Update app.py

app.py

@@ -81,10 +81,10 @@ def draw_ocr_on_image(image_pil, ocr_data):
     draw = ImageDraw.Draw(image_pil)
     
     try:
-        avg_height = sum(d['h'] for d in ocr_data if d['h'] > 0) / len(ocr_data) if ocr_data and any(d['h'] > 0 for d in ocr_data) else 10
+        avg_height = sum(d['h'] for d in ocr_data if d['h'] > 0) / len([d for d in ocr_data if d['h'] > 0]) if ocr_data and any(d['h'] > 0 for d in ocr_data) else 10
         font_size = max(8, int(avg_height * 0.6)) 
         font = ImageFont.truetype("arial.ttf", font_size)
-    except (IOError, ZeroDivisionError): # Added ZeroDivisionError for safety
+    except (IOError, ZeroDivisionError):
         font = ImageFont.load_default()
         font_size = 10 
         print("Arial font not found or issue with height calculation, using default font.")
@@ -108,7 +108,6 @@ def process_image_and_xml(image_path, xml_path, show_overlay):
     if image_path is None:
         return None, "Please upload an image.", None
     if xml_path is None:
-        # If image_path is not None, we can still show the image
         try:
             img_pil_orig = Image.open(image_path).convert("RGB")
         except Exception as e:
@@ -127,7 +126,11 @@ def process_image_and_xml(image_path, xml_path, show_overlay):
         img_for_overlay = img_pil.copy()
         overlay_image_pil = draw_ocr_on_image(img_for_overlay, ocr_box_data)
     elif show_overlay and not ocr_box_data and not (isinstance(extracted_text, str) and extracted_text.startswith("Error")):
-        extracted_text += "\n(No bounding box data found or parsed for overlay)"
+        if isinstance(extracted_text, str):
+            extracted_text += "\n(No bounding box data found or parsed for overlay)"
+        else: # Should not happen if parse_alto_xml returns (str, list)
+            extracted_text = "(No bounding box data found or parsed for overlay)"
+
 
     return img_pil, extracted_text, overlay_image_pil
 
@@ -144,7 +147,6 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
 
     with gr.Row():
         with gr.Column(scale=1):
-            # Corrected: type="filepath"
             image_input = gr.File(label="Upload Image (PNG, JPG, etc.)", type="filepath")
             xml_input = gr.File(label="Upload ALTO XML File (.xml)", type="filepath")
             show_overlay_checkbox = gr.Checkbox(label="Show OCR Overlay on Image", value=False)
@@ -159,17 +161,11 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
     output_image_overlay = gr.Image(label="Image with OCR Overlay", type="pil", interactive=False, visible=True)
 
     def update_interface(image_filepath, xml_filepath, show_overlay_val):
-        # image_filepath and xml_filepath are now strings (paths) or None
-        
-        # Initial check for None inputs to provide guidance
         if image_filepath is None and xml_filepath is None:
             return None, "Please upload an image and an XML file.", None
         if image_filepath is None:
-            # xml_filepath might be present, but we need the image first
             return None, "Please upload an image file.", None
-        # xml_filepath can be None here, process_image_and_xml handles it
-
-        # Call the main processing function
+        
         img, text, overlay_img = process_image_and_xml(image_filepath, xml_filepath, show_overlay_val)
         
         return img, text, overlay_img
@@ -189,7 +185,8 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
     gr.Markdown("---")
     gr.Markdown("### Example ALTO XML Snippet (for `String` element extraction):")
     gr.Code(
-        language="xml",
+        # Corrected: language="text" or omit language parameter
+        language="text",  # Or language=None
         value="""
 <alto xmlns="http://www.loc.gov/standards/alto/v3/alto.xsd">
   <Description>...</Description>
@@ -218,24 +215,24 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
 
 if __name__ == "__main__":
     try:
-        # from PIL import Image as PImage # Already imported as Image
         img = Image.new('RGB', (2394, 3612), color = 'lightgray')
         img.save("dummy_image.png")
         print("Created dummy_image.png for testing.")
         
-        # Make sure your example XML is named 189819724.34.xml and is in the same directory
-        # Or, create it if it doesn't exist with the content you provided
         example_xml_filename = "189819724.34.xml"
         if not os.path.exists(example_xml_filename):
             print(f"WARNING: Example XML '{example_xml_filename}' not found. Please create it or upload your own.")
-            # You could also write the example XML content here if needed for testing
-            # with open(example_xml_filename, "w") as f:
-            #     f.write("""<?xml version="1.0" encoding="UTF-8"?> ...your full XML... </alto>""")
-
+            # Example of writing the provided XML if it's missing for easier testing
+            # Note: The provided XML is partial, so this would be a placeholder
+            # If you have the full XML, you can paste it here.
+            # For now, just a warning.
+            # with open(example_xml_filename, "w", encoding="utf-8") as f:
+            #     f.write('<?xml version="1.0" encoding="UTF-8"?>\n<alto xmlns="http://www.loc.gov/standards/alto/v3/alto.xsd"><Description><MeasurementUnit>pixel</MeasurementUnit><sourceImageInformation><fileName>./data/pdfs/189742210.pdf</fileName></sourceImageInformation><OCRProcessing ID="IdOcr"><ocrProcessingStep><processingDateTime>Wed Nov 20 19:03:43 2019\n</processingDateTime><processingSoftware><softwareCreator>CONTRIBUTORS</softwareCreator><softwareName>pdfalto</softwareName><softwareVersion>0.1</softwareVersion></processingSoftware></ocrProcessingStep></OCRProcessing></Description><Styles><TextStyle ID="font0" FONTFAMILY="helvetica" FONTSIZE="21.000" FONTTYPE="sans-serif" FONTWIDTH="proportional" FONTCOLOR="#WWWWWW" FONTSTYLE="italics"/><TextStyle ID="font1" FONTFAMILY="helvetica" FONTSIZE="10.000" FONTTYPE="sans-serif" FONTWIDTH="proportional" FONTCOLOR="#WWWWWW" FONTSTYLE="italics"/><TextStyle ID="font2" FONTFAMILY="helvetica" FONTSIZE="10.000" FONTTYPE="sans-serif" FONTWIDTH="proportional" FONTCOLOR="#WWWWWW" FONTSTYLE=""/></Styles><Layout><Page ID="Page13" PHYSICAL_IMG_NR="13" WIDTH="2394" HEIGHT="3612"><PrintSpace><TextLine WIDTH="684" HEIGHT="108" ID="p13_t1" HPOS="465" VPOS="196"><String ID="p13_w1" CONTENT="Introduction" HPOS="465" VPOS="196" WIDTH="684" HEIGHT="108" STYLEREFS="font0"/></TextLine></PrintSpace></Page></Layout></alto>') # Shortened for brevity
+            # print(f"Created a placeholder '{example_xml_filename}'. Please replace with full XML if needed.")
 
     except ImportError:
         print("Pillow not installed, can't create dummy image.")
     except Exception as e:
-        print(f"Error creating dummy image: {e}")
+        print(f"Error creating dummy image or example XML: {e}")
 
     demo.launch()