-
Notifications
You must be signed in to change notification settings - Fork 27
/
functions.py
executable file
·88 lines (71 loc) · 2.94 KB
/
functions.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
#!/usr/bin/env python3
# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this
# file, You can obtain one at http://mozilla.org/MPL/2.0/.
import markovify
from bs4 import BeautifulSoup
import re, multiprocessing, sqlite3, shutil, os, html
def make_sentence(output, cfg):
class nlt_fixed(markovify.NewlineText): #modified version of NewlineText that never rejects sentences
def test_sentence_input(self, sentence):
return True #all sentences are valid <3
shutil.copyfile("toots.db", "toots-copy.db") #create a copy of the database because reply.py will be using the main one
db = sqlite3.connect("toots-copy.db")
db.text_factory = str
c = db.cursor()
if cfg['learn_from_cw']:
toots = c.execute("SELECT content FROM `toots` ORDER BY RANDOM() LIMIT 10000").fetchall()
else:
toots = c.execute("SELECT content FROM `toots` WHERE cw = 0 ORDER BY RANDOM() LIMIT 10000").fetchall()
if len(toots) == 0:
output.send("Database is empty! Try running main.py.")
return
model = nlt_fixed(
"\n".join([toot[0] for toot in toots])
)
db.close()
os.remove("toots-copy.db")
toots_str = None
sentence = None
tries = 0
while sentence is None and tries < 10:
sentence = model.make_short_sentence(500, tries=10000)
tries = tries + 1
# optionally remove mentions
if cfg['mention_handling'] == 1:
sentence = re.sub(r"^\S*@\u200B\S*\s?", "", sentence)
elif cfg['mention_handling'] == 0:
sentence = re.sub(r"\S*@\u200B\S*\s?", "", sentence)
output.send(sentence)
def make_toot(cfg):
toot = None
pin, pout = multiprocessing.Pipe(False)
p = multiprocessing.Process(target = make_sentence, args = [pout, cfg])
p.start()
p.join(5) #wait 5 seconds to get something
if p.is_alive(): #if it's still trying to make a toot after 5 seconds
p.terminate()
p.join()
else:
toot = pin.recv()
if toot == None:
toot = "Toot generation failed! Contact Lynne ([email protected]) for assistance."
return toot
def extract_toot(toot):
toot = html.unescape(toot) # convert HTML escape codes to text
soup = BeautifulSoup(toot, "html.parser")
for lb in soup.select("br"): # replace <br> with linebreak
lb.replace_with("\n")
for p in soup.select("p"): # ditto for <p>
p.replace_with("\n")
for ht in soup.select("a.hashtag"): # convert hashtags from links to text
ht.unwrap()
for link in soup.select("a"): #ocnvert <a href='https://example.com>example.com</a> to just https://example.com
if 'href' in link:
# apparently not all a tags have a href, which is understandable if you're doing normal web stuff, but on a social media platform??
link.replace_with(link["href"])
text = soup.get_text()
text = re.sub(r"https://([^/]+)/(@[^\s]+)", r"\2@\1", text) # put mastodon-style mentions back in
text = re.sub(r"https://([^/]+)/users/([^\s/]+)", r"@\2@\1", text) # put pleroma-style mentions back in
text = text.rstrip("\n") # remove trailing newline(s)
return text