commit 3ec32d82889d1fea9aa4f5682a8ab88074e1605c · edavis.dev/bsky-tools

+10 -35
bsky-activity.py
···

       3
       3
        
       import asyncio

     

       4
       4
        
       from datetime import datetime, timezone

     

       5
       5
        
       from io import BytesIO

     

       6
       6
       +
       import json

     

       6
       7
        
       import os

     

       7
       8
        
       import sqlite3

     

       8
       9
        
       import sys

     
···

       29
       30
        
       ])

     

       30
       31
        
       

     

       31
       32
        
       async def bsky_activity():

     

       32
       32
       -
           redis_cnx = redis.Redis()

     

       33
       33
       -
           relay_url = 'wss://bsky.network/xrpc/com.atproto.sync.subscribeRepos'

     

       34
       34
       -
           firehose_seq = redis_cnx.get('dev.edavis.muninsky.seq')

     

       35
       35
       -
           if firehose_seq:

     

       36
       36
       -
               relay_url += f'?cursor={firehose_seq.decode()}'

     

       33
       33
       +
           relay_url = 'ws://localhost:6008/subscribe'

     

       37
       34
        
       

     

       38
       35
        
           sys.stdout.write(f'opening websocket connection to {relay_url}\n')

     

       39
       36
        
           sys.stdout.flush()

     

       40
       37
        
       

     

       41
       38
        
           async with websockets.connect(relay_url, ping_timeout=60) as firehose:

     

       42
       39
        
               while True:

     

       43
       43
       -
                   frame = BytesIO(await firehose.recv())

     

       44
       44
       -
                   header = dag_cbor.decode(frame, allow_concat=True)

     

       45
       45
       -
                   if header['op'] != 1 or header['t'] != '#commit':

     

       46
       46
       -
                       continue

     

       40
       40
       +
                   payload = BytesIO(await firehose.recv())

     

       47
       41
        
       

     

       48
       48
       -
                   payload = dag_cbor.decode(frame)

     

       49
       49
       -
                   if payload['tooBig']:

     

       50
       50
       -
                       # TODO(ejd): figure out how to get blocks out-of-band

     

       51
       51
       -
                       continue

     

       52
       52
       -
       

     

       53
       53
       -
                   # TODO(ejd): figure out how to validate blocks

     

       54
       54
       -
                   blocks = payload.pop('blocks')

     

       55
       55
       -
                   car_parsed = CAR.from_bytes(blocks)

     

       56
       56
       -
       

     

       57
       57
       -
                   message = payload.copy()

     

       58
       58
       -
                   del message['ops']

     

       59
       59
       -
                   message['commit'] = message['commit'].encode('base32')

     

       60
       60
       -
       

     

       61
       61
       -
                   for commit_op in payload['ops']:

     

       62
       62
       -
                       op = commit_op.copy()

     

       63
       63
       -
                       if op['cid'] is not None:

     

       64
       64
       -
                           op['cid'] = op['cid'].encode('base32')

     

       65
       65
       -
                           op['record'] = car_parsed.blocks.get(op['cid'])

     

       66
       66
       -
       

     

       67
       67
       -
                       yield message, op

     

       42
       42
       +
                   yield json.load(payload)

     

       68
       43
        
       

     

       69
       44
        
       async def main():

     

       70
       45
        
           redis_cnx = redis.Redis()

     
···

       89
       64
        
           sys.stdout.flush()

     

       90
       65
        
       

     

       91
       66
        
           op_count = 0

     

       92
       92
       -
           async for commit, op in bsky_activity():

     

       93
       93
       -
               if op['action'] != 'create':

     

       67
       67
       +
           async for payload in bsky_activity():

     

       68
       68
       +
               if payload['opType'] != 'c':

     

       94
       69
        
                   continue

     

       95
       70
        
       

     

       96
       96
       -
               collection, _ = op['path'].split('/')

     

       71
       71
       +
               collection = payload['collection']

     

       97
       72
        
               if collection not in app_bsky_allowlist:

     

       98
       73
        
                   continue

     

       99
       74
        
       

     

       100
       100
       -
               repo_did = commit['repo']

     

       101
       101
       -
               repo_update_time = datetime.strptime(commit['time'], '%Y-%m-%dT%H:%M:%S.%fZ').replace(tzinfo=timezone.utc)

     

       75
       75
       +
               repo_did = payload['did']

     

       76
       76
       +
               repo_update_time = datetime.now(timezone.utc)

     

       102
       77
        
               db_cnx.execute(

     

       103
       78
        
                   'insert into users values (:did, :ts) on conflict (did) do update set ts = :ts',

     

       104
       79
        
                   {'did': repo_did, 'ts': repo_update_time.timestamp()}

     
···

       111
       86
        
               op_count += 1

     

       112
       87
        
               if op_count % 500 == 0:

     

       113
       88
        
                   now = datetime.now(timezone.utc)

     

       114
       114
       -
                   payload_seq = commit['seq']

     

       89
       89
       +
                   payload_seq = payload['seq']

     

       115
       90
        
                   payload_lag = now - repo_update_time

     

       116
       91
        
       

     

       117
       92
        
                   sys.stdout.write(f'seq: {payload_seq}, lag: {payload_lag.total_seconds()}\n')