markdownfix

src/app.py

@@ -60,9 +60,9 @@ sample_data = [
         "pad": True,
     },
     {
-        "preview_image": str(SAMPLES_PATH / "code.jpg"),
+        "preview_image": str(SAMPLES_PATH / "7666.jpg"),
         "prompts": ["Convert code to text."],
-        "image": str(SAMPLES_PATH / "code.jpg"),
+        "image": str(SAMPLES_PATH / "7666.jpg"),
         "name": "Code Recognition",
         "pad": True,
     },
@@ -527,29 +527,25 @@ with gr.Blocks(fill_height=True, css_paths=css_file_path, head_paths=head_file_p
         else:
             cb[-1] = gr.ChatMessage(role="assistant", content=answer)
 
-        if "convert this page to docling" in msg.lower() or ("convert" in msg.lower() and "otsl" in msg.lower()):
+        if "convert" in msg.lower() and "otsl" in msg.lower():
             try:
                 doctags_doc = DocTagsDocument.from_doctags_and_image_pairs([answer], [Image.open(image_path)])
                 doc = DoclingDocument.load_from_doctags(doctags_doc, document_name="Document")
                 markdown_output = doc.export_to_markdown()
-                response = gr.ChatMessage(
+                md_response = gr.ChatMessage(
                     role="assistant",
                     content=f"\nConverted to Markdown using docling.\n\n**MD Output:**\n\n{markdown_output}",
                 )
-                cb.append(response)
+                cb.append(md_response)
             except Exception as e:
                 error_response = gr.ChatMessage(role="assistant", content=f"Error creating markdown output: {e!s}")
                 cb.append(error_response)
+
         elif "convert formula to latex" in msg.lower():
             try:
                 doctags_doc = DocTagsDocument.from_doctags_and_image_pairs([answer], [Image.open(image_path)])
                 doc = DoclingDocument.load_from_doctags(doctags_doc, document_name="Document")
                 markdown_output = doc.export_to_markdown()
-                if markdown_output.count("$$") >= 2:
-                    parts = markdown_output.split("$$", 2)
-                    formula = parts[1].strip()
-                    wrapped = f"$$\n\\begin{{aligned}}\n{formula}\n\\end{{aligned}}\n$$"
-                    markdown_output = parts[0] + wrapped + parts[2]
                 md_response = gr.ChatMessage(
                     role="assistant",
                     content=f"\nConverted to Markdown using docling.\n\n**LaTeX Output:**\n\n{markdown_output}",
@@ -559,6 +555,20 @@ with gr.Blocks(fill_height=True, css_paths=css_file_path, head_paths=head_file_p
                 error_response = gr.ChatMessage(role="assistant", content=f"Error creating LaTeX output: {e!s}")
                 cb.append(error_response)
 
+        elif "extract all key-value pairs" in msg.lower() and "doctag format" in msg.lower():
+            try:
+                doctags_doc = DocTagsDocument.from_doctags_and_image_pairs([answer], [Image.open(image_path)])
+                doc = DoclingDocument.load_from_doctags(doctags_doc, document_name="Document")
+                html_output = doc.export_to_html()
+                html_response = gr.ChatMessage(
+                    role="assistant",
+                    content=f"\nExtracted key-value pairs using docling.\n\n**HTML Output:**\n\n{html_output}",
+                )
+                cb.append(html_response)
+            except Exception as e:
+                error_response = gr.ChatMessage(role="assistant", content=f"Error creating HTML output: {e!s}")
+                cb.append(error_response)
+
         if has_loc_tags:
             try:
                 annotated_image = draw_bounding_boxes(image_path, answer, is_doctag_response=has_doctag)