Update app.py

app.py

@@ -37,7 +37,7 @@ def extract_text_from_images(image_paths):
 
 # Step 3: Rebuild the PDF with Extracted Text
 def rebuild_pdf(original_pdf_path, text_data, output_pdf_path):
-    """Overlay extracted text onto the original PDF with enhanced alignment."""
+    """Overlay extracted text onto the original PDF with enhanced alignment and font embedding."""
     with fitz.open(original_pdf_path) as pdf:
         output_pdf = fitz.open()
 
@@ -52,25 +52,26 @@ def rebuild_pdf(original_pdf_path, text_data, output_pdf_path):
             for bbox, text, conf in text_data[page_num]:
                 if conf > 0.6:  # Confidence threshold
                     (x_min, y_min), (x_max, y_max) = bbox[0], bbox[2]
-                    rect = fitz.Rect(x_min, y_min, x_max, y_max)
-
-                    # Adjust text placement for alignment
-                    adjusted_x = x_min + 2  # Fine-tune as needed
-                    adjusted_y = y_min + 2  # Fine-tune as needed
-                    adjusted_rect = fitz.Rect(adjusted_x, adjusted_y, x_max, y_max)
+                    
+                    # Apply dynamic alignment adjustments
+                    adjusted_x_min = x_min + 2  # Shift horizontally
+                    adjusted_y_min = y_min + 3  # Shift vertically
+                    adjusted_rect = fitz.Rect(adjusted_x_min, adjusted_y_min, x_max, y_max)
 
+                    # Insert text with font embedding
                     new_page.insert_textbox(
                         adjusted_rect,  # Use adjusted bounding box
                         text,
-                        fontsize=10,  # Adjust font size as needed
-                        fontname="helv",  # Use Helvetica for compatibility
-                        color=(0, 0, 0),  # Black color for text
+                        fontsize=10,  # Adjust font size for better fit
+                        fontname="helv",  # Use Helvetica for better compatibility
+                        fontfile="path/to/font.ttf",  # Embed font (replace with actual font path)
+                        color=(0, 0, 0),  # Black text
                     )
 
             # Add original diagrams and graphics
             new_page.show_pdf_page(page.rect, pdf, page_num)
 
-        # Save rebuilt PDF with embedded fonts
+        # Save rebuilt PDF with optimized structure
         output_pdf.save(output_pdf_path, garbage=4)