support & in utils.parse.sed, change sed.py to use utils.parse.sed

modules/message_filter.py

@@ -30,7 +30,7 @@ class Module(ModuleManager.BaseModule):
 
         filters = self._get_filters(event["server"], target)
         for filter in filters:
-            sed = utils.parse.sed.parse_sed(filter)
+            sed = utils.parse.sed.parse(filter)
             type, out = utils.parse.sed.sed(sed, message)
 
             if type == "m" and out:

modules/sed.py

@@ -4,9 +4,7 @@
 import re, traceback
 from src import ModuleManager, utils
 
-REGEX_SPLIT = re.compile("(?<!\\\\)/")
 REGEX_SED = re.compile("^(?:(\\S+)[:,] )?s/")
-SED_AMPERSAND = re.compile(r"((?:^|[^\\])(?:\\\\)*)&")
 
 @utils.export("channelset",
     utils.BoolSetting("sed","Disable/Enable sed in a channel"))
@@ -21,72 +19,38 @@ class Module(ModuleManager.BaseModule):
     @utils.kwarg("command", "sed")
     @utils.kwarg("pattern", REGEX_SED)
     def channel_message(self, event):
-        sed_split = re.split(REGEX_SPLIT, event["message"], 3)
+        for_user = event["match"].group(1)
-        if len(sed_split) > 2:
+        sed_s = event["message"]
-            if not self._closest_setting(event, "sed", False):
+        if for_user:
-                return
+            sed_s = sed_s.split(" ", 1)[1]
+        if not self._closest_setting(event, "sed", False):
+            return
 
-            regex_flags = 0
+        try:
-            flags = (sed_split[3:] or [""])[0].split(" ", 1)[0]
+            sed = utils.parse.sed.parse(event["message"])
-            count = None
+        except:
+            traceback.print_exc()
+            event["stderr"].write("Invalid regex in pattern")
+            return
+        sed.replace = utils.irc.bold(sed.replace)
 
-            last_flag = ""
+        if self._closest_setting(event, "sed-sender-only", False):
-            for flag in flags:
+            for_user = event["user"].nickname
-                if flag.isdigit():
-                    if last_flag.isdigit():
-                        count = int(str(count) + flag)
-                    elif not count:
-                        count = int(flag)
-                elif flag == "i":
-                    regex_flags |= re.I
-                elif flag == "g":
-                    count = 0
-                last_flag = flag
-            if count == None:
-                count = 1
 
-            try:
+        match_line = None
-                pattern = re.compile(sed_split[1], regex_flags)
+        match_message = None
-            except:
+        with utils.deadline():
-                traceback.print_exc()
+            for line in event["target"].buffer.get_all(for_user):
-                event["stderr"].write("Invalid regex in pattern")
+                if not line.from_self:
-                return
+                    match = sed.match(line.message)
+                    if not match == line.message:
+                        match_line = line
+                        match_message = match
+                        break
 
-            for_user = event["match"].group(1)
+        if match_line:
-            if self._closest_setting(event, "sed-sender-only", False):
+            if match_line.action:
-                for_user = event["user"].nickname
+                format = "* %s %s"
-
+            else:
-            match_line = None
+                format = "<%s> %s"
-            match = None
+            event["stdout"].write(format % (match_line.sender, match_message))
-            match_message = None
-            with utils.deadline():
-                for line in event["target"].buffer.get_all(for_user):
-                    if not line.from_self:
-                        message = line.notes.get("sed-line", line.message)
-                        match = pattern.search(message)
-                        if match and not REGEX_SED.match(message):
-                            match_line = line
-                            match = match.group(0)
-                            match_message = message
-                            break
-
-            if match:
-                replace = sed_split[2]
-                replace = replace.replace("\\/", "/")
-
-                with utils.deadline():
-                    for found in SED_AMPERSAND.finditer(replace):
-                        found = found.group(1)
-                        replace.replace(found, "%s%s" % (found, match))
-
-                replace_color = utils.irc.bold(replace)
-
-                new_message = re.sub(pattern, replace, message, count)
-                new_message_color = re.sub(pattern, utils.irc.bold(replace),
-                    message, count)
-                if match_line.action:
-                    prefix = "* %s" % match_line.sender
-                else:
-                    prefix = "<%s>" % match_line.sender
-                match_line.notes["sed-line"] = new_message
-                event["stdout"].write("%s %s" % (prefix, new_message_color))

src/utils/parse/sed.py

@@ -1,5 +1,18 @@
 import dataclasses, re, typing
 
+def _tokens(s: str, token: str) -> typing.List[int]:
+    backslash = False
+    tokens = []
+    for i, c in enumerate(s):
+        if not backslash:
+            if c == token:
+                tokens.append(i)
+            elif c == "\\":
+                backslash = True
+        else:
+            backslash = False
+    return tokens
+
 class Sed(object):
     type: str
     def match(self, s: str) -> typing.Optional[str]:
@@ -13,7 +26,17 @@ class SedReplace(Sed):
     count: int
 
     def match(self, s):
-        return self.pattern.sub(self.replace, s, self.count)
+        matches = list(self.pattern.finditer(s))
+        if not self.count == 0:
+            matches = matches[:self.count]
+
+        for match in matches:
+            replace_copy = self.replace
+            for token in reversed(_tokens(replace_copy, "&")):
+                replace_copy = (
+                    replace_copy[:token]+match.group(0)+replace_copy[token+1:])
+            s = s.replace(match.group(0), replace_copy, 1)
+        return s
 
 @dataclasses.dataclass
 class SedMatch(Sed):
@@ -26,24 +49,15 @@ class SedMatch(Sed):
             return match.group(0)
         return None
 
-def _sed_split(s):
+def _sed_split(s: str) -> typing.List[str]:
-    backslash = False
+    tokens = _tokens(s, "/")
-    forward_slash = []
+    if tokens and (not tokens[-1] == (len(s)-1)):
-    for i, c in enumerate(s):
+        tokens.append(len(s))
-        if not backslash:
-            if c == "/":
-                forward_slash.append(i)
-            if c == "\\":
-                backslash = True
-        else:
-            backslash = False
-    if forward_slash and (not forward_slash[-1] == (len(s)-1)):
-        forward_slash.append(len(s))
 
     last = 0
     out = []
-    for i in forward_slash:
+    for i in tokens:
-        out.append(s[last:i])
+        out.append(s[last:i].replace("\\/", "/"))
         last = i+1
     return out
 
@@ -56,13 +70,13 @@ def _sed_flags(s: str) -> typing.Tuple[int, int]:
         re_flags |= re.I
     return count, re_flags
 
-def parse_sed(sed_s: str) -> typing.Optional[Sed]:
+def parse(sed_s: str) -> typing.Optional[Sed]:
     type, pattern, *args = _sed_split(sed_s)
     if type == "s":
         replace, *args = args
         count, flags = _sed_flags((args or [""])[0])
-        pattern = re.compile(pattern, flags)
+        pattern_re = re.compile(pattern, flags)
-        return SedReplace(type, pattern, replace, count)
+        return SedReplace(type, pattern_re, replace, count)
     elif type == "m":
         count, flags = _sed_flags((args or [""])[0])
         return SedMatch(type, re.compile(pattern, flags))