bbox[2] - bbox[0]
|
| 256 |
-
except Exception:
|
| 257 |
-
tw = len(text) * 24
|
| 258 |
-
draw.text(((1080 - tw) // 2, 330), text, fill=(255, 255, 255), font=font)
|
| 259 |
-
os.makedirs(os.path.dirname(out_path), exist_ok=True)
|
| 260 |
-
img.save(out_path, quality=90)
|
| 261 |
-
return out_path
|
| 262 |
-
except Exception:
|
| 263 |
-
return out_path
|
| 264 |
-
|
| 265 |
-
|
| 266 |
-
def scrape_any_url(url: str) -> dict:
|
| 267 |
-
"""Scrape article content from any URL."""
|
| 268 |
-
if not url or not url.startswith("http"):
|
| 269 |
-
return {"title": "", "text": "", "summary": "", "image": "", "og_image": "", "via": ""}
|
| 270 |
-
try:
|
| 271 |
-
r = requests.get(url, headers=HEADERS, timeout=15, allow_redirects=True)
|
| 272 |
-
if r.status_code != 200 or not r.text:
|
| 273 |
-
return {"title": "", "text": "", "summary": "", "image": "", "og_image": "", "via": _domain(url)}
|
| 274 |
-
r.encoding = "utf-8"
|
| 275 |
-
soup = BeautifulSoup(r.text, "lxml")
|
| 276 |
-
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", ".breadcrumb"]):
|
| 277 |
-
tag.decompose()
|
| 278 |
-
title = ""
|
| 279 |
-
ogt = soup.find("meta", property="og:title")
|
| 280 |
-
if ogt:
|
| 281 |
-
title = ogt.get("content", "")
|
| 282 |
-
h1 = soup.find("h1")
|
| 283 |
-
if not title and h1:
|
| 284 |
-
title = h1.get_text(strip=True)
|
| 285 |
-
if not title:
|
| 286 |
-
t = soup.find("title")
|
| 287 |
-
if t:
|
| 288 |
-
title = t.get_text(strip=True)
|
| 289 |
-
og_image = ""
|
| 290 |
-
ogi = soup.find("meta", property="og:image")
|
| 291 |
-
if ogi:
|
| 292 |
-
og_image = ogi.get("content", "")
|
| 293 |
-
if og_image.startswith("//"):
|
| 294 |
-
og_image = "https:" + og_image
|
| 295 |
-
summary = ""
|
| 296 |
-
ogd = soup.find("meta", property="og:description") or soup.find("meta", attrs={"name": "description"})
|
| 297 |
-
if ogd:
|
| 298 |
-
summary = ogd.get("content", "")[:500]
|
| 299 |
-
body_text = []
|
| 300 |
-
for sel in ["article", ".singular-content", ".detail-content", ".fck_detail", ".content-detail", ".knc-content", "main", ".cms-body", ".article__body", ".post-content", ".entry-content"]:
|
| 301 |
-
el = soup.select_one(sel)
|
| 302 |
-
if el and len(el.find_all("p")) >= 2:
|
| 303 |
-
for p in el.find_all("p"):
|
| 304 |
-
t = _clean_text(p.get_text(strip=True))
|
| 305 |
-
if t and len(t) > 30:
|
| 306 |
-
body_text.append(t)
|
| 307 |
-
break
|
| 308 |
-
if not body_text and soup.body:
|
| 309 |
-
for p in soup.body.find_all("p"):
|
| 310 |
-
t = _clean_text(p.get_text(strip=True))
|
| 311 |
-
if t and len(t) > 30:
|
| 312 |
-
body_text.append(t)
|
| 313 |
-
text = "\n".join(body_text)
|
| 314 |
-
return {"title": _clean_text(title), "text": text, "summary": _clean_text(summary), "image": og_image, "og_image": og_image, "via": _domain(url), "url": url}
|
| 315 |
-
except Exception as e:
|
| 316 |
-
return {"title": "", "text": "", "summary": "", "image": "", "og_image": "", "via": _domain(url)}
|
| 317 |
-
|
| 318 |
-
|
| 319 |
-
def make_post(title: str, text: str, img: str, url: str, kind: str = "auto", sources: list = None) -> dict:
|
| 320 |
-
"""Create a wall post dict."""
|
| 321 |
-
import random as _r2
|
| 322 |
-
now = int(time.time() * 1000)
|
| 323 |
-
return {
|
| 324 |
-
"id": str(now) + str(_r2.randint(100, 999)),
|
| 325 |
-
"title": (title or "Bài viết")[:200],
|
| 326 |
-
"text": (text or "")[:5000],
|
| 327 |
-
"img": img or "",
|
| 328 |
-
"url": url or "",
|
| 329 |
-
"kind": kind or "auto",
|
| 330 |
-
"sources": sources or [],
|
| 331 |
-
"created": now,
|
| 332 |
-
"created_str": datetime.now(_VN_TZ).strftime("%H:%M %d/%m/%Y"),
|
| 333 |
-
}
|
|
|
|
| 197 |
break
|
| 198 |
if chunks:
|
| 199 |
return "\n".join("• " + c for c in chunks)
|
| 200 |
+
return "• Không có đủ nội dung để tóm tắt."
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
app_v2_entry.py
CHANGED
|
@@ -226,7 +226,7 @@ _STOP=set('và của các những một được trong với cho tại sau trư
|
|
| 226 |
def _has_kw(topic,title):
|
| 227 |
tl=topic.lower();tt=(title or'').lower()
|
| 228 |
if tl in tt:return True
|
| 229 |
-
words=[w for w in re.findall(r'[A-Za-zÀ-ỹ0-9]+',tl) if len(w)>2 and w
|
| 230 |
if not words:return True
|
| 231 |
return any(w in tt for w in words)
|
| 232 |
|
|
@@ -498,9 +498,9 @@ def _get_hot_topics():
|
|
| 498 |
is_dup=any(len(set(e.split())&set(key.split()))/max(len(set(e.split())),len(set(key.split())),1)>0.6 for e in seen)
|
| 499 |
if is_dup:continue
|
| 500 |
seen.add(key);topics.append({'label':'#'+re.sub(r'\s+','',display[key].title()),'topic':display[key],'count':count})
|
| 501 |
-
if len
|
| 502 |
for kw in['World Cup 2026','Kinh tế Việt Nam','Bóng đá châu Âu','Công nghệ AI','Giá vàng','Thời tiết']:
|
| 503 |
-
if len
|
| 504 |
if not any(kw.lower() in s for s in seen):topics.append({'label':'#'+re.sub(r'\s+','',kw.title()),'topic':kw,'count':0})
|
| 505 |
_hot_cache.update({'t':now,'d':topics[:24]});return topics[:24]
|
| 506 |
|
|
@@ -810,7 +810,7 @@ def _wl(eid:int):return JSONResponse(scrape_lineups(eid))
|
|
| 810 |
@app.get('/api/wc2026/match/{eid}')
|
| 811 |
def _wm(eid:int):return JSONResponse(scrape_match_detail(eid))
|
| 812 |
|
| 813 |
-
DATA_DIR='/data' if os.path.isdir('/data') else os.path.join(os.path.dirname(os.path.abspath(__file__)),
|
| 814 |
os.makedirs(DATA_DIR,exist_ok=True)
|
| 815 |
IF=os.path.join(DATA_DIR,'interactions_v2.json')
|
| 816 |
CF=os.path.join(DATA_DIR,'comments_v2.json')
|
|
@@ -821,54 +821,43 @@ os.makedirs(WALL_VIDEO_DIR,exist_ok=True)
|
|
| 821 |
_il=threading.Lock();_cl=threading.Lock();_wl_lock=threading.Lock()
|
| 822 |
def _lj(p):
|
| 823 |
try:
|
| 824 |
-
p=str(p)
|
| 825 |
if os.path.exists(p):return json.load(open(p,'r',encoding='utf-8'))
|
| 826 |
except:pass
|
| 827 |
-
return
|
| 828 |
def _sj(p,d):
|
| 829 |
-
try:
|
| 830 |
-
p=str(p)
|
| 831 |
-
os.makedirs(os.path.dirname(p),exist_ok=True)
|
| 832 |
-
tmp=p+'.tmp'
|
| 833 |
-
with open(tmp,'w',encoding='utf-8') as f:
|
| 834 |
-
json.dump(d,f,ensure_ascii=False)
|
| 835 |
-
os.replace(tmp,p)
|
| 836 |
except:pass
|
| 837 |
|
| 838 |
-
|
| 839 |
-
|
| 840 |
-
|
|
|
|
|
|
|
| 841 |
|
| 842 |
-
|
| 843 |
-
|
| 844 |
-
|
| 845 |
-
posts = posts[:200]
|
| 846 |
-
_save_wall_posts(posts)
|
| 847 |
-
return posts
|
| 848 |
|
| 849 |
-
@app.get('/api/
|
| 850 |
-
def
|
| 851 |
-
return JSONResponse({
|
| 852 |
|
| 853 |
-
@app.post(
|
| 854 |
-
def
|
| 855 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 856 |
|
| 857 |
-
|
| 858 |
-
|
| 859 |
-
|
| 860 |
-
|
| 861 |
-
|
| 862 |
-
|
| 863 |
-
|
| 864 |
-
if pid == post_id_s or pid.startswith(post_id_s):
|
| 865 |
-
post = p
|
| 866 |
-
break
|
| 867 |
-
if not post:
|
| 868 |
-
return JSONResponse({'error': 'Không tìm thấy bài viết', 'post_id': post_id}, status_code=404)
|
| 869 |
-
if post.get('video'):
|
| 870 |
-
return JSONResponse({'video': post['video'], 'post': post})
|
| 871 |
-
return JSONResponse({'error': 'Chưa có video cho bài này. Vui lòng upload video trước.'}, status_code=409)
|
| 872 |
|
| 873 |
@app.get('/api/wall')
|
| 874 |
def api_wall():
|
|
@@ -1004,7 +993,7 @@ _UNIQUE_CHARS = {
|
|
| 1004 |
_STOPWORDS = {
|
| 1005 |
'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'},
|
| 1006 |
'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'},
|
| 1007 |
-
'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', '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'},
|
| 1008 |
'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'},
|
| 1009 |
}
|
| 1010 |
|
|
@@ -1056,9 +1045,1313 @@ _EMOTION_KEYWORDS = {
|
|
| 1056 |
'vi': ['vui', 'hạnh phúc', 'tuyệt vời', 'tuyệt', 'ý nghĩa', 'đẹp', 'thích', 'yêu', 'vui vẻ', 'hân hoan', 'phấn khích', 'chiến thắng', 'thành công'],
|
| 1057 |
},
|
| 1058 |
'sad': {
|
| 1059 |
-
'en': ['sad', 'unhappy', 'terrible', 'awful', 'horrible', 'miserable', 'depressed', '
|
| 1060 |
-
'pt': ['triste', 'infeliz', 'terrível', '
|
| 1061 |
-
'es': ['triste', 'infeliz', 'terrible', '
|
| 1062 |
-
'vi': ['buồn', '
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1063 |
},
|
| 1064 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 226 |
def _has_kw(topic,title):
|
| 227 |
tl=topic.lower();tt=(title or'').lower()
|
| 228 |
if tl in tt:return True
|
| 229 |
+
words=[w for w in re.findall(r'[A-Za-zÀ-ỹ0-9]+',tl) if len(w)>2 and w not in _STOP]
|
| 230 |
if not words:return True
|
| 231 |
return any(w in tt for w in words)
|
| 232 |
|
|
|
|
| 498 |
is_dup=any(len(set(e.split())&set(key.split()))/max(len(set(e.split())),len(set(key.split())),1)>0.6 for e in seen)
|
| 499 |
if is_dup:continue
|
| 500 |
seen.add(key);topics.append({'label':'#'+re.sub(r'\s+','',display[key].title()),'topic':display[key],'count':count})
|
| 501 |
+
if len(topics)>=20:break
|
| 502 |
for kw in['World Cup 2026','Kinh tế Việt Nam','Bóng đá châu Âu','Công nghệ AI','Giá vàng','Thời tiết']:
|
| 503 |
+
if len(topics)>=24:break
|
| 504 |
if not any(kw.lower() in s for s in seen):topics.append({'label':'#'+re.sub(r'\s+','',kw.title()),'topic':kw,'count':0})
|
| 505 |
_hot_cache.update({'t':now,'d':topics[:24]});return topics[:24]
|
| 506 |
|
|
|
|
| 810 |
@app.get('/api/wc2026/match/{eid}')
|
| 811 |
def _wm(eid:int):return JSONResponse(scrape_match_detail(eid))
|
| 812 |
|
| 813 |
+
DATA_DIR='/data' if os.path.isdir('/data') else os.path.join(os.path.dirname(os.path.abspath(__file__)),'data')
|
| 814 |
os.makedirs(DATA_DIR,exist_ok=True)
|
| 815 |
IF=os.path.join(DATA_DIR,'interactions_v2.json')
|
| 816 |
CF=os.path.join(DATA_DIR,'comments_v2.json')
|
|
|
|
| 821 |
_il=threading.Lock();_cl=threading.Lock();_wl_lock=threading.Lock()
|
| 822 |
def _lj(p):
|
| 823 |
try:
|
|
|
|
| 824 |
if os.path.exists(p):return json.load(open(p,'r',encoding='utf-8'))
|
| 825 |
except:pass
|
| 826 |
+
return{}
|
| 827 |
def _sj(p,d):
|
| 828 |
+
try:open(p+'.tmp','w',encoding='utf-8').write(json.dumps(d,ensure_ascii=False));os.replace(p+'.tmp',p)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 829 |
except:pass
|
| 830 |
|
| 831 |
+
@app.post('/api/v2/interact')
|
| 832 |
+
async def _int(request:Request):
|
| 833 |
+
b=await request.json();v=str(b.get('id','')).strip();t=str(b.get('type','')).strip()
|
| 834 |
+
if not v or t not in('view','like'):return JSONResponse({'error':'x'},status_code=400)
|
| 835 |
+
with _il:db=_lj(IF);db.setdefault(v,{'views':0,'likes':0,'comments':0});db[v][t+'s']+=1;_sj(IF,db);return JSONResponse(db[v])
|
| 836 |
|
| 837 |
+
@app.get('/api/v2/interactions')
|
| 838 |
+
def _gi(id:str=Query(...)):
|
| 839 |
+
with _il:return JSONResponse(_lj(IF).get(id.strip(),{'views':0,'likes':0,'comments':0}))
|
|
|
|
|
|
|
|
|
|
| 840 |
|
| 841 |
+
@app.get('/api/v2/comments')
|
| 842 |
+
def _gc(id:str=Query(...)):
|
| 843 |
+
with _cl:return JSONResponse({'comments':_lj(CF).get(id.strip(),[])})
|
| 844 |
|
| 845 |
+
@app.post('/api/v2/comment')
|
| 846 |
+
async def _pc(request:Request):
|
| 847 |
+
b=await request.json();v=str(b.get('id','')).strip();tx=str(b.get('text','')).strip()[:500]
|
| 848 |
+
if not v or not tx:return JSONResponse({'error':'x'},status_code=400)
|
| 849 |
+
c={'text':tx,'time':time.strftime('%H:%M %d/%m',time.localtime()),'ts':int(time.time())}
|
| 850 |
+
with _cl:db=_lj(CF);db.setdefault(v,[]);db[v].append(c);db[v]=db[v][-200:];_sj(CF,db);cms=db[v]
|
| 851 |
+
with _il:idb=_lj(IF);idb.setdefault(v,{'views':0,'likes':0,'comments':0});idb[v]['comments']=len(cms);_sj(IF,idb)
|
| 852 |
+
return JSONResponse({'comments':cms})
|
| 853 |
|
| 854 |
+
def _load_wall_posts():
|
| 855 |
+
with _wl_lock:
|
| 856 |
+
return _lj(WALL_FILE)
|
| 857 |
+
|
| 858 |
+
def _save_wall_posts(posts):
|
| 859 |
+
with _wl_lock:
|
| 860 |
+
_sj(WALL_FILE, posts)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 861 |
|
| 862 |
@app.get('/api/wall')
|
| 863 |
def api_wall():
|
|
|
|
| 993 |
_STOPWORDS = {
|
| 994 |
'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'},
|
| 995 |
'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'},
|
| 996 |
+
'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'},
|
| 997 |
'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'},
|
| 998 |
}
|
| 999 |
|
|
|
|
| 1045 |
'vi': ['vui', 'hạnh phúc', 'tuyệt vời', 'tuyệt', 'ý nghĩa', 'đẹp', 'thích', 'yêu', 'vui vẻ', 'hân hoan', 'phấn khích', 'chiến thắng', 'thành công'],
|
| 1046 |
},
|
| 1047 |
'sad': {
|
| 1048 |
+
'en': ['sad', 'unhappy', 'terrible', 'awful', 'horrible', 'miserable', 'depressed', 'grief', 'sorrow', 'tragic', 'unfortunate', 'painful', 'death', 'die', 'kill'],
|
| 1049 |
+
'pt': ['triste', 'infeliz', 'terrível', 'horrível', 'miserável', 'deprimido', 'dor', 'trágico', 'infelizmente', 'penoso', 'morte', 'morrer'],
|
| 1050 |
+
'es': ['triste', 'infeliz', 'terrible', 'horrible', 'miserable', 'deprimido', 'dolor', 'trágico', 'desafortunado', 'penoso', 'muerte', 'morir'],
|
| 1051 |
+
'vi': ['buồn', 'không vui', 'tồi tệ', 'kinh khủng', 'đau khổ', 'đau buồn', 'bi thương', 'khốn nạn', 'đau đớn', 'thảm họa', 'chết', 'mất'],
|
| 1052 |
+
},
|
| 1053 |
+
'excited': {
|
| 1054 |
+
'en': ['excited', 'thrilling', 'amazing', 'wow', 'incredible', 'unbelievable', 'awesome', 'exhilarating', 'electrifying', 'breathtaking', 'breakthrough', 'record'],
|
| 1055 |
+
'pt': ['animado', 'emocionante', 'incrível', 'impressionante', 'sensacional', 'eletrizante', 'empolgante', 'recorde'],
|
| 1056 |
+
'es': ['emocionante', 'increíble', 'impresionante', 'sensacional', 'electrizante', 'emocionado', 'entusiasmado', 'récord'],
|
| 1057 |
+
'vi': ['hào hứng', 'phấn khích', 'thú vị', 'tuyệt cú mèo', 'đỉnh cao', 'ngoạn mục', 'sục sôi', 'kỷ lục', 'đột phá'],
|
| 1058 |
+
},
|
| 1059 |
+
'humorous': {
|
| 1060 |
+
'en': ['funny', 'hilarious', 'joke', 'laugh', 'comedy', 'humor', 'amusing', 'witty', 'sarcastic', 'ironic', 'ridiculous', 'absurd', 'lol', 'haha'],
|
| 1061 |
+
'pt': ['engraçado', 'hilário', 'piada', 'rir', 'comédia', 'humor', 'divertido', 'irônico', 'ridículo', 'absurdo', 'kkk'],
|
| 1062 |
+
'es': ['gracioso', 'hilarante', 'broma', 'risa', 'comedia', 'humor', 'divertido', 'irónico', 'ridículo', 'absurdo', 'jaja'],
|
| 1063 |
+
'vi': ['hài hước', 'buồn cười', 'đùa', 'cười', 'hài', 'vui nhộn', 'hóm hỉnh', 'mỉa mai', 'lố bịch', 'vô lý', 'haha'],
|
| 1064 |
+
},
|
| 1065 |
+
'serious': {
|
| 1066 |
+
'en': ['serious', 'critical', 'important', 'urgent', 'severe', 'grave', 'significant', 'crucial', 'vital', 'essential', 'alarming', 'concerning', 'crisis', 'war', 'conflict'],
|
| 1067 |
+
'pt': ['sério', 'crítico', 'importante', 'urgente', 'grave', 'significativo', 'crucial', 'vital', 'essencial', 'preocupante', 'crise', 'guerra', 'conflito'],
|
| 1068 |
+
'es': ['serio', 'crítico', 'importante', 'urgente', 'grave', 'significativo', 'crucial', 'vital', 'esencial', 'preocupante', 'crisis', 'guerra', 'conflicto'],
|
| 1069 |
+
'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'],
|
| 1070 |
+
},
|
| 1071 |
+
}
|
| 1072 |
+
|
| 1073 |
+
def detect_emotion(text, language='vietnamese'):
|
| 1074 |
+
"""Detect emotion from text using keyword matching."""
|
| 1075 |
+
if not text:
|
| 1076 |
+
return 'neutral'
|
| 1077 |
+
text_lower = text.lower()
|
| 1078 |
+
|
| 1079 |
+
scores = {}
|
| 1080 |
+
for emotion, lang_keywords in _EMOTION_KEYWORDS.items():
|
| 1081 |
+
keywords = lang_keywords.get(language, lang_keywords.get('en', []))
|
| 1082 |
+
score = sum(1 for kw in keywords if kw in text_lower)
|
| 1083 |
+
scores[emotion] = score
|
| 1084 |
+
|
| 1085 |
+
if max(scores.values()) == 0:
|
| 1086 |
+
return 'neutral'
|
| 1087 |
+
|
| 1088 |
+
return max(scores, key=scores.get)
|
| 1089 |
+
|
| 1090 |
+
def detect_language_and_emotion(title, text):
|
| 1091 |
+
"""Detect both language and emotion from article content."""
|
| 1092 |
+
combined = f"{title} {text}"
|
| 1093 |
+
lang = detect_language(combined)
|
| 1094 |
+
emotion = detect_emotion(combined, lang)
|
| 1095 |
+
return lang, emotion
|
| 1096 |
+
|
| 1097 |
+
# Voice selection based on language and emotion (using MultilingualNeural voices)
|
| 1098 |
+
VOICE_BY_LANG_EMOTION = {
|
| 1099 |
+
'vietnamese': {
|
| 1100 |
+
'happy': ('vi-VN-HoaiMyNeural', 'vui'),
|
| 1101 |
+
'sad': ('vi-VN-NamMinhNeural', 'buồn'),
|
| 1102 |
+
'excited': ('vi-VN-HoaiMyNeural', 'hào hứng'),
|
| 1103 |
+
'humorous': ('vi-VN-HoaiMyNeural', 'vui'),
|
| 1104 |
+
'serious': ('vi-VN-NamMinhNeural', 'nghiêm túc'),
|
| 1105 |
+
'neutral': ('vi-VN-HoaiMyNeural', 'trung_tinh'),
|
| 1106 |
+
},
|
| 1107 |
+
'portuguese': {
|
| 1108 |
+
'happy': ('pt-BR-ThalitaMultilingualNeural', 'feliz'),
|
| 1109 |
+
'sad': ('pt-BR-ThalitaMultilingualNeural', 'triste'),
|
| 1110 |
+
'excited': ('pt-BR-ThalitaMultilingualNeural', 'animado'),
|
| 1111 |
+
'humorous': ('pt-BR-ThalitaMultilingualNeural', 'engraçado'),
|
| 1112 |
+
'serious': ('pt-BR-ThalitaMultilingualNeural', 'sério'),
|
| 1113 |
+
'neutral': ('pt-BR-ThalitaMultilingualNeural', 'neutro'),
|
| 1114 |
+
},
|
| 1115 |
+
'english': {
|
| 1116 |
+
'happy': ('en-US-AndrewMultilingualNeural', 'happy'),
|
| 1117 |
+
'sad': ('en-AU-WilliamMultilingualNeural', 'sad'),
|
| 1118 |
+
'excited': ('en-US-AndrewMultilingualNeural', 'excited'),
|
| 1119 |
+
'humorous': ('en-US-AndrewMultilingualNeural', 'funny'),
|
| 1120 |
+
'serious': ('en-AU-WilliamMultilingualNeural', 'serious'),
|
| 1121 |
+
'neutral': ('en-US-AndrewMultilingualNeural', 'neutral'),
|
| 1122 |
+
},
|
| 1123 |
+
'french': {
|
| 1124 |
+
'happy': ('fr-FR-VivienneMultilingualNeural', 'heureux'),
|
| 1125 |
+
'sad': ('fr-FR-RemyMultilingualNeural', 'triste'),
|
| 1126 |
+
'excited': ('fr-FR-VivienneMultilingualNeural', 'excité'),
|
| 1127 |
+
'humorous': ('fr-FR-VivienneMultilingualNeural', 'drôle'),
|
| 1128 |
+
'serious': ('fr-FR-RemyMultilingualNeural', 'sérieux'),
|
| 1129 |
+
'neutral': ('fr-FR-VivienneMultilingualNeural', 'neutre'),
|
| 1130 |
+
},
|
| 1131 |
+
'german': {
|
| 1132 |
+
'happy': ('de-DE-SeraphinaMultilingualNeural', 'glücklich'),
|
| 1133 |
+
'sad': ('de-DE-FlorianMultilingualNeural', 'traurig'),
|
| 1134 |
+
'excited': ('de-DE-SeraphinaMultilingualNeural', 'aufgeregt'),
|
| 1135 |
+
'humorous': ('de-DE-SeraphinaMultilingualNeural', 'lustig'),
|
| 1136 |
+
'serious': ('de-DE-FlorianMultilingualNeural', 'ernst'),
|
| 1137 |
+
'neutral': ('de-DE-SeraphinaMultilingualNeural', 'neutral'),
|
| 1138 |
+
},
|
| 1139 |
+
'korean': {
|
| 1140 |
+
'happy': ('ko-KR-HyunsuMultilingualNeural', '행복'),
|
| 1141 |
+
'sad': ('ko-KR-HyunsuMultilingualNeural', '슬픔'),
|
| 1142 |
+
'excited': ('ko-KR-HyunsuMultilingualNeural', '흥분'),
|
| 1143 |
+
'humorous': ('ko-KR-HyunsuMultilingualNeural', '유쾌'),
|
| 1144 |
+
'serious': ('ko-KR-HyunsuMultilingualNeural', '진지'),
|
| 1145 |
+
'neutral': ('ko-KR-HyunsuMultilingualNeural', '중립'),
|
| 1146 |
+
},
|
| 1147 |
+
'italian': {
|
| 1148 |
+
'happy': ('it-IT-GiuseppeMultilingualNeural', 'felice'),
|
| 1149 |
+
'sad': ('it-IT-GiuseppeMultilingualNeural', 'triste'),
|
| 1150 |
+
'excited': ('it-IT-GiuseppeMultilingualNeural', 'emozionato'),
|
| 1151 |
+
'humorous': ('it-IT-GiuseppeMultilingualNeural', 'divertente'),
|
| 1152 |
+
'serious': ('it-IT-GiuseppeMultilingualNeural', 'serio'),
|
| 1153 |
+
'neutral': ('it-IT-GiuseppeMultilingualNeural', 'neutro'),
|
| 1154 |
},
|
| 1155 |
}
|
| 1156 |
+
|
| 1157 |
+
# All valid voice IDs (new MultilingualNeural format)
|
| 1158 |
+
VALID_VOICES = {
|
| 1159 |
+
'vi-VN-HoaiMyNeural', 'vi-VN-NamMinhNeural',
|
| 1160 |
+
'en-US-AndrewMultilingualNeural', 'en-AU-WilliamMultilingualNeural',
|
| 1161 |
+
'pt-BR-ThalitaMultilingualNeural',
|
| 1162 |
+
'fr-FR-VivienneMultilingualNeural', 'fr-FR-RemyMultilingualNeural',
|
| 1163 |
+
'de-DE-SeraphinaMultilingualNeural', 'de-DE-FlorianMultilingualNeural',
|
| 1164 |
+
'ko-KR-HyunsuMultilingualNeural',
|
| 1165 |
+
'it-IT-GiuseppeMultilingualNeural',
|
| 1166 |
+
}
|
| 1167 |
+
|
| 1168 |
+
def get_voice_for_content(title, text, preferred_voice=None):
|
| 1169 |
+
"""Get appropriate voice based on content language and emotion."""
|
| 1170 |
+
# Accept the new MultilingualNeural voices directly
|
| 1171 |
+
if preferred_voice and preferred_voice in VALID_VOICES:
|
| 1172 |
+
return preferred_voice
|
| 1173 |
+
|
| 1174 |
+
# Also accept old shorthand voice IDs and map them to new format
|
| 1175 |
+
old_voice_map = {
|
| 1176 |
+
'hoaimy': 'vi-VN-HoaiMyNeural',
|
| 1177 |
+
'namminh': 'vi-VN-NamMinhNeural',
|
| 1178 |
+
'andrew': 'en-US-AndrewMultilingualNeural',
|
| 1179 |
+
'jenny': 'en-US-AndrewMultilingualNeural',
|
| 1180 |
+
'thalita': 'pt-BR-ThalitaMultilingualNeural',
|
| 1181 |
+
'pt_thalita': 'pt-BR-ThalitaMultilingualNeural',
|
| 1182 |
+
'pt_francisco': 'pt-BR-ThalitaMultilingualNeural',
|
| 1183 |
+
'ela': 'en-US-AndrewMultilingualNeural',
|
| 1184 |
+
'es_carlos': 'en-US-AndrewMultilingualNeural',
|
| 1185 |
+
'denise': 'fr-FR-VivienneMultilingualNeural',
|
| 1186 |
+
'katja': 'de-DE-SeraphinaMultilingualNeural',
|
| 1187 |
+
'nanami': 'en-US-AndrewMultilingualNeural',
|
| 1188 |
+
'sunhee': 'ko-KR-HyunsuMultilingualNeural',
|
| 1189 |
+
'xiaochen': 'en-US-AndrewMultilingualNeural',
|
| 1190 |
+
}
|
| 1191 |
+
if preferred_voice and preferred_voice in old_voice_map:
|
| 1192 |
+
return old_voice_map[preferred_voice]
|
| 1193 |
+
|
| 1194 |
+
lang, emotion = detect_language_and_emotion(title, text)
|
| 1195 |
+
lang_map = VOICE_BY_LANG_EMOTION.get(lang, VOICE_BY_LANG_EMOTION['vietnamese'])
|
| 1196 |
+
voice, _ = lang_map.get(emotion, lang_map['neutral'])
|
| 1197 |
+
return voice
|
| 1198 |
+
|
| 1199 |
+
|
| 1200 |
+
def _is_relevant_image(img_url, title, text):
|
| 1201 |
+
"""Check if an image is relevant to the article content."""
|
| 1202 |
+
if not img_url:
|
| 1203 |
+
return False
|
| 1204 |
+
skip_patterns = ['pixel', 'analytics', 'tracking', '1x1.gif', 'spacer.gif',
|
| 1205 |
+
'logo', 'icon', 'avatar', 'emoji', 'smiley', 'sprite',
|
| 1206 |
+
'advertisement', 'ad-banner', 'sponsored', 'banner-ads']
|
| 1207 |
+
img_lower = img_url.lower()
|
| 1208 |
+
for p in skip_patterns:
|
| 1209 |
+
if p in img_lower:
|
| 1210 |
+
return False
|
| 1211 |
+
if not any(img_lower.endswith(ext) for ext in ['.jpg', '.jpeg', '.png', '.webp', '.gif']):
|
| 1212 |
+
return False
|
| 1213 |
+
return True
|
| 1214 |
+
|
| 1215 |
+
|
| 1216 |
+
def _filter_relevant_images(images, title, text, max_images=8):
|
| 1217 |
+
"""Filter and rank images by relevance to article content."""
|
| 1218 |
+
if not images:
|
| 1219 |
+
return []
|
| 1220 |
+
seen = set()
|
| 1221 |
+
relevant = []
|
| 1222 |
+
for img in images:
|
| 1223 |
+
if img in seen:
|
| 1224 |
+
continue
|
| 1225 |
+
seen.add(img)
|
| 1226 |
+
if _is_relevant_image(img, title, text):
|
| 1227 |
+
relevant.append(img)
|
| 1228 |
+
return relevant[:max_images]
|
| 1229 |
+
|
| 1230 |
+
|
| 1231 |
+
def _scrape_article_for_rewrite(url):
|
| 1232 |
+
"""Scrape article: extract title, paragraphs, RELEVANT images, OG image."""
|
| 1233 |
+
try:
|
| 1234 |
+
r = req.get(url, headers=_UA_RW, timeout=15, allow_redirects=True)
|
| 1235 |
+
r.encoding = 'utf-8'
|
| 1236 |
+
soup = BeautifulSoup(r.text, 'lxml')
|
| 1237 |
+
for tag in soup.find_all(['script', 'style', 'nav', 'footer', 'aside', 'form']):
|
| 1238 |
+
tag.decompose()
|
| 1239 |
+
h1 = soup.find('h1')
|
| 1240 |
+
ogt = soup.find('meta', property='og:title')
|
| 1241 |
+
title = (h1.get_text(strip=True) if h1 else '') or (ogt.get('content', '') if ogt else '')
|
| 1242 |
+
ogi = soup.find('meta', property='og:image')
|
| 1243 |
+
og_img = ogi.get('content', '') if ogi else ''
|
| 1244 |
+
if og_img and og_img.startswith('//'):
|
| 1245 |
+
og_img = 'https:' + og_img
|
| 1246 |
+
block = None
|
| 1247 |
+
for sel in ['article', '.singular-content', '.detail-content', '.fck_detail', '.content-detail', '.knc-content', 'main', '.cms-body', '.article__body']:
|
| 1248 |
+
el = soup.select_one(sel)
|
| 1249 |
+
if el and len(el.find_all('p')) >= 2:
|
| 1250 |
+
block = el
|
| 1251 |
+
break
|
| 1252 |
+
if not block:
|
| 1253 |
+
block = soup.body or soup
|
| 1254 |
+
paragraphs = []
|
| 1255 |
+
all_images = []
|
| 1256 |
+
seen_imgs = set()
|
| 1257 |
+
if og_img and og_img not in seen_imgs:
|
| 1258 |
+
all_images.append(og_img)
|
| 1259 |
+
seen_imgs.add(og_img)
|
| 1260 |
+
for el in block.find_all(['p', 'h2', 'h3', 'figure', 'img'], recursive=True):
|
| 1261 |
+
if el.name == 'p':
|
| 1262 |
+
t = _clean(el.get_text(strip=True))
|
| 1263 |
+
if t and len(t) > 40:
|
| 1264 |
+
paragraphs.append(t)
|
| 1265 |
+
elif el.name in ('figure', 'img'):
|
| 1266 |
+
im = el if el.name == 'img' else el.find('img')
|
| 1267 |
+
if im:
|
| 1268 |
+
src = im.get('data-src') or im.get('src') or im.get('data-original') or ''
|
| 1269 |
+
if src and 'base64' not in src:
|
| 1270 |
+
if src.startswith('//'):
|
| 1271 |
+
src = 'https:' + src
|
| 1272 |
+
if src not in seen_imgs:
|
| 1273 |
+
all_images.append(src)
|
| 1274 |
+
seen_imgs.add(src)
|
| 1275 |
+
# Filter to relevant images only
|
| 1276 |
+
relevant_images = _filter_relevant_images(all_images, title, ' '.join(paragraphs[:5]))
|
| 1277 |
+
return {'title': _clean(title), 'paragraphs': paragraphs, 'images': relevant_images, 'og_img': og_img}
|
| 1278 |
+
except Exception:
|
| 1279 |
+
return None
|
| 1280 |
+
|
| 1281 |
+
|
| 1282 |
+
def _extract_key_points_rw(paragraphs, max_points=5):
|
| 1283 |
+
r"""Extract key points from paragraphs - extracts ALL sentences, not just first one.
|
| 1284 |
+
|
| 1285 |
+
Fixes: Original regex `^(.+?[.!?])\s` only captured first sentence per paragraph.
|
| 1286 |
+
Now splits on all sentence boundaries and takes valid sentences until max_points.
|
| 1287 |
+
"""
|
| 1288 |
+
points = []
|
| 1289 |
+
|
| 1290 |
+
for p in paragraphs:
|
| 1291 |
+
if len(points) >= max_points:
|
| 1292 |
+
break
|
| 1293 |
+
|
| 1294 |
+
p = _clean(p)
|
| 1295 |
+
if not p:
|
| 1296 |
+
continue
|
| 1297 |
+
|
| 1298 |
+
# Split paragraph into sentences using Vietnamese + English punctuation
|
| 1299 |
+
sentences = re.split(r'(?<=[.!?])\s+(?=[A-ZÀ-Ỹ0-9])', p)
|
| 1300 |
+
sentences = [s.strip() for s in sentences if s.strip()]
|
| 1301 |
+
|
| 1302 |
+
for sentence in sentences:
|
| 1303 |
+
if len(points) >= max_points:
|
| 1304 |
+
break
|
| 1305 |
+
|
| 1306 |
+
# Clean sentence - remove extra whitespace
|
| 1307 |
+
sentence = _clean(sentence)
|
| 1308 |
+
|
| 1309 |
+
if len(sentence) < 30:
|
| 1310 |
+
continue
|
| 1311 |
+
|
| 1312 |
+
# Check for duplicates
|
| 1313 |
+
if any(sentence[:60] in existing for existing in points):
|
| 1314 |
+
continue
|
| 1315 |
+
|
| 1316 |
+
# Ensure sentence ends with punctuation
|
| 1317 |
+
if not sentence.endswith(('.', '!', '?')):
|
| 1318 |
+
sentence = sentence + '.'
|
| 1319 |
+
|
| 1320 |
+
points.append(sentence)
|
| 1321 |
+
|
| 1322 |
+
# If no valid sentences found, take chunks from raw text
|
| 1323 |
+
if not points:
|
| 1324 |
+
raw = '\n'.join(paragraphs)
|
| 1325 |
+
for i in range(0, min(len(raw), max_points * 300), 280):
|
| 1326 |
+
chunk = _clean(raw[i:i+280])
|
| 1327 |
+
if len(chunk) >= 30 and chunk not in points:
|
| 1328 |
+
points.append(chunk + ('.' if not chunk.endswith('.') else ''))
|
| 1329 |
+
if len(points) >= max_points:
|
| 1330 |
+
break
|
| 1331 |
+
|
| 1332 |
+
return points
|
| 1333 |
+
|
| 1334 |
+
|
| 1335 |
+
@app.post("/api/rewrite_slide")
|
| 1336 |
+
async def api_rewrite_slide(request: Request):
|
| 1337 |
+
"""Fast rewrite as SLIDES - no AI needed, instant response."""
|
| 1338 |
+
body = await request.json()
|
| 1339 |
+
url = _clean(body.get("url", ""))
|
| 1340 |
+
context = body.get("context", "")
|
| 1341 |
+
preferred_voice = body.get("voice", "") # Accept custom voice selection
|
| 1342 |
+
if not url and not context:
|
| 1343 |
+
return JSONResponse({"error": "Cần URL hoặc nội dung"}, status_code=400)
|
| 1344 |
+
data = None
|
| 1345 |
+
if url and url.startswith("http"):
|
| 1346 |
+
data = _scrape_article_for_rewrite(url)
|
| 1347 |
+
if not data and context:
|
| 1348 |
+
paragraphs = [_clean(p) for p in context.split('\n') if len(_clean(p)) > 40]
|
| 1349 |
+
data = {'title': paragraphs[0][:80] if paragraphs else 'Bài viết', 'paragraphs': paragraphs, 'images': [], 'og_img': ''}
|
| 1350 |
+
if not data or not data.get('paragraphs'):
|
| 1351 |
+
return JSONResponse({"error": "Không đọc được bài viết"}, status_code=422)
|
| 1352 |
+
points = _extract_key_points_rw(data['paragraphs'], max_points=12)
|
| 1353 |
+
if not points:
|
| 1354 |
+
return JSONResponse({"error": "Không tìm được ý chính"}, status_code=422)
|
| 1355 |
+
images = data.get('images', [])
|
| 1356 |
+
slides = []
|
| 1357 |
+
for i, point in enumerate(points):
|
| 1358 |
+
img = images[i] if i < len(images) else (images[-1] if images else '')
|
| 1359 |
+
if img and 'cdnphoto.dantri' in img:
|
| 1360 |
+
img = '/api/proxy/img?url=' + _quote2(img, safe='')
|
| 1361 |
+
slides.append({'text': point, 'image': img, 'index': i + 1})
|
| 1362 |
+
summary_text = '\n\n'.join([f"• {s['text']}" for s in slides])
|
| 1363 |
+
|
| 1364 |
+
# Auto-detect language and emotion
|
| 1365 |
+
lang, emotion = detect_language_and_emotion(data['title'], summary_text)
|
| 1366 |
+
# Use preferred voice if provided, otherwise auto-detect
|
| 1367 |
+
voice = preferred_voice if preferred_voice else get_voice_for_content(data['title'], summary_text)
|
| 1368 |
+
|
| 1369 |
+
post = {
|
| 1370 |
+
"id": str(int(time.time() * 1000)) + str(_random2.randint(100, 999)),
|
| 1371 |
+
"title": data['title'],
|
| 1372 |
+
"text": summary_text,
|
| 1373 |
+
"img": images[0] if images else '',
|
| 1374 |
+
"url": url,
|
| 1375 |
+
"kind": "slide_summary",
|
| 1376 |
+
"slides": slides,
|
| 1377 |
+
"images": images[:10],
|
| 1378 |
+
"video": "",
|
| 1379 |
+
"voice": voice,
|
| 1380 |
+
"emotion": emotion,
|
| 1381 |
+
"language": lang,
|
| 1382 |
+
"ts": int(time.time())
|
| 1383 |
+
}
|
| 1384 |
+
posts = _load_wall_posts()
|
| 1385 |
+
posts.insert(0, post)
|
| 1386 |
+
_save_wall_posts(posts)
|
| 1387 |
+
return JSONResponse({"post": post, "slides": slides})
|
| 1388 |
+
|
| 1389 |
+
|
| 1390 |
+
@app.post("/api/rewrite_share")
|
| 1391 |
+
async def api_rewrite_share(request: Request):
|
| 1392 |
+
"""Rewrite article and post to Tường AI with SLIDES + AI text."""
|
| 1393 |
+
body = await request.json()
|
| 1394 |
+
url = _clean(body.get("url", ""))
|
| 1395 |
+
ctx = _clean(body.get("context", ""))
|
| 1396 |
+
preferred_voice = body.get("voice", "") # Accept custom voice selection
|
| 1397 |
+
if not url and not ctx:
|
| 1398 |
+
return JSONResponse({"error": "Cần URL hoặc nội dung"}, status_code=400)
|
| 1399 |
+
data = None
|
| 1400 |
+
if url and url.startswith("http"):
|
| 1401 |
+
data = _scrape_article_for_rewrite(url)
|
| 1402 |
+
if not data and ctx:
|
| 1403 |
+
paragraphs = [_clean(p) for p in ctx.split('\n') if len(_clean(p)) > 40]
|
| 1404 |
+
data = {'title': paragraphs[0][:80] if paragraphs else 'Bài viết', 'paragraphs': paragraphs, 'images': [], 'og_img': ''}
|
| 1405 |
+
if not data or not data.get('paragraphs'):
|
| 1406 |
+
return JSONResponse({"error": "Không đọc được bài viết"}, status_code=422)
|
| 1407 |
+
raw_text = '\n'.join(data['paragraphs'])
|
| 1408 |
+
if len(raw_text) < 50:
|
| 1409 |
+
raw_text = ctx[:14000]
|
| 1410 |
+
if len(raw_text) < 50:
|
| 1411 |
+
return JSONResponse({"error": "Bài viết quá ngắn"}, status_code=422)
|
| 1412 |
+
domain = ''
|
| 1413 |
+
try:
|
| 1414 |
+
from urllib.parse import urlparse
|
| 1415 |
+
domain = urlparse(url).netloc.replace('www.', '')
|
| 1416 |
+
except:
|
| 1417 |
+
pass
|
| 1418 |
+
|
| 1419 |
+
# Generate AI summary text
|
| 1420 |
+
ai_text = None
|
| 1421 |
+
try:
|
| 1422 |
+
import ai_ext
|
| 1423 |
+
if hasattr(ai_ext, 'qwen_generate'):
|
| 1424 |
+
prompt = f'Tóm tắt đăng Tường AI:\nTiêu đề: {data["title"]}\n{raw_text[:14000]}\n\n4-6 ý chính. Cuối ghi nguồn.'
|
| 1425 |
+
ai_text = await ai_ext.qwen_generate(prompt, max_tokens=1000)
|
| 1426 |
+
except Exception:
|
| 1427 |
+
pass
|
| 1428 |
+
if not ai_text or len(ai_text) < 80:
|
| 1429 |
+
key_pts = _extract_key_points_rw(data['paragraphs'], max_points=12)
|
| 1430 |
+
if key_pts:
|
| 1431 |
+
ai_text = '\n\n'.join([f"• {p}" for p in key_pts])
|
| 1432 |
+
else:
|
| 1433 |
+
ai_text = f"Tóm tắt: {data['title']}\n\n{raw_text[:1200]}\n\nNguồn: {domain}"
|
| 1434 |
+
|
| 1435 |
+
# Build slides from key points (FIX: include slides in rewrite_share too!)
|
| 1436 |
+
points = _extract_key_points_rw(data['paragraphs'], max_points=12)
|
| 1437 |
+
images = data.get('images', [])
|
| 1438 |
+
slides = []
|
| 1439 |
+
for i, point in enumerate(points):
|
| 1440 |
+
img = images[i] if i < len(images) else (images[-1] if images else '')
|
| 1441 |
+
if img and 'cdnphoto.dantri' in img:
|
| 1442 |
+
img = '/api/proxy/img?url=' + _quote2(img, safe='')
|
| 1443 |
+
slides.append({'text': point, 'image': img, 'index': i + 1})
|
| 1444 |
+
|
| 1445 |
+
# Auto-detect language and emotion
|
| 1446 |
+
lang, emotion = detect_language_and_emotion(data['title'], ai_text)
|
| 1447 |
+
# Use preferred voice if provided, otherwise auto-detect
|
| 1448 |
+
voice = preferred_voice if preferred_voice else get_voice_for_content(data['title'], ai_text)
|
| 1449 |
+
|
| 1450 |
+
post = {
|
| 1451 |
+
"id": str(int(time.time() * 1000)) + str(_random2.randint(100, 999)),
|
| 1452 |
+
"title": data['title'],
|
| 1453 |
+
"text": ai_text,
|
| 1454 |
+
"img": images[0] if images else '',
|
| 1455 |
+
"url": url,
|
| 1456 |
+
"kind": "rewrite",
|
| 1457 |
+
"slides": slides,
|
| 1458 |
+
"images": images[:10],
|
| 1459 |
+
"video": "",
|
| 1460 |
+
"voice": voice,
|
| 1461 |
+
"emotion": emotion,
|
| 1462 |
+
"language": lang,
|
| 1463 |
+
"ts": int(time.time())
|
| 1464 |
+
}
|
| 1465 |
+
posts = _load_wall_posts()
|
| 1466 |
+
posts.insert(0, post)
|
| 1467 |
+
_save_wall_posts(posts)
|
| 1468 |
+
return JSONResponse({"post": post, "slides": slides})
|
| 1469 |
+
|
| 1470 |
+
|
| 1471 |
+
@app.post("/api/url_wall")
|
| 1472 |
+
async def api_url_wall(request: Request):
|
| 1473 |
+
"""Submit URL to add to Tường AI."""
|
| 1474 |
+
body = await request.json()
|
| 1475 |
+
url = _clean(body.get("url", ""))
|
| 1476 |
+
if not url or not url.startswith('http'):
|
| 1477 |
+
return JSONResponse({"error": "URL không hợp lệ"}, status_code=400)
|
| 1478 |
+
# Reuse rewrite_share logic
|
| 1479 |
+
req._body = json.dumps({"url": url}).encode()
|
| 1480 |
+
return await api_rewrite_share(request)
|
| 1481 |
+
|
| 1482 |
+
|
| 1483 |
+
# ===== PERSONAL OPINION POST v2: AI tổng hợp bài viết từ quan điểm + nguồn tin HOT =====
|
| 1484 |
+
|
| 1485 |
+
# ===== KEYWORD EXTRACTION FROM OPINION =====
|
| 1486 |
+
_STOP_WORDS_EX = set("""
|
| 1487 |
+
và của các những một được trong với cho tại sau trước khi không người
|
| 1488 |
+
việt nam hôm nay mới nhất nóng tin tức cập nhật theo từ đến là có thì
|
| 1489 |
+
này đã để về lại nên cũng rất như vì do nếu sẽ nếu thế nhưng mà vẫn
|
| 1490 |
+
đang vào ra hơn đây đó nào cả cùng đã từng hãy còn chỉ cũng đều khiến
|
| 1491 |
+
được đã bị bởi qua những lúc cái gì cô chú bác anh chị em bạn tôi mình
|
| 1492 |
+
ông bà thầy vì vậy chính phải ấy đấy đâu đó thôi nhé đấy ạ nhỉ
|
| 1493 |
+
ngày tháng năm giờ phút giây tuần tháng quý
|
| 1494 |
+
""".strip().split())
|
| 1495 |
+
|
| 1496 |
+
def _extract_keywords_from_opinion(text, max_keywords=5):
|
| 1497 |
+
"""Extract meaningful keywords from user's opinion for news search."""
|
| 1498 |
+
if not text:
|
| 1499 |
+
return []
|
| 1500 |
+
text = text.lower()
|
| 1501 |
+
text = re.sub(r'https?://\S+', '', text)
|
| 1502 |
+
text = re.sub(r'[^\w\sÀ-ỹ]', ' ', text)
|
| 1503 |
+
text = re.sub(r'\s+', ' ', text).strip()
|
| 1504 |
+
words = [w for w in text.split() if len(w) > 2 and w not in _STOP_WORDS_EX]
|
| 1505 |
+
word_scores = {}
|
| 1506 |
+
for w in words:
|
| 1507 |
+
word_scores[w] = word_scores.get(w, 0) + 1
|
| 1508 |
+
sorted_words = sorted(word_scores.items(), key=lambda x: -x[1])
|
| 1509 |
+
top_words = [w for w, s in sorted_words[:max_keywords]]
|
| 1510 |
+
phrases = []
|
| 1511 |
+
for i in range(len(words) - 1):
|
| 1512 |
+
phrase = words[i] + ' ' + words[i + 1]
|
| 1513 |
+
if len(phrase) > 5:
|
| 1514 |
+
phrases.append(phrase)
|
| 1515 |
+
phrase_scores = {}
|
| 1516 |
+
for p in phrases:
|
| 1517 |
+
phrase_scores[p] = phrase_scores.get(p, 0) + 1
|
| 1518 |
+
sorted_phrases = sorted(phrase_scores.items(), key=lambda x: -x[1])
|
| 1519 |
+
top_phrases = [p for p, s in sorted_phrases[:3]]
|
| 1520 |
+
result = []
|
| 1521 |
+
for p in top_phrases:
|
| 1522 |
+
if p not in result:
|
| 1523 |
+
result.append(p)
|
| 1524 |
+
for w in top_words:
|
| 1525 |
+
if w not in result:
|
| 1526 |
+
result.append(w)
|
| 1527 |
+
return result[:max_keywords]
|
| 1528 |
+
|
| 1529 |
+
|
| 1530 |
+
@app.post("/api/personal_post/preview")
|
| 1531 |
+
async def api_personal_post_preview(request: Request):
|
| 1532 |
+
"""Preview personal post: fetch full articles, let AI compose logical article with images."""
|
| 1533 |
+
body = await request.json()
|
| 1534 |
+
opinion = _clean(body.get("opinion", ""))
|
| 1535 |
+
selected_topics = body.get("selected_topics", []) or []
|
| 1536 |
+
selected_sources = body.get("selected_sources", []) or []
|
| 1537 |
+
|
| 1538 |
+
if not opinion or len(opinion) < 10:
|
| 1539 |
+
return JSONResponse({"error": "Quan điểm cá nhân quá ngắn (cần ít nhất 10 ký tự)"}, status_code=400)
|
| 1540 |
+
|
| 1541 |
+
# Lấy keywords từ QUAN ĐIỂM CÁ NHÂN để tìm nguồn tin chính xác
|
| 1542 |
+
keywords = _extract_keywords_from_opinion(opinion, max_keywords=5)
|
| 1543 |
+
if keywords:
|
| 1544 |
+
selected_topics = keywords[:3]
|
| 1545 |
+
else:
|
| 1546 |
+
# Fallback: hot topics
|
| 1547 |
+
hot = _get_hot_topics()
|
| 1548 |
+
selected_topics = [t.get("topic", "") for t in hot[:3] if t.get("topic")]
|
| 1549 |
+
|
| 1550 |
+
# Tìm nguồn tin
|
| 1551 |
+
all_sources = []
|
| 1552 |
+
seen_urls = set()
|
| 1553 |
+
for topic in selected_topics[:3]:
|
| 1554 |
+
sources = _search_all(topic, limit=5)
|
| 1555 |
+
for s in sources:
|
| 1556 |
+
if s.get("url") and s["url"] not in seen_urls:
|
| 1557 |
+
seen_urls.add(s["url"])
|
| 1558 |
+
all_sources.append(s)
|
| 1559 |
+
if len(all_sources) >= 6:
|
| 1560 |
+
break
|
| 1561 |
+
if len(all_sources) >= 6:
|
| 1562 |
+
break
|
| 1563 |
+
|
| 1564 |
+
for src in selected_sources:
|
| 1565 |
+
if src.get("url") and src["url"] not in seen_urls:
|
| 1566 |
+
all_sources.insert(0, src)
|
| 1567 |
+
|
| 1568 |
+
# Scrape nội dung đầy đủ từng nguồn (paragraphs + images)
|
| 1569 |
+
source_details = []
|
| 1570 |
+
source_images = []
|
| 1571 |
+
for src in all_sources[:5]:
|
| 1572 |
+
url = src.get("url", "")
|
| 1573 |
+
if not url:
|
| 1574 |
+
continue
|
| 1575 |
+
try:
|
| 1576 |
+
art = _scrape_article_for_rewrite(url)
|
| 1577 |
+
if art:
|
| 1578 |
+
src_detail = {
|
| 1579 |
+
"title": art.get("title", src.get("title", "")),
|
| 1580 |
+
"url": url,
|
| 1581 |
+
"via": src.get("via", ""),
|
| 1582 |
+
"paragraphs": art.get("paragraphs", [])[:8],
|
| 1583 |
+
"images": art.get("images", [])[:3],
|
| 1584 |
+
"og_image": art.get("og_img", "")
|
| 1585 |
+
}
|
| 1586 |
+
source_details.append(src_detail)
|
| 1587 |
+
# Collect images for proxy
|
| 1588 |
+
for img in art.get("images", [])[:2]:
|
| 1589 |
+
if any(x in img for x in ["cdnphoto.dantri", "vnexpress", "vcdn", "refooty"]):
|
| 1590 |
+
img = "/api/proxy/img?url=" + _quote2(img, safe="")
|
| 1591 |
+
source_images.append(img)
|
| 1592 |
+
except:
|
| 1593 |
+
pass
|
| 1594 |
+
if len(source_details) >= 5:
|
| 1595 |
+
break
|
| 1596 |
+
|
| 1597 |
+
# Tạo title từ opinion
|
| 1598 |
+
opinion_words = re.findall(r"[A-Za-zÀ-ỹ0-9]+", opinion)
|
| 1599 |
+
title_words = opinion_words[:8] if len(opinion_words) >= 8 else opinion_words[:4]
|
| 1600 |
+
title = " ".join([w[0].upper() + w[1:] for w in title_words]) if title_words else "Quan điểm cá nhân"
|
| 1601 |
+
title = title[:80]
|
| 1602 |
+
|
| 1603 |
+
# AI sinh bài viết hoàn chỉnh
|
| 1604 |
+
ai_text = None
|
| 1605 |
+
try:
|
| 1606 |
+
import ai_ext
|
| 1607 |
+
if hasattr(ai_ext, 'qwen_generate'):
|
| 1608 |
+
# Build detailed context from source articles
|
| 1609 |
+
source_context = ""
|
| 1610 |
+
for i, sd in enumerate(source_details[:5]):
|
| 1611 |
+
src_title = sd.get("title", "")
|
| 1612 |
+
src_via = sd.get("via", "")
|
| 1613 |
+
src_paras = sd.get("paragraphs", [])
|
| 1614 |
+
source_context += f"\n=== Nguồn {i+1}: {src_title} ({src_via}) ===\n"
|
| 1615 |
+
for j, p in enumerate(src_paras[:4]):
|
| 1616 |
+
source_context += f" - {p[:300]}\n"
|
| 1617 |
+
|
| 1618 |
+
prompt = (
|
| 1619 |
+
"QUAN ĐIỂM: " + opinion[:500] + "\nNGUỒN: " + source_context[:1000] + "\n\n"
|
| 1620 |
+
"=== NGUỒN TIN THAM KHẢO ===\n" + source_context + "\n\n"
|
| 1621 |
+
"=== YÊU CẦU VIẾT BÀI THEO SLIDE ===\n"
|
| 1622 |
+
"Viết bài thành 5-6 ĐOẠN VĂN NGẮN, mỗi đoạn là 1 SLIDE.\n"
|
| 1623 |
+
"\n"
|
| 1624 |
+
"QUAN TRỌNG NHẤT: MỗI SLIDE PHẢI KẾT HỢP QUAN ĐIỂM CÁ NHÂN + NỘI DUNG NGUỒN TIN, KHÔNG PHẢI CHỈ NÓI VỀ NGUỒN TIN.\n"
|
| 1625 |
+
"\n"
|
| 1626 |
+
"SLIDE 1 - MỞ ĐẦU:\n"
|
| 1627 |
+
"- NHIỆN HỮU QUAN ĐIỂM CÁ NHÂN LÊN ĐẦU\n"
|
| 1628 |
+
"- Giới thiệu chủ đề, nêu rõ quan điểm của bạn (dựa vào QUAN ĐIỂM CÁ NHÂN ở trên)\n"
|
| 1629 |
+
"- 2-4 câu hoàn chỉnh\n"
|
| 1630 |
+
"\n"
|
| 1631 |
+
"SLIDE 2-3-4-5 - PHÂN TÍCH:\n"
|
| 1632 |
+
"- Mỗi slide: B�Commencer bằng QUAN ĐIỂM CÁ NHÂN, sau đó dẫn chứng từ 1 nguồn tin\n"
|
| 1633 |
+
"- Ví dụ: \"Theo quan điểm của tôi, đây là vấn đề cần lưu ý. Theo VnExpress...\"\n"
|
| 1634 |
+
"- Dẫn chứng từ nguồn (ghi rõ tên báo: Theo VnExpress, Theo Thanh Niên...)\n"
|
| 1635 |
+
"- 2-4 câu hoàn chỉnh mỗi slide\n"
|
| 1636 |
+
"\n"
|
| 1637 |
+
"SLIDE 6 - KẾT LUẬN:\n"
|
| 1638 |
+
"- Tổng kết quan điểm cá nhân, đưa ra nhận định cuối cùng\n"
|
| 1639 |
+
"- 2-3 câu hoàn chỉnh\n"
|
| 1640 |
+
"\n"
|
| 1641 |
+
"Định dạng đầu ra:\n"
|
| 1642 |
+
"---SLIDE 1---\n"
|
| 1643 |
+
"[nội dung đoạn văn slide 1]\n"
|
| 1644 |
+
"---SLIDE 2---\n"
|
| 1645 |
+
"[n��i dung đoạn văn slide 2]\n"
|
| 1646 |
+
"...v.v...\n"
|
| 1647 |
+
"\n"
|
| 1648 |
+
"QUAN TRỌNG:\n"
|
| 1649 |
+
"- Mỗi slide là 1 đoạn văn HOÀN CHỈNH, 2-4 câu\n"
|
| 1650 |
+
"- PHẢI KẾT THÚC BẰNG DẤU CHẤM (.) HOẢN TOÀN\n"
|
| 1651 |
+
"- Kết hợp QUAN ĐIỂM CÁ NHÂN với NỘI DUNG NGUỒN TIN\n"
|
| 1652 |
+
"- Không gạch đầu dòng, không bullet points\n"
|
| 1653 |
+
"- Viết liền mạch tự nhiên, giọng văn báo chí\n"
|
| 1654 |
+
"- Mỗi slide phải khác nhau, không lặp ý\n"
|
| 1655 |
+
"- Độ dài: 300-600 từ"
|
| 1656 |
+
)
|
| 1657 |
+
ai_text = None # Không dùng AI, để code tự kết hợp opinion + source
|
| 1658 |
+
except:
|
| 1659 |
+
pass
|
| 1660 |
+
|
| 1661 |
+
if not ai_text or len(ai_text) < 100:
|
| 1662 |
+
# Fallback: build article manually
|
| 1663 |
+
ai_text = "## " + title + "\n\n"
|
| 1664 |
+
ai_text += opinion + "\n\n"
|
| 1665 |
+
for i, sd in enumerate(source_details[:5]):
|
| 1666 |
+
ai_text += "### " + sd.get("title", f"Nguồn {i+1}") + "\n"
|
| 1667 |
+
for p in sd.get("paragraphs", [])[:3]:
|
| 1668 |
+
ai_text += p[:250] + "\n"
|
| 1669 |
+
ai_text += "*Nguồn: " + sd.get("via", "") + "*\n\n"
|
| 1670 |
+
ai_text += "\n---\n*Bài viết tổng hợp từ quan điểm cá nhân và các nguồn tin liên quan*"
|
| 1671 |
+
|
| 1672 |
+
# Parse slides từ AI output (format: ---SLIDE N--- content)
|
| 1673 |
+
slides = []
|
| 1674 |
+
if ai_text:
|
| 1675 |
+
# Try to parse the ---SLIDE--- format
|
| 1676 |
+
pattern = r'---SLIDE\s*(\d+)---\s*\n(.*?)(?=---SLIDE|\Z)'
|
| 1677 |
+
matches = re.findall(pattern, ai_text, re.DOTALL)
|
| 1678 |
+
|
| 1679 |
+
if matches:
|
| 1680 |
+
for idx, (num, content) in enumerate(matches):
|
| 1681 |
+
# Normalize: ensure complete sentences
|
| 1682 |
+
text = _ensure_sentence_complete(content)
|
| 1683 |
+
if len(text) > 40:
|
| 1684 |
+
img = source_images[idx] if idx < len(source_images) else ""
|
| 1685 |
+
slides.append({"text": text, "image": img, "index": idx + 1})
|
| 1686 |
+
|
| 1687 |
+
# If we have parsed slides, ensure minimum 3
|
| 1688 |
+
if len(slides) < 3:
|
| 1689 |
+
# Use parsed slides as base, fill remaining from AI text
|
| 1690 |
+
used_indices = set()
|
| 1691 |
+
for s in slides:
|
| 1692 |
+
used_indices.add(s['index'] - 1)
|
| 1693 |
+
|
| 1694 |
+
# Split remaining AI text into more slides
|
| 1695 |
+
sentences = re.split(r'(?<=[.!?])\s+', ai_text)
|
| 1696 |
+
current_chunk = ""
|
| 1697 |
+
next_idx = len(slides)
|
| 1698 |
+
|
| 1699 |
+
for sent in sentences:
|
| 1700 |
+
sent = _ensure_sentence_complete(sent)
|
| 1701 |
+
if len(sent) < 20:
|
| 1702 |
+
continue
|
| 1703 |
+
|
| 1704 |
+
# Skip if this sentence is already in parsed slides
|
| 1705 |
+
found = False
|
| 1706 |
+
for slide in slides:
|
| 1707 |
+
if sent[:50] in slide['text']:
|
| 1708 |
+
found = True
|
| 1709 |
+
break
|
| 1710 |
+
|
| 1711 |
+
if found:
|
| 1712 |
+
continue
|
| 1713 |
+
|
| 1714 |
+
if current_chunk and len(current_chunk + " " + sent) <= 380:
|
| 1715 |
+
current_chunk += " " + sent
|
| 1716 |
+
else:
|
| 1717 |
+
if len(current_chunk) > 50:
|
| 1718 |
+
img = source_images[next_idx] if next_idx < len(source_images) else ""
|
| 1719 |
+
slides.append({"text": current_chunk, "image": img, "index": next_idx + 1})
|
| 1720 |
+
current_chunk = sent
|
| 1721 |
+
next_idx += 1
|
| 1722 |
+
|
| 1723 |
+
# Add final chunk
|
| 1724 |
+
if len(current_chunk) > 50 and next_idx < 6:
|
| 1725 |
+
img = source_images[next_idx] if next_idx < len(source_images) else ""
|
| 1726 |
+
slides.append({"text": current_chunk, "image": img, "index": next_idx + 1})
|
| 1727 |
+
|
| 1728 |
+
# Ultimate fallback: create slides from opinion + source
|
| 1729 |
+
if len(slides) < 2:
|
| 1730 |
+
slides = []
|
| 1731 |
+
# Slide 1: opinion
|
| 1732 |
+
if opinion and len(opinion) > 20:
|
| 1733 |
+
slides.append({"text": opinion[:450], "image": source_images[0] if source_images else "", "index": 1})
|
| 1734 |
+
|
| 1735 |
+
# Slide 2-6: from AI text or sources
|
| 1736 |
+
if ai_text:
|
| 1737 |
+
sentences = re.split(r'(?<=[.!?])\s+', ai_text)
|
| 1738 |
+
for i, sent in enumerate(sentences[:5]):
|
| 1739 |
+
text = _ensure_sentence_complete(_clean(sent))
|
| 1740 |
+
if len(text) > 60:
|
| 1741 |
+
if len(slides) < 6:
|
| 1742 |
+
img = source_images[len(slides)] if len(slides) < len(source_images) else ""
|
| 1743 |
+
slides.append({"text": text, "image": img, "index": len(slides) + 1})
|
| 1744 |
+
|
| 1745 |
+
# Fill remaining with key points from sources - KẾT HỢP VỚI QUAN ĐIỂM CÁ NHÂN
|
| 1746 |
+
src_idx = len(slides)
|
| 1747 |
+
while len(slides) < 4 and src_idx < len(source_details):
|
| 1748 |
+
paragraphs = source_details[src_idx].get("paragraphs", [])
|
| 1749 |
+
src_title = source_details[src_idx].get("title", "")
|
| 1750 |
+
src_via = source_details[src_idx].get("via", "")
|
| 1751 |
+
for p in paragraphs[:2]:
|
| 1752 |
+
if len(p) > 60 and len(slides) < 6:
|
| 1753 |
+
# Kết hợp opinion với nội dung source
|
| 1754 |
+
combined = f"Theo góc nhìn của tôi, {opinion[:100]}... Theo {src_via}: {p[:250]}"
|
| 1755 |
+
img = source_images[len(slides)] if len(slides) < len(source_images) else ""
|
| 1756 |
+
slides.append({"text": _ensure_sentence_complete(combined), "image": img, "index": len(slides) + 1})
|
| 1757 |
+
break # Mỗi nguồn 1 slide
|
| 1758 |
+
src_idx += 1
|
| 1759 |
+
|
| 1760 |
+
# Final fallback: ensure at least 2-3 slides
|
| 1761 |
+
while len(slides) < 3:
|
| 1762 |
+
idx = len(slides)
|
| 1763 |
+
if idx == 0 and opinion:
|
| 1764 |
+
slides.append({"text": opinion[:400], "image": "", "index": 1})
|
| 1765 |
+
elif ai_text:
|
| 1766 |
+
slides.append({"text": ai_text[idx*300:(idx+1)*300], "image": "", "index": idx + 1})
|
| 1767 |
+
else:
|
| 1768 |
+
slides.append({"text": f"Nguồn tham khảo {idx + 1}", "image": "", "index": idx + 1})
|
| 1769 |
+
|
| 1770 |
+
preview = {
|
| 1771 |
+
"title": title,
|
| 1772 |
+
"text": ai_text,
|
| 1773 |
+
"opinion": opinion,
|
| 1774 |
+
"images": source_images[:10],
|
| 1775 |
+
"sources": source_details[:5],
|
| 1776 |
+
"slides": slides[:6] # Max 6 slides
|
| 1777 |
+
}
|
| 1778 |
+
|
| 1779 |
+
return JSONResponse({"preview": preview})
|
| 1780 |
+
|
| 1781 |
+
|
| 1782 |
+
@app.post("/api/personal_post")
|
| 1783 |
+
async def api_personal_post(request: Request):
|
| 1784 |
+
"""Create and save personal opinion post."""
|
| 1785 |
+
body = await request.json()
|
| 1786 |
+
opinion = _clean(body.get("opinion", ""))
|
| 1787 |
+
selected_topics = body.get("selected_topics", []) or []
|
| 1788 |
+
selected_sources = body.get("selected_sources", []) or []
|
| 1789 |
+
custom_title = body.get("custom_title", "")
|
| 1790 |
+
custom_slides = body.get("custom_slides", [])
|
| 1791 |
+
|
| 1792 |
+
if not opinion or len(opinion) < 10:
|
| 1793 |
+
return JSONResponse({"error": "Quan điểm cá nhân quá ngắn (cần ít nhất 10 ký tự)"}, status_code=400)
|
| 1794 |
+
|
| 1795 |
+
if not selected_topics:
|
| 1796 |
+
# Lấy keywords từ QUAN ĐIỂM CÁ NHÂN để tìm nguồn tin chính xác
|
| 1797 |
+
keywords = _extract_keywords_from_opinion(opinion, max_keywords=5)
|
| 1798 |
+
if keywords:
|
| 1799 |
+
selected_topics = keywords[:3]
|
| 1800 |
+
else:
|
| 1801 |
+
hot = _get_hot_topics()
|
| 1802 |
+
selected_topics = [t.get("topic", "") for t in hot[:3] if t.get("topic")]
|
| 1803 |
+
|
| 1804 |
+
all_sources = []
|
| 1805 |
+
seen_urls = set()
|
| 1806 |
+
for topic in selected_topics[:3]:
|
| 1807 |
+
sources = _search_all(topic, limit=5)
|
| 1808 |
+
for s in sources:
|
| 1809 |
+
if s.get("url") and s["url"] not in seen_urls:
|
| 1810 |
+
seen_urls.add(s["url"])
|
| 1811 |
+
all_sources.append(s)
|
| 1812 |
+
if len(all_sources) >= 6:
|
| 1813 |
+
break
|
| 1814 |
+
if len(all_sources) >= 6:
|
| 1815 |
+
break
|
| 1816 |
+
|
| 1817 |
+
for src in selected_sources:
|
| 1818 |
+
if src.get("url") and src["url"] not in seen_urls:
|
| 1819 |
+
all_sources.insert(0, src)
|
| 1820 |
+
|
| 1821 |
+
source_details = []
|
| 1822 |
+
source_images = []
|
| 1823 |
+
for src in all_sources[:5]:
|
| 1824 |
+
url = src.get("url", "")
|
| 1825 |
+
if not url:
|
| 1826 |
+
continue
|
| 1827 |
+
try:
|
| 1828 |
+
art = _scrape_article_for_rewrite(url)
|
| 1829 |
+
if art:
|
| 1830 |
+
src_detail = {
|
| 1831 |
+
"title": art.get("title", src.get("title", "")),
|
| 1832 |
+
"url": url,
|
| 1833 |
+
"via": src.get("via", ""),
|
| 1834 |
+
"paragraphs": art.get("paragraphs", [])[:6],
|
| 1835 |
+
"images": art.get("images", [])[:2],
|
| 1836 |
+
"og_image": art.get("og_img", "")
|
| 1837 |
+
}
|
| 1838 |
+
source_details.append(src_detail)
|
| 1839 |
+
for img in art.get("images", [])[:2]:
|
| 1840 |
+
if any(x in img for x in ["cdnphoto.dantri", "vnexpress", "vcdn", "refooty"]):
|
| 1841 |
+
img = "/api/proxy/img?url=" + _quote2(img, safe="")
|
| 1842 |
+
source_images.append(img)
|
| 1843 |
+
except:
|
| 1844 |
+
pass
|
| 1845 |
+
|
| 1846 |
+
# Title
|
| 1847 |
+
if custom_title:
|
| 1848 |
+
title = custom_title[:80]
|
| 1849 |
+
else:
|
| 1850 |
+
opinion_words = re.findall(r"[A-Za-zÀ-ỹ0-9]+", opinion)
|
| 1851 |
+
title_words = opinion_words[:8] if len(opinion_words) >= 8 else opinion_words[:4]
|
| 1852 |
+
title = " ".join([w[0].upper() + w[1:] for w in title_words]) if title_words else "Quan điểm cá nhân"
|
| 1853 |
+
title = title[:80]
|
| 1854 |
+
|
| 1855 |
+
# AI sinh bài
|
| 1856 |
+
ai_text = None
|
| 1857 |
+
try:
|
| 1858 |
+
import ai_ext
|
| 1859 |
+
if hasattr(ai_ext, 'qwen_generate'):
|
| 1860 |
+
source_context = ""
|
| 1861 |
+
for i, sd in enumerate(source_details[:5]):
|
| 1862 |
+
src_title = sd.get("title", "")
|
| 1863 |
+
src_via = sd.get("via", "")
|
| 1864 |
+
src_paras = sd.get("paragraphs", [])
|
| 1865 |
+
source_context += f"\nNguồn {i+1}: {src_title} ({src_via})\n"
|
| 1866 |
+
for j, p in enumerate(src_paras[:3]):
|
| 1867 |
+
source_context += f" - {p[:300]}\n"
|
| 1868 |
+
prompt = (
|
| 1869 |
+
"QUAN ĐIỂM: " + opinion[:500] + "\nNGUỒN: " + source_context[:1000] + "\n\n"
|
| 1870 |
+
"=== NGUỒN TIN ===\n" + source_context + "\n\n"
|
| 1871 |
+
"=== YÊU CẦU VIẾT BÀI THEO SLIDE ===\n"
|
| 1872 |
+
"Viết bài thành 5-6 ĐOẠN VĂN NG���N, mỗi đoạn là 1 SLIDE.\n"
|
| 1873 |
+
"\n"
|
| 1874 |
+
"SLIDE 1 - MỞ ĐẦU: Giới thiệu chủ đề, nêu quan điểm cá nhân (2-4 câu hoàn chỉnh)\n"
|
| 1875 |
+
"SLIDE 2-3-4-5 - PHÂN TÍCH: Mỗi slide dùng 1 nguồn tin cụ thể, kết hợp quan điểm cá nhân, ghi rõ nguồn (Theo VnExpress...), 2-4 câu hoàn chỉnh, thành 1 đoạn văn hoàn chỉnh\n"
|
| 1876 |
+
"SLIDE 6 - KẾT LUẬN: Tổng kết quan điểm, nhận định cuối cùng (2-3 câu hoàn chỉnh)\n"
|
| 1877 |
+
"\n"
|
| 1878 |
+
"Định dạng:\n"
|
| 1879 |
+
"---SLIDE 1---\n[đoạn văn hoàn chỉnh kết thúc bằng dấu chấm]\n---SLIDE 2---\n[đoạn văn hoàn chỉnh kết thúc bằng dấu chấm]\n...\n"
|
| 1880 |
+
"\n"
|
| 1881 |
+
"QUAN TRỌNG: Mỗi slide là 1 đoạn văn HOÀN CHỈNH, 2-4 câu, PHẢI KẾT THÚC BẰNG DẤU CHẤM (.). Kết hợp QUAN ĐIỂM + NGUỒN TIN. Không gạch đầu dòng. Viết liền mạch. 300-600 từ."
|
| 1882 |
+
)
|
| 1883 |
+
ai_text = None # Không dùng AI, để code tự kết hợp opinion + source
|
| 1884 |
+
except:
|
| 1885 |
+
pass
|
| 1886 |
+
|
| 1887 |
+
if not ai_text or len(ai_text) < 100:
|
| 1888 |
+
ai_text = "## " + title + "\n\n" + opinion + "\n\n"
|
| 1889 |
+
for i, sd in enumerate(source_details[:5]):
|
| 1890 |
+
ai_text += "### " + sd.get("title", "") + "\n"
|
| 1891 |
+
for p in sd.get("paragraphs", [])[:2]:
|
| 1892 |
+
ai_text += p[:250] + "\n"
|
| 1893 |
+
ai_text += "\n---\n*Nguồn: " + sd.get("via", "") + "*\n\n"
|
| 1894 |
+
|
| 1895 |
+
# Tạo slides
|
| 1896 |
+
if custom_slides and len(custom_slides) > 0:
|
| 1897 |
+
slides = []
|
| 1898 |
+
for i, slide in enumerate(custom_slides):
|
| 1899 |
+
slides.append({
|
| 1900 |
+
"text": slide.get("text", ""),
|
| 1901 |
+
"image": slide.get("image", ""),
|
| 1902 |
+
"index": i + 1
|
| 1903 |
+
})
|
| 1904 |
+
else:
|
| 1905 |
+
slides = []
|
| 1906 |
+
# Parse từ AI output (format: ---SLIDE N---)
|
| 1907 |
+
if ai_text:
|
| 1908 |
+
pattern = r'---SLIDE\s*(\d+)---\s*\n(.*?)(?=---SLIDE|\Z)'
|
| 1909 |
+
matches = re.findall(pattern, ai_text, re.DOTALL)
|
| 1910 |
+
if matches:
|
| 1911 |
+
for idx, (num, content) in enumerate(matches):
|
| 1912 |
+
# Normalize: ensure complete sentences
|
| 1913 |
+
text = _ensure_sentence_complete(content)
|
| 1914 |
+
if len(text) > 30:
|
| 1915 |
+
img = source_images[idx] if idx < len(source_images) else ""
|
| 1916 |
+
slides.append({"text": text, "image": img, "index": idx + 1})
|
| 1917 |
+
|
| 1918 |
+
# Fallback: split by paragraphs
|
| 1919 |
+
if len(slides) < 3:
|
| 1920 |
+
paragraphs = [p.strip() for p in re.split(r'\n\n+', ai_text) if p.strip()]
|
| 1921 |
+
slides = []
|
| 1922 |
+
para_count = 0
|
| 1923 |
+
for p in paragraphs:
|
| 1924 |
+
p = p.strip()
|
| 1925 |
+
if p.startswith('#') or p.startswith('---') or p.startswith('*Nguồn'):
|
| 1926 |
+
continue
|
| 1927 |
+
# Normalize: ensure complete sentences
|
| 1928 |
+
p_normalized = _ensure_sentence_complete(p)
|
| 1929 |
+
if len(p_normalized) > 50:
|
| 1930 |
+
img = source_images[para_count] if para_count < len(source_images) else ""
|
| 1931 |
+
slides.append({"text": p_normalized, "image": img, "index": para_count + 1})
|
| 1932 |
+
para_count += 1
|
| 1933 |
+
if para_count >= 6:
|
| 1934 |
+
break
|
| 1935 |
+
|
| 1936 |
+
if len(slides) < 2:
|
| 1937 |
+
slides = []
|
| 1938 |
+
# Slide 1: QUAN ĐIỂM CÁ NHÂN (BẮT BUỘC)
|
| 1939 |
+
slides.append({"text": f"Theo quan điểm cá nhân: {opinion[:400]}", "image": source_images[0] if source_images else "", "index": 1})
|
| 1940 |
+
|
| 1941 |
+
# Slide 2-6: KẾT HỢP QUAN ĐIỂM + SOURCE
|
| 1942 |
+
for i in range(min(5, len(source_details))):
|
| 1943 |
+
if len(slides) >= 6:
|
| 1944 |
+
break
|
| 1945 |
+
src = source_details[i]
|
| 1946 |
+
src_via = src.get("via", "")
|
| 1947 |
+
src_paras = src.get("paragraphs", [])
|
| 1948 |
+
|
| 1949 |
+
src_text = ""
|
| 1950 |
+
for p in src_paras[:2]:
|
| 1951 |
+
p = p.strip()[:280]
|
| 1952 |
+
if len(p) > 50:
|
| 1953 |
+
src_text = p
|
| 1954 |
+
break
|
| 1955 |
+
|
| 1956 |
+
if src_text:
|
| 1957 |
+
combined = f"Theo góc nhìn cá nhân, {opinion[:60]}. Theo {src_via}: {src_text}"
|
| 1958 |
+
img = source_images[len(slides)] if len(slides) < len(source_images) else (source_images[-1] if source_images else "")
|
| 1959 |
+
slides.append({"text": _ensure_sentence_complete(combined), "image": img, "index": len(slides) + 1})
|
| 1960 |
+
|
| 1961 |
+
lang, emotion = detect_language_and_emotion(title, ai_text)
|
| 1962 |
+
voice = get_voice_for_content(title, ai_text)
|
| 1963 |
+
|
| 1964 |
+
post = {
|
| 1965 |
+
"id": str(int(time.time() * 1000)) + str(_random2.randint(100, 999)),
|
| 1966 |
+
"title": title,
|
| 1967 |
+
"text": ai_text,
|
| 1968 |
+
"img": source_images[0] if source_images else "",
|
| 1969 |
+
"url": "",
|
| 1970 |
+
"kind": "personal_opinion",
|
| 1971 |
+
"slides": slides,
|
| 1972 |
+
"images": source_images[:10],
|
| 1973 |
+
"video": "",
|
| 1974 |
+
"voice": voice,
|
| 1975 |
+
"emotion": emotion,
|
| 1976 |
+
"language": lang,
|
| 1977 |
+
"ts": int(time.time()),
|
| 1978 |
+
"sources": source_details[:5]
|
| 1979 |
+
}
|
| 1980 |
+
|
| 1981 |
+
posts = _load_wall_posts()
|
| 1982 |
+
posts.insert(0, post)
|
| 1983 |
+
_save_wall_posts(posts)
|
| 1984 |
+
|
| 1985 |
+
return JSONResponse({"post": post, "slides": slides})
|
| 1986 |
+
|
| 1987 |
+
|
| 1988 |
+
# ===== END PERSONAL OPINION POST v2 =====
|
| 1989 |
+
|
| 1990 |
+
def _bg():
|
| 1991 |
+
time.sleep(15)
|
| 1992 |
+
while True:
|
| 1993 |
+
try:get_wc2026_all()
|
| 1994 |
+
except:pass
|
| 1995 |
+
time.sleep(90)
|
| 1996 |
+
threading.Thread(target=_bg,daemon=True).start()
|
| 1997 |
+
|
| 1998 |
+
# ===== AUTO SCHEDULER: rewrite AI + short at 7/13/19 VN time =====
|
| 1999 |
+
_AUTO_SCHEDULE_TIMES = [(7, '07:00'), (13, '13:00'), (19, '19:00')]
|
| 2000 |
+
_AUTO_LOG = os.path.join(DATA_DIR, 'auto_rewrite_log.json')
|
| 2001 |
+
|
| 2002 |
+
def _load_auto_log():
|
| 2003 |
+
try:
|
| 2004 |
+
if os.path.exists(_AUTO_LOG):
|
| 2005 |
+
with open(_AUTO_LOG, 'r') as f:
|
| 2006 |
+
return json.load(f)
|
| 2007 |
+
except: pass
|
| 2008 |
+
return {}
|
| 2009 |
+
|
| 2010 |
+
def _save_auto_log(log):
|
| 2011 |
+
try:
|
| 2012 |
+
tmp = _AUTO_LOG + '.tmp'
|
| 2013 |
+
with open(tmp, 'w') as f:
|
| 2014 |
+
json.dump(log, f)
|
| 2015 |
+
os.replace(tmp, _AUTO_LOG)
|
| 2016 |
+
except: pass
|
| 2017 |
+
|
| 2018 |
+
async def _auto_fetch_short(post_id):
|
| 2019 |
+
"""Try to auto-generate a short for a post."""
|
| 2020 |
+
try:
|
| 2021 |
+
import httpx
|
| 2022 |
+
async with httpx.AsyncClient(timeout=180) as cl:
|
| 2023 |
+
r = await cl.post(
|
| 2024 |
+
f"http://localhost:7860/api/ai/short/{post_id}",
|
| 2025 |
+
json={"voice":"vi-VN-HoaiMyNeural","emotion":"neutral","speed":1.2},
|
| 2026 |
+
headers={"Content-Type":"application/json"}
|
| 2027 |
+
)
|
| 2028 |
+
if r.status_code < 300:
|
| 2029 |
+
sj = r.json()
|
| 2030 |
+
if sj.get('video'):
|
| 2031 |
+
posts = _load_wall_posts()
|
| 2032 |
+
for p in posts:
|
| 2033 |
+
if p.get('id') == post_id:
|
| 2034 |
+
p['video'] = sj['video']
|
| 2035 |
+
break
|
| 2036 |
+
_save_wall_posts(posts)
|
| 2037 |
+
return True
|
| 2038 |
+
except: pass
|
| 2039 |
+
return False
|
| 2040 |
+
|
| 2041 |
+
async def _auto_rewrite_one(topic, slot_label, used_urls=None, post_index=0):
|
| 2042 |
+
"""Rewrite one topic: find articles, summarize, post to wall, trigger short.
|
| 2043 |
+
used_urls: shared set to avoid duplicate articles across topics.
|
| 2044 |
+
post_index: 0-based index to create multiple posts per topic (0,1,2 = up to 3 posts)."""
|
| 2045 |
+
from urllib.parse import quote as _q
|
| 2046 |
+
# Get MORE items to support 1-3 posts per topic
|
| 2047 |
+
items = _search_all(topic, limit=12)
|
| 2048 |
+
# Skip URLs already used by another topic
|
| 2049 |
+
if used_urls is not None:
|
| 2050 |
+
filtered = [it for it in items if it.get('url') not in used_urls]
|
| 2051 |
+
if filtered:
|
| 2052 |
+
items = filtered
|
| 2053 |
+
if not items or post_index >= len(items):
|
| 2054 |
+
return False
|
| 2055 |
+
|
| 2056 |
+
# Get article at post_index (0,1,2 for multiple posts)
|
| 2057 |
+
item = items[post_index] # post_index allows multiple articles per topic
|
| 2058 |
+
url = item.get('url', '')
|
| 2059 |
+
title = item.get('title', topic)
|
| 2060 |
+
if url and used_urls is not None:
|
| 2061 |
+
used_urls.add(url)
|
| 2062 |
+
if not url.startswith('http'):
|
| 2063 |
+
return False
|
| 2064 |
+
|
| 2065 |
+
data = _scrape_article_for_rewrite(url)
|
| 2066 |
+
if not data or not data.get('paragraphs'):
|
| 2067 |
+
return False
|
| 2068 |
+
|
| 2069 |
+
raw_text = '\n'.join(data['paragraphs'])
|
| 2070 |
+
ai_text = None
|
| 2071 |
+
|
| 2072 |
+
# Try AI generation
|
| 2073 |
+
try:
|
| 2074 |
+
import ai_ext
|
| 2075 |
+
prompt = f"Tóm tắt tin tức (tự động {slot_label}):\nTiêu đề: {data['title']}\n{raw_text[:10000]}\n\n4-6 ý chính dạng bullet. Cuối ghi nguồn."
|
| 2076 |
+
ai_text = await ai_ext.qwen_generate(prompt, max_tokens=1000)
|
| 2077 |
+
except: pass
|
| 2078 |
+
|
| 2079 |
+
if not ai_text or len(ai_text) < 80:
|
| 2080 |
+
pts = data['paragraphs'][:6]
|
| 2081 |
+
ai_text = '\n\n'.join([f"• {p[:300]}" for p in pts])
|
| 2082 |
+
via = item.get('via', '') or urlparse(url).netloc.replace('www.', '')
|
| 2083 |
+
ai_text += f"\n\nNguồn tham khảo: {via}"
|
| 2084 |
+
|
| 2085 |
+
# Build slides
|
| 2086 |
+
images = data.get('images', [])
|
| 2087 |
+
pts = data['paragraphs'][:10]
|
| 2088 |
+
slides = []
|
| 2089 |
+
for i, p in enumerate(pts[:8]):
|
| 2090 |
+
img = images[i] if i < len(images) else (images[-1] if images else data.get('og_img', ''))
|
| 2091 |
+
slides.append({'text': p[:300], 'image': img, 'index': i + 1})
|
| 2092 |
+
|
| 2093 |
+
post_id = str(int(time.time() * 1000)) + str(_random2.randint(100, 999))
|
| 2094 |
+
post = {
|
| 2095 |
+
"id": post_id, "title": data.get('title', title)[:200],
|
| 2096 |
+
"text": ai_text, "img": images[0] if images else data.get('og_img', ''),
|
| 2097 |
+
"url": url, "kind": "auto_rewrite", "slides": slides,
|
| 2098 |
+
"images": images[:10], "video": "",
|
| 2099 |
+
"voice": "vi-VN-HoaiMyNeural", "emotion": "neutral",
|
| 2100 |
+
"language": "vietnamese", "ts": int(time.time()),
|
| 2101 |
+
"auto_scheduled": True, "slot": slot_label,
|
| 2102 |
+
}
|
| 2103 |
+
|
| 2104 |
+
posts = _load_wall_posts()
|
| 2105 |
+
posts.insert(0, post)
|
| 2106 |
+
_save_wall_posts(posts)
|
| 2107 |
+
|
| 2108 |
+
# Trigger short generation async
|
| 2109 |
+
threading.Thread(target=lambda: asyncio.run(_auto_fetch_short(post_id)), daemon=True).start()
|
| 2110 |
+
return True
|
| 2111 |
+
|
| 2112 |
+
async def _do_scheduled_run(slot_label):
|
| 2113 |
+
"""Main scheduled run: 1-3 posts from 3 different HOT topics (3-9 total), no duplicates."""
|
| 2114 |
+
print(f"[auto] Starting scheduled rewrite for {slot_label}")
|
| 2115 |
+
|
| 2116 |
+
# Get top hot topics, skip duplicates
|
| 2117 |
+
all_topics = _get_hot_topics()
|
| 2118 |
+
seen_topics = set()
|
| 2119 |
+
unique_topics = []
|
| 2120 |
+
for t in all_topics:
|
| 2121 |
+
kw = t.get('topic', '').lower().strip()
|
| 2122 |
+
if kw and len(kw) > 5 and kw not in seen_topics:
|
| 2123 |
+
is_dup = False
|
| 2124 |
+
for s in seen_topics:
|
| 2125 |
+
# Check if one topic is substring of another
|
| 2126 |
+
if kw in s or s in kw:
|
| 2127 |
+
is_dup = True
|
| 2128 |
+
break
|
| 2129 |
+
if not is_dup:
|
| 2130 |
+
seen_topics.add(kw)
|
| 2131 |
+
unique_topics.append(t)
|
| 2132 |
+
if len(unique_topics) >= 3:
|
| 2133 |
+
break
|
| 2134 |
+
|
| 2135 |
+
job_topics = [t['topic'] for t in unique_topics[:3] if t.get('topic')]
|
| 2136 |
+
if not job_topics:
|
| 2137 |
+
print(f"[auto] No hot topics found, skipping")
|
| 2138 |
+
return
|
| 2139 |
+
|
| 2140 |
+
print(f"[auto] Running 3 topics: {job_topics}")
|
| 2141 |
+
|
| 2142 |
+
# Track used URLs to avoid cross-topic duplicates
|
| 2143 |
+
_used_urls = set()
|
| 2144 |
+
results = []
|
| 2145 |
+
|
| 2146 |
+
# Process each topic, create 1-3 posts per topic
|
| 2147 |
+
for jt in job_topics:
|
| 2148 |
+
for post_idx in range(3): # Try up to 3 posts per topic
|
| 2149 |
+
try:
|
| 2150 |
+
ok = await asyncio.wait_for(_auto_rewrite_one(jt, slot_label, _used_urls, post_idx), timeout=120)
|
| 2151 |
+
if ok:
|
| 2152 |
+
results.append((jt, post_idx, True))
|
| 2153 |
+
print(f"[auto] Created post {post_idx+1} for '{jt}'")
|
| 2154 |
+
else:
|
| 2155 |
+
# No more articles for this topic
|
| 2156 |
+
break
|
| 2157 |
+
except Exception as e:
|
| 2158 |
+
print(f"[auto] Error on '{jt}' post {post_idx}: {e}")
|
| 2159 |
+
results.append((jt, post_idx, False))
|
| 2160 |
+
await asyncio.sleep(1) # Small delay between posts
|
| 2161 |
+
|
| 2162 |
+
# Ensure at least 3 posts total (fallback if needed)
|
| 2163 |
+
successful_posts = sum(1 for _, _, ok in results if ok)
|
| 2164 |
+
print(f"[auto] Done {slot_label}: {successful_posts} posts created")
|
| 2165 |
+
|
| 2166 |
+
# Log
|
| 2167 |
+
from datetime import datetime, timezone, timedelta
|
| 2168 |
+
VN_TZ_SCHED = timezone(timedelta(hours=7))
|
| 2169 |
+
today_str = datetime.now(VN_TZ_SCHED).strftime('%Y-%m-%d')
|
| 2170 |
+
log = _load_auto_log()
|
| 2171 |
+
if today_str not in log: log[today_str] = {}
|
| 2172 |
+
log[today_str][slot_label] = {
|
| 2173 |
+
'time': datetime.now(VN_TZ_SCHED).strftime('%H:%M:%S'),
|
| 2174 |
+
'count': successful_posts,
|
| 2175 |
+
'total': len(job_topics),
|
| 2176 |
+
}
|
| 2177 |
+
_save_auto_log(log)
|
| 2178 |
+
|
| 2179 |
+
def _scheduler_loop():
|
| 2180 |
+
"""Check every 60s; trigger at 7:00, 13:00, 19:00 VN time.
|
| 2181 |
+
On startup, check for any missed slots today and run them immediately."""
|
| 2182 |
+
time.sleep(35)
|
| 2183 |
+
from datetime import datetime, timezone, timedelta
|
| 2184 |
+
VN_TZ_SCHED = timezone(timedelta(hours=7))
|
| 2185 |
+
|
| 2186 |
+
_last_run_date = ""
|
| 2187 |
+
_last_run_slots = set()
|
| 2188 |
+
|
| 2189 |
+
# On startup: check log for missed slots today
|
| 2190 |
+
try:
|
| 2191 |
+
start_now = datetime.now(VN_TZ_SCHED)
|
| 2192 |
+
today_str = start_now.strftime('%Y-%m-%d')
|
| 2193 |
+
current_hour = start_now.hour
|
| 2194 |
+
current_minute = start_now.minute
|
| 2195 |
+
log = _load_auto_log()
|
| 2196 |
+
today_log = log.get(today_str, {})
|
| 2197 |
+
for h, label in _AUTO_SCHEDULE_TIMES:
|
| 2198 |
+
# Run if slot is past (either strictly earlier hour, or same hour but window has passed)
|
| 2199 |
+
should_run = False
|
| 2200 |
+
if h < current_hour:
|
| 2201 |
+
should_run = True
|
| 2202 |
+
elif h == current_hour and current_minute > 10:
|
| 2203 |
+
should_run = True
|
| 2204 |
+
if should_run and label not in today_log:
|
| 2205 |
+
print(f"[auto] Detected missed slot {label} (h={h} < now={current_hour}:{current_minute}), running catch-up now")
|
| 2206 |
+
_run_scheduled_sync(label)
|
| 2207 |
+
_last_run_slots.add(label)
|
| 2208 |
+
except Exception as e:
|
| 2209 |
+
print(f"[auto] Catch-up check error: {e}")
|
| 2210 |
+
|
| 2211 |
+
while True:
|
| 2212 |
+
try:
|
| 2213 |
+
now = datetime.now(VN_TZ_SCHED)
|
| 2214 |
+
today = now.strftime('%Y-%m-%d')
|
| 2215 |
+
hour = now.hour
|
| 2216 |
+
minute = now.minute
|
| 2217 |
+
|
| 2218 |
+
if today != _last_run_date:
|
| 2219 |
+
_last_run_date = today
|
| 2220 |
+
_last_run_slots = set()
|
| 2221 |
+
|
| 2222 |
+
slot = None
|
| 2223 |
+
for h, label in _AUTO_SCHEDULE_TIMES:
|
| 2224 |
+
if hour == h and 0 <= minute < 5:
|
| 2225 |
+
slot = label
|
| 2226 |
+
break
|
| 2227 |
+
|
| 2228 |
+
if slot and slot not in _last_run_slots:
|
| 2229 |
+
_last_run_slots.add(slot)
|
| 2230 |
+
_run_scheduled_sync(slot)
|
| 2231 |
+
except Exception as e:
|
| 2232 |
+
print(f"[auto] Loop error: {e}")
|
| 2233 |
+
|
| 2234 |
+
time.sleep(60)
|
| 2235 |
+
|
| 2236 |
+
threading.Thread(target=_scheduler_loop, daemon=True, name='auto-rewrite-scheduler').start()
|
| 2237 |
+
|
| 2238 |
+
@app.get('/api/debug/auto_schedule')
|
| 2239 |
+
async def debug_auto_schedule(slot: str = '07:00'):
|
| 2240 |
+
"""Manually trigger auto scheduler for debugging."""
|
| 2241 |
+
try:
|
| 2242 |
+
# Check if we can access the data directory
|
| 2243 |
+
log = _load_auto_log()
|
| 2244 |
+
topics = _get_hot_topics()[:3]
|
| 2245 |
+
job_topics = [t['topic'] for t in topics if t.get('topic')]
|
| 2246 |
+
return JSONResponse({
|
| 2247 |
+
"slot": slot,
|
| 2248 |
+
"log": log,
|
| 2249 |
+
"hot_topics": job_topics,
|
| 2250 |
+
"wall_posts_count": len(_load_wall_posts()),
|
| 2251 |
+
"data_dir_writable": os.access(DATA_DIR, os.W_OK) if os.path.isdir(DATA_DIR) else False,
|
| 2252 |
+
"data_dir_exists": os.path.isdir(DATA_DIR),
|
| 2253 |
+
})
|
| 2254 |
+
except Exception as e:
|
| 2255 |
+
return JSONResponse({"error": str(e)}, status_code=500)
|
| 2256 |
+
|
| 2257 |
+
def _run_scheduled_sync(slot):
|
| 2258 |
+
"""Run _do_scheduled_run in a separate event loop (for background thread)."""
|
| 2259 |
+
loop = asyncio.new_event_loop()
|
| 2260 |
+
asyncio.set_event_loop(loop)
|
| 2261 |
+
try:
|
| 2262 |
+
loop.run_until_complete(_do_scheduled_run(slot))
|
| 2263 |
+
except Exception as e:
|
| 2264 |
+
print(f"[auto] Background run error: {e}")
|
| 2265 |
+
finally:
|
| 2266 |
+
loop.close()
|
| 2267 |
+
|
| 2268 |
+
@app.get('/api/debug/trigger_auto')
|
| 2269 |
+
async def debug_trigger_auto(slot: str = '19:00'):
|
| 2270 |
+
"""Trigger _do_scheduled_run in background thread (non-blocking)."""
|
| 2271 |
+
threading.Thread(target=_run_scheduled_sync, args=(slot,), daemon=True).start()
|
| 2272 |
+
return JSONResponse({"status": "started", "slot": slot})
|
| 2273 |
+
|
| 2274 |
+
# ===== SHORTS RSS PROXY ENDPOINT =====
|
| 2275 |
+
@app.get("/api/shorts/rss")
|
| 2276 |
+
def shorts_rss():
|
| 2277 |
+
"""Get shorts from YouTube RSS feeds server-side"""
|
| 2278 |
+
import xml.etree.ElementTree as ET
|
| 2279 |
+
import html as html_lib2
|
| 2280 |
+
import re as re2
|
| 2281 |
+
|
| 2282 |
+
YOUTUBE_CHANNELS = {
|
| 2283 |
+
"baodantri7941": "UC_x5TKhOgd6GhYvv5z4I3jg",
|
| 2284 |
+
"baosuckhoedoisongboyte": "UCBsY5fXTQLkF_JnH9kLkL4g",
|
| 2285 |
+
}
|
| 2286 |
+
|
| 2287 |
+
shorts = []
|
| 2288 |
+
seen = set()
|
| 2289 |
+
|
| 2290 |
+
for handle, channel_id in YOUTUBE_CHANNELS.items():
|
| 2291 |
+
try:
|
| 2292 |
+
rss_url = f"https://www.youtube.com/feeds/videos.xml?channel_id={channel_id}"
|
| 2293 |
+
r = req.get(rss_url, headers=HEADERS, timeout=15)
|
| 2294 |
+
if r.status_code != 200:
|
| 2295 |
+
continue
|
| 2296 |
+
|
| 2297 |
+
root = ET.fromstring(r.text)
|
| 2298 |
+
ns = {
|
| 2299 |
+
'atom': 'http://www.w3.org/2005/Atom',
|
| 2300 |
+
'yt': 'http://www.youtube.com/xml/schemas/2015',
|
| 2301 |
+
'media': 'http://search.yahoo.com/mrss/'
|
| 2302 |
+
}
|
| 2303 |
+
|
| 2304 |
+
for entry in root.findall('atom:entry', ns)[:30]:
|
| 2305 |
+
title_el = entry.find('atom:title', ns)
|
| 2306 |
+
title = html_lib2.unescape(title_el.text) if title_el is not None and title_el.text else ''
|
| 2307 |
+
|
| 2308 |
+
link_el = entry.find('atom:link', ns)
|
| 2309 |
+
link = link_el.get('href', '') if link_el is not None else ''
|
| 2310 |
+
|
| 2311 |
+
vid_el = entry.find('yt:videoId', ns)
|
| 2312 |
+
vid = vid_el.text if vid_el is not None else ''
|
| 2313 |
+
|
| 2314 |
+
if not vid or vid in seen:
|
| 2315 |
+
continue
|
| 2316 |
+
|
| 2317 |
+
# Check if it's a short
|
| 2318 |
+
is_short = '#shorts' in title.lower() or '#short' in title.lower() or '/shorts/' in link
|
| 2319 |
+
|
| 2320 |
+
if not is_short:
|
| 2321 |
+
desc_el = entry.find('media:description', ns)
|
| 2322 |
+
if desc_el is not None and desc_el.text:
|
| 2323 |
+
if '#shorts' in desc_el.text.lower():
|
| 2324 |
+
is_short = True
|
| 2325 |
+
|
| 2326 |
+
if not is_short:
|
| 2327 |
+
continue
|
| 2328 |
+
|
| 2329 |
+
seen.add(vid)
|
| 2330 |
+
|
| 2331 |
+
# Get thumbnail
|
| 2332 |
+
thumb = f"https://i.ytimg.com/vi/{vid}/hqdefault.jpg"
|
| 2333 |
+
media_group = entry.find('media:group', ns)
|
| 2334 |
+
if media_group is not None:
|
| 2335 |
+
thumb_el = media_group.find('media:thumbnail', ns)
|
| 2336 |
+
if thumb_el is not None:
|
| 2337 |
+
thumb = thumb_el.get('url', thumb)
|
| 2338 |
+
|
| 2339 |
+
shorts.append({
|
| 2340 |
+
'id': vid,
|
| 2341 |
+
'title': title.replace('#shorts', '').replace('#short', '').strip()[:120],
|
| 2342 |
+
'img': thumb,
|
| 2343 |
+
'link': f'https://www.youtube.com/shorts/{vid}',
|
| 2344 |
+
'channel': handle,
|
| 2345 |
+
'source': 'yt'
|
| 2346 |
+
})
|
| 2347 |
+
|
| 2348 |
+
if len(shorts) >= 40:
|
| 2349 |
+
break
|
| 2350 |
+
|
| 2351 |
+
except Exception as e:
|
| 2352 |
+
print(f"RSS error for {handle}: {e}")
|
| 2353 |
+
continue
|
| 2354 |
+
|
| 2355 |
+
return {"shorts": shorts, "count": len(shorts)}
|
| 2356 |
+
|
| 2357 |
+
app.mount('/static',StaticFiles(directory=STATIC_DIR),name='vnews_static')
|