schema_definitions.py

"""
Schema definitions for PDF field extraction.

This module loads the real estate alias IDs from a JSON file that is exported
from the TypeScript source (real-estate-property-alias-ids.ts) using the
export-alias-ids.js script.

To update the schema:
1. Modify the TypeScript source file
2. Run: node scripts/export-alias-ids.js
3. The JSON file will be updated automatically
"""

import json
import os

# Valid field types that the Vision model can detect
VALID_FIELD_TYPES = [
    "checkbox", 
    "date", 
    "dollar", 
    "email", 
    "fullName", 
    "initial", 
    "longText", 
    "number", 
    "phone", 
    "shortText", 
    "signature", 
    "usAddress",
    "ssn",
    "iban"
]


def load_real_estate_schema():
    """
    Load the real estate schema from the exported JSON file.
    Falls back to an empty dict if the file doesn't exist.
    """
    json_path = os.path.join(os.path.dirname(__file__), "real_estate_alias_ids.json")
    
    if not os.path.exists(json_path):
        print(f"Warning: {json_path} not found. Run 'node scripts/export-alias-ids.js' to generate it.")
        return {}
    
    try:
        with open(json_path, "r") as f:
            data = json.load(f)
            
        # Transform the JSON format to match the expected format
        # JSON has: { key: { description, type, name } }
        # We need: { key: { desc, type } }
        schema = {}
        for key, value in data.items():
            schema[key] = {
                "desc": value.get("description", ""),
                "type": value.get("type", "shortText"),
                "name": value.get("name", key)
            }
        
        return schema
    except Exception as e:
        print(f"Error loading real estate schema: {e}")
        return {}


# Load the schema on module import
REAL_ESTATE_SCHEMA_MAP = load_real_estate_schema()

# Print info about loaded schema
if REAL_ESTATE_SCHEMA_MAP:
    print(f"Loaded {len(REAL_ESTATE_SCHEMA_MAP)} alias IDs from real_estate_alias_ids.json")
else:
    print("Warning: No alias IDs loaded. Schema mapping will create dynamic fields for all detections.")