X-Git-Url: https://plomlompom.com/repos/?a=blobdiff_plain;ds=inline;f=plomlombot.py;h=9e9d0a44ce21bc1d07204dcb9780c54ac1b31d0e;hb=9260bc740d86b0b83d90c31eea2f898dfb2cf594;hp=747ab077700fa772d1c54cc62b61290c1a7bcad1;hpb=24cf06720f7e0c13e6075f69426529a2119085ba;p=plomlombot-irc.git diff --git a/plomlombot.py b/plomlombot.py old mode 100644 new mode 100755 index 747ab07..9e9d0a4 --- a/plomlombot.py +++ b/plomlombot.py @@ -1,102 +1,217 @@ +#!/usr/bin/python3 + +import argparse import socket -import datetime +import datetime import select import time import re import urllib.request +import http.client import html +import html.parser + +# Defaults, may be overwritten by command line arguments. +SERVER = "irc.freenode.net" +PORT = 6667 +TIMEOUT = 240 +USERNAME = "plomlombot" +NICKNAME = USERNAME + + +class HTMLParser(html.parser.HTMLParser): + def __init__(self, html, tag): + super().__init__() + self._tag_to_check = tag + self._tag = "" + self.data = "" + self.feed(html) + def handle_starttag(self, tag, attrs): + if self.data == "" and tag == self._tag_to_check: + self._tag = tag + def handle_endtag(self, tag): + self._tag = "" + def handle_data(self, data): + if self._tag != "": + self.data = data + + +class ExceptionForRestart(Exception): + pass -servername = "" -timeout = 480 -username = "plomlombot" -nickname = username -channel = "#zrolaps" class IO: - def __init__(self, server, port): + + def __init__(self, server, port, timeout): + self.timeout = timeout self.socket = socket.socket() self.socket.connect((server, port)) self.socket.setblocking(0) self.line_buffer = [] self.rune_buffer = "" self.last_pong = time.time() - def _pingtest(self): - if self.last_pong + timeout < time.time(): - raise RuntimeError("server not answering") - self.send_line("PING " + nickname + " " + servername) - def send_line(self, msg_orig): - msg = msg_orig + "\r\n" + self.servername = self.recv_line(send_ping=False).split(" ")[0][1:] + + def _pingtest(self, send_ping=True): + if self.last_pong + self.timeout < time.time(): + print("SERVER NOT ANSWERING") + raise ExceptionForRestart + if send_ping: + self.send_line("PING " + self.servername) + + def send_line(self, msg): + msg = msg.replace("\r", " ") + msg = msg.replace("\n", " ") + if len(msg.encode("utf-8")) > 510: + print("NOT SENT LINE TO SERVER (too long): " + msg) + print("LINE TO SERVER: " + + str(datetime.datetime.now()) + ": " + msg) + msg = msg + "\r\n" msg_len = len(msg) total_sent_len = 0 while total_sent_len < msg_len: sent_len = self.socket.send(bytes(msg[total_sent_len:], "UTF-8")) if sent_len == 0: - raise RuntimeError("socket connection broken") + print("SOCKET CONNECTION BROKEN") + raise ExceptionForRestart total_sent_len += sent_len - print("LINE TO SERVER: " - + str(datetime.datetime.now()) + ": " + msg_orig) - def recv_line_wrapped(self): + + def _recv_line_wrapped(self, send_ping=True): if len(self.line_buffer) > 0: return self.line_buffer.pop(0) while True: - ready = select.select([self.socket], [], [], int(timeout / 2)) + ready = select.select([self.socket], [], [], int(self.timeout / 2)) if not ready[0]: - self._pingtest() + self._pingtest(send_ping) return None self.last_pong = time.time() - received_runes = self.socket.recv(1024).decode("UTF-8") + received_bytes = self.socket.recv(1024) + try: + received_runes = received_bytes.decode("UTF-8") + except UnicodeDecodeError: + received_runes = received_bytes.decode("latin1") if len(received_runes) == 0: - raise RuntimeError("socket connection broken") - self.rune_buffer += received_runes + print("SOCKET CONNECTION BROKEN") + raise ExceptionForRestart + self.rune_buffer += received_runes lines_split = str.split(self.rune_buffer, "\r\n") self.line_buffer += lines_split[:-1] self.rune_buffer = lines_split[-1] if len(self.line_buffer) > 0: return self.line_buffer.pop(0) - def recv_line(self): - line = self.recv_line_wrapped() + + def recv_line(self, send_ping=True): + line = self._recv_line_wrapped(send_ping) if line: print("LINE FROM SERVER " + str(datetime.datetime.now()) + ": " + - line) + line) return line -io = IO("irc.freenode.net", 6667) -io.send_line("NICK " + nickname) -io.send_line("USER " + username + " 0 * : ") -io.send_line("JOIN " + channel) -servername = io.recv_line().split(" ")[0][1:] -while 1: - line = io.recv_line() - if not line: - continue - tokens = line.split(" ") - if len(tokens) > 1: - if tokens[1] == "PRIVMSG": - sender = "" - for rune in tokens[0]: - if rune == "!": - break - if rune != ":": - sender += rune - receiver = "" - for rune in tokens[2]: - if rune == "!": - break - if rune != ":": - receiver += rune - target = sender - if receiver != nickname: - target = receiver - msg = str.join(" ", tokens[3:])[1:] - matches = re.findall("(https?://[^\s]+)", msg) + +def init_session(server, port, timeout, nickname, username, channel): + print("CONNECTING TO " + server) + io = IO(server, port, timeout) + io.send_line("NICK " + nickname) + io.send_line("USER " + username + " 0 * : ") + io.send_line("JOIN " + channel) + return io + + +def lineparser_loop(io, nickname): + + def act_on_privmsg(tokens): + + def url_check(msg): + + def notice(msg): + io.send_line("NOTICE " + target + " :" + msg) + + matches = re.findall("(https?://[^\s>]+)", msg) for i in range(len(matches)): url = matches[i] - webpage = urllib.request.urlopen(url) + request = urllib.request.Request(url, headers={ + "User-Agent": "plomlombot" + }) + try: + webpage = urllib.request.urlopen(request, timeout=15) + except (urllib.error.HTTPError, urllib.error.URLError, + UnicodeError, http.client.BadStatusLine) as error: + notice("TROUBLE FOLLOWING URL: " + str(error)) + continue charset = webpage.info().get_content_charset() + if not charset: + notice("TROUBLE READING PAGE TITLE: no charset in header") + continue + content_type = webpage.info().get_content_type() + if content_type not in ('text/html', 'text/xml', + 'application/xhtml+xml'): + notice("TROUBLE READING PAGE TITLE: bad content type " + + content_type) + continue content = webpage.read().decode(charset) - title = str(content).split('