pdf-mass-cleanuptools/pdf_processor.py
2025-02-19 21:42:34 +00:00

145 lines
No EOL
5.3 KiB
Python

from pdf2image import convert_from_path
import os
import json
from pathlib import Path
import tempfile
import base64
import anthropic
from typing import List, Dict
import time
class PDFProcessor:
def __init__(self, input_dir: str, output_dir: str, api_key: str):
self.input_dir = Path(input_dir)
self.output_dir = Path(output_dir)
self.output_dir.mkdir(parents=True, exist_ok=True)
self.temp_dir = Path(tempfile.mkdtemp())
self.client = anthropic.Client(api_key=api_key)
def encode_image(self, image_path: str) -> str:
"""Convert image to base64 for API"""
with open(image_path, 'rb') as image_file:
return base64.b64encode(image_file.read()).decode('utf-8')
def analyze_image(self, image_path: str) -> Dict:
"""Analyze a single image using Claude Vision API"""
try:
with open(image_path, 'rb') as img:
message = self.client.messages.create(
model="claude-3-opus-20240229",
max_tokens=1000,
messages=[{
"role": "user",
"content": [
{
"type": "text",
"text": """Analyze this magazine cover and extract the following metadata:
1. Magazine Title
2. Issue Date/Publication Date
3. Publisher
4. Issue Number
Format your response as JSON with these exact keys:
{
"title": string,
"date": string,
"publisher": string,
"issue_number": string,
"confidence": "high|medium|low"
}
If any field cannot be determined, use null. Set confidence based on how clear the information is."""
},
{
"type": "image",
"source": {
"type": "base64",
"media_type": "image/jpeg",
"data": self.encode_image(image_path)
}
}
]
}]
)
# Parse the JSON response
response_text = message.content[0].text
metadata = json.loads(response_text)
return metadata
except Exception as e:
print(f"Error analyzing image {image_path}: {str(e)}")
return {
"title": None,
"date": None,
"publisher": None,
"issue_number": None,
"confidence": "error"
}
def process_pdfs(self) -> List[Dict]:
"""Process all PDFs in the input directory"""
pdf_files = list(self.input_dir.glob('*.pdf'))
results = []
for pdf_path in pdf_files:
try:
result = self.process_single_pdf(pdf_path)
results.append(result)
# Small delay to respect API rate limits
time.sleep(1)
except Exception as e:
print(f"Error processing {pdf_path}: {str(e)}")
results.append({
'pdf_path': str(pdf_path),
'status': 'error',
'error': str(e)
})
# Save results to JSON
with open(self.output_dir / 'processing_results.json', 'w', encoding='utf-8') as f:
json.dump(results, f, indent=4, ensure_ascii=False)
return results
def process_single_pdf(self, pdf_path: Path) -> Dict:
"""Process a single PDF file"""
print(f"Processing: {pdf_path}")
# Convert first page to image
images = convert_from_path(pdf_path, first_page=1, last_page=1)
if not images:
raise Exception("Could not extract first page")
# Save first page image
first_page = images[0]
image_path = self.temp_dir / f"{pdf_path.stem}_page1.jpg"
first_page.save(str(image_path), 'JPEG')
# Analyze the image
metadata = self.analyze_image(str(image_path))
return {
'pdf_path': str(pdf_path),
'image_path': str(image_path),
'metadata': metadata,
'status': 'completed'
}
def main():
# Get API key from environment variable
api_key = os.getenv('ANTHROPIC_API_KEY')
if not api_key:
raise ValueError("ANTHROPIC_API_KEY environment variable not set")
input_dir = "path/to/pdfs"
output_dir = "path/to/output"
processor = PDFProcessor(input_dir, output_dir, api_key)
results = processor.process_pdfs()
print(f"\nProcessed {len(results)} PDF files")
print(f"Results saved to: {processor.output_dir}/processing_results.json")
if __name__ == "__main__":
main()