this repo has no description
edavis.dev
1import logging
2import re
3
4import apsw
5import apsw.ext
6
7from . import BaseFeed
8
9# https://en.wikipedia.org/wiki/Seven_dirty_words
10SDW_REGEX = re.compile(r'^(shit|piss|fuck|cunt|cocksucker|motherfucker|tits)[!,./;?~ ]*$', re.I|re.A)
11
12class SevenDirtyWordsFeed(BaseFeed):
13 FEED_URI = 'at://did:plc:4nsduwlpivpuur4mqkbfvm6a/app.bsky.feed.generator/sdw'
14
15 def __init__(self):
16 self.db_cnx = apsw.Connection('db/sdw.db')
17 self.db_cnx.pragma('journal_mode', 'WAL')
18 self.db_cnx.pragma('wal_autocheckpoint', '0')
19
20 with self.db_cnx:
21 self.db_cnx.execute("""
22 create table if not exists posts (uri text, create_ts timestamp);
23 create unique index if not exists create_ts_idx on posts(create_ts);
24 """)
25
26 self.logger = logging.getLogger('feeds.sdw')
27
28 def process_commit(self, commit):
29 if commit['opType'] != 'c':
30 return
31
32 if commit['collection'] != 'app.bsky.feed.post':
33 return
34
35 record = commit.get('record')
36 if record is None:
37 return
38
39 conds = [
40 record.get('reply') is None,
41 record.get('embed') is None,
42 record.get('facets') is None,
43 SDW_REGEX.search(record['text']) is not None,
44 ]
45
46 if not all(conds):
47 return
48
49 repo = commit['did']
50 rkey = commit['rkey']
51 post_uri = f'at://{repo}/app.bsky.feed.post/{rkey}'
52 ts = self.safe_timestamp(record.get('createdAt')).timestamp()
53 self.transaction_begin(self.db_cnx)
54 self.db_cnx.execute(
55 'insert into posts (uri, create_ts) values (:uri, :ts)',
56 dict(uri=post_uri, ts=ts)
57 )
58
59 def commit_changes(self):
60 self.logger.debug('committing changes')
61 self.transaction_commit(self.db_cnx)
62 self.wal_checkpoint(self.db_cnx, 'RESTART')
63
64 def serve_feed(self, limit, offset, langs):
65 cur = self.db_cnx.execute("""
66 select uri
67 from posts
68 order by create_ts desc
69 limit :limit
70 offset :offset
71 """, dict(limit=limit, offset=offset))
72 return [uri for (uri,) in cur]
73
74 def serve_feed_debug(self, limit, offset, langs):
75 query = "select * from posts order by create_ts desc limit :limit offset :offset"
76 bindings = dict(limit=limit, offset=offset)
77 return apsw.ext.format_query_table(
78 self.db_cnx, query, bindings,
79 string_sanitize=2, text_width=9999, use_unicode=True
80 )