commit 86047b21c7959b2343cebdf098cc5b969e0f96bd · edavis.dev/bsky-tools

feed_manager.py

···

       6
        
       from feeds.homeruns import HomeRunsTeamFeed

     

       7
        
       from feeds.norazone_interesting import NoraZoneInteresting

     

       8
        
       from feeds.sevendirtywords import SevenDirtyWordsFeed

     

       0
        
       
     

       9
        
       

     

       10
        
       class FeedManager:

     

       11
        
           def __init__(self):

     
···

       48
        
       feed_manager.register(HomeRunsTeamFeed)

     

       49
        
       feed_manager.register(NoraZoneInteresting)

     

       50
        
       feed_manager.register(SevenDirtyWordsFeed)

     

       0

···

       6
        
       from feeds.homeruns import HomeRunsTeamFeed

     

       7
        
       from feeds.norazone_interesting import NoraZoneInteresting

     

       8
        
       from feeds.sevendirtywords import SevenDirtyWordsFeed

     

       9
       +
       from feeds.ratio import RatioFeed

     

       10
        
       

     

       11
        
       class FeedManager:

     

       12
        
           def __init__(self):

     
···

       49
        
       feed_manager.register(HomeRunsTeamFeed)

     

       50
        
       feed_manager.register(NoraZoneInteresting)

     

       51
        
       feed_manager.register(SevenDirtyWordsFeed)

     

       52
       +
       feed_manager.register(RatioFeed)

+144

feeds/ratio.py

···

       1
       +
       import logging

     

       2
       +
       

     

       3
       +
       import apsw

     

       4
       +
       import apsw.ext

     

       5
       +
       

     

       6
       +
       from . import BaseFeed

     

       7
       +
       

     

       8
       +
       class RatioFeed(BaseFeed):

     

       9
       +
           FEED_URI = 'at://did:plc:4nsduwlpivpuur4mqkbfvm6a/app.bsky.feed.generator/ratio'

     

       10
       +
           SERVE_FEED_QUERY = """

     

       11
       +
           with served as (

     

       12
       +
             select

     

       13
       +
               uri,

     

       14
       +
               create_ts,

     

       15
       +
               ( unixepoch('now') - create_ts ) as age_seconds,

     

       16
       +
               replies,

     

       17
       +
               quoteposts,

     

       18
       +
               likes,

     

       19
       +
               reposts,

     

       20
       +
               ( replies + quoteposts ) / ( likes + reposts ) as ratio,

     

       21
       +
               exp( -1 * ( ( unixepoch('now') - create_ts ) / ( 3600.0 * 16 ) ) ) as decay

     

       22
       +
             from posts

     

       23
       +
           )

     

       24
       +
           select

     

       25
       +
             *,

     

       26
       +
             ( ratio * decay ) as score

     

       27
       +
           from served

     

       28
       +
           where replies > 15 and ratio > 2.5

     

       29
       +
           order by score desc

     

       30
       +
           limit :limit offset :offset

     

       31
       +
           """

     

       32
       +
           DELETE_OLD_POSTS_QUERY = """

     

       33
       +
           delete from posts

     

       34
       +
           where

     

       35
       +
             create_ts < unixepoch('now', '-5 days')

     

       36
       +
           """

     

       37
       +
       

     

       38
       +
           def __init__(self):

     

       39
       +
               self.db_cnx = apsw.Connection('db/ratio.db')

     

       40
       +
               self.db_cnx.pragma('journal_mode', 'WAL')

     

       41
       +
               self.db_cnx.pragma('wal_autocheckpoint', '0')

     

       42
       +
       

     

       43
       +
               with self.db_cnx:

     

       44
       +
                   self.db_cnx.execute("""

     

       45
       +
                   create table if not exists posts (

     

       46
       +
                     uri text, create_ts timestamp,

     

       47
       +
                     replies float, likes float, reposts float, quoteposts float

     

       48
       +
                   );

     

       49
       +
                   create unique index if not exists uri_idx on posts(uri);

     

       50
       +
                   """)

     

       51
       +
       

     

       52
       +
               self.logger = logging.getLogger('feeds.ratio')

     

       53
       +
       

     

       54
       +
           def process_commit(self, commit):

     

       55
       +
               if commit['opType'] != 'c':

     

       56
       +
                   return

     

       57
       +
       

     

       58
       +
               subject_uri = None

     

       59
       +
               is_reply = False

     

       60
       +
               is_quotepost = False

     

       61
       +
       

     

       62
       +
               if commit['collection'] in {'app.bsky.feed.like', 'app.bsky.feed.repost'}:

     

       63
       +
                   record = commit.get('record')

     

       64
       +
                   ts = self.safe_timestamp(record.get('createdAt')).timestamp()

     

       65
       +
                   try:

     

       66
       +
                       subject_uri = record['subject']['uri']

     

       67
       +
                   except KeyError:

     

       68
       +
                       return

     

       69
       +
               elif commit['collection'] == 'app.bsky.feed.post':

     

       70
       +
                   record = commit.get('record')

     

       71
       +
                   ts = self.safe_timestamp(record.get('createdAt')).timestamp()

     

       72
       +
                   if record.get('reply') is not None:

     

       73
       +
                       is_reply = True

     

       74
       +
                       try:

     

       75
       +
                           subject_uri = record['reply']['parent']['uri']

     

       76
       +
                       except KeyError:

     

       77
       +
                           return

     

       78
       +
       

     

       79
       +
                       # only count non-OP replies

     

       80
       +
                       if subject_uri.startswith('at://' + commit['did']):

     

       81
       +
                           return

     

       82
       +
       

     

       83
       +
                   elif record.get('embed') is not None:

     

       84
       +
                       is_quotepost = True

     

       85
       +
                       t = record['embed']['$type']

     

       86
       +
                       if t == 'app.bsky.embed.record':

     

       87
       +
                           try:

     

       88
       +
                               subject_uri = record['embed']['record']['uri']

     

       89
       +
                           except KeyError:

     

       90
       +
                               return

     

       91
       +
                       elif t == 'app.bsky.embed.recordWithMedia':

     

       92
       +
                           try:

     

       93
       +
                               subject_uri = record['embed']['record']['record']['uri']

     

       94
       +
                           except KeyError:

     

       95
       +
                               return

     

       96
       +
       

     

       97
       +
               if subject_uri is None:

     

       98
       +
                   return

     

       99
       +
       

     

       100
       +
               params = {

     

       101
       +
                   'uri': subject_uri,

     

       102
       +
                   'ts': ts,

     

       103
       +
                   'is_reply': int(is_reply),

     

       104
       +
                   'is_like': int(commit['collection'] == 'app.bsky.feed.like'),

     

       105
       +
                   'is_repost': int(commit['collection'] == 'app.bsky.feed.repost'),

     

       106
       +
                   'is_quotepost': int(is_quotepost),

     

       107
       +
               }

     

       108
       +
       

     

       109
       +
               self.transaction_begin(self.db_cnx)

     

       110
       +
       

     

       111
       +
               self.db_cnx.execute("""

     

       112
       +
               insert into posts(uri, create_ts, replies, likes, reposts, quoteposts)

     

       113
       +
               values (:uri, :ts,

     

       114
       +
                       case when :is_reply then 1 else 0 end,

     

       115
       +
                       case when :is_like then 1 else 0 end,

     

       116
       +
                       case when :is_repost then 1 else 0 end,

     

       117
       +
                       case when :is_quotepost then 1 else 0 end)

     

       118
       +
               on conflict(uri)

     

       119
       +
               do update set

     

       120
       +
                 replies = replies + case when :is_reply then 1 else 0 end,

     

       121
       +
                 likes = likes + case when :is_like then 1 else 0 end,

     

       122
       +
                 reposts = reposts + case when :is_repost then 1 else 0 end,

     

       123
       +
                 quoteposts = quoteposts + case when :is_quotepost then 1 else 0 end

     

       124
       +
               """, params)

     

       125
       +
       

     

       126
       +
           def delete_old_posts(self):

     

       127
       +
               self.db_cnx.execute(self.DELETE_OLD_POSTS_QUERY)

     

       128
       +
       

     

       129
       +
           def commit_changes(self):

     

       130
       +
               self.logger.debug('committing changes')

     

       131
       +
               self.delete_old_posts()

     

       132
       +
               self.transaction_commit(self.db_cnx)

     

       133
       +
               self.wal_checkpoint(self.db_cnx, 'RESTART')

     

       134
       +
       

     

       135
       +
           def serve_feed(self, limit, offset, langs):

     

       136
       +
               cur = self.db_cnx.execute(self.SERVE_FEED_QUERY, dict(limit=limit, offset=offset))

     

       137
       +
               return [row[0] for row in cur]

     

       138
       +
       

     

       139
       +
           def serve_feed_debug(self, limit, offset, langs):

     

       140
       +
               bindings = dict(limit=limit, offset=offset)

     

       141
       +
               return apsw.ext.format_query_table(

     

       142
       +
                   self.db_cnx, self.SERVE_FEED_QUERY, bindings,

     

       143
       +
                   string_sanitize=2, text_width=9999, use_unicode=True

     

       144
       +
               )