mirror of
https://github.com/Lynnesbian/FediBooks/
synced 2024-11-25 16:48:58 +00:00
Compare commits
3 commits
ea8e46c07c
...
a9fcda794c
Author | SHA1 | Date | |
---|---|---|---|
a9fcda794c | |||
af1f8c6d8b | |||
be9685a881 |
2 changed files with 14 additions and 4 deletions
14
scrape.py
14
scrape.py
|
@ -9,6 +9,14 @@ import functions
|
||||||
cfg = json.load(open('config.json'))
|
cfg = json.load(open('config.json'))
|
||||||
|
|
||||||
def scrape_posts(account):
|
def scrape_posts(account):
|
||||||
|
db = MySQLdb.connect(
|
||||||
|
host = cfg['db_host'],
|
||||||
|
user=cfg['db_user'],
|
||||||
|
passwd=cfg['db_pass'],
|
||||||
|
db=cfg['db_name'],
|
||||||
|
use_unicode=True,
|
||||||
|
charset="utf8mb4"
|
||||||
|
)
|
||||||
handle = account[0]
|
handle = account[0]
|
||||||
outbox = account[1]
|
outbox = account[1]
|
||||||
print("Scraping {}".format(handle))
|
print("Scraping {}".format(handle))
|
||||||
|
@ -78,12 +86,14 @@ def scrape_posts(account):
|
||||||
|
|
||||||
if r.status_code == 429:
|
if r.status_code == 429:
|
||||||
# we are now being ratelimited, move on to the next user
|
# we are now being ratelimited, move on to the next user
|
||||||
|
print("Hit rate limit while scraping {}".format(handle))
|
||||||
done = True
|
done = True
|
||||||
else:
|
else:
|
||||||
j = r.json()
|
j = r.json()
|
||||||
|
|
||||||
db.commit()
|
db.commit()
|
||||||
|
|
||||||
|
db.commit()
|
||||||
print("Finished scraping {}".format(handle))
|
print("Finished scraping {}".format(handle))
|
||||||
|
|
||||||
print("Establishing DB connection")
|
print("Establishing DB connection")
|
||||||
|
@ -101,9 +111,9 @@ cursor = db.cursor()
|
||||||
print("Downloading posts")
|
print("Downloading posts")
|
||||||
cursor.execute("SELECT `handle`, `outbox` FROM `fedi_accounts` ORDER BY RAND()")
|
cursor.execute("SELECT `handle`, `outbox` FROM `fedi_accounts` ORDER BY RAND()")
|
||||||
accounts = cursor.fetchall()
|
accounts = cursor.fetchall()
|
||||||
|
cursor.close()
|
||||||
|
db.close()
|
||||||
with Pool(cfg['service_threads']) as p:
|
with Pool(cfg['service_threads']) as p:
|
||||||
p.map(scrape_posts, accounts)
|
p.map(scrape_posts, accounts)
|
||||||
|
|
||||||
db.commit()
|
|
||||||
|
|
||||||
print("Done!")
|
print("Done!")
|
||||||
|
|
4
webui.py
4
webui.py
|
@ -287,7 +287,7 @@ def bot_accounts_add():
|
||||||
|
|
||||||
if r.status_code == 200:
|
if r.status_code == 200:
|
||||||
j = r.json()
|
j = r.json()
|
||||||
if 'is_pro' in j['contact_account']:
|
if 'contact_account' in j and 'is_pro' in j['contact_account']:
|
||||||
# gab instance
|
# gab instance
|
||||||
error = "Gab instances are not supported."
|
error = "Gab instances are not supported."
|
||||||
return render_template("bot_accounts_add.html", error = error)
|
return render_template("bot_accounts_add.html", error = error)
|
||||||
|
@ -395,7 +395,7 @@ def bot_create():
|
||||||
session['instance_type'] = "Pleroma"
|
session['instance_type'] = "Pleroma"
|
||||||
session['step'] += 1
|
session['step'] += 1
|
||||||
else:
|
else:
|
||||||
if 'is_pro' in j['contact_account']:
|
if 'contact_account' in j and 'is_pro' in j['contact_account']:
|
||||||
# gab instance
|
# gab instance
|
||||||
session['error'] = "Gab instances are not supported."
|
session['error'] = "Gab instances are not supported."
|
||||||
else:
|
else:
|
||||||
|
|
Loading…
Reference in a new issue