home · contact · privacy
Use html5lib for BeautifulSoup parsing.
[plomlombot-irc.git] / plomlombot.py
index 11a8206ab4691eaf500cedbb0873efeb532da561..6f6d33994255504e2ac4903cbe254d623db5fbdc 100755 (executable)
@@ -19,6 +19,7 @@ PORT = 6667
 TIMEOUT = 240
 USERNAME = "plomlombot"
 NICKNAME = USERNAME
+TWTFILE = ""
 
 
 class ExceptionForRestart(Exception):
@@ -93,7 +94,7 @@ class IO:
         return line
 
 
-def handle_command(command, argument, notice, target):
+def handle_command(command, argument, notice, target, session):
     hash_string = hashlib.md5(target.encode("utf-8")).hexdigest()
     quotesfile_name = "quotes_" + hash_string
 
@@ -161,7 +162,7 @@ def handle_command(command, argument, notice, target):
         notice("QUOTE #" + str(i + 1) + ": " + lines[i])
 
     def markov():
-        from random import shuffle
+        from random import choice, shuffle
         select_length = 2
         selections = []
 
@@ -171,7 +172,8 @@ def handle_command(command, argument, notice, target):
                 for selection in selections:
                     add = True
                     for j in range(i):
-                        if snippet[j] != selection[j]:
+                        j += 1
+                        if snippet[-j] != selection[-(j+1)]:
                             add = False
                             break
                     if add:
@@ -180,24 +182,51 @@ def handle_command(command, argument, notice, target):
                     break
             if [] == usable_selections:
                 usable_selections = selections
-            shuffle(usable_selections)
-            return usable_selections[0][select_length]
+            selection = choice(usable_selections)
+            return selection[select_length]
 
         hash_string = hashlib.md5(target.encode("utf-8")).hexdigest()
         markovfeed_name = "markovfeed_" + hash_string
         if not os.access(markovfeed_name, os.F_OK):
             notice("NOT ENOUGH TEXT TO MARKOV.")
             return
+
+        # Lowercase incoming lines, ensure they end in a sentence end mark.
         file = open(markovfeed_name, "r")
         lines = file.readlines()
         file.close()
         tokens = []
+        sentence_end_markers = ".!?)("
         for line in lines:
-            line = line.replace("\n", "")
+            line = line.lower().replace("\n", "")
+            if line[-1] not in sentence_end_markers:
+                line += "."
             tokens += line.split()
         if len(tokens) <= select_length:
             notice("NOT ENOUGH TEXT TO MARKOV.")
             return
+
+        # Replace URLs with escape string for now, so that the Markov selector
+        # won't see them as different strings. Stash replaced URLs in urls.
+        urls = []
+        url_escape = "\nURL"
+        url_starts = ["http://", "https://", "<http://", "<https://"]
+        for i in range(len(tokens)):
+            for url_start in url_starts:
+                if tokens[i][:len(url_start)] == url_start:
+                    length = len(tokens[i])
+                    if url_start[0] == "<":
+                        try:
+                            length = tokens[i].index(">") + 1
+                        except ValueError:
+                            pass
+                    urls += [tokens[i][:length]]
+                    tokens[i] = url_escape + tokens[i][length:]
+                    break
+
+        # For each snippet of select_length, use markov() to find continuation
+        # token from selections. Replace present users' names with malkovich.
+        # Start snippets with the beginning of a sentence, if possible.
         for i in range(len(tokens) - select_length):
             token_list = []
             for j in range(select_length + 1):
@@ -206,16 +235,58 @@ def handle_command(command, argument, notice, target):
         snippet = []
         for i in range(select_length):
             snippet += [""]
+        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]
+                break
         msg = ""
+        malkovich = "malkovich"
         while 1:
             new_end = markov(snippet)
+            for name in session.users_in_chan:
+                if new_end[:len(name)] == name.lower():
+                    new_end = malkovich + new_end[len(name):]
+                    break
             if len(msg) + len(new_end) > 200:
                 break
             msg += new_end + " "
             for i in range(select_length - 1):
                 snippet[i] = snippet[i + 1]
             snippet[select_length - 1] = new_end
-        notice(msg.lower() + "malkovich.")
+
+        # Replace occurences of url escape string with random choice from urls.
+        while True:
+            index = msg.find(url_escape)
+            if index < 0:
+                break
+            msg = msg.replace(url_escape, choice(urls), 1)
+
+        # More meaningful ways to randomly end sentences.
+        notice(msg + malkovich + ".")
+
+    def twt():
+        def try_open(mode):
+            try:
+                twtfile = open(session.twtfile, mode)
+            except (PermissionError, FileNotFoundError) as err:
+                notice("CAN'T ACCESS OR CREATE TWT FILE: " + str(err))
+                return None
+            return twtfile
+
+        from datetime import datetime
+        if not os.access(session.twtfile, os.F_OK):
+            twtfile = try_open("w")
+            if None == twtfile:
+                return
+            twtfile.close()
+        twtfile = try_open("a")
+        if None == twtfile:
+            return
+        twtfile.write(datetime.utcnow().isoformat() + "\t" + argument + "\n")
+        twtfile.close()
+        notice("WROTE TWT.")
 
     if "addquote" == command:
         addquote()
@@ -223,6 +294,8 @@ def handle_command(command, argument, notice, target):
         quote()
     elif "markov" == command:
         markov()
+    elif "twt" == command:
+        twt()
 
 
 def handle_url(url, notice, show_url=False):
@@ -243,12 +316,13 @@ def handle_url(url, notice, show_url=False):
     except (requests.exceptions.TooManyRedirects,
             requests.exceptions.ConnectionError,
             requests.exceptions.InvalidURL,
+            UnicodeError,
             requests.exceptions.InvalidSchema) as error:
         notice("TROUBLE FOLLOWING URL: " + str(error))
         return
     if mobile_twitter_hack(url):
         return
-    title = bs4.BeautifulSoup(r.text).title
+    title = bs4.BeautifulSoup(r.text, "html5lib").title
     if title:
         prefix = "PAGE TITLE: "
         if show_url:
@@ -260,12 +334,15 @@ def handle_url(url, notice, show_url=False):
 
 class Session:
 
-    def __init__(self, io, username, nickname, channel):
+    def __init__(self, io, username, nickname, channel, twtfile):
         self.io = io
         self.nickname = nickname
+        self.channel = channel
+        self.users_in_chan = []
+        self.twtfile = twtfile
         self.io.send_line("NICK " + self.nickname)
         self.io.send_line("USER " + username + " 0 * : ")
-        self.io.send_line("JOIN " + channel)
+        self.io.send_line("JOIN " + self.channel)
 
     def loop(self):
 
@@ -282,7 +359,7 @@ class Session:
                 if "!" == msg[0]:
                     tokens = msg[1:].split()
                     argument = str.join(" ", tokens[1:])
-                    handle_command(tokens[0], argument, notice, target)
+                    handle_command(tokens[0], argument, notice, target, self)
                     return
                 hash_string = hashlib.md5(target.encode("utf-8")).hexdigest()
                 markovfeed_name = "markovfeed_" + hash_string
@@ -308,6 +385,14 @@ class Session:
             msg = str.join(" ", tokens[3:])[1:]
             handle_input(msg, target)
 
+        def name_from_join_or_part(tokens):
+            token = tokens[0][1:]
+            index_cut = token.find("@")
+            index_ex = token.find("!")
+            if index_ex > 0 and index_ex < index_cut:
+                index_cut = index_ex
+            return token[:index_cut]
+
         while True:
             line = self.io.recv_line()
             if not line:
@@ -318,7 +403,17 @@ class Session:
                     self.io.send_line("PONG " + tokens[1])
                 elif tokens[1] == "PRIVMSG":
                     handle_privmsg(tokens)
-
+                elif tokens[1] == "353":
+                    names = tokens[5:]
+                    names[0] = names[0][1:]
+                    self.users_in_chan += names
+                elif tokens[1] == "JOIN":
+                    name = name_from_join_or_part(tokens)
+                    if name != self.nickname:
+                        self.users_in_chan += [name]
+                elif tokens[1] == "PART":
+                    name = name_from_join_or_part(tokens)
+                    del(self.users_in_chan[self.users_in_chan.index(name)])
 
 def parse_command_line_arguments():
     parser = argparse.ArgumentParser()
@@ -329,7 +424,7 @@ def parse_command_line_arguments():
     parser.add_argument("-p, --port", action="store", dest="port", type=int,
                         default=PORT, help="port to connect to (default : "
                         + str(PORT) + ")")
-    parser.add_argument("-t, --timeout", action="store", dest="timeout",
+    parser.add_argument("-w, --wait", action="store", dest="timeout",
                         type=int, default=TIMEOUT,
                         help="timeout in seconds after which to attempt " +
                         "reconnect (default: " + str(TIMEOUT) + ")")
@@ -339,6 +434,9 @@ def parse_command_line_arguments():
     parser.add_argument("-n, --nickname", action="store", dest="nickname",
                         default=NICKNAME, help="nickname to use (default: "
                         + NICKNAME + ")")
+    parser.add_argument("-t, --twtfile", action="store", dest="twtfile",
+                        default=TWTFILE, help="twtfile to use (default: "
+                        + TWTFILE + ")")
     parser.add_argument("CHANNEL", action="store", help="channel to join")
     opts, unknown = parser.parse_known_args()
     return opts
@@ -348,7 +446,8 @@ opts = parse_command_line_arguments()
 while True:
     try:
         io = IO(opts.server, opts.port, opts.timeout)
-        session = Session(io, opts.username, opts.nickname, opts.CHANNEL)
+        session = Session(io, opts.username, opts.nickname, opts.CHANNEL,
+            opts.twtfile)
         session.loop()
     except ExceptionForRestart:
         io.socket.close()