commit f6f272e6d8775481f78bd987d0110926b32a26a4 · edavis.dev/bsky-tools

+87

bsky-activity.py

···

       1
       +
       #!/usr/bin/env python3

     

       2
       +
       

     

       3
       +
       import dag_cbor

     

       4
       +
       import os

     

       5
       +
       import redis

     

       6
       +
       import sqlite3

     

       7
       +
       import sys

     

       8
       +
       from datetime import datetime, timezone

     

       9
       +
       from firehose_utils import commit_ops

     

       10
       +
       from io import BytesIO

     

       11
       +
       

     

       12
       +
       app_bsky_allowlist = set([

     

       13
       +
           'app.bsky.actor.profile',

     

       14
       +
           'app.bsky.feed.generator',

     

       15
       +
           'app.bsky.feed.like',

     

       16
       +
           'app.bsky.feed.post',

     

       17
       +
           'app.bsky.feed.repost',

     

       18
       +
           'app.bsky.feed.threadgate',

     

       19
       +
           'app.bsky.graph.block',

     

       20
       +
           'app.bsky.graph.follow',

     

       21
       +
           'app.bsky.graph.list',

     

       22
       +
           'app.bsky.graph.listblock',

     

       23
       +
           'app.bsky.graph.listitem',

     

       24
       +
           'app.bsky.labeler.service',

     

       25
       +
       ])

     

       26
       +
       

     

       27
       +
       def main():

     

       28
       +
           redis_cnx = redis.Redis()

     

       29
       +
           redis_pipe = redis_cnx.pipeline()

     

       30
       +
           redis_sub = redis_cnx.pubsub(ignore_subscribe_messages=True)

     

       31
       +
       

     

       32
       +
           db_fname = '/opt/muninsky/users.db'

     

       33
       +
           db_fname = 'users.db'

     

       34
       +
       

     

       35
       +
           db_cnx = sqlite3.connect(db_fname)

     

       36
       +
           with db_cnx:

     

       37
       +
               db_cnx.executescript("""

     

       38
       +
               PRAGMA journal_mode = WAL;

     

       39
       +
               PRAGMA synchronous = off;

     

       40
       +
               CREATE TABLE IF NOT EXISTS users (did TEXT, ts TIMESTAMP);

     

       41
       +
               CREATE UNIQUE INDEX IF NOT EXISTS did_idx on users(did);

     

       42
       +
               CREATE INDEX IF NOT EXISTS ts_idx on users(ts);

     

       43
       +
               """)

     

       44
       +
       

     

       45
       +
           op_count = 0

     

       46
       +
           redis_sub.subscribe('bsky-tools:firehose:stream')

     

       47
       +
           for event in redis_sub.listen():

     

       48
       +
               frame = BytesIO(event['data'])

     

       49
       +
               header = dag_cbor.decode(frame, allow_concat=True)

     

       50
       +
               if header['op'] != 1 or header['t'] != '#commit':

     

       51
       +
                   continue

     

       52
       +
       

     

       53
       +
               payload = dag_cbor.decode(frame)

     

       54
       +
               if payload['tooBig']:

     

       55
       +
                   # TODO(ejd): how handle these?

     

       56
       +
                   continue

     

       57
       +
       

     

       58
       +
               for op in commit_ops(payload):

     

       59
       +
                   if op['action'] != 'create':

     

       60
       +
                       continue

     

       61
       +
       

     

       62
       +
                   collection, _ = op['path'].split('/')

     

       63
       +
                   if collection not in app_bsky_allowlist:

     

       64
       +
                       continue

     

       65
       +
       

     

       66
       +
                   repo_did = payload['repo']

     

       67
       +
                   ts = datetime.now(timezone.utc).timestamp()

     

       68
       +
                   db_cnx.execute(

     

       69
       +
                       'insert into users values (:did, :ts) on conflict (did) do update set ts = :ts',

     

       70
       +
                       {'did': repo_did, 'ts': ts}

     

       71
       +
                   )

     

       72
       +
       

     

       73
       +
                   redis_pipe \

     

       74
       +
                       .incr(collection) \

     

       75
       +
                       .incr('dev.edavis.muninsky.ops')

     

       76
       +
       

     

       77
       +
                   op_count += 1

     

       78
       +
                   if op_count % 500 == 0:

     

       79
       +
                       payload_seq = payload['seq']

     

       80
       +
                       sys.stdout.write(f'checkpoint: seq: {payload_seq}\n')

     

       81
       +
                       redis_pipe.set('dev.edavis.muninsky.seq', payload_seq)

     

       82
       +
                       redis_pipe.execute()

     

       83
       +
                       db_cnx.commit()

     

       84
       +
                       sys.stdout.flush()

     

       85
       +
       

     

       86
       +
       if __name__ == '__main__':

     

       87
       +
           main()

+11

firehose_utils.py

···

       1
       +
       from atproto import CAR

     

       2
       +
       

     

       3
       +
       def commit_ops(payload):

     

       4
       +
           # TODO(ejd): figure out how to validate blocks

     

       5
       +
           car_parsed = CAR.from_bytes(payload['blocks'])

     

       6
       +
           for op in payload['ops']:

     

       7
       +
               repo_op = op.copy()

     

       8
       +
               if op['cid'] is not None:

     

       9
       +
                   repo_op['cid'] = op['cid'].encode('base32')

     

       10
       +
                   repo_op['record'] = car_parsed.blocks[repo_op['cid']]

     

       11
       +
               yield repo_op