Upload app_v2_entry.py

app_v2_entry.py

@@ -1,4 +1,4 @@
-"""VNEWS v2 Entry Point - with AI endpoints + fast bongda proxy"""
+"""VNEWS v2 Entry Point - with fast bongda proxy"""
 import sys, os
 from main import app, HEADERS, BONGDA_HEADERS, fetch_bongda_api, HL_LEAGUES
 
@@ -12,10 +12,10 @@ from fastapi.staticfiles import StaticFiles
 from starlette.routing import Mount
 from fastapi import Query, Request
 import requests as req
-from urllib.parse import quote, urlparse
 from bs4 import BeautifulSoup
 import re, html as html_lib, json, threading, time
 from concurrent.futures import ThreadPoolExecutor, as_completed
+from urllib.parse import quote
 
 HL_LEAGUES['friendly'] = {"path": "giai-khac/friendly", "name": "Giao hữu", "emoji": "🤝"}
 
@@ -25,7 +25,162 @@ app.routes[:]=[r for r in app.routes if not isinstance(r, Mount)]
 app.router.routes=[r for r in app.router.routes if not isinstance(r, Mount)]
 
 def _clean(s): return re.sub(r"\s+"," ",html_lib.unescape(str(s or""))).strip()
+
+# Cache for match details (5 min TTL)
+_match_cache = {}
+
+# === FAST BONGDA PROXY ENDPOINT ===
+def _get_match_detail(event_id, slug=None):
+    """Internal function to scrape match detail from bongda.com.vn"""
+    headers = {"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36", "Accept": "text/html", "Referer": "https://bongda.com.vn/"}
+    
+    if slug:
+        url = f"https://bongda.com.vn/tran-dau/{event_id}/centre/{slug}"
+    else:
+        url = f"https://bongda.com.vn/tran-dau/{event_id}"
+    
+    resp = req.get(url, headers=headers, timeout=15, allow_redirects=True)
+    if resp.status_code != 200:
+        return None
+    
+    soup = BeautifulSoup(resp.text, 'html.parser')
+    result = {"event_id": event_id, "found": False, "sections": []}
+    info = {}
+    
+    tel = soup.select_one('.teams')
+    if tel:
+        he = tel.select_one('.team.home')
+        if he:
+            p_tags = [p for p in he.select('p') if not p.get('class') or 'logo' not in p.get('class', [])]
+            if p_tags: info['home_team'] = _clean(p_tags[0].get_text())
+            lo = he.select_one('img')
+            if lo: info['home_logo'] = lo.get('src', '')
+        ae = tel.select_one('.team.away')
+        if ae:
+            p_tags = ae.select('p')
+            team_ps = [p for p in p_tags if not p.get('class') or 'logo' not in p.get('class', [])]
+            if team_ps: info['away_team'] = _clean(team_ps[-1].get_text())
+            lo = ae.select_one('img')
+            if lo: info['away_logo'] = lo.get('src', '')
+        sc = tel.select_one('.score')
+        if sc:
+            parts = [_clean(p.get_text()) for p in sc.select('p')]
+            if len(parts) >= 2: info['score'] = f"{parts[0]} - {parts[1]}"
+            lb = sc.select_one('.label')
+            if lb: info['status_label'] = _clean(lb.get_text())
+    
+    if info.get('home_team') and info.get('away_team'):
+        result['info'] = info
+        result['found'] = True
+        result['sections'].append('info')
+    
+    events = []
+    for ev in soup.select('.events .period .event'):
+        ev_cls = ' '.join(ev.get('class', []))
+        ev_data = {'team': 'home' if 'home' in ev_cls else 'away', 'period': '', 'type': 'unknown', 'time': '', 'players': ''}
+        
+        parent = ev.parent
+        if parent:
+            h2 = parent.find('h2')
+            if h2: ev_data['period'] = _clean(h2.get_text())
+        
+        if ev.select_one('[class*="goal"]'): ev_data['type'] = 'goal'
+        elif ev.select_one('[class*="redcard"]'): ev_data['type'] = 'redcard'
+        elif ev.select_one('[class*="yellowcard"]'): ev_data['type'] = 'yellowcard'
+        elif ev.select_one('[class*="substitution"]'): ev_data['type'] = 'substitution'
+        
+        players_el = ev.select_one('.players')
+        if players_el:
+            pl_text = _clean(players_el.get_text(' ', strip=True))
+            m = re.match(r"(\d+)'(.*)", pl_text)
+            if m:
+                ev_data['time'] = f"{m.group(1)}'"
+                ev_data['players'] = m.group(2)
+            else:
+                ev_data['players'] = pl_text
+        events.append(ev_data)
+    
+    if events:
+        result['events'] = events
+        result['sections'].append('events')
+    
+    pred = soup.select_one('.prediction-card')
+    if pred:
+        team_info = pred.select_one('.team-info')
+        if team_info:
+            teams = team_info.select('.team')
+            pred_data = {}
+            if len(teams) >= 2:
+                pred_data['home_name'] = _clean(teams[0].select_one('.team-name').get_text()) if teams[0].select_one('.team-name') else ''
+                pred_data['away_name'] = _clean(teams[1].select_one('.team-name').get_text()) if teams[1].select_one('.team-name') else ''
+                divider = team_info.select_one('.divider')
+                if divider: pred_data['result'] = _clean(divider.get_text())
+                vc = pred.select_one('.vote-count')
+                if vc: pred_data['vote_count'] = _clean(vc.get_text())
+            result['prediction'] = pred_data
+    
+    recent = []
+    ml = soup.select_one('.matches-list')
+    if ml:
+        for item in ml.select('.match-detail, .match-item, li'):
+            de = item.select_one('.date, .time')
+            le = item.select_one('.league')
+            he_item = item.select_one('.home, .team-home')
+            ae_item = item.select_one('.away, .team-away')
+            se = item.select_one('.score, .result')
+            if he_item or ae_item:
+                recent.append({'date': _clean(de.get_text()) if de else '', 'league': _clean(le.get_text()) if le else '', 'home': _clean(he_item.get_text()) if he_item else '', 'away': _clean(ae_item.get_text()) if ae_item else '', 'score': _clean(se.get_text()) if se else 'vs'})
+    if recent:
+        result['recent_matches'] = recent
+        result['sections'].append('recent')
+    
+    try:
+        api_h = {"User-Agent": "Mozilla/5.0", "Accept": "application/json", "X-Requested-With": "XMLHttpRequest", "Referer": "https://bongda.com.vn/"}
+        ar = req.get(f"https://bongda.com.vn/api/fixtures/h2h-stats?event_id={event_id}", headers=api_h, timeout=10)
+        if ar.status_code == 200:
+            ad = ar.json()
+            if ad.get('status') == 'success' and ad.get('html'):
+                asp = BeautifulSoup(ad['html'], 'html.parser')
+                ast = {}
+                for row in asp.select('li, tr'):
+                    cells = row.select('td, span, p')
+                    if len(cells) >= 3:
+                        lb = _clean(cells[0].get_text())
+                        if lb: ast[lb] = {'home': _clean(cells[1].get_text()), 'away': _clean(cells[2].get_text())}
+                if ast:
+                    result['h2h_stats_parsed'] = ast
+                    result['sections'].append('h2h_stats')
+    except: pass
+    
+    return result
+
+@app.get('/api/proxy/bongda')
+def proxy_bongda(event_id: int = Query(default=None), slug: str = Query(default=None)):
+    if event_id is None:
+        return JSONResponse({'error': 'event_id required'}, status_code=400)
+    
+    cache_key = f"{event_id}_{slug}"
+    now = time.time()
+    cached = _match_cache.get(cache_key)
+    if cached and now - cached.get('_ts', 0) < 300:
+        return JSONResponse(cached)
+    
+    try:
+        result = _get_match_detail(event_id, slug)
+        if result:
+            result['_ts'] = now
+            _match_cache[cache_key] = result
+            return JSONResponse(result)
+    except Exception as e:
+        err = {"event_id": event_id, "found": False, "error": str(e), "_ts": now}
+        _match_cache[cache_key] = err
+        return JSONResponse(err)
+    
+    return JSONResponse({"event_id": event_id, "found": False})
+
+# === Rest of endpoints (existing) ===
 _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())
+
 def _has_kw(topic,title):
     tl=topic.lower();tt=(title or'').lower()
     if tl in tt:return True
@@ -33,7 +188,6 @@ def _has_kw(topic,title):
     if not words:return True
     return any(w in tt for w in words)
 
-# Search functions...
 def _s_vnexpress(topic,limit=8):
     items=[]
     try:
@@ -236,149 +390,6 @@ def _st():return JSONResponse({'persistent':os.path.isdir('/data') and os.access
 @app.get('/s')
 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>')
 
-# === BONGDA PROXY (v5-fast) ===
-@app.get('/api/proxy/bongda')
-def proxy_bongda(event_id: int = Query(default=None)):
-    """Fast server-side proxy for bongda.com.vn match detail."""
-    if event_id is None:
-        return JSONResponse({'error': 'event_id required'}, status_code=400)
-    
-    result = {"event_id": event_id, "found": False, "sections": []}
-    
-    try:
-        url = f"https://bongda.com.vn/tran-dau/{event_id}/centre/"
-        headers = {"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36", "Accept": "text/html", "Referer": "https://bongda.com.vn/"}
-        resp = req.get(url, headers=headers, timeout=15, allow_redirects=True)
-        
-        if resp.status_code != 200:
-            return JSONResponse(result)
-        
-        soup = BeautifulSoup(resp.text, 'html.parser')
-        info = {}
-        
-        tel = soup.select_one('.teams')
-        if tel:
-            he = tel.select_one('.team.home')
-            if he:
-                ne = he.select_one('p:not(.logo)') or he.find('p')
-                if ne: info['home_team'] = _clean(ne.get_text())
-                lo = he.select_one('img')
-                if lo: info['home_logo'] = lo.get('src', '')
-            ae = tel.select_one('.team.away')
-            if ae:
-                ne = ae.select_one('p:not(.logo)') or ae.find('p')
-                if ne: info['away_team'] = _clean(ne.get_text())
-                lo = ae.select_one('img')
-                if lo: info['away_logo'] = lo.get('src', '')
-            sc = tel.select_one('.score')
-            if sc:
-                parts = [_clean(p.get_text()) for p in sc.select('p')]
-                if len(parts) >= 2: info['score'] = f"{parts[0]} - {parts[1]}"
-                lb = sc.select_one('.label')
-                if lb: info['status_label'] = _clean(lb.get_text())
-        
-        if info.get('home_team') and info.get('away_team'):
-            result['info'] = info
-            result['found'] = True
-            result['sections'].append('info')
-        
-        events = []
-        ev_div = soup.select_one('.events')
-        if ev_div:
-            period = ''
-            for child in ev_div.children:
-                if not hasattr(child, 'name') or not child.name: continue
-                cls = ' '.join(child.get('class', []))
-                if 'period' in cls:
-                    h2 = child.find('h2')
-                    if h2: period = _clean(h2.get_text())
-                    for ev in child.children:
-                        if not hasattr(ev, 'name') or not ev.name: continue
-                        ev_cls = ' '.join(ev.get('class', []))
-                        if 'event' not in ev_cls: continue
-                        
-                        ev_data = {
-                            'team': 'home' if 'home' in ev_cls else 'away',
-                            'period': period,
-                            'type': 'unknown',
-                            'time': '',
-                            'players': '',
-                        }
-                        
-                        type_el = ev.select_one('.event-type')
-                        if type_el:
-                            if type_el.select_one('[class*="redcard"]'): ev_data['type'] = 'redcard'
-                            elif type_el.select_one('[class*="yellowcard"]'): ev_data['type'] = 'yellowcard'
-                            elif type_el.select_one('[class*="goal"]'): ev_data['type'] = 'goal'
-                            elif type_el.select_one('[class*="substitution"]'): ev_data['type'] = 'substitution'
-                        
-                        players_el = ev.select_one('.players')
-                        if players_el:
-                            time_el = players_el.select_one('.event-time')
-                            if time_el: ev_data['time'] = _clean(time_el.get_text())
-                            ev_data['players'] = _clean(players_el.get_text(' ', strip=True))
-                        
-                        events.append(ev_data)
-        
-        if events:
-            result['events'] = events
-            result['sections'].append('events')
-        
-        pred = soup.select_one('.prediction-card')
-        if pred:
-            team_info = pred.select_one('.team-info')
-            if team_info:
-                teams = team_info.select('.team')
-                if len(teams) >= 2:
-                    pred_data = {}
-                    pred_data['home_name'] = _clean(teams[0].select_one('.team-name').get_text()) if teams[0].select_one('.team-name') else ''
-                    pred_data['away_name'] = _clean(teams[1].select_one('.team-name').get_text()) if teams[1].select_one('.team-name') else ''
-                    divider = team_info.select_one('.divider')
-                    if divider: pred_data['result'] = _clean(divider.get_text())
-                    vc = pred.select_one('.vote-count')
-                    if vc: pred_data['vote_count'] = _clean(vc.get_text())
-                    result['prediction'] = pred_data
-        
-        recent = []
-        ml = soup.select_one('.matches-list')
-        if ml:
-            for item in ml.select('.match-detail, .match-item, li'):
-                de = item.select_one('.date, .time')
-                le = item.select_one('.league')
-                he_item = item.select_one('.home, .team-home')
-                ae_item = item.select_one('.away, .team-away')
-                se = item.select_one('.score, .result')
-                if he_item or ae_item:
-                    recent.append({
-                        'date': _clean(de.get_text()) if de else '', 'league': _clean(le.get_text()) if le else '',
-                        'home': _clean(he_item.get_text()) if he_item else '', 'away': _clean(ae_item.get_text()) if ae_item else '',
-                        'score': _clean(se.get_text()) if se else 'vs',
-                    })
-        if recent:
-            result['recent_matches'] = recent
-            result['sections'].append('recent')
-        
-        try:
-            api_h = {"User-Agent": "Mozilla/5.0", "Accept": "application/json", "X-Requested-With": "XMLHttpRequest", "Referer": "https://bongda.com.vn/"}
-            ar = req.get(f"https://bongda.com.vn/api/fixtures/h2h-stats?event_id={event_id}", headers=api_h, timeout=10)
-            if ar.status_code == 200:
-                ad = ar.json()
-                if ad.get('status') == 'success' and ad.get('html'):
-                    asp = BeautifulSoup(ad['html'], 'html.parser')
-                    ast = {}
-                    for row in asp.select('li, tr'):
-                        cells = row.select('td, span, p')
-                        if len(cells) >= 3:
-                            lb = _clean(cells[0].get_text())
-                            if lb: ast[lb] = {'home': _clean(cells[1].get_text()), 'away': _clean(cells[2].get_text())}
-                    if ast: result['h2h_stats_parsed'] = ast; result['sections'].append('h2h_stats')
-        except: pass
-        
-        return JSONResponse(result)
-    except Exception as e:
-        return JSONResponse({"event_id": event_id, "found": False, "error": str(e)})
-
-# === WC2026 endpoints ===
 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)
 @app.get('/api/wc2026')
 def _w():return JSONResponse(get_wc2026_all())
@@ -444,9 +455,4 @@ def _bg():
         time.sleep(90)
 threading.Thread(target=_bg,daemon=True).start()
 
-app.mount('/static',StaticFiles(directory=STATIC_DIR),name='vnews_static')
-
-# _run.py helper
-RUN_PY_CONTENT = 'from app_v2_entry import app'
-with open('/app/_run.py', 'w') as f:
-    f.write(RUN_PY_CONTENT)
+app.mount('/static',StaticFiles(directory=STATIC_DIR),name='vnews_static')