import requests
import json

# Replace with your JSON URL
json_url = "https://download.geofabrik.de/index-v1.json"

# Fetch JSON data from the URL
response = requests.get(json_url)
data = response.json()

# List to store extracted values
pbf_urls = []

# Recursive function to traverse the JSON
def extract_pbf_urls(obj):
    if isinstance(obj, dict):
        # Check if 'urls' is in the dictionary and it is a dictionary
        if "urls" in obj and isinstance(obj["urls"], dict):
            # Check if 'pbf' key exists and store its value
            if "pbf" in obj["urls"]:
                pbf_urls.append(obj["urls"]["pbf"])
        # Continue traversal
        for value in obj.values():
            extract_pbf_urls(value)
    elif isinstance(obj, list):
        for item in obj:
            extract_pbf_urls(item)

# Run the extraction
extract_pbf_urls(data)

# Sort the URLs
pbf_urls.sort()

# Save to a flat JSON file
with open("pbf_urls.json", "w") as f:
    json.dump(pbf_urls, f, indent=4)

print("Extracted URLs saved to pbf_urls.json")