Enhance URL handling in transformers by patching multiple functions for comprehensive URL fixing; improve requests.get handling for malformed URLs

app.py

@@ -23,15 +23,20 @@ def patch_transformers_url():
     """Fix URL scheme issue in transformers 4.17.0 with comprehensive URL handling"""
     try:
         import transformers.file_utils
+        import requests
+        
+        # Patch multiple functions for comprehensive URL fixing
         original_get_from_cache = transformers.file_utils.get_from_cache
         
         def patched_get_from_cache(url, *args, **kwargs):
             # Comprehensive URL fixing for various formats
             if isinstance(url, str):
+                original_url = url
                 # Handle different types of malformed URLs
-                if url.startswith('/api/'):
+                if url.startswith('/api/resolve-cache/') or url.startswith('/api/'):
                     # Fix relative API URLs - ensure proper base URL
                     url = 'https://huggingface.co' + url
+                    print(f"Fixed relative API URL: {original_url} -> {url}")
                 elif url.startswith('//'):
                     # Fix protocol-relative URLs
                     url = 'https:' + url
@@ -67,8 +72,51 @@ def patch_transformers_url():
                         print(f"All download attempts failed for {url}: {e}")
                         raise
         
+        # Also patch cached_path function which might be causing the issue
+        if hasattr(transformers.file_utils, 'cached_path'):
+            original_cached_path = transformers.file_utils.cached_path
+            
+            def patched_cached_path(url_or_filename, *args, **kwargs):
+                if isinstance(url_or_filename, str):
+                    if url_or_filename.startswith('/api/resolve-cache/') or url_or_filename.startswith('/api/'):
+                        url_or_filename = 'https://huggingface.co' + url_or_filename
+                        print(f"Fixed cached_path URL: {url_or_filename}")
+                return original_cached_path(url_or_filename, *args, **kwargs)
+            
+            transformers.file_utils.cached_path = patched_cached_path
+        
+        # Patch http_get function to handle malformed URLs at the lowest level
+        if hasattr(transformers.file_utils, 'http_get'):
+            original_http_get = transformers.file_utils.http_get
+            
+            def patched_http_get(url, *args, **kwargs):
+                if isinstance(url, str):
+                    if url.startswith('/api/resolve-cache/') or url.startswith('/api/'):
+                        url = 'https://huggingface.co' + url
+                        print(f"Fixed http_get URL: {url}")
+                return original_http_get(url, *args, **kwargs)
+            
+            transformers.file_utils.http_get = patched_http_get
+        
+        # Patch requests.get at the lowest level to catch any remaining malformed URLs
+        original_requests_get = requests.get
+        
+        def patched_requests_get(url, *args, **kwargs):
+            if isinstance(url, str):
+                if url.startswith('/api/resolve-cache/') or url.startswith('/api/'):
+                    original_url = url
+                    url = 'https://huggingface.co' + url
+                    print(f"Fixed requests.get URL: {original_url} -> {url}")
+                elif not url.startswith(('http://', 'https://', 'ftp://')):
+                    if url.startswith('/'):
+                        url = 'https://huggingface.co' + url
+                        print(f"Fixed relative URL in requests.get: {url}")
+            return original_requests_get(url, *args, **kwargs)
+        
+        requests.get = patched_requests_get
+        
         transformers.file_utils.get_from_cache = patched_get_from_cache
-        print("Applied enhanced URL patch for transformers")
+        print("Applied comprehensive URL patch for transformers and requests")
     except Exception as e:
         print(f"Warning: Could not patch transformers URL handling: {e}")