home · contact · privacy
Add toggle for markovfile logging.
[plomlombot-irc.git] / plomlombot.py
index 741b01c65beb60b8c3a0ab0221bf781eae5989fa..9bf6ce0ee0319ead6b96d68ffb4b49527e833620 100755 (executable)
@@ -55,9 +55,59 @@ class Line:
                     self.receiver += rune
 
 
+class Log:
+
+    def __init__(self, chandir, nickname, username, channel, rmlogs):
+        self.nickname = nickname
+        self.username = username
+        self.channel = channel
+        self.chandir = chandir
+        self.rmlogcycle = rmlogs
+        self.rawlogdir = chandir + "raw_logs/"
+        self.logdir = chandir + "logs/"
+        if not os.path.exists(self.logdir):
+            os.makedirs(self.logdir)
+        if not os.path.exists(self.rawlogdir):
+            os.makedirs(self.rawlogdir)
+
+    def log(self, line, sent=False):
+        identity = ""
+        separator = " > "
+        if sent:
+            separator = " "
+            line = Line("< " + line)
+            line.sender = self.nickname
+            identity = self.username + "@localhost"
+        else:
+            if type(line) == str:
+                line = Line(line)
+        now = datetime.datetime.utcnow()
+        form = "%Y-%m-%d %H:%M:%S UTC"
+        write_to_file(self.rawlogdir + now.strftime("%Y-%m-%d") + ".txt", "a",
+                      now.strftime(form) + separator + line.line + "\n")
+        to_log = irclog.format_logline(line, self.channel, identity)
+        if to_log != None:
+            write_to_file(self.logdir + now.strftime("%Y-%m-%d") + ".txt", "a",
+                          now.strftime(form) + " " + to_log + "\n")
+
+    def rmlogs(self):
+        if self.rmlogcycle > 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.rmlogcycle:
+                    os.remove(f)
+
+    def separator_line(self):
+        now = datetime.datetime.utcnow()
+        write_to_file(self.logdir + now.strftime("%Y-%m-%d") + ".txt", "a",
+                      "-----------------------\n")
+
+
 class IO:
 
     def __init__(self, server, port, timeout):
+        self.log = None
         self.timeout = timeout
         self.socket = socket.socket()
         try:
@@ -68,7 +118,10 @@ class IO:
         self.line_buffer = []
         self.rune_buffer = ""
         self.last_pong = time.time()
-        self.servername = self.recv_line(send_ping=False).split(" ")[0][1:]
+        line = self.recv_line(send_ping=False)
+        if not line or len(line) < 1:
+            raise ExceptionForRestart
+        self.servername = line.split(" ")[0][1:]
 
     def _pingtest(self, send_ping=True):
         if self.last_pong + self.timeout < time.time():
@@ -84,6 +137,8 @@ class IO:
             print("NOT SENT LINE TO SERVER (too long): " + msg)
         print("LINE TO SERVER: "
               + str(datetime.datetime.now()) + ": " + msg)
+        if self.log != None:
+            self.log.log(msg, True)
         msg = msg + "\r\n"
         msg_len = len(msg)
         total_sent_len = 0
@@ -121,6 +176,8 @@ class IO:
     def recv_line(self, send_ping=True):
         line = self._recv_line_wrapped(send_ping)
         if line:
+            if self.log != None:
+                self.log.log(line)
             print("LINE FROM SERVER " + str(datetime.datetime.now()) + ": " +
                   line)
         return line
@@ -136,12 +193,12 @@ def handle_command(command, argument, notice, target, session):
         quotesfile = open(session.quotesfile, "r")
         lines = quotesfile.readlines()
         quotesfile.close()
-        notice("ADDED QUOTE #" + str(len(lines) - 1))
+        notice("added quote #" + str(len(lines) - 1))
 
     def quote():
 
         def help():
-            notice("SYNTAX: !quote [int] OR !quote search QUERY")
+            notice("syntax: !quote [int] OR !quote search QUERY")
             notice("QUERY may be a boolean grouping of quoted or unquoted " +
                    "search terms, examples:")
             notice("!quote search foo")
@@ -159,7 +216,7 @@ def handle_command(command, argument, notice, target, session):
             help()
             return
         if not os.access(session.quotesfile, os.F_OK):
-            notice("NO QUOTES AVAILABLE")
+            notice("no quotes available")
             return
         quotesfile = open(session.quotesfile, "r")
         lines = quotesfile.readlines()
@@ -168,7 +225,7 @@ def handle_command(command, argument, notice, target, session):
         if len(tokens) == 1:
             i = int(tokens[0])
             if i == 0 or i > len(lines):
-                notice("THERE'S NO QUOTE OF THAT INDEX")
+                notice("there's no quote of that index")
                 return
             i = i - 1
         elif len(tokens) > 1:
@@ -176,25 +233,25 @@ def handle_command(command, argument, notice, target, session):
             try:
                 results = plomsearch.search(query, lines)
             except plomsearch.LogicParserError as err:
-                notice("FAILED QUERY PARSING: " + str(err))
+                notice("failed query parsing: " + str(err))
                 return
             if len(results) == 0:
-                notice("NO QUOTES MATCHING QUERY")
+                notice("no quotes matching query")
             else:
                 if len(results) > 3:
-                    notice("SHOWING 3 OF " + str(len(results)) + " QUOTES")
+                    notice("showing 3 of " + str(len(results)) + " quotes")
                 for result in results[:3]:
-                    notice("QUOTE #" + str(result[0] + 1) + ": "
+                    notice("quote #" + str(result[0] + 1) + ": "
                            + result[1][:-1])
             return
         else:
             i = random.randrange(len(lines))
-        notice("QUOTE #" + str(i + 1) + ": " + lines[i][:-1])
+        notice("quote #" + str(i + 1) + ": " + lines[i][:-1])
 
     def markov():
-        from random import choice, shuffle
-        select_length = 2
-        selections = []
+
+        def help():
+            notice("syntax: !markov [integer from 1 to infinite]")
 
         def markov(snippet):
             usable_selections = []
@@ -215,8 +272,26 @@ def handle_command(command, argument, notice, target, session):
             selection = choice(usable_selections)
             return selection[select_length]
 
+        if "" == argument:
+            tokens = []
+        else:
+            tokens = argument.split(" ")
+        if (len(tokens) > 1 or (len(tokens) == 1 and not tokens[0].isdigit())):
+            help()
+            return
+
+        from random import choice, shuffle
+        select_length = 2
+        if len(tokens) == 1:
+            n = int(tokens[0])
+            if n > 0:
+                select_length = n
+            else:
+                notice("bad value, using default: " + str(select_length))
+        selections = []
+
         if not os.access(session.markovfile, os.F_OK):
-            notice("NOT ENOUGH TEXT TO MARKOV.")
+            notice("not enough text to markov for selection length")
             return
 
         # Lowercase incoming lines, ensure they end in a sentence end mark.
@@ -230,8 +305,8 @@ def handle_command(command, argument, notice, target, session):
             if line[-1] not in sentence_end_markers:
                 line += "."
             tokens += line.split()
-        if len(tokens) <= select_length:
-            notice("NOT ENOUGH TEXT TO MARKOV.")
+        if len(tokens) - 1 <= select_length:
+            notice("not enough text to markov")
             return
 
         # Replace URLs with escape string for now, so that the Markov selector
@@ -266,8 +341,8 @@ def handle_command(command, argument, notice, target, session):
         shuffle(selections)
         for i in range(len(selections)):
             if selections[i][0][-1] in sentence_end_markers:
-                for i in range(select_length):
-                    snippet[i] = selections[i][i + 1]
+                for j in range(select_length):
+                    snippet[j] = selections[j][j + 1]
                 break
         msg = ""
         malkovich = "malkovich"
@@ -299,7 +374,7 @@ def handle_command(command, argument, notice, target, session):
             try:
                 twtfile = open(session.twtfile, mode)
             except (PermissionError, FileNotFoundError) as err:
-                notice("CAN'T ACCESS OR CREATE TWT FILE: " + str(err))
+                notice("can't access or create twt file: " + str(err))
                 return None
             return twtfile
 
@@ -314,7 +389,7 @@ def handle_command(command, argument, notice, target, session):
             return
         twtfile.write(datetime.utcnow().isoformat() + "\t" + argument + "\n")
         twtfile.close()
-        notice("WROTE TWT.")
+        notice("wrote twt.")
 
     if "addquote" == command:
         addquote()
@@ -348,7 +423,7 @@ def handle_url(url, notice, show_url=False):
     signal.signal(signal.SIGALRM, timeout_handler)
     signal.alarm(15)
     try:
-        r = requests.get(url, stream=True)
+        r = requests.get(url, headers = {'User-Agent': 'plomlombot'}, stream=True)
         r.raw.decode_content = True
         text = r.raw.read(10000000+1)
         if len(text) > 10000000:
@@ -361,68 +436,49 @@ def handle_url(url, notice, show_url=False):
             ValueError,
             requests.exceptions.InvalidSchema) as error:
         signal.alarm(0)
-        notice("TROUBLE FOLLOWING URL: " + str(error))
+        notice("trouble following url: " + str(error))
         return False
     signal.alarm(0)
     if mobile_twitter_hack(url):
         return True
     title = bs4.BeautifulSoup(text, "html5lib").title
     if title and title.string:
-        prefix = "PAGE TITLE: "
+        prefix = "page title: "
         if show_url:
-            prefix = "PAGE TITLE FOR <" + url + ">: "
+            prefix = "page title for <" + url + ">: "
         notice(prefix + title.string.strip())
     else:
-        notice("PAGE HAS NO TITLE TAG")
+        notice("page has no title tag")
     return True
 
 
 class Session:
 
-    def __init__(self, io, username, nickname, channel, twtfile, dbdir, rmlogs):
+    def __init__(self, io, username, nickname, channel, twtfile, dbdir, rmlogs,
+                 markov_input):
         self.io = io
         self.nickname = nickname
-        self.username = username
-        self.channel = channel
         self.users_in_chan = []
         self.twtfile = twtfile
-        self.dbdir = dbdir
-        self.rmlogs = rmlogs
+        hash_channel = hashlib.md5(channel.encode("utf-8")).hexdigest()
+        chandir = dbdir + "/" + hash_channel + "/"
+        self.markov_input = markov_input
+        self.markovfile = chandir + "markovfeed"
+        self.quotesfile = chandir + "quotes"
+        self.log = Log(chandir, self.nickname, username, channel, rmlogs)
         self.io.send_line("NICK " + self.nickname)
-        self.io.send_line("USER " + self.username + " 0 * : ")
-        self.io.send_line("JOIN " + self.channel)
-        hash_channel = hashlib.md5(self.channel.encode("utf-8")).hexdigest()
-        self.chandir = self.dbdir + "/" + hash_channel + "/"
-        self.rawlogdir = self.chandir + "raw_logs/"
-        self.logdir = self.chandir + "logs/"
-        if not os.path.exists(self.logdir):
-            os.makedirs(self.logdir)
-        if not os.path.exists(self.rawlogdir):
-            os.makedirs(self.rawlogdir)
-        self.markovfile = self.chandir + "markovfeed"
-        self.quotesfile = self.chandir + "quotes"
+        self.io.send_line("USER " + username + " 0 * : ")
+        self.io.send_line("JOIN " + channel)
+        self.io.log = self.log
+        self.log.separator_line()
 
     def loop(self):
 
-        def log(line):
-            if type(line) == str:
-                line = Line(":" + self.nickname + "!~" + self.username +
-                            "@localhost" + " " + line)
-            now = datetime.datetime.utcnow()
-            form = "%Y-%m-%d %H:%M:%S UTC\t"
-            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:
-                write_to_file(self.logdir + now.strftime("%Y-%m-%d") + ".txt",
-                              "a", now.strftime(form) + " " + to_log + "\n")
-
         def handle_privmsg(line):
 
             def notice(msg):
                 line = "NOTICE " + target + " :" + msg
                 self.io.send_line(line)
-                log(line)
 
             target = line.sender
             if line.receiver != self.nickname:
@@ -434,31 +490,23 @@ class Session:
                 if handle_url(matches[i], notice):
                     url_count += 1
                     if url_count == 3:
-                        notice("MAXIMUM NUMBER OF URLs TO PARSE PER MESSAGE "
-                               "REACHED")
+                        notice("maximum number of urls to parse per message "
+                               "reached")
                         break
-            if "!" == msg[0]:
+            if "!" == msg[0] and len(msg) > 1:
                 tokens = msg[1:].split()
                 argument = str.join(" ", tokens[1:])
                 handle_command(tokens[0], argument, notice, target, self)
                 return
-            write_to_file(self.markovfile, "a", msg + "\n")
+            if self.markov_input:
+                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:
-            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)
+            self.log.rmlogs()
             line = self.io.recv_line()
             if not line:
                 continue
             line = Line(line)
-            log(line)
             if len(line.tokens) > 1:
                 if line.tokens[0] == "PING":
                     self.io.send_line("PONG " + line.tokens[1])
@@ -507,6 +555,9 @@ def parse_command_line_arguments():
                         type=int, default=0,
                         help="maximum age in seconds for logfiles in logs/ "
                         "(0 means: never delete, and is default)")
+    parser.add_argument("-m, --markov_store", action="store_true",
+                        dest="markov_store",
+                        help="log channel discussions for !markov input")
     parser.add_argument("CHANNEL", action="store", help="channel to join")
     opts, unknown = parser.parse_known_args()
     return opts
@@ -517,9 +568,9 @@ while True:
     try:
         io = IO(opts.server, opts.port, opts.timeout)
         hash_server = hashlib.md5(opts.server.encode("utf-8")).hexdigest()
-        dbdir = opts.dbdir + "/" + hash_server 
+        dbdir = opts.dbdir + "/" + hash_server
         session = Session(io, opts.username, opts.nickname, opts.CHANNEL,
-            opts.twtfile, dbdir, opts.rmlogs)
+                          opts.twtfile, dbdir, opts.rmlogs, opts.markov_input)
         session.loop()
     except ExceptionForRestart:
         io.socket.close()