home
·
contact
·
privacy
projects
/
plomlombot-irc.git
/ blobdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
|
commitdiff
|
tree
raw
|
inline
| side by side
Handle URLs for too large websites / files.
[plomlombot-irc.git]
/
plomlombot.py
diff --git
a/plomlombot.py
b/plomlombot.py
index 61da87140fca80d902042d933b0b235fe25dcd0c..aa7061cac656f3f6b5f768e40787ccb6fbc475c4 100755
(executable)
--- a/
plomlombot.py
+++ b/
plomlombot.py
@@
-24,6
+24,12
@@
TWTFILE = ""
DBDIR = os.path.expanduser("~/plomlombot_db")
DBDIR = os.path.expanduser("~/plomlombot_db")
+def write_to_file(path, mode, text):
+ f = open(path, mode)
+ f.write(text)
+ f.close()
+
+
class ExceptionForRestart(Exception):
pass
class ExceptionForRestart(Exception):
pass
@@
-123,12
+129,9
@@
def handle_command(command, argument, notice, target, session):
def addquote():
if not os.access(session.quotesfile, os.F_OK):
def addquote():
if not os.access(session.quotesfile, os.F_OK):
- quotesfile = open(session.quotesfile, "w")
- quotesfile.write("QUOTES FOR " + target + ":\n")
- quotesfile.close()
- quotesfile = open(session.quotesfile, "a")
- quotesfile.write(argument + "\n")
- quotesfile.close()
+ write_to_file(session.quotesfile, "w",
+ "QUOTES FOR " + target + ":\n")
+ write_to_file(session.quotesfile, "a", argument + "\n")
quotesfile = open(session.quotesfile, "r")
lines = quotesfile.readlines()
quotesfile.close()
quotesfile = open(session.quotesfile, "r")
lines = quotesfile.readlines()
quotesfile.close()
@@
-177,9
+180,11
@@
def handle_command(command, argument, notice, target, session):
if len(results) == 0:
notice("NO QUOTES MATCHING QUERY")
else:
if len(results) == 0:
notice("NO QUOTES MATCHING QUERY")
else:
- for result in results:
- notice("QUOTE #" + str(result[0] + 1) + " : "
- + result[1][-1])
+ if len(results) > 3:
+ notice("SHOWING 3 OF " + str(len(results)) + " QUOTES")
+ for result in results[:3]:
+ notice("QUOTE #" + str(result[0] + 1) + ": "
+ + result[1][:-1])
return
else:
i = random.randrange(len(lines))
return
else:
i = random.randrange(len(lines))
@@
-334,17
+339,22
@@
def handle_url(url, notice, show_url=False):
return True
try:
return True
try:
- r = requests.get(url, timeout=15)
+ r = requests.get(url, timeout=5, stream=True)
+ r.raw.decode_content = True
+ text = r.raw.read(10000000+1)
+ if len(text) > 10000000:
+ raise ValueError('Too large a response')
except (requests.exceptions.TooManyRedirects,
requests.exceptions.ConnectionError,
requests.exceptions.InvalidURL,
UnicodeError,
except (requests.exceptions.TooManyRedirects,
requests.exceptions.ConnectionError,
requests.exceptions.InvalidURL,
UnicodeError,
+ ValueError,
requests.exceptions.InvalidSchema) as error:
notice("TROUBLE FOLLOWING URL: " + str(error))
return
if mobile_twitter_hack(url):
return
requests.exceptions.InvalidSchema) as error:
notice("TROUBLE FOLLOWING URL: " + str(error))
return
if mobile_twitter_hack(url):
return
- title = bs4.BeautifulSoup(
r.
text, "html5lib").title
+ title = bs4.BeautifulSoup(text, "html5lib").title
if title and title.string:
prefix = "PAGE TITLE: "
if show_url:
if title and title.string:
prefix = "PAGE TITLE: "
if show_url:
@@
-356,7
+366,7
@@
def handle_url(url, notice, show_url=False):
class Session:
class Session:
- def __init__(self, io, username, nickname, channel, twtfile, dbdir):
+ def __init__(self, io, username, nickname, channel, twtfile, dbdir
, rmlogs
):
self.io = io
self.nickname = nickname
self.username = username
self.io = io
self.nickname = nickname
self.username = username
@@
-364,6
+374,7
@@
class Session:
self.users_in_chan = []
self.twtfile = twtfile
self.dbdir = dbdir
self.users_in_chan = []
self.twtfile = twtfile
self.dbdir = dbdir
+ self.rmlogs = rmlogs
self.io.send_line("NICK " + self.nickname)
self.io.send_line("USER " + self.username + " 0 * : ")
self.io.send_line("JOIN " + self.channel)
self.io.send_line("NICK " + self.nickname)
self.io.send_line("USER " + self.username + " 0 * : ")
self.io.send_line("JOIN " + self.channel)
@@
-385,15
+396,13
@@
class Session:
line = Line(":" + self.nickname + "!~" + self.username +
"@localhost" + " " + line)
now = datetime.datetime.utcnow()
line = Line(":" + self.nickname + "!~" + self.username +
"@localhost" + " " + line)
now = datetime.datetime.utcnow()
- logfile = open(self.rawlogdir + now.strftime("%Y-%m-%d") + ".txt", "a")
form = "%Y-%m-%d %H:%M:%S UTC\t"
form = "%Y-%m-%d %H:%M:%S UTC\t"
- logfile.write(now.strftime(form) + " " + line.line + "\n")
-
logfile.close(
)
+ write_to_file(self.rawlogdir + now.strftime("%Y-%m-%d") + ".txt",
+
"a", now.strftime(form) + " " + line.line + "\n"
)
to_log = irclog.format_logline(line, self.channel)
if to_log != None:
to_log = irclog.format_logline(line, self.channel)
if to_log != None:
- logfile = open(self.logdir + now.strftime("%Y-%m-%d") + ".txt", "a")
- logfile.write(now.strftime(form) + " " + to_log + "\n")
- logfile.close()
+ write_to_file(self.logdir + now.strftime("%Y-%m-%d") + ".txt",
+ "a", now.strftime(form) + " " + to_log + "\n")
def handle_privmsg(line):
def handle_privmsg(line):
@@
-414,11
+423,18
@@
class Session:
argument = str.join(" ", tokens[1:])
handle_command(tokens[0], argument, notice, target, self)
return
argument = str.join(" ", tokens[1:])
handle_command(tokens[0], argument, notice, target, self)
return
- file = open(self.markovfile, "a")
- file.write(msg + "\n")
- file.close()
+ write_to_file(self.markovfile, "a", msg + "\n")
+ now = datetime.datetime.utcnow()
+ write_to_file(self.logdir + now.strftime("%Y-%m-%d") + ".txt", "a",
+ "-----------------------\n")
while True:
while True:
+ if self.rmlogs > 0:
+ for f in os.listdir(self.logdir):
+ f = os.path.join(self.logdir, f)
+ if os.path.isfile(f) and \
+ os.stat(f).st_mtime < time.time() - self.rmlogs:
+ os.remove(f)
line = self.io.recv_line()
if not line:
continue
line = self.io.recv_line()
if not line:
continue
@@
-455,7
+471,7
@@
def parse_command_line_arguments():
+ str(PORT) + ")")
parser.add_argument("-w, --wait", action="store", dest="timeout",
type=int, default=TIMEOUT,
+ str(PORT) + ")")
parser.add_argument("-w, --wait", action="store", dest="timeout",
type=int, default=TIMEOUT,
- help="timeout in seconds after which to attempt "
+
+ help="timeout in seconds after which to attempt "
"reconnect (default: " + str(TIMEOUT) + ")")
parser.add_argument("-u, --username", action="store", dest="username",
default=USERNAME, help="username to use (default: "
"reconnect (default: " + str(TIMEOUT) + ")")
parser.add_argument("-u, --username", action="store", dest="username",
default=USERNAME, help="username to use (default: "
@@
-468,6
+484,10
@@
def parse_command_line_arguments():
+ TWTFILE + ")")
parser.add_argument("-d, --dbdir", action="store", dest="dbdir",
default=DBDIR, help="directory to store DB files in")
+ TWTFILE + ")")
parser.add_argument("-d, --dbdir", action="store", dest="dbdir",
default=DBDIR, help="directory to store DB files in")
+ parser.add_argument("-r, --rmlogs", action="store", dest="rmlogs",
+ type=int, default=0,
+ help="maximum age in seconds for logfiles in logs/ "
+ "(0 means: never delete, and is default)")
parser.add_argument("CHANNEL", action="store", help="channel to join")
opts, unknown = parser.parse_known_args()
return opts
parser.add_argument("CHANNEL", action="store", help="channel to join")
opts, unknown = parser.parse_known_args()
return opts
@@
-480,7
+500,7
@@
while True:
hash_server = hashlib.md5(opts.server.encode("utf-8")).hexdigest()
dbdir = opts.dbdir + "/" + hash_server
session = Session(io, opts.username, opts.nickname, opts.CHANNEL,
hash_server = hashlib.md5(opts.server.encode("utf-8")).hexdigest()
dbdir = opts.dbdir + "/" + hash_server
session = Session(io, opts.username, opts.nickname, opts.CHANNEL,
- opts.twtfile, dbdir)
+ opts.twtfile, dbdir
, opts.rmlogs
)
session.loop()
except ExceptionForRestart:
io.socket.close()
session.loop()
except ExceptionForRestart:
io.socket.close()