Update app.py

app.py

@@ -34,9 +34,9 @@ def align_images(img1, img2):
     aligned_img = cv2.warpPerspective(img2, matrix, (img1.shape[1], img1.shape[0]))
     return aligned_img
 
-# Compare images for visual changes
-def compare_images(img1, img2):
-    diff = cv2.absdiff(img1, img2)
+# Perform pixel-based comparison for visual changes
+def compare_visual_changes(orig_img, edit_img, start_position):
+    diff = cv2.absdiff(orig_img, edit_img)
     gray_diff = cv2.cvtColor(diff, cv2.COLOR_BGR2GRAY)
 
     # Apply Gaussian blur to reduce noise
@@ -45,124 +45,107 @@ def compare_images(img1, img2):
     # Apply thresholding
     _, thresh = cv2.threshold(blurred_diff, 40, 255, cv2.THRESH_BINARY)
 
-    # Morphological operations to smooth out noise
+    # Morphological operations to clean noise
     kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (5, 5))
     cleaned = cv2.morphologyEx(thresh, cv2.MORPH_CLOSE, kernel)
 
-    return cleaned
-
-# Compare text and generate differences
-def compare_text(orig_text, edit_text, start_position):
-    diff = difflib.ndiff(orig_text.splitlines(), edit_text.splitlines())
-    text_changes = []
-    position_counter = start_position
-    for line in diff:
-        if line.startswith("+ "):  # Added text
-            text_changes.append((position_counter, f'"{line[2:]}" added at {position_counter}'))
-        elif line.startswith("- "):  # Removed text
-            text_changes.append((position_counter, f'"{line[2:]}" removed at {position_counter}'))
-        position_counter += 1
-    return text_changes, position_counter
-
-# Highlight visual changes
-def highlight_visual_changes(img1, img2, mask, start_position):
-    overlay = img2.copy()
-    contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
+    contours, _ = cv2.findContours(cleaned, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
+    overlay = edit_img.copy()
     visual_changes = []
+    position_counter = start_position
     font = cv2.FONT_HERSHEY_SIMPLEX
     font_scale = 0.8
     thickness = 2
-    position_counter = start_position
 
     for cnt in contours:
-        if cv2.contourArea(cnt) > 100:  # Filter based on area to reduce false positives
+        if cv2.contourArea(cnt) > 100:  # Filter out small regions
             x, y, w, h = cv2.boundingRect(cnt)
-            cv2.rectangle(overlay, (x, y), (x + w, y + h), (0, 0, 255), 2)  # Red for changes
+            cv2.rectangle(overlay, (x, y), (x + w, y + h), (0, 0, 255), 2)  # Red bounding box
             cv2.putText(overlay, str(position_counter), (x, y - 10), font, font_scale, (0, 255, 0), thickness)
             visual_changes.append((position_counter, f'Visual change detected at position {position_counter}'))
             position_counter += 1
 
     return overlay, visual_changes, position_counter
 
+# Perform OCR-based comparison for text changes
+def compare_text_changes(orig_img, edit_img, start_position):
+    orig_text = pytesseract.image_to_string(orig_img)
+    edit_text = pytesseract.image_to_string(edit_img)
+    diff = difflib.ndiff(orig_text.splitlines(), edit_text.splitlines())
+    text_changes = []
+    position_counter = start_position
+
+    for line in diff:
+        if line.startswith("+ "):  # Added text
+            text_changes.append((position_counter, f'"{line[2:]}" added at position {position_counter}'))
+        elif line.startswith("- "):  # Removed text
+            text_changes.append((position_counter, f'"{line[2:]}" removed at position {position_counter}'))
+        position_counter += 1
+
+    return text_changes, position_counter
+
 # Sanitize text for PDF compatibility
 def sanitize_text(text):
-    """Sanitize text for FPDF by replacing unsupported characters."""
     return text.encode('latin-1', errors='replace').decode('latin-1')
 
-# Generate visual changes report
-def generate_visual_report(original_images, edited_images, combined_images, visual_changes):
-    output_path = "outputs/visual_changes.pdf"
-    pdf_visual = FPDF()
-    for img in combined_images:
-        temp_path = "temp_image_visual.png"
+# Generate PDF report
+def generate_report(images, changes, title, output_path):
+    pdf = FPDF()
+    for img in images:
+        temp_path = "temp_image.png"
         cv2.imwrite(temp_path, img)
-        pdf_visual.add_page()
-        pdf_visual.image(temp_path, x=10, y=10, w=190)
+        pdf.add_page()
+        pdf.image(temp_path, x=10, y=10, w=190)
         os.remove(temp_path)
-    pdf_visual.add_page()
-    pdf_visual.set_font("Arial", size=12)
-    pdf_visual.cell(0, 10, "Visual Changes", ln=True, align="C")
-    pdf_visual.ln(10)
-    for _, change in visual_changes:
-        pdf_visual.cell(0, 10, sanitize_text(change), ln=True)
-    pdf_visual.output(output_path)
-    return output_path
 
-# Generate text changes report
-def generate_text_report(original_images, edited_images, combined_images, text_changes):
-    output_path = "outputs/text_changes.pdf"
-    pdf_text = FPDF()
-    for img in combined_images:
-        temp_path = "temp_image_text.png"
-        cv2.imwrite(temp_path, img)
-        pdf_text.add_page()
-        pdf_text.image(temp_path, x=10, y=10, w=190)
-        os.remove(temp_path)
-    pdf_text.add_page()
-    pdf_text.set_font("Arial", size=12)
-    pdf_text.cell(0, 10, "Text Changes", ln=True, align="C")
-    pdf_text.ln(10)
-    for _, change in text_changes:
-        pdf_text.cell(0, 10, sanitize_text(change), ln=True)
-    pdf_text.output(output_path)
+    pdf.add_page()
+    pdf.set_font("Arial", size=12)
+    pdf.cell(0, 10, sanitize_text(title), ln=True, align="C")
+    pdf.ln(10)
+    for _, change in changes:
+        pdf.cell(0, 10, sanitize_text(change), ln=True)
+
+    pdf.output(output_path)
     return output_path
 
-# Generate separate PDFs for visual and text changes
+# Perform visual and text comparisons separately
 def generate_separate_comparisons(original_pdf, edited_pdf):
     original_images = convert_pdf_to_images(original_pdf)
     edited_images = convert_pdf_to_images(edited_pdf)
-    combined_images = []
+
+    # Visual comparison
+    visual_combined_images = []
     visual_changes = []
-    text_changes = []
     position_counter = 1
-
     for orig_img, edit_img in zip(original_images, edited_images):
         aligned_img = align_images(orig_img, edit_img)
-
-        # Visual comparison
-        diff_mask = compare_images(orig_img, aligned_img)
-        highlighted_img, page_visual_changes, position_counter = highlight_visual_changes(
-            orig_img, edit_img, diff_mask, position_counter
+        highlighted_img, page_visual_changes, position_counter = compare_visual_changes(
+            orig_img, aligned_img, position_counter
         )
         visual_changes.extend(page_visual_changes)
+        visual_combined_images.append(np.hstack((orig_img, highlighted_img)))
 
-        # Text comparison
-        orig_text = pytesseract.image_to_string(orig_img)
-        edit_text = pytesseract.image_to_string(edit_img)
-        page_text_changes, position_counter = compare_text(orig_text, edit_text, position_counter)
-        text_changes.extend(page_text_changes)
+    # Generate visual changes report
+    visual_report_path = generate_report(
+        visual_combined_images, visual_changes, "Visual Changes", "outputs/visual_changes.pdf"
+    )
 
-        # Combine images for side-by-side display
-        height = min(orig_img.shape[0], highlighted_img.shape[0])
-        orig_img_resized = orig_img[:height]
-        highlighted_img_resized = highlighted_img[:height]
-        combined_images.append(np.hstack((orig_img_resized, highlighted_img_resized)))
+    # Text comparison
+    text_combined_images = []
+    text_changes = []
+    position_counter = 1
+    for orig_img, edit_img in zip(original_images, edited_images):
+        aligned_img = align_images(orig_img, edit_img)
+        page_text_changes, position_counter = compare_text_changes(orig_img, aligned_img, position_counter)
+        text_changes.extend(page_text_changes)
+        text_combined_images.append(np.hstack((orig_img, aligned_img)))
 
-    # Generate separate reports
-    visual_report = generate_visual_report(original_images, edited_images, combined_images, visual_changes)
-    text_report = generate_text_report(original_images, edited_images, combined_images, text_changes)
+    # Generate text changes report
+    text_report_path = generate_report(
+        text_combined_images, text_changes, "Text Changes", "outputs/text_changes.pdf"
+    )
 
-    return visual_report, text_report
+    return visual_report_path, text_report_path
 
 # Gradio interface function
 def pdf_comparison(original_pdf, edited_pdf):