simon987/Discord-Channel-scraper
1
0
Fork 0
mirror of https://github.com/simon987/Discord-Channel-scraper.git synced 2025-04-10 14:16:48 +00:00
Code Issues Projects Releases Wiki Activity

PEP 8 compliance

Browse Source
This commit is contained in:
Simon 2018-12-01 10:57:14 -05:00
parent bc4094923a
commit 433aa66a10
1 changed files with 47 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

76
scrape-logs.py
View File

@ -1,14 +1,12 @@
#!/usr/bin/python
import discord
import asyncio
import getpass
import argparse
import re
import logging
logging.basicConfig(
level = "WARNING",
style = "{",
level="WARNING",
style="{",
format="[{asctime}] [{process}] [{levelname}] {filename}:{lineno} {msg}"
)
@ -16,42 +14,62 @@ log = logging.getLogger(__name__)
parser = argparse.ArgumentParser(description='Scrapes messages from a Discord channel.')
parser.add_argument('--username','-u', action='store', help='Username to login under. If not specified, username will be prompted for.')
#parser.add_argument('--password','-p', action='store', help='Password to login under. If not specified, password will be prompted for.')
parser.add_argument('--flag','-f', action='store', default="!yank", help='An alternative to specifing the server and channel, specify a piece of regex which when matched against a message sent by the target user, will trigger scraping of the channel the message was posted in. Useful for private messages and private chats. Default value is "!yank", activates by default if no server is specified.')
parser.add_argument('--quiet','-q', action='store_true', help='Supress messages in Discord')
parser.add_argument('--server','--guild','-s', action='store', help='Discord server name to scrape from (user must be a member of the server and have history privileges). This field is case sensitive. If channel is not specified the entire server will be scraped.')
parser.add_argument('--channel','-c', action='store', help='Discord channel name to scrape from (user must have history privileges for the particular channel). This field is case sensitive.')
parser.add_argument('--limit','-l', action='store', default=1000000, type=int, help='Number of messages to save. Default is 1000000')
parser.add_argument('--output','-o', action='store', help="Outputs all messages into a single file. If not specified, messages are saved under the format: <channel name>.txt.")
parser.add_argument('--logging', action='store', choices=[10,20,30,40,50], default=20, help='Change the logging level. Defaults to 20, info.')
parser.add_argument('--username', '-u', action='store', help='Username to login under. If not specified, '
'username will be prompted for.')
# parser.add_argument('--password','-p', action='store', help='Password to login under. If not specified,
# password will be prompted for.')
parser.add_argument('--flag', '-f', action='store', default="!yank", help='An alternative to specifying the server and'
' channel, specify a piece of regex which'
' when matched against a message sent by the'
' target user, will trigger scraping of the'
' channel the message was posted in. Useful'
' for private messages and private chats.'
' Default value is "!yank", activates by'
' default if no server is specified.')
parser.add_argument('--quiet', '-q', action='store_true', help='Suppress messages in Discord')
parser.add_argument('--server', '--guild','-s', action='store', help='Discord server name to scrape from '
'(user must be a member of the server and'
' have history privileges). This field is case'
' sensitive. If channel is not specified the '
'entire server will be scraped.')
parser.add_argument('--channel', '-c', action='store', help='Discord channel name to scrape from '
'(user must have history privileges for the particular'
' channel). This field is case sensitive.')
parser.add_argument('--limit', '-l', action='store', default=1000000, type=int, help='Number of messages to save.'
' Default is 1000000')
parser.add_argument('--output', '-o', action='store', help="Outputs all messages into a single file."
" If not specified, messages are saved under the format:"
" <channel name>.txt.")
parser.add_argument('--logging', action='store', choices=[10, 20, 30, 40, 50], default=20, help='Change the logging '
'level. Defaults to 20, info.')
args = parser.parse_args()
log.setLevel(args.logging)
#prompt for username
if (not args.username):
# prompt for username
if not args.username:
args.username = input("Username: ")
password = getpass.getpass("Password for user {0}: ".format(args.username))
client = discord.Client()
async def getLogs(channel):
async def get_logs(channel):
try:
if not args.quiet:
await client.send_message(channel,"Getting the logs for channel {0}".format(channel.name))
await client.send_message(channel, "Getting the logs for channel {0}".format(channel.name))
log.info("Getting the logs for channel {0}".format(channel.name))
with open("{0}.txt".format(channel.name),'w') as f:
with open("{0}.txt".format(channel.name), 'w') as f:
async for line in client.logs_from(channel, limit=args.limit):
for i in line.attachments:
try:
f.write('{0}::file:{1}\n'.format(line.author.name,i['url']))
f.write('{0}::file:{1}\n'.format(line.author.name, i['url']))
except:
continue
try:
f.write('{0}: {1}\n'.format(line.author.name,line.content)) #line is of the message type
f.write('{0}: {1}\n'.format(line.author.name, line.content)) # line is of the message type
except:
continue
if not args.quiet:
@ -60,20 +78,21 @@ async def getLogs(channel):
except Exception as e:
if not args.quiet:
await client.send_message(channel,'Failed saving logs: {}'.format(e.message))
log.error("Error while downloading channel {0}: {1}".format(channel.name,e.message))
log.error("Error while downloading channel {0}: {1}".format(channel.name, e.message))
#Strangely, this will work once we are logged in
# Strangely, this will work once we are logged in
@client.async_event
async def on_message(message):
try:
log.debug(str(message.channel.server.name) + " -> " + str(message.channel.name) + ' - ' + str(message.author) + ': ' + str(message.content))
log.debug(str(message.channel.server.name) + " -> " + str(message.channel.name) + ' - ' + str(message.author) +
': ' + str(message.content))
except:
log.debug("Private message - " + str(message.author) + ': ' + str(message.content))
log.debug("Private message - " + str(message.author) + ': ' + str(message.content))
if not args.server and not args.channel:
if args.flag == message.content[:len(args.flag)]:
await getLogs(message.channel)
await get_logs(message.channel)
# if (not args.server) and message.author.id == client.user.id and re.compile(args.flag).match(message.content):
# print("Matched {}".format(args.flag))
# await getLogs(message.channel)
@ -84,20 +103,19 @@ async def on_ready():
log.info("Logged in as user {0}".format(client.user.name))
if args.server and args.channel:
channel = ""
try:
channel = discord.utils.get(client.get_all_channels(), server__name=args.server, name=args.channel)
except:
channel = ""
if channel:
await getLogs(channel)
await get_logs(channel)
else:
log.error("Could not find channel {0} in server {1}".format(args.channel, args.server))
await client.logout()
elif args.server:
log.info("Downloading messages for all channels in server {0}".format(arg.server))
log.info("Downloading messages for all channels in server {0}".format(args.server))
for channel in server.channels:
await getLogs(channel)
await get_logs(channel)
await client.logout()
else:
log.info('Entering flag mode with flag "{0}"'.format(args.flag))