pleroma-ebooks/generators/markov.py
io c22a493dff license compliance stuff and other changes
relicense to AGPLv3
comply with the MPL's file copyleft
some other changes in the direction of supporting GPT-2 which were hard to untangle from the rest of the changes (sorry)
2021-07-26 04:52:44 +00:00

45 lines
1.5 KiB
Python

# SPDX-License-Identifier: MPL-2.0
import sqlite3
import markovify
def make_sentence(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
db = sqlite3.connect("toots.db")
db.text_factory = str
c = db.cursor()
if cfg['learn_from_cw']:
ignored_cws_query_params = "(" + ",".join("?" * len(cfg["ignored_cws"])) + ")"
toots = c.execute(f"SELECT content FROM `toots` WHERE cw IS NULL OR CW NOT IN {ignored_cws_query_params} ORDER BY RANDOM() LIMIT 10000", cfg["ignored_cws"]).fetchall()
else:
toots = c.execute("SELECT content FROM `toots` WHERE cw IS NULL ORDER BY RANDOM() LIMIT 10000").fetchall()
if len(toots) == 0:
raise ValueError("Database is empty! Try running main.py.")
nlt = markovify.NewlineText if cfg['overlap_ratio_enabled'] else nlt_fixed
model = nlt("\n".join(toot[0].replace('\n', ' ') for toot in toots))
db.close()
if cfg['limit_length']:
sentence_len = randint(cfg['length_lower_limit'], cfg['length_upper_limit'])
sentence = None
tries = 0
for tries in range(10):
if (sentence := model.make_short_sentence(
max_chars=500,
tries=10000,
max_overlap_ratio=cfg['overlap_ratio'] if cfg['overlap_ratio_enabled'] else 0.7,
max_words=sentence_len if cfg['limit_length'] else None
)) is not None:
break
else:
raise ValueError("Failed 10 times to produce a sentence!")
return sentence