commit 1b816c2d7819a35c4cfc9e92993438f23dde2d8a · edavis.dev/bsky-tools

+27 -36
bsky-activity.py
···

       1
       1
        
       #!/usr/bin/env python3

     

       2
       2
        
       

     

       3
       3
       -
       import dag_cbor

     

       4
       3
        
       import os

     

       5
       4
        
       import redis

     

       6
       5
        
       import sqlite3

     

       7
       6
        
       import sys

     

       8
       7
        
       from datetime import datetime, timezone

     

       9
       9
       -
       from firehose_utils import commit_ops

     

       10
       10
       -
       from io import BytesIO

     

       8
       8
       +
       from firehose_utils import subscribe_commits

     

       11
       9
        
       

     

       12
       10
        
       app_bsky_allowlist = set([

     

       13
       11
        
           'app.bsky.actor.profile',

     
···

       27
       25
        
       def main():

     

       28
       26
        
           redis_cnx = redis.Redis()

     

       29
       27
        
           redis_pipe = redis_cnx.pipeline()

     

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

     

       31
       28
        
       

     

       32
       29
        
           if os.path.exists('/opt/muninsky/users.db'):

     

       33
       30
        
               db_fname = '/opt/muninsky/users.db'

     
···

       43
       40
        
               CREATE UNIQUE INDEX IF NOT EXISTS did_idx on users(did);

     

       44
       41
        
               CREATE INDEX IF NOT EXISTS ts_idx on users(ts);

     

       45
       42
        
               """)

     

       43
       43
       +
       

     

       44
       44
       +
           sys.stdout.write('starting up\n')

     

       45
       45
       +
           sys.stdout.flush()

     

       46
       46
        
       

     

       47
       47
        
           op_count = 0

     

       48
       48
       -
           redis_sub.subscribe('bsky-tools:firehose:stream')

     

       49
       49
       -
           for event in redis_sub.listen():

     

       50
       50
       -
               frame = BytesIO(event['data'])

     

       51
       51
       -
               header = dag_cbor.decode(frame, allow_concat=True)

     

       52
       52
       -
               if header['op'] != 1 or header['t'] != '#commit':

     

       48
       48
       +
           for commit, op in subscribe_commits():

     

       49
       49
       +
               if op['action'] != 'create':

     

       53
       50
        
                   continue

     

       54
       51
        
       

     

       55
       55
       -
               payload = dag_cbor.decode(frame)

     

       56
       56
       -
               if payload['tooBig']:

     

       57
       57
       -
                   # TODO(ejd): how handle these?

     

       52
       52
       +
               collection, _ = op['path'].split('/')

     

       53
       53
       +
               if collection not in app_bsky_allowlist:

     

       58
       54
        
                   continue

     

       59
       55
        
       

     

       60
       60
       -
               for op in commit_ops(payload):

     

       61
       61
       -
                   if op['action'] != 'create':

     

       62
       62
       -
                       continue

     

       56
       56
       +
               repo_did = commit['repo']

     

       57
       57
       +
               now = datetime.now(timezone.utc)

     

       58
       58
       +
               db_cnx.execute(

     

       59
       59
       +
                   'insert into users values (:did, :ts) on conflict (did) do update set ts = :ts',

     

       60
       60
       +
                   {'did': repo_did, 'ts': now.timestamp()}

     

       61
       61
       +
               )

     

       63
       62
        
       

     

       64
       64
       -
                   collection, _ = op['path'].split('/')

     

       65
       65
       -
                   if collection not in app_bsky_allowlist:

     

       66
       66
       -
                       continue

     

       63
       63
       +
               redis_pipe \

     

       64
       64
       +
                   .incr(collection) \

     

       65
       65
       +
                   .incr('dev.edavis.muninsky.ops')

     

       67
       66
        
       

     

       68
       68
       -
                   repo_did = payload['repo']

     

       69
       69
       -
                   ts = datetime.now(timezone.utc).timestamp()

     

       70
       70
       -
                   db_cnx.execute(

     

       71
       71
       -
                       'insert into users values (:did, :ts) on conflict (did) do update set ts = :ts',

     

       72
       72
       -
                       {'did': repo_did, 'ts': ts}

     

       73
       73
       -
                   )

     

       67
       67
       +
               op_count += 1

     

       68
       68
       +
               if op_count % 1000 == 0:

     

       69
       69
       +
                   payload_seq = commit['seq']

     

       70
       70
       +
                   payload_time = datetime.strptime(commit['time'], '%Y-%m-%dT%H:%M:%S.%fZ').replace(tzinfo=timezone.utc)

     

       71
       71
       +
                   payload_lag = now - payload_time

     

       74
       72
        
       

     

       75
       75
       -
                   redis_pipe \

     

       76
       76
       -
                       .incr(collection) \

     

       77
       77
       -
                       .incr('dev.edavis.muninsky.ops')

     

       78
       78
       -
       

     

       79
       79
       -
                   op_count += 1

     

       80
       80
       -
                   if op_count % 500 == 0:

     

       81
       81
       -
                       payload_seq = payload['seq']

     

       82
       82
       -
                       sys.stdout.write(f'checkpoint: seq: {payload_seq}\n')

     

       83
       83
       -
                       redis_pipe.set('dev.edavis.muninsky.seq', payload_seq)

     

       84
       84
       -
                       redis_pipe.execute()

     

       85
       85
       -
                       db_cnx.commit()

     

       86
       86
       -
                       sys.stdout.flush()

     

       73
       73
       +
                   sys.stdout.write(f'seq: {payload_seq}, lag: {payload_lag.total_seconds()}\n')

     

       74
       74
       +
                   redis_pipe.set('dev.edavis.muninsky.seq', payload_seq)

     

       75
       75
       +
                   redis_pipe.execute()

     

       76
       76
       +
                   db_cnx.commit()

     

       77
       77
       +
                   sys.stdout.flush()

     

       87
       78
        
       

     

       88
       79
        
       if __name__ == '__main__':

     

       89
       80
        
           main()