fix: detect Format A by domain differences, remove duplicate detection code

app/downloaders/anime_sites/animesama.py

@@ -778,33 +778,42 @@ class AnimeSamaDownloader(BaseAnimeSite):
 
                     if eps_matches:
                         # Determine the format by looking at the data
-                        # Format A: each epsX array is one source with all episodes (few arrays, many URLs each)
+                        # Format A: each epsX array is one SOURCE with all episodes (different domains per array)
-                        # Format B: each epsX array is one episode with multiple sources (many arrays or similar counts)
+                        # Format B: each epsX array is one EPISODE with multiple sources (same domains across arrays)
                         
                         eps1_urls = re.findall(r"'(https?://[^']+)'", eps_matches[0][1])
                         num_episode_arrays = len(eps_matches)
                         
                         is_format_a = True  # Default
                         
-                        if num_episode_arrays > 5:
+                        if num_episode_arrays >= 2:
-                            # Many arrays = Format B (each array = one episode)
+                            # Extract domains from first URLs of each array
-                            is_format_a = False
+                            def get_domain(url):
-                        elif num_episode_arrays >= 2:
+                                return url.split('/')[2] if '/' in url else url
-                            # Check URL counts - if similar, it's Format B
+                            
-                            url_counts = []
+                            domains_per_array = []
                             for eps_num, urls_text in eps_matches:
                                 urls = re.findall(r"'(https?://[^']+)'", urls_text)
-                                url_counts.append(len(urls))
+                                if urls:
+                                    domains = set(get_domain(u) for u in urls[:3])  # Sample first 3
+                                    domains_per_array.append(domains)
                             
-                            if url_counts and max(url_counts) > 0:
+                            # Check if domains are different across arrays
-                                avg_count = sum(url_counts) / len(url_counts)
+                            # If each array has completely different domains → Format A (each = source)
-                                variance = max(url_counts) / avg_count if avg_count > 0 else 1
+                            # If arrays share domains → Format B (each = episode with multiple sources)
+                            all_domains = set()
+                            for domains in domains_per_array:
+                                all_domains.update(domains)
                             
-                                # Similar counts (< 1.5x variance) = Format B
+                            # If total unique domains ≈ sum of domains per array → Format A
-                                if variance < 1.5:
+                            # If total unique domains << sum of domains per array → Format B (shared)
-                                    is_format_a = False
+                            total_domain_count = sum(len(d) for d in domains_per_array)
+                            if len(all_domains) < total_domain_count * 0.7:
+                                # Domains are shared across arrays → Format B
+                                is_format_a = False
                         
-                        logger.debug(f"Detected format {'A (source-based)' if is_format_a else 'B (episode-based)'} - {num_episode_arrays} arrays")
+                        
+                        # No more host preference!
 
                         # No more host preference! Just collect all available URLs for each episode
                         # The download system will automatically detect and use the appropriate downloader