SedRegex: add support for custom delimiters

Via a new "plugins.SedRegex.delimiters" setting. This defaults to all symbols to match the existing behaviour.

Fixes #1610.

Author: jlu5

plugins/SedRegex/config.py

@@ -49,6 +49,10 @@ def configure(advanced):
 
 conf.registerChannelValue(SedRegex, 'displayErrors',
     registry.Boolean(True, _("""Should errors be displayed?""")))
+conf.registerChannelValue(SedRegex, 'delimiters',
+    registry.String('', _("""List of delimiters to match sed expressions on.
+        Multiple delimiters can be specified as a single string: e.g. "/@".
+        If empty, defaults to all symbols.""")))
 conf.registerChannelValue(SedRegex, 'boldReplacementText',
     registry.Boolean(True, _("""Should the replacement text be bolded?""")))
 conf.registerChannelValue(SedRegex, 'enable',

plugins/SedRegex/plugin.py

@@ -47,7 +47,7 @@
 except ImportError:
     _ = lambda x: x
 
-from .sedregex import SED_REGEX
+from .sedregex import makeSedRegex
 
 TAG_SEEN = 'SedRegex.seen'
 TAG_IS_REGEX = 'SedRegex.isRegex'
@@ -92,7 +92,7 @@ class SedRegex(callbacks.Plugin):
     public = True
 
     @staticmethod
-    def _unpack_sed(expr):
+    def _unpack_sed(sedRegex, expr):
         if '\0' in expr:
             raise ValueError('Expression can\'t contain NUL')
 
@@ -107,7 +107,7 @@ def _unpack_sed(expr):
 
             escaped_expr += c
 
-        match = SED_REGEX.search(escaped_expr)
+        match = sedRegex.search(escaped_expr)
 
         if not match:
             return
@@ -145,7 +145,13 @@ def doPrivmsg(self, irc, msg):
         # SedRegex was enabled.
         msg.tag(TAG_SEEN)
 
-        regexMatch = SED_REGEX.match(msg.args[1])
+
+        delimiters = self.registryValue('delimiters', msg.channel, irc.network)
+        if delimiters:
+            delimiters = re.escape(delimiters)
+        sedRegex = makeSedRegex(delimiters)
+        text = msg.args[1]
+        regexMatch = sedRegex.match(text)
         if not regexMatch:
             return
 
@@ -154,7 +160,7 @@ def doPrivmsg(self, irc, msg):
         msg.tag(TAG_IS_REGEX)
 
         try:
-            (pattern, replacement, count, flags) = self._unpack_sed(msg.args[1])
+            (pattern, replacement, count, flags) = self._unpack_sed(sedRegex, text)
         except Exception as e:
             self.log.warning(_("SedRegex parser error: %s"), e, exc_info=True)
             if self.registryValue('displayErrors', msg.channel, irc.network):
@@ -172,7 +178,7 @@ def doPrivmsg(self, irc, msg):
         regex_timeout = self.registryValue('processTimeout')
         try:
             message = process(self._replacer_process, irc, msg,
-                    target, pattern, replacement, count, iterable,
+                    target, pattern, replacement, count, iterable, sedRegex,
                     timeout=regex_timeout, pn=self.name(), cn='replacer')
         except ProcessTimeoutError:
             irc.error(_("Search timed out."))
@@ -187,7 +193,7 @@ def doPrivmsg(self, irc, msg):
         else:
             irc.reply(message, prefixNick=False)
 
-    def _replacer_process(self, irc, msg, target, pattern, replacement, count, messages):
+    def _replacer_process(self, irc, msg, target, pattern, replacement, count, messages, sedRegex):
         for m in messages:
             if m.command in ('PRIVMSG', 'NOTICE') and \
                     ircutils.strEqual(m.args[0], msg.args[0]) and m.tagged('receivedBy') == irc:
@@ -209,7 +215,7 @@ def _replacer_process(self, irc, msg, target, pattern, replacement, count, messa
                 # so we only need to do this check once per message.
                 if not m.tagged(TAG_SEEN):
                     m.tag(TAG_SEEN)
-                    if SED_REGEX.match(m.args[1]):
+                    if sedRegex.match(m.args[1]):
                         m.tag(TAG_IS_REGEX)
                 # Ignore messages containing a regexp if ignoreRegex is on.
                 if self.registryValue('ignoreRegex', msg.channel, irc.network) and m.tagged(TAG_IS_REGEX):

plugins/SedRegex/sedregex.py

@@ -2,25 +2,29 @@
 
 import re
 
-SED_REGEX = re.compile(
-    # This part matches an optional nick followed by ":" or ",", used to direct replacement
-    # at a particular user.
-    r"^(?:(?P<nick>.+?)[:,] )?"
+def makeSedRegex(delimiters=''):
+    delimiters = delimiters or r"^\w\s"
 
-    # Match and save the delimiter (any one symbol) as a named group
-    r"s(?P<delim>[^\w\s])"
+    return re.compile(
+        # This part matches an optional nick followed by ":" or ",", used to direct replacement
+        # at a particular user.
+        r"^(?:(?P<nick>.+?)[:,] )?"
 
-    # Match the pattern to replace, which can be any string up to the first instance of the delimiter
-    r"(?P<pattern>(?:(?!(?P=delim)).)*)(?P=delim)"
+        # Match and save the delimiter (any one symbol) as a named group
+        fr"s(?P<delim>[{delimiters}])"
 
-    # Ditto with the replacement
-    r"(?P<replacement>(?:(?!(?P=delim)).)*)"
+        # Match the pattern to replace, which can be any string up to the first instance of the
+        # delimiter
+        r"(?P<pattern>(?:(?!(?P=delim)).)*)(?P=delim)"
 
-    # Optional final delimiter plus flags at the end
-    r"(?:(?P=delim)(?P<flags>[a-z]*))?"
-)
+        # Ditto with the replacement
+        r"(?P<replacement>(?:(?!(?P=delim)).)*)"
+
+        # Optional final delimiter plus flags at the end
+        r"(?:(?P=delim)(?P<flags>[a-z]*))?"
+    )
 
 if __name__ == '__main__':
     print("This is the full regex used by the plugin; paste it into your favourite regex tester "
           "for debugging:")
-    print(SED_REGEX)
+    print(makeSedRegex())

plugins/SedRegex/test.py

@@ -151,6 +151,33 @@ def testNonSlashSeparator(self):
         m = self.getMsg(' ')
         self.assertIn('4 * 2 = 8', str(m))
 
+    def testCustomDelimiters(self):
+        with conf.supybot.plugins.sedregex.delimiters.context("@."):
+            self.feedMsg('test')
+            self.feedMsg('s@t@b')
+            m = self.getMsg(' ')
+            self.assertIn('best', str(m))
+
+            self.feedMsg('s.t.w')
+            m = self.getMsg(' ')
+            self.assertIn('west', str(m))
+
+            # / is not in the delimiters list, so it is ignored
+            self.getMsg('s/t/r')
+            for msg in self.irc.state.history:
+                self.assertNotIn("rest", str(msg))
+
+        # These would fail if the delimiters set isn't escaped correctly
+        with conf.supybot.plugins.sedregex.delimiters.context("]["):
+            self.feedMsg('test')
+            self.feedMsg('s]t]f')
+            m = self.getMsg(' ')
+            self.assertIn('fest', str(m))
+
+            self.feedMsg('s[t[qu[')
+            m = self.getMsg(' ')
+            self.assertIn('quest', str(m))
+
     def testWeirdSeparatorsFail(self):
         self.feedMsg("can't touch this", frm=self.__class__.other)
         # Only symbols are allowed as separators