debug_analysis.py

#!/usr/bin/env python3

import json
from pathlib import Path

def analyze_wto_vs_cto():
    print("=" * 60)
    print("WTO vs CTO Analysis")
    print("=" * 60)
    
    # Analyze WTO extracted document
    wto_file = Path("extracted/0d76c981-9eb0-4d2c-86b5-57cffb4ec7cd.json")
    if wto_file.exists():
        with open(wto_file) as f:
            wto_data = json.load(f)
            
        print("WTO STRUCTURE:")
        print(f"- Images count: {len(wto_data.get('images', []))}")
        print(f"- Has attachment_details: {bool(wto_data.get('attachment_details'))}")
        
        if wto_data.get('images'):
            first_image = wto_data['images'][0]
            print(f"- First image structure: {list(first_image.keys())}")
            print(f"- Has attachment_id: {bool(first_image.get('attachment_id'))}")
        
        # Check content
        content = wto_data.get('content_markdown', '')
        image_refs_local = content.count('images/0d76c981-9eb0-4d2c-86b5-57cffb4ec7cd/')
        image_refs_api = content.count('api/attachments.redirect?id=')
        print(f"- Local image refs in content: {image_refs_local}")
        print(f"- API image refs in content: {image_refs_api}")
    
    print("\n" + "-" * 40)
    
    # Analyze CTO reference
    cto_file = Path("todo.reference.json")
    if cto_file.exists():
        with open(cto_file) as f:
            cto_data = json.load(f)
            
        print("CTO REFERENCE STRUCTURE:")
        print(f"- Attachments count: {len(cto_data.get('attachments', []))}")
        print(f"- Has attachment_details: {bool(cto_data.get('attachment_details'))}")
        
        if cto_data.get('attachment_details'):
            details = cto_data['attachment_details']
            print(f"- Attachment details count: {len(details)}")
            first_key = list(details.keys())[0]
            first_detail = details[first_key]
            print(f"- First detail structure: {list(first_detail.keys())}")
            print(f"- Has attachment_id: {bool(first_detail.get('attachment_id'))}")
            print(f"- Has api_url: {bool(first_detail.get('api_url'))}")
            print(f"- Uploaded status: {first_detail.get('uploaded')}")
        
        # Check content
        content = cto_data.get('md_content', '')
        image_refs_api = content.count('api/attachments.redirect?id=')
        print(f"- API image refs in content: {image_refs_api}")
    
    print("\n" + "=" * 60)
    print("ANALYSIS COMPLETE")
    print("=" * 60)

if __name__ == "__main__":
    analyze_wto_vs_cto()