push

utils/llm_handler.py

@@ -239,22 +239,74 @@ Include:
 
     def check_educational(self, transcript: str, title: str = "") -> bool:
         """
-        FIX: Pehle sirf obvious non-educational content reject karo.
-        Doubt ho to ALLOW karo — reject mat karo.
+        Step 1: Title se check karo (agar available ho)
+        Step 2: Transcript keywords check karo
+        Step 3: Har case mein AI se final confirm karo
         """
 
-        # Sirf ye cheezein clearly reject karo
-        hard_reject_keywords = [
+        title_lower = title.lower().strip()
+        text = (transcript[:3000] + " " + title).lower()
+
+        # ─── STEP 1: TITLE check (sirf tab jab title available ho) ───
+        if title_lower:
+            title_reject_patterns = [
+                # Songs / Music
+                'official music video', 'official video', 'official audio',
+                'lyrics video', 'lyric video', 'music video',
+                'live performance', 'live at', 'concert', 'official mv',
+                '(mv)', 'audio song', 'full song', 'new song',
+                # Vlogs
+                'vlog', 'day in my life', 'daily routine', 'morning routine',
+                'night routine', 'week in my life', 'come with me',
+                'grwm', 'get ready with me',
+                # Cartoons / Entertainment
+                'cartoon', 'anime', 'episode', 'season', 'animated series',
+                'full episode', 'dubbed', 'subtitled',
+                # Other entertainment
+                'reaction video', 'reacting to', 'prank', 'challenge',
+                'funny moments', 'highlights', 'compilation',
+                'trailer', 'teaser', 'behind the scenes',
+            ]
+            for pattern in title_reject_patterns:
+                if pattern in title_lower:
+                    return False
+
+        # ─── STEP 2: TRANSCRIPT keyword check ───
+
+        # Ye hon to → TURANT REJECT (1 bhi kafi hai)
+        strong_reject_single = [
+            'official music video', 'music video',
+            'verse 1', 'verse 2', 'chorus', 'pre-chorus', 'bridge',  # song structure
+            'subscribe to my channel', 'like share subscribe',
+            'aaj ka vlog', 'aaj ki vlog', 'today\'s vlog',
+            'cartoon network', 'nickelodeon',
+        ]
+        for kw in strong_reject_single:
+            if kw in text:
+                return False
+
+        # Ye 2+ milein to → REJECT
+        soft_reject_keywords = [
+            # Song related
+            'lyrics', 'singer', 'hook', 'beat drop', 'rap', 'singing',
+            # Vlog related
+            'vlog', 'daily routine', 'morning routine', 'outfit', 'haul',
+            'grwm', 'storytime', 'q&a', 'q and a',
+            # Cartoon related
+            'cartoon', 'anime', 'animation', 'animated', 'episode', 'character',
+            'villain', 'superhero', 'pixar', 'disney',
+            # Cooking
             'recipe', 'cooking', 'ingredient', 'tablespoon', 'teaspoon',
-            'karahi', 'biryani', 'pakana', 'khana banana',
-            'drama serial', 'episode dekho', 'actor', 'actress',
-            'song lyrics', 'music video', 'concert',
-            'makeup tutorial', 'skincare routine', 'fashion haul',
-            'funny prank', 'comedy sketch',
-            'gaming gameplay', 'game stream',
+            'karahi', 'biryani', 'pakana', 'khana banana', 'masala',
+            # Other
+            'makeup', 'skincare', 'prank', 'challenge', 'funny',
+            'gameplay', 'game stream', 'drama serial',
         ]
+        reject_count = sum(1 for kw in soft_reject_keywords if kw in text)
+        if reject_count >= 2:
+            return False
 
-        # Ye hain to definitely educational hai
+        # ─── STEP 3: Educational keywords → ALLOW ───
         hard_accept_keywords = [
             'lecture', 'lesson', 'chapter', 'tutorial', 'course',
             'university', 'college', 'school', 'class',
@@ -267,20 +319,44 @@ Include:
             'accounting', 'finance', 'networking', 'compiler',
             'parh', 'seekhna', 'samajhna', 'taleem', 'ilm', 'sabaq',
             'teacher', 'student', 'syllabus', 'notes',
-            'explain', 'definition', 'concept', 'theory',
+            'explain', 'definition', 'concept', 'theory', 'topic',
         ]
-
-        text = (transcript[:3000] + " " + title).lower()
-
-        # Koi bhi hard_accept mila → turant allow
         for kw in hard_accept_keywords:
             if kw in text:
                 return True
 
-        # Sirf tab reject karo jab 3+ hard_reject keywords hon
-        reject_count = sum(1 for kw in hard_reject_keywords if kw in text)
-        if reject_count >= 3:
-            return False
-
-        # Baaki sab cases mein → ALLOW (doubt ka faida user ko do)
-        return True
+        # ─── STEP 4: Kuch clear nahi → AI se ZAROOR pocho ───
+        # Title ho ya na ho — AI transcript se analyse karega
+        try:
+            title_info = f"Title: {title}" if title_lower else "Title: (not available)"
+            system_prompt = """You are a strict content classifier. Return ONLY 'yes' or 'no'. No explanation.
+
+RETURN 'yes' ONLY IF the content is clearly one of:
+- University or school lecture
+- Coding or programming tutorial
+- Academic subject explanation (math, science, history, etc.)
+- Professional skill training
+
+RETURN 'no' IF the content is:
+- Song, music video, lyrics
+- Vlog, daily routine, lifestyle content
+- Cartoon, anime, animated show
+- Cooking, recipe
+- Drama, TV show, movie
+- Gaming, gameplay
+- Comedy, prank, challenge
+- News, entertainment
+- Fashion, beauty, makeup
+- Travel, tourism"""
+
+            user_prompt = f"""{title_info}
+Transcript (first 1000 chars):
+{transcript[:1000]}
+
+Is this educational content? Answer yes or no only."""
+
+            response = self._call_llm(system_prompt, user_prompt, max_tokens=5)
+            return 'yes' in response.lower().strip()
+        except:
+            # AI fail ho jaye to safe side → reject
+            return False