pdf_auto_crop_page
Automatically crop PDF pages to remove blank margins by detecting content boundaries, optimizing document layout and reducing file size.
Instructions
Automatically crop a PDF page to remove blank margins by detecting content boundaries.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| pdf_path | Yes | ||
| page_number | No | ||
| padding | No |
Output Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| result | Yes |
Implementation Reference
- Core handler function decorated with @mcp.tool(). Opens PDF, detects content bounds from words, images, drawings; computes union rect; applies conservative padding (min 20pt, asymmetric); crops pages if significant margins; saves to timestamped file.
@mcp.tool() async def pdf_auto_crop_page( pdf_path: str, page_number: Optional[int] = None, padding: float = 10.0 ) -> str: """Automatically crop a PDF page to remove blank margins by detecting content boundaries.""" if not os.path.exists(pdf_path): return f"Error: PDF file not found: {pdf_path}" if not validate_pdf_file(pdf_path): return f"Error: Invalid PDF file: {pdf_path}" try: # Open PDF document doc = fitz.open(pdf_path) # Determine pages to process if page_number is not None: if not validate_page_number(doc, page_number): doc.close() return f"Error: Invalid page number {page_number}. Document has {len(doc)} pages." pages_to_process = [page_number] else: pages_to_process = list(range(len(doc))) cropped_pages = 0 for page_num in pages_to_process: page = doc[page_num] # Get text at word level for tighter bounds words = page.get_text("words") text_rects = [word[:4] for word in words if len(word) >= 4] # Get image rectangles images = page.get_images() image_rects = [img[:4] for img in images if len(img) >= 4] # Get drawing objects (lines, shapes, paths) - NO external dependencies drawing_rects = [] try: drawings = page.get_drawings() for drawing in drawings: if 'rect' in drawing: drawing_rects.append(drawing['rect']) except Exception: pass # Combine all rectangles all_rects = text_rects + image_rects + drawing_rects # Filter out invalid rectangles (outside page bounds or with invalid coordinates) page_rect = page.rect valid_rects = [] for rect in all_rects: if len(rect) >= 4: try: r = fitz.Rect(rect[:4]) # Check if rectangle is valid and within reasonable bounds if (r.is_valid and r.x0 >= 0 and r.y0 >= 0 and r.x1 <= page_rect.width and r.y1 <= page_rect.height and r.width > 0 and r.height > 0): valid_rects.append(rect[:4]) except Exception: continue all_rects = valid_rects if all_rects: # Calculate union of all content rectangles content_rect = fitz.Rect(all_rects[0]) for rect in all_rects[1:]: content_rect |= fitz.Rect(rect) # More conservative padding strategy to preserve section flow # Use asymmetric padding: less aggressive on sides, more generous on top/bottom page_rect = page.rect # Calculate how much we can crop while preserving document flow # Only crop if there's significant margin (at least 2 points on each side) margin_threshold = 2.0 # Check if margins are significant enough to warrant cropping left_margin = content_rect.x0 right_margin = page_rect.width - content_rect.x1 top_margin = content_rect.y0 bottom_margin = page_rect.height - content_rect.y1 # Only crop if margins are substantial if (left_margin > margin_threshold or right_margin > margin_threshold or top_margin > margin_threshold or bottom_margin > margin_threshold): # Conservative padding: preserve more space for better flow conservative_padding = max(padding, 20.0) # At least 20 points padding # Asymmetric padding: less on sides, more on top/bottom for better section flow content_rect = content_rect + [ -min(conservative_padding * 0.5, left_margin * 0.8), # left: 50% of padding or 80% of margin -min(conservative_padding, bottom_margin * 0.8), # bottom: full padding or 80% of margin min(conservative_padding * 0.5, right_margin * 0.8), # right: 50% of padding or 80% of margin min(conservative_padding, top_margin * 0.8) # top: full padding or 80% of margin ] # Ensure the crop box is within page bounds content_rect.intersect(page_rect) # Apply crop if there's any reduction in size if (content_rect.width < page_rect.width or content_rect.height < page_rect.height): page.set_cropbox(content_rect) cropped_pages += 1 else: # No content found, skip this page continue if cropped_pages == 0: doc.close() return "No content found to crop on any pages." # Generate output filename output_path = generate_output_filename(pdf_path, "auto_cropped") # Save the modified PDF doc.save(output_path) doc.close() page_info = f"page {page_number + 1}" if page_number is not None else f"{cropped_pages} pages" return f"Successfully auto-cropped {page_info}. Output saved to: {output_path}" except Exception as e: return f"Error auto-cropping PDF: {str(e)}" - pdf_manipulation_mcp_server/pdf_server.py:19-19 (registration)Initialization of the FastMCP server instance. All tools including pdf_auto_crop_page are registered via @mcp.tool() decorators on their handler functions.
mcp = FastMCP("pdf-manipulation") - Utility to generate timestamped output filename used by pdf_auto_crop_page to save results without overwriting input.
def generate_output_filename(input_path: str, suffix: str = "modified") -> str: """Generate a new filename with timestamp to avoid overwriting originals.""" path = Path(input_path) timestamp = datetime.now().strftime("%Y%m%d_%H%M%S") return str(path.parent / f"{path.stem}_{suffix}_{timestamp}{path.suffix}") - Validates if input file is a valid PDF by attempting to open it with PyMuPDF; used early in handler.
def validate_pdf_file(pdf_path: str) -> bool: """Validate that the file is a valid PDF.""" try: doc = fitz.open(pdf_path) doc.close() return True except Exception: return False - Checks if specified page number is valid for the document; used when page_number param provided.
def validate_page_number(doc: fitz.Document, page_num: int) -> bool: """Validate that the page number exists in the document.""" return 0 <= page_num < len(doc)