Update scripts/generate_playlist.py

scripts/generate_playlist.py

@@ -1,7 +1,8 @@
 #!/usr/bin/env python3
 """
-IPTV Traditional Broadcasters Only - Simplified Version
-Clean, minimal code with no redundancy
+IPTV Enhanced Country Detection - Updated Version
+Uses 3-point analysis: Channel Name + EPG ID + Logo URL
+Then filters to keep only legitimate countries
 """
 
 import os
@@ -14,56 +15,188 @@ script_dir = Path(__file__).parent
 root_dir = script_dir.parent
 os.chdir(root_dir)
 
-def is_streaming_platform(all_text):
-    """Check if channel is from streaming platform (simplified)."""
-    platforms = ["plex.tv", "pluto.tv", "jmp2.uk/plu-", "sam-", "samsung", "tubi", "xumo", "roku", "youtube"]
-    return any(platform in all_text for platform in platforms)
+def detect_country_from_channel_content(channel_name, epg_id="", logo_url="", stream_url=""):
+    """
+    Enhanced country detection using 3-point analysis
+    Priority: EPG ID > Logo URL > Channel Name > Stream URL
+    """
     
-def detect_traditional_broadcaster_country(channel_name, epg_id="", logo_url="", stream_url=""):
-    """Detect country for traditional broadcasters only."""
+    # Combine all text for analysis
     all_text = f"{channel_name.lower()} {epg_id.lower()} {logo_url.lower()} {stream_url.lower()}"
-    channel_lower = channel_name.lower()
     
-    # Skip streaming platforms
-    if is_streaming_platform(all_text):
-        return "Uncategorized"
+    # STEP 1: Check for streaming services first (these go to Uncategorized)
+    streaming_services = [
+        "plex", "pluto", "tubi", "samsung", "xumo", "stirr", "crackle", "imdb tv",
+        "daddylive", "drew247", "aixmedia", "moveonjoy", "drewlive24", "udptv",
+        "a1xs.vip", "zekonew", "forcedtoplay", "cdn1host", "tvpass.org",
+        "jmp2.uk/plu-", "provider-static.plex.tv", "images.pluto.tv"
+    ]
     
-    # EPG ID detection (most reliable)
-    epg_countries = {
-        ".ca": "🇨🇦 Canada", ".us": "🇺🇸 United States", ".uk": "🇬🇧 United Kingdom",
-        ".ph": "🇵🇭 Philippines", ".au": "🇦🇺 Australia", ".jp": "🇯🇵 Japan",
-        ".my": "🇲🇾 Malaysia", ".de": "🇩🇪 Germany", ".fr": "🇫🇷 France",
-        ".es": "🇪🇸 Spain", ".it": "🇮🇹 Italy", ".br": "🇧🇷 Brazil"
+    for service in streaming_services:
+        if service in all_text:
+            return "Uncategorized"
+    
+    # STEP 2: EPG ID detection (most reliable)
+    epg_patterns = {
+        ".ca": "🇨🇦 Canada",
+        ".us": "🇺🇸 United States", 
+        ".uk": "🇬🇧 United Kingdom",
+        ".ph": "🇵🇭 Philippines",
+        ".au": "🇦🇺 Australia",
+        ".jp": "🇯🇵 Japan",
+        ".my": "🇲🇾 Malaysia",
+        ".de": "🇩🇪 Germany",
+        ".fr": "🇫🇷 France",
+        ".es": "🇪🇸 Spain",
+        ".it": "🇮🇹 Italy",
+        ".br": "🇧🇷 Brazil",
+        ".nl": "🇳🇱 Netherlands"
     }
     
-    for domain, country in epg_countries.items():
+    for domain, country in epg_patterns.items():
         if domain in epg_id.lower():
             return country
     
-    # Essential broadcaster patterns (simplified)
-    broadcasters = {
-        "🇨🇦 Canada": ["tsn", "cbc", "ctv", "global", "sportsnet", "w network"],
-        "🇺🇸 United States": ["cbs", "nbc", "abc", "fox", "cnn", "espn", "discovery", "hgtv", "mtv", "c-span"],
-        "🇬🇧 United Kingdom": ["bbc", "itv", "sky", "channel 4", "channel 5"],
-        "🇵🇭 Philippines": ["abs-cbn", "gma", "anc", "tv5"],
-        "🇦🇺 Australia": ["abc australia", "nine network", "seven network"],
-        "🇯🇵 Japan": ["nhk", "fuji tv", "animax"],
-        "🇲🇾 Malaysia": ["tv1", "tv2", "astro"],
-        "🇩🇪 Germany": ["ard", "zdf", "rtl", "sat.1"],
-        "🇫🇷 France": ["tf1", "france 2", "canal+"],
-        "🇪🇸 Spain": ["antena 3", "telecinco", "tve"],
-        "🇮🇹 Italy": ["rai", "mediaset", "canale 5"],
-        "🇧🇷 Brazil": ["globo", "sbt", "record"],
-        "🇲🇽 Mexico": ["televisa", "tv azteca"],
-        "🇷🇺 Russia": ["первый", "россия", "нтв"]
+    # STEP 3: Specific channel fixes (from your channel_processor.txt)
+    channel_lower = channel_name.lower()
+    
+    # Canadian sports channels (TSN series)
+    if any(x in channel_lower for x in ["tsn 1", "tsn 2", "tsn 3", "tsn 4", "tsn 5", "tsn1", "tsn2", "tsn3", "tsn4", "tsn5"]):
+        return "🇨🇦 Canada"
+    
+    # CBC News Toronto (Canadian)
+    if "cbc news toronto" in channel_lower:
+        return "🇨🇦 Canada"
+    
+    # US channels that were misclassified
+    if any(x in channel_lower for x in ["tv land", "tvland", "we tv", "wetv", "all weddings we tv", "cheaters", "cheers", "christmas 365"]):
+        return "🇺🇸 United States"
+    
+    # UK shows/channels
+    if "come dine with me" in channel_lower:
+        return "🇬🇧 United Kingdom"
+    
+    # Philippines news channels
+    if any(x in channel_lower for x in ["anc global", "anc ph"]):
+        return "🇵🇭 Philippines"
+    
+    # Japan anime channels
+    if "animax" in channel_lower:
+        return "🇯🇵 Japan"
+    
+    # STEP 4: Logo URL analysis
+    logo_patterns = {
+        "🇨🇦 Canada": ["/canada/", "/ca/", "canada.", "canadian"],
+        "🇺🇸 United States": ["/usa/", "/us/", "united-states", "american"],
+        "🇬🇧 United Kingdom": ["/uk/", "/united-kingdom/", "british", "england"],
+        "🇩🇪 Germany": ["/germany/", "/de/", "german", "deutschland"],
+        "🇫🇷 France": ["/france/", "/fr/", "french", "français"],
+        "🇮🇹 Italy": ["/italy/", "/it/", "italian", "italiano"],
+        "🇪🇸 Spain": ["/spain/", "/es/", "spanish", "español"],
+        "🇳🇱 Netherlands": ["/netherlands/", "/nl/", "dutch", "nederland"],
+        "🇦🇺 Australia": ["/australia/", "/au/", "australian", "aussie"],
+        "🇯🇵 Japan": ["/japan/", "/jp/", "japanese", "日本"],
+        "🇰🇷 South Korea": ["/korea/", "/kr/", "korean", "한국"],
+        "🇮🇳 India": ["/india/", "/in/", "indian", "भारत"],
+        "🇧🇷 Brazil": ["/brazil/", "/br/", "brazilian", "brasil"],
+        "🇲🇽 Mexico": ["/mexico/", "/mx/", "mexican", "méxico"],
+        "🇦🇷 Argentina": ["/argentina/", "/ar/", "argentinian", "argentina"],
+        "🇵🇭 Philippines": ["/philippines/", "/ph/", "filipino", "pilipinas"]
     }
     
-    for country, keywords in broadcasters.items():
-        if any(keyword in all_text for keyword in keywords):
-            return country
+    for country, patterns in logo_patterns.items():
+        for pattern in patterns:
+            if pattern in logo_url.lower():
+                return country
+    
+    # STEP 5: Enhanced broadcaster patterns
+    broadcaster_patterns = {
+        "🇨🇦 Canada": [
+            "cbc", "tsn", "ctv", "global", "sportsnet", "citytv", "aptn", "teletoon", "ytv",
+            "discovery canada", "history canada", "slice", "w network", "oln", "hgtv canada",
+            "food network canada", "showcase", "crave", "super channel", "hollywood suite"
+        ],
+        "🇺🇸 United States": [
+            "cbs", "nbc", "abc", "fox", "cnn", "espn", "amc", "mtv", "comedy central",
+            "discovery usa", "history usa", "tlc usa", "hgtv usa", "food network usa", "paramount",
+            "nickelodeon usa", "cartoon network usa", "disney usa", "lifetime", "e!", "bravo usa"
+        ],
+        "🇬🇧 United Kingdom": [
+            "bbc", "itv", "channel 4", "channel 5", "sky", "dave", "really", "yesterday",
+            "discovery uk", "history uk", "tlc uk", "living", "alibi", "gold", "drama"
+        ],
+        "🇩🇪 Germany": [
+            "ard", "zdf", "rtl", "pro7", "sat.1", "vox", "kabel eins", "super rtl", "rtl2",
+            "discovery germany", "history germany", "tlc germany", "dmax", "sixx", "tele 5"
+        ],
+        "🇫🇷 France": [
+            "tf1", "france 2", "france 3", "france 5", "m6", "canal+", "arte", "w9", "tmc",
+            "discovery france", "history france", "tlc france", "planete+", "ushuaia tv"
+        ],
+        "🇮🇹 Italy": [
+            "rai", "canale 5", "italia 1", "rete 4", "la7", "tv8", "nove", "20 mediaset",
+            "discovery italia", "history italia", "dmax italia", "real time", "giallo"
+        ],
+        "🇪🇸 Spain": [
+            "tve", "la 1", "la 2", "antena 3", "cuatro", "telecinco", "la sexta", "nova",
+            "discovery spain", "history spain", "dmax spain", "mega", "neox", "clan"
+        ],
+        "🇳🇱 Netherlands": [
+            "npo", "rtl 4", "rtl 5", "rtl 7", "sbs6", "veronica", "net5", "rtl z",
+            "discovery netherlands", "history netherlands", "tlc netherlands"
+        ],
+        "🇦🇺 Australia": [
+            "abc australia", "nine network", "seven network", "ten", "foxtel",
+            "discovery australia", "history australia", "lifestyle"
+        ],
+        "🇯🇵 Japan": [
+            "nhk", "fuji tv", "tbs", "tv asahi", "tv tokyo", "nippon tv", "animax"
+        ],
+        "🇰🇷 South Korea": [
+            "kbs", "mbc", "sbs", "jtbc", "tvn", "ocn"
+        ],
+        "🇮🇳 India": [
+            "zee", "star plus", "colors", "sony tv", "& tv", "discovery india"
+        ],
+        "🇧🇷 Brazil": [
+            "globo", "sbt", "record", "band", "discovery brasil"
+        ],
+        "🇲🇽 Mexico": [
+            "televisa", "tv azteca", "once tv", "discovery mexico"
+        ],
+        "🇦🇷 Argentina": [
+            "telefe", "canal 13", "america tv", "discovery argentina"
+        ],
+        "🇵🇭 Philippines": [
+            "abs-cbn", "gma", "anc", "tv5", "pba rush"
+        ]
+    }
+    
+    for country, keywords in broadcaster_patterns.items():
+        for keyword in keywords:
+            if keyword in all_text:
+                return country
     
     return "Uncategorized"
 
+
+def is_valid_country_group(group_name):
+    """Check if group name is a valid country (not a streaming service)"""
+    valid_countries = [
+        "🇺🇸 United States", "🇨🇦 Canada", "🇬🇧 United Kingdom", "🇩🇪 Germany", 
+        "🇫🇷 France", "🇮🇹 Italy", "🇪🇸 Spain", "🇳🇱 Netherlands", "🇧🇪 Belgium",
+        "🇦🇹 Austria", "🇨🇭 Switzerland", "🇸🇪 Sweden", "🇳🇴 Norway", "🇩🇰 Denmark",
+        "🇫🇮 Finland", "🇵🇱 Poland", "🇨🇿 Czech Republic", "🇭🇺 Hungary", "🇵🇹 Portugal",
+        "🇬🇷 Greece", "🇷🇴 Romania", "🇧🇬 Bulgaria", "🇭🇷 Croatia", "🇷🇸 Serbia",
+        "🇦🇺 Australia", "🇯🇵 Japan", "🇰🇷 South Korea", "🇮🇳 India", "🇨🇳 China",
+        "🇧🇷 Brazil", "🇲🇽 Mexico", "🇦🇷 Argentina", "🇨🇱 Chile", "🇨🇴 Colombia",
+        "🇷🇺 Russia", "🇹🇷 Turkey", "🇸🇦 Saudi Arabia", "🇦🇪 UAE", "🇪🇬 Egypt",
+        "🇿🇦 South Africa", "🇳🇬 Nigeria", "🇰🇪 Kenya", "🇮🇱 Israel", "🇹🇭 Thailand",
+        "🇻🇳 Vietnam", "🇵🇭 Philippines", "🇮🇩 Indonesia", "🇲🇾 Malaysia", "🇸🇬 Singapore"
+    ]
+    return group_name in valid_countries
+
+
 def load_channels():
     """Load channels from channels.txt."""
     if not os.path.exists('channels.txt'):
@@ -95,12 +228,20 @@ def load_channels():
         print(f"❌ Error loading channels: {e}")
         return []
 
+
 def reorganize_channels(channels):
-    """Reorganize channels by country (traditional only)."""
-    print("📺 Organizing traditional broadcasters by country...")
+    """Enhanced reorganization with 3-point analysis."""
+    print("🔍 Enhanced Country Detection with 3-Point Analysis")
+    print("📊 Analyzing: Channel Name + EPG ID + Logo URL")
+    print("-" * 60)
     
     changes = 0
-    stats = {}
+    stats = {
+        'country_detected': 0,
+        'sent_to_uncategorized': 0,
+        'kept_existing_country': 0
+    }
+    country_counts = {}
     
     for channel in channels:
         old_group = channel.get('Group', 'Uncategorized')
@@ -109,24 +250,46 @@ def reorganize_channels(channels):
         logo = channel.get('Logo', '')
         stream_url = channel.get('Stream URL', '')
         
-        new_group = detect_traditional_broadcaster_country(stream_name, epg_id, logo, stream_url)
+        # Detect country using enhanced 3-point analysis
+        detected_country = detect_country_from_channel_content(stream_name, epg_id, logo, stream_url)
         
-        if old_group != new_group:
-            if new_group == "Uncategorized":
+        # Decide final group
+        if is_valid_country_group(old_group) and detected_country != "Uncategorized":
+            # Keep existing valid country
+            final_group = old_group
+            stats['kept_existing_country'] += 1
+        elif detected_country != "Uncategorized":
+            # Use detected country
+            final_group = detected_country
+            stats['country_detected'] += 1
+            if old_group != detected_country:
+                print(f"🔍 Fixed: '{stream_name}' {old_group} → {detected_country}")
+                changes += 1
+        else:
+            # Send to Uncategorized
+            final_group = "Uncategorized"
+            stats['sent_to_uncategorized'] += 1
+            if old_group != "Uncategorized":
                 print(f"📱 Platform: '{stream_name}' → Uncategorized")
-            else:
-                print(f"📺 Fixed: '{stream_name}' {old_group} → {new_group}")
-            channel['Group'] = new_group
-            changes += 1
+                changes += 1
         
-        stats[new_group] = stats.get(new_group, 0) + 1
+        channel['Group'] = final_group
+        country_counts[final_group] = country_counts.get(final_group, 0) + 1
     
-    print(f"\n✅ Changes made: {changes}")
-    print(f"📺 Traditional broadcasters: {sum(v for k, v in stats.items() if k != 'Uncategorized')}")
-    print(f"📱 Streaming/Unknown: {stats.get('Uncategorized', 0)}")
+    print(f"\n📊 PROCESSING RESULTS:")
+    print(f"✅ Changes made: {changes}")
+    print(f"🔍 Country detected: {stats['country_detected']}")
+    print(f"✅ Kept existing countries: {stats['kept_existing_country']}")
+    print(f"📱 Sent to Uncategorized: {stats['sent_to_uncategorized']}")
+    
+    print(f"\n🌍 FINAL GROUP DISTRIBUTION:")
+    sorted_countries = sorted(country_counts.items(), key=lambda x: (x[0] == "Uncategorized", -x[1]))
+    for country, count in sorted_countries:
+        print(f"   {country}: {count} channels")
     
     return channels
 
+
 def save_channels(channels):
     """Save channels to file."""
     # Backup
@@ -152,6 +315,7 @@ def save_channels(channels):
         print(f"❌ Save error: {e}")
         return False
 
+
 def generate_m3u(channels):
     """Generate M3U playlist."""
     try:
@@ -179,19 +343,23 @@ def generate_m3u(channels):
         print(f"❌ M3U error: {e}")
         return False
 
+
 def main():
     """Main function."""
-    print("📺 Traditional Broadcasters Only - Country Organization")
-    print("=" * 60)
+    print("🎯 Enhanced IPTV Country Detection - 3-Point Analysis")
+    print("=" * 70)
+    print("🔍 Analyzes: Channel Name + EPG ID + Logo URL")
+    print("🎯 Filters: Only countries remain, streaming services → Uncategorized")
+    print("=" * 70)
     
     channels = load_channels()
     if not channels:
         return False
     
-    # Reorganize
+    # Enhanced reorganization
     channels = reorganize_channels(channels)
     
-    # Sort: Countries first, then Uncategorized last
+    # Sort: Countries first (alphabetically), then Uncategorized last
     channels.sort(key=lambda x: (
         "zzz" if x.get('Group') == "Uncategorized" else x.get('Group', ''),
         x.get('Stream name', '')
@@ -212,12 +380,15 @@ def main():
     except:
         pass
     
-    print("\n🎉 COMPLETED!")
-    print("✅ Traditional broadcasters organized by country")
-    print("✅ All streaming platforms → Uncategorized")
+    print("\n🎉 ENHANCED PROCESSING COMPLETE!")
+    print("✅ 3-point analysis applied to all channels")
+    print("✅ Countries detected from EPG ID, Logo URL, and Channel Names")
+    print("✅ Streaming services filtered to Uncategorized")
+    print("✅ Clean country-organized playlist generated")
     
     return True
 
+
 if __name__ == "__main__":
     success = main()
     exit(0 if success else 1)