@@ -141,26 +141,29 @@ class html: if ctype.startswith('audio/'): return "eu não tenho ouvidos, seu insensível!" - title_pattern = re.compile(r"<[Tt][Ii][Tt][Ll][Ee][^>]*?>(.*?)</[Tt][Ii][Tt][Ll][Ee]>", re.UNICODE|re.MULTILINE|re.DOTALL) - data = '' - while True: - if len(data) > DATA_LIMIT: - break - - d = self.urlObj.read(DATA_CHUNK) - if not d: - break - - data += d - - title_search = title_pattern.search(data) - if title_search is not None: - title = title_search.group(1) - title = title.strip().replace("\n"," ").replace("\r", " ") - title = re.sub("&#?\w+;", "", title) - return "[ %s ]" % (title) - - return None + if 'html' in ctype or 'xml' in ctype: + title_pattern = re.compile(r"<title[^>]*?>(.*?)< */ *title *>", re.UNICODE|re.MULTILINE|re.DOTALL|re.IGNORECASE) + data = '' + while True: + if len(data) > DATA_LIMIT: + break + + d = self.urlObj.read(DATA_CHUNK) + if not d: + break + + data += d + + title_search = title_pattern.search(data) + if title_search is not None: + title = title_search.group(1) + title = title.strip().replace("\n"," ").replace("\r", " ") + title = re.sub("&#?\w+;", "", title) + return "[ %s ]" % (title) + # no title found + return None + + return "%s? o que é isso?" % (ctype) password = sys.argv[1] bot_sql.py 9dfc3423627ee950c30c87e9dfad20ed46628ca5 Eduardo Habkost ehabkost@raisama.net http://github.com/ehabkost/PythonBot/commit/92573b90ab18eaf4821dd2222de48f765cfb0336 92573b90ab18eaf4821dd2222de48f765cfb0336 2009-11-03T10:02:50-08:00 2009-11-03T10:02:50-08:00 look for <title> only on html or xml documents Signed-off-by: Eduardo Habkost <ehabkost@raisama.net> 39ca26db7fe3a37de770303b4fbd1c580768f7e8 Eduardo Habkost ehabkost@raisama.net