SedRegex: Split _replacer_process into smaller functions

That each either access global state or apply the user-provided regexp.

A future commit will add an alternative implementation of _replacer_process
for Windows that calls them differently in order to access global state
in the main process and only applies regexps in the child process.

Author: progval

plugins/SedRegex/plugin.py

@@ -58,6 +58,60 @@
 class SearchNotFoundError(Exception):
     pass
 
+
+def filter_messages(network, msg, target, messages, ignoreRegex, sedRegex):
+    """Applies all filters but the user-provided regexp, so it is safe to
+    run in the main process."""
+    for m in messages:
+        if m.command in ('PRIVMSG', 'NOTICE') and \
+                ircutils.strEqual(m.args[0], msg.args[0]) and \
+                m.tagged('receivedBy') is not None and \
+                m.tagged('receivedBy').network == network :
+            if target and m.nick != target:
+                continue
+            # Don't snarf ignored users' messages unless specifically
+            # told to.
+            if ircdb.checkIgnored(m.prefix) and not target:
+                continue
+
+            # Test messages sent before SedRegex was activated. Mark them all as seen
+            # so we only need to do this check once per message.
+            if not m.tagged(TAG_SEEN):
+                m.tag(TAG_SEEN)
+                if sedRegex.match(m.args[1]):
+                    m.tag(TAG_IS_REGEX)
+            # Ignore messages containing a regexp if ignoreRegex is on.
+            if ignoreRegex and m.tagged(TAG_IS_REGEX):
+                continue
+
+            yield m
+
+def get_first_matching_message(pattern, messages):
+    for m in messages:
+        # When running substitutions, ignore the "* nick" part of any actions.
+        action = ircmsgs.isAction(m)
+        if action:
+            text = ircmsgs.unAction(m)
+        else:
+            text = m.args[1]
+
+        replace_result = pattern.search(text)
+        if replace_result:
+            return m
+
+def apply_substitution(pattern, replacement, m, count):
+    action = ircmsgs.isAction(m)
+    if action:
+        text = ircmsgs.unAction(m)
+    else:
+        text = m.args[1]
+
+    subst = pattern.sub(replacement, text, count)
+    if action:  # If the message was an ACTION, prepend the nick back.
+        subst = '* %s %s' % (m.nick, subst)
+
+    return axe_spaces(subst)
+
 class SedRegex(callbacks.Plugin):
     """
     Enable SedRegex on the desired channels:
@@ -175,6 +229,8 @@ def doPrivmsg(self, irc, msg):
             return
 
         regex_timeout = self.registryValue('processTimeout')
+        if self.registryValue('boldReplacementText', msg.channel, irc.network):
+            replacement = ircutils.bold(replacement)
         try:
             message = process(self._replacer_process, irc, msg,
                     target, pattern, replacement, count, iterable, sedRegex,
@@ -192,63 +248,35 @@ def doPrivmsg(self, irc, msg):
         else:
             irc.reply(message, prefixNick=False)
 
-    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:
-                if target and m.nick != target:
-                    continue
-                # Don't snarf ignored users' messages unless specifically
-                # told to.
-                if ircdb.checkIgnored(m.prefix) and not target:
-                    continue
-
-                # When running substitutions, ignore the "* nick" part of any actions.
-                action = ircmsgs.isAction(m)
-                if action:
-                    text = ircmsgs.unAction(m)
-                else:
-                    text = m.args[1]
-
-                # Test messages sent before SedRegex was activated. Mark them all as seen
-                # so we only need to do this check once per message.
-                if not m.tagged(TAG_SEEN):
-                    m.tag(TAG_SEEN)
-                    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):
-                    self.log.debug("Skipping message %s because it is tagged as isRegex", m.args[1])
-                    continue
-
-                try:
-                    replace_result = pattern.search(text)
-                    if replace_result:
-                        if self.registryValue('boldReplacementText',
-                                              msg.channel, irc.network):
-                            replacement = ircutils.bold(replacement)
-                        subst = pattern.sub(replacement, text, count)
-                        if action:  # If the message was an ACTION, prepend the nick back.
-                            subst = '* %s %s' % (m.nick, subst)
-
-                        subst = axe_spaces(subst)
+    def _format_result(self, irc, msg, m, subst):
+        if m.nick == msg.nick:
+            fmt = self.registryValue('format', msg.channel, irc.network)
+            env = {'replacement': subst}
+        else:
+            fmt = self.registryValue('format.other', msg.channel, irc.network)
+            env = {'otherNick': msg.nick, 'replacement': subst}
 
-                        if m.nick == msg.nick:
-                            fmt = self.registryValue('format', msg.channel, irc.network)
-                            env = {'replacement': subst}
-                        else:
-                            fmt = self.registryValue('format.other', msg.channel, irc.network)
-                            env = {'otherNick': msg.nick, 'replacement': subst}
+        return ircutils.standardSubstitute(irc, m, fmt, env)
 
-                        return ircutils.standardSubstitute(irc, m, fmt, env)
+    def _replacer_process(self, irc, msg, target, pattern, replacement, count,
+                          messages, sedRegex):
+        ignoreRegex = self.registryValue('ignoreRegex', msg.channel, irc.network)
+        messages = filter_messages(irc.network, msg, target, messages,
+                                   ignoreRegex, sedRegex)
 
-                except Exception as e:
-                    self.log.warning(_("SedRegex error: %s"), e, exc_info=True)
-                    raise
+        try:
+            m = get_first_matching_message(pattern, messages)
+            if m:
+                subst = apply_substitution(pattern, replacement, m, count)
+                return self._format_result(irc, msg, m, subst)
+        except Exception as e:
+            self.log.warning(_("SedRegex error: %s"), e, exc_info=True)
+            raise
 
         self.log.debug(_("SedRegex: Search %r not found in the last %i messages of %s."),
                          msg.args[1], len(irc.state.history), msg.args[0])
         raise SearchNotFoundError()
+
     doNotice = doPrivmsg
 
 Class = SedRegex