1
0
Fork 0
mirror of https://github.com/Lynnesbian/FediBooks/ synced 2024-11-25 16:48:58 +00:00

Compare commits

...

5 commits

2 changed files with 75 additions and 18 deletions

View file

@ -2,12 +2,17 @@
from mastodon import Mastodon
import MySQLdb
import requests
import markovify
from multiprocessing import Pool
import json, re
import functions
cfg = json.load(open('config.json'))
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
def scrape_posts(account):
handle = account[0]
outbox = account[1]
@ -82,10 +87,67 @@ def scrape_posts(account):
db.commit()
c.close()
print("Finished {}".format(handle))
def make_post(bot):
pass
def make_post(handle):
handle = handle[0]
print("Generating post for {}".format(handle))
c = db.cursor()
c.execute("""
SELECT
learn_from_cw, client_id, client_secret, secret
FROM
bots, credentials
WHERE
bots.credentials_id = (SELECT
credentials_id
FROM
bots
WHERE
handle = %s)
""", (handle,))
bot = c.fetchone()
client = Mastodon(
client_id = bot[1],
client_secret = bot[2],
access_token = bot[3],
api_base_url = "https://{}".format(handle.split("@")[2])
)
# by default, only select posts that don't have CWs.
# if learn_from_cw, then also select posts with CWs
cw_list = [False]
if bot[0]:
cw_list = [False, True]
# select 1000 random posts for the bot to learn from
c.execute("SELECT content FROM posts WHERE fedi_id IN (SELECT fedi_id FROM bot_learned_accounts WHERE bot_id = %s) AND cw IN %s ORDER BY RAND() LIMIT 1000", (handle, cw_list))
# this line is a little gross/optimised but here's what it does
# 1. fetch all of the results from the above query
# 2. turn (('this',), ('format')) into ('this', 'format')
# 3. convert the tuple to a list
# 4. join the list into a string separated by newlines
posts = "\n".join(list(sum(c.fetchall(), ())))
model = nlt_fixed(posts)
tries = 0
sentence = None
# even with such a high tries value for markovify, it still sometimes returns none.
# so we implement our own tries function as well, and try ten times.
while sentence is None and tries < 10:
sentence = model.make_short_sentence(500, tries = 10000)
tries += 1
# TODO: mention handling
if sentence == None:
# TODO: send an error email
pass
else:
client.status_post(sentence)
# TODO: update date of last post
print("Establishing DB connection")
db = MySQLdb.connect(
@ -95,24 +157,19 @@ db = MySQLdb.connect(
db=cfg['db_name']
)
print("Downloading posts")
print("Cleaning up database")
# delete any fedi accounts we no longer need
cursor = db.cursor()
cursor.execute("DELETE FROM fedi_accounts WHERE handle NOT IN (SELECT fedi_id FROM bot_learned_accounts);")
print("Downloading posts")
cursor.execute("SELECT `handle`, `outbox` FROM `fedi_accounts` ORDER BY RAND()")
accounts = cursor.fetchall()
with Pool(8) as p:
p.map(scrape_posts, accounts)
# with Pool(8) as p:
# p.map(scrape_posts, accounts)
print("Generating posts")
cursor.execute("""
SELECT
bots.handle, credentials.client_id, credentials.client_secret, credentials.secret
FROM
bots,
credentials
WHERE
bots.credentials_id = credentials.id
AND bots.enabled = TRUE;
""")
cursor.execute("SELECT handle FROM bots WHERE enabled = TRUE")
bots = cursor.fetchall()
with Pool(8) as p:

View file

@ -184,7 +184,7 @@ def bot_accounts_add():
if r.status_code == 200:
# success!!
c = mysql.connection.cursor()
c.execute("INSERT INTO `fedi_accounts` (`handle`, `outbox`) VALUES (%s, %s)", (request.form['account'], outbox))
c.execute("REPLACE INTO `fedi_accounts` (`handle`, `outbox`) VALUES (%s, %s)", (request.form['account'], outbox))
c.execute("INSERT INTO `bot_learned_accounts` (`bot_id`, `fedi_id`) VALUES (%s, %s)", (session['bot'], request.form['account']))
c.close()
mysql.connection.commit()
@ -307,7 +307,7 @@ def bot_create():
# authentication success!!
c = mysql.connection.cursor()
c.execute("INSERT INTO `credentials` (client_id, client_secret, secret) VALUES (%s, %s, %s)", (session['client_id'], session['client_secret'], session['code']))
c.execute("INSERT INTO `credentials` (client_id, client_secret, secret) VALUES (%s, %s, %s)", (session['client_id'], session['client_secret'], session['secret']))
credentials_id = c.lastrowid
mysql.connection.commit()