Merge pull request #103 from enoch3712/102-fix-vision-not-working-properly

fix with vision refactoring

Author: enoch3712

extract_thinker/document_loader/document_loader.py

@@ -69,25 +69,27 @@ def load(self, source: Union[str, BytesIO]) -> Any:
         """Enhanced load method that handles vision mode."""
         if not self.can_handle(source):
             raise ValueError("Unsupported file type or stream.")
-            
+        
         response = {}
 
+        # Always process text content
+        content = self.load_content_from_file(source) if isinstance(source, str) else self.load_content_from_stream(source)
+        
+        # Merge content with response
+        if content is not None:
+            if isinstance(content, dict):
+                response.update(content)
+            else:
+                response['content'] = content
+        
+        # If vision mode is enabled, add images
         if self.vision_mode:
             if not self.can_handle_vision(source):
                 raise ValueError("Source cannot be processed in vision mode. Only PDFs and images are supported.")
 
             # Convert to images and add to response
             response['images'] = self.convert_to_images(source)
 
-        # Normal processing
-        content = self.load_content_from_file(source) if isinstance(source, str) else self.load_content_from_stream(source)
-        
-        # Merge content with response
-        if isinstance(content, dict):
-            response.update(content)
-        else:
-            response['content'] = content
-            
         return response
 
     def getContent(self) -> Any:

extract_thinker/document_loader/document_loader_llm_image.py

@@ -12,20 +12,22 @@ def __init__(self, content=None, cache_ttl=300, llm=None):
         self.llm = llm
 
     def load_content_from_file(self, file_path: str) -> Union[str, object]:
-        images = self.convert_to_images(file_path)
-        results = []
-        for _, image_bytes in images.items():
-            image_stream = BytesIO(image_bytes)
-            results.append({"image": image_stream})
-        return results
+        pass
+        # images = self.convert_to_images(file_path)
+        # results = []
+        # for _, image_bytes in images.items():
+        #     image_stream = BytesIO(image_bytes)
+        #     results.append({"image": image_stream})
+        # return results
 
     def load_content_from_stream(self, stream: BytesIO) -> Union[str, object]:
-        images = self.convert_to_images(stream)
-        results = []
-        for _, image_bytes in images.items():
-            image_stream = BytesIO(image_bytes)
-            results.append({"image": image_stream})
-        return results
+        pass
+        # images = self.convert_to_images(stream)
+        # results = []
+        # for _, image_bytes in images.items():
+        #     image_stream = BytesIO(image_bytes)
+        #     results.append({"image": image_stream})
+        # return results
 
     def load_content_from_stream_list(self, stream: BytesIO) -> List[Any]:
         return self.load_content_from_stream(stream)