rothbard/backfill_is_archived.py

#!/usr/bin/env python3
"""
One-off script to backfill is_archived field on all projects in Firestore.

This sets is_archived = True for projects where phase_name == "Archived",
and is_archived = False for all other projects.

Usage:
    python backfill_is_archived.py
"""

import os
from firebase_admin import credentials, initialize_app, firestore

# Path to your staging service account JSON
CREDENTIALS_PATH = "./rothbard-staging2-12345-firebase-adminsdk-fbsvc-7f95268383.json"

def main():
    # Initialize Firebase Admin with staging credentials
    cred = credentials.Certificate(CREDENTIALS_PATH)
    app = initialize_app(cred, name='backfill-is-archived')
    db = firestore.client(app=app)

    projects_ref = db.collection("projects")
    docs = list(projects_ref.stream())

    total = len(docs)
    archived_count = 0
    updated_count = 0
    batch_size = 500

    print(f"Found {total} projects. Processing in batches of {batch_size}...")

    for i in range(0, total, batch_size):
        batch = db.batch()
        batch_docs = docs[i:i + batch_size]

        for doc in batch_docs:
            data = doc.to_dict()
            phase_name = data.get("phase_name", "")
            is_archived = (phase_name == "Archived")

            if is_archived:
                archived_count += 1

            # Only update if the field is missing or different
            if data.get("is_archived") != is_archived:
                ref = projects_ref.document(doc.id)
                batch.update(ref, {"is_archived": is_archived})
                updated_count += 1

        batch.commit()
        print(f"  Committed batch {i//batch_size + 1}/{(total + batch_size - 1)//batch_size}")

    print(f"\nDone!")
    print(f"  Total projects: {total}")
    print(f"  Projects with phase_name == 'Archived': {archived_count}")
    print(f"  Documents updated: {updated_count}")

if __name__ == "__main__":
    main()