extract_program_metadata.py

import json
import re

def clean_html(raw_html):
    if not raw_html: return ""
    cleanr = re.compile('<.*?>')
    cleantext = re.sub(cleanr, '', raw_html)
    return cleantext.strip()

try:
    with open("nyc_benefits.json", "r") as f:
        data = json.load(f)
        
    metadata = {}
    
    print(f"Loaded {len(data)} entries from nyc_benefits.json")
    
    for entry in data:
        code = entry.get("program_code")
        if code and code != "NULL":
            name = entry.get("program_name", "Unknown")
            # Prefer plain language name if available?
            plain = entry.get("plain_language_program_name")
            if plain and plain != "NULL":
                 # Maybe append plain name? 
                 # name = f"{name} ({plain})"
                 pass
                 
            # Description: brief excerpt is usually better for a list
            desc_html = entry.get("brief_excerpt", "")
            if desc_html == "NULL": desc_html = entry.get("program_description", "")
            
            desc = clean_html(desc_html)
            
            metadata[code] = {
                "name": name,
                "description": desc
            }
            
    print(f"Extracted metadata for {len(metadata)} programs.")
    
    with open("program_metadata.json", "w") as f:
        json.dump(metadata, f, indent=4)
        print("Wrote program_metadata.json")

except Exception as e:
    print(f"Error: {e}")