X-Git-Url: https://plomlompom.com/repos/process?a=blobdiff_plain;f=plomlombot.py;h=8ac8f01c00cf0cc4e00ad6e4a664a743926ac2f6;hb=73d956fc4322397624652d1a63341f3092988679;hp=b99468857704b1417855544e91df9a2d3e9bad9a;hpb=12695b212e2e4d7150ebb72d3f63d5bcd38d445e;p=plomlombot-irc.git diff --git a/plomlombot.py b/plomlombot.py old mode 100644 new mode 100755 index b994688..8ac8f01 --- a/plomlombot.py +++ b/plomlombot.py @@ -1,3 +1,5 @@ +#!/usr/bin/python3 + import argparse import socket import datetime @@ -5,6 +7,7 @@ import select import time import re import urllib.request +import http.client import html # Defaults, may be overwritten by command line arguments. @@ -14,9 +17,11 @@ TIMEOUT = 240 USERNAME = "plomlombot" NICKNAME = USERNAME + class ExceptionForRestart(Exception): pass + class IO: def __init__(self, server, port, timeout): @@ -42,7 +47,7 @@ class IO: 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) + + str(datetime.datetime.now()) + ": " + msg) msg = msg + "\r\n" msg_len = len(msg) total_sent_len = 0 @@ -62,11 +67,15 @@ class IO: 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: print("SOCKET CONNECTION BROKEN") raise ExceptionForRestart - self.rune_buffer += received_runes + 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] @@ -77,9 +86,10 @@ class IO: line = self._recv_line_wrapped(send_ping) if line: print("LINE FROM SERVER " + str(datetime.datetime.now()) + ": " + - line) + line) return line + def init_session(server, port, timeout, nickname, username, channel): print("CONNECTING TO " + server) io = IO(server, port, timeout) @@ -88,33 +98,42 @@ def init_session(server, port, timeout, nickname, username, channel): io.send_line("JOIN " + channel) return io + def lineparser_loop(io, nickname): def act_on_privmsg(tokens): def url_check(msg): - matches = re.findall("(https?://[^\s]+)", 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] + request = urllib.request.Request(url, headers={ + "User-Agent": "plomlombot" + }) try: - webpage = urllib.request.urlopen(url, timeout=15) - except urllib.error.HTTPError as error: - print("TROUBLE FOLLOWING URL: " + str(error)) + 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: - charset="utf-8" + notice("TROUBLE READING PAGE TITLE: no charset in header") + continue content_type = webpage.info().get_content_type() - if not content_type in ('text/html', 'text/xml', - 'application/xhtml+xml'): - print("TROUBLE INTERPRETING URL: bad content type " - + 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('