"""
Image processing utilities for PDF MCP server
Simplified algorithm: configurable filtering + maximum size limit
"""
from PIL import Image
import io
from typing import Tuple, Optional
from .config import config
def crop_image_to_max_dimension(
image_bytes: bytes,
max_dimension: int = None
) -> Tuple[Optional[bytes], int, int]:
"""
Simplified image scaling
Algorithm:
1. If either dimension < config.min_image_dimension → discard (return None)
2. Scale down large images only (maintain aspect ratio)
3. Keep small images as is
4. LANCZOS resampling (high quality)
Args:
image_bytes: Original image as bytes
max_dimension: Maximum width or height (default: config.max_image_dimension)
Returns:
Tuple of (scaled_image_bytes, new_width, new_height)
or None if image is too small
"""
# Use config values as defaults
if max_dimension is None:
max_dimension = config.max_image_dimension
min_dim = config.min_image_dimension
try:
# Load image
img = Image.open(io.BytesIO(image_bytes))
width, height = img.size
# Discard if either dimension is below minimum
if width < min_dim or height < min_dim:
return None, width, height
# Return as is if already small enough
if width <= max_dimension and height <= max_dimension:
return image_bytes, width, height
# Scale down large images only (maintain aspect ratio)
scale = min(max_dimension / width, max_dimension / height)
new_width = int(width * scale)
new_height = int(height * scale)
# LANCZOS resampling (high quality downscaling)
img_resized = img.resize((new_width, new_height), Image.Resampling.LANCZOS)
# Convert back to bytes
output = io.BytesIO()
img_format = img.format or 'PNG'
# Save with quality from config
if img_format.upper() in ['JPEG', 'JPG']:
img_resized.save(output, format='JPEG', quality=config.jpeg_quality)
else:
img_resized.save(output, format=img_format)
# Close images to free resources
img.close()
img_resized.close()
return output.getvalue(), new_width, new_height
except Exception:
# Return original if processing fails
min_dim = config.min_image_dimension
try:
img = Image.open(io.BytesIO(image_bytes))
width, height = img.size
# Check size
if width < min_dim or height < min_dim:
return None, width, height
return image_bytes, width, height
except Exception:
return None, 0, 0
def is_header_footer_image(img_data: dict) -> bool:
"""
Image filtering
Filtering conditions:
1. If either dimension < config.min_image_dimension → discard
2. Extreme aspect ratio (> config.max_aspect_ratio) → discard (dividing lines, etc.)
Args:
img_data: Dictionary with 'width', 'height', 'image' keys
Returns:
True if image should be filtered out
"""
try:
width = img_data.get("width", 0)
height = img_data.get("height", 0)
min_dim = config.min_image_dimension
# 1. Discard if either dimension is below minimum size
if width < min_dim or height < min_dim:
return True
# 2. Extreme aspect ratio filter (configurable, 0=disabled)
if config.max_aspect_ratio > 0 and width > 0 and height > 0:
aspect_ratio = max(width / height, height / width)
if aspect_ratio > config.max_aspect_ratio:
return True # Consider as dividing line if > 15:1
return False
except Exception:
return False
