commit 4066b9620b9883592896ef821e7228ce7a5dcd50 · edavis.dev/bsky-tools

feed_manager.py

···

       5
       5
        
       from feeds.popular import PopularFeed

     

       6
       6
        
       from feeds.homeruns import HomeRunsTeamFeed

     

       7
       7
        
       from feeds.norazone_interesting import NoraZoneInteresting

     

       8
       8
       +
       from feeds.sevendirtywords import SevenDirtyWordsFeed

     

       8
       9
        
       

     

       9
       10
        
       class FeedManager:

     

       10
       11
        
           def __init__(self):

     
···

       46
       47
        
       feed_manager.register(PopularFeed)

     

       47
       48
        
       feed_manager.register(HomeRunsTeamFeed)

     

       48
       49
        
       feed_manager.register(NoraZoneInteresting)

     

       50
       50
       +
       feed_manager.register(SevenDirtyWordsFeed)

+75

feeds/sevendirtywords.py

···

       1
       1
       +
       import logging

     

       2
       2
       +
       import re

     

       3
       3
       +
       

     

       4
       4
       +
       import apsw

     

       5
       5
       +
       import apsw.ext

     

       6
       6
       +
       

     

       7
       7
       +
       from . import BaseFeed

     

       8
       8
       +
       

     

       9
       9
       +
       class SevenDirtyWordsFeed(BaseFeed):

     

       10
       10
       +
           FEED_URI = 'at://did:plc:4nsduwlpivpuur4mqkbfvm6a/app.bsky.feed.generator/sdw'

     

       11
       11
       +
       

     

       12
       12
       +
           def __init__(self):

     

       13
       13
       +
               self.db_cnx = apsw.Connection('db/sdw.db')

     

       14
       14
       +
               self.db_cnx.pragma('journal_mode', 'WAL')

     

       15
       15
       +
               self.db_cnx.pragma('wal_autocheckpoint', '0')

     

       16
       16
       +
       

     

       17
       17
       +
               with self.db_cnx:

     

       18
       18
       +
                   self.db_cnx.execute("""

     

       19
       19
       +
                   create table if not exists posts (uri text, create_ts timestamp);

     

       20
       20
       +
                   create unique index if not exists create_ts_idx on posts(create_ts);

     

       21
       21
       +
                   """)

     

       22
       22
       +
       

     

       23
       23
       +
               self.logger = logging.getLogger('feeds.sdw')

     

       24
       24
       +
       

     

       25
       25
       +
           def process_commit(self, commit):

     

       26
       26
       +
               if commit['opType'] != 'c':

     

       27
       27
       +
                   return

     

       28
       28
       +
       

     

       29
       29
       +
               if commit['collection'] != 'app.bsky.feed.post':

     

       30
       30
       +
                   return

     

       31
       31
       +
       

     

       32
       32
       +
               record = commit.get('record')

     

       33
       33
       +
               if record is None:

     

       34
       34
       +
                   return

     

       35
       35
       +
       

     

       36
       36
       +
               # https://en.wikipedia.org/wiki/Seven_dirty_words

     

       37
       37
       +
               if re.search(r'\b(shit|piss|fuck|cunt|cocksucker|motherfucker|tits)\b', record['text'], re.I) is not None:

     

       38
       38
       +
                   repo = commit['did']

     

       39
       39
       +
                   rkey = commit['rkey']

     

       40
       40
       +
                   post_uri = f'at://{repo}/app.bsky.feed.post/{rkey}'

     

       41
       41
       +
                   ts = self.safe_timestamp(record.get('createdAt')).timestamp()

     

       42
       42
       +
                   self.transaction_begin(self.db_cnx)

     

       43
       43
       +
                   self.db_cnx.execute(

     

       44
       44
       +
                       'insert into posts (uri, create_ts) values (:uri, :ts)',

     

       45
       45
       +
                       dict(uri=post_uri, ts=ts)

     

       46
       46
       +
                   )

     

       47
       47
       +
       

     

       48
       48
       +
           def delete_old_posts(self):

     

       49
       49
       +
               self.db_cnx.execute(

     

       50
       50
       +
                   "delete from posts where create_ts < unixepoch('-24 hours')"

     

       51
       51
       +
               )

     

       52
       52
       +
       

     

       53
       53
       +
           def commit_changes(self):

     

       54
       54
       +
               self.logger.debug('committing changes')

     

       55
       55
       +
               self.delete_old_posts()

     

       56
       56
       +
               self.transaction_commit(self.db_cnx)

     

       57
       57
       +
               self.wal_checkpoint(self.db_cnx, 'RESTART')

     

       58
       58
       +
       

     

       59
       59
       +
           def serve_feed(self, limit, offset, langs):

     

       60
       60
       +
               cur = self.db_cnx.execute("""

     

       61
       61
       +
               select uri

     

       62
       62
       +
               from posts

     

       63
       63
       +
               order by create_ts desc

     

       64
       64
       +
               limit :limit

     

       65
       65
       +
               offset :offset

     

       66
       66
       +
               """, dict(limit=limit, offset=offset))

     

       67
       67
       +
               return [uri for (uri,) in cur]

     

       68
       68
       +
       

     

       69
       69
       +
           def serve_feed_debug(self, limit, offset, langs):

     

       70
       70
       +
               query = "select * from posts order by create_ts desc limit :limit offset :offset"

     

       71
       71
       +
               bindings = dict(limit=limit, offset=offset)

     

       72
       72
       +
               return apsw.ext.format_query_table(

     

       73
       73
       +
                   self.db_cnx, query, bindings,

     

       74
       74
       +
                   string_sanitize=2, text_width=9999, use_unicode=True

     

       75
       75
       +
               )