|
| 1 | +""" |
| 2 | +One-time migration: read all User joinDates, build monthly cumulative history, |
| 3 | +store in Setting entity 'user_count_history'. |
| 4 | +
|
| 5 | +Usage (production): |
| 6 | + GOOGLE_CLOUD_PROJECT=glowscript python ide/shell_cmds/buildUserHistory.py |
| 7 | +
|
| 8 | +Usage (local emulator): |
| 9 | + DATASTORE_EMULATOR_HOST=localhost:8081 python ide/shell_cmds/buildUserHistory.py |
| 10 | +
|
| 11 | +Cost: ~$0.16 (reads ~315k User entities). Run once only. |
| 12 | +""" |
| 13 | + |
| 14 | +from collections import defaultdict |
| 15 | +from datetime import datetime, timezone |
| 16 | +import json |
| 17 | +import os |
| 18 | + |
| 19 | +from google.cloud import ndb |
| 20 | +from ide.models import User, Setting |
| 21 | + |
| 22 | + |
| 23 | +def build_history(client): |
| 24 | + monthly_new = defaultdict(int) |
| 25 | + count = 0 |
| 26 | + skipped = 0 |
| 27 | + |
| 28 | + with client.context(): |
| 29 | + for user in User.query(): # NDB iterates in batches — does not load all into memory |
| 30 | + count += 1 |
| 31 | + if count % 10000 == 0: |
| 32 | + print(f" {count} users processed...") |
| 33 | + |
| 34 | + join_date = getattr(user, 'joinDate', None) |
| 35 | + if join_date: |
| 36 | + month_str = join_date.strftime('%Y-%m') |
| 37 | + monthly_new[month_str] += 1 |
| 38 | + else: |
| 39 | + skipped += 1 |
| 40 | + |
| 41 | + print(f"Done: {count} users, {skipped} skipped (no joinDate)") |
| 42 | + |
| 43 | + sorted_months = sorted(monthly_new.keys()) |
| 44 | + cumulative = 0 |
| 45 | + points = [] |
| 46 | + for month in sorted_months: |
| 47 | + cumulative += monthly_new[month] |
| 48 | + points.append({'month': month, 'count': cumulative}) |
| 49 | + |
| 50 | + history = { |
| 51 | + 'updated': datetime.now(timezone.utc).strftime('%Y-%m-%d'), |
| 52 | + 'points': points, |
| 53 | + } |
| 54 | + |
| 55 | + existing = ndb.Key('Setting', 'user_count_history').get() |
| 56 | + if not existing: |
| 57 | + existing = Setting(id='user_count_history') |
| 58 | + existing.value = json.dumps(history) |
| 59 | + existing.put() |
| 60 | + |
| 61 | + print(f"Stored {len(points)} monthly data points") |
| 62 | + if points: |
| 63 | + print(f"Latest: {points[-1]}") |
| 64 | + |
| 65 | + |
| 66 | +if __name__ == '__main__': |
| 67 | + project = os.environ.get('GOOGLE_CLOUD_PROJECT', 'glowscript') |
| 68 | + emulator = os.environ.get('DATASTORE_EMULATOR_HOST') |
| 69 | + client = ndb.Client(project='glowscript-dev' if emulator else project) |
| 70 | + print(f"Connecting to {'emulator at ' + emulator if emulator else 'production Datastore'}...") |
| 71 | + build_history(client) |
0 commit comments