From 6774cce09995cb874bab32bd0c980b4e271c5093 Mon Sep 17 00:00:00 2001
From: Jan-Philipp Litza <jpl@plutex.de>
Date: Tue, 4 Jan 2022 11:04:03 +0100
Subject: [PATCH] Support multiple regexp matches per line

By using the match group's indices, this also avoids replacing unrelated
chunks of the line.
---
 anonip.py | 47 ++++++++++++++++++++++++++++++-----------------
 tests.py  |  6 ++++++
 2 files changed, 36 insertions(+), 17 deletions(-)

diff --git a/anonip.py b/anonip.py
index 727f422..d51c150 100755
--- a/anonip.py
+++ b/anonip.py
@@ -180,6 +180,35 @@ def process_ip(self, ip):
                     )
             return trunc_ip
 
+    def process_regex_match(self, match):
+        """
+        This function processes a single regex match.
+
+        It returns the anonymized match as string and can be called with re.sub.
+
+        :param match: re.Match
+        :return: str
+        """
+        ret = []
+        last_pos = 0
+
+        for i, g in enumerate(match.groups(), start=1):
+            if not g:
+                continue
+            ip_str, ip = self.extract_ip(g)
+            replacement = (
+                self.process_ip(ip) if ip
+                else self.replace or g
+            )
+            ret.extend((
+                match.group(0)[last_pos:match.start(i) - match.start(0)],
+                str(replacement),
+            ))
+            last_pos = match.end(i) - match.start(0)
+
+        ret.append(match[0][last_pos:])
+        return "".join(ret)
+
     def process_line_regex(self, line):
         """
         This function processes a single line based on the provided regex.
@@ -189,23 +218,7 @@ def process_line_regex(self, line):
         :param line: str
         :return: str
         """
-        match = re.match(self.regex, line)
-        if not match:
-            logger.debug("Regex did not match!")
-            return line
-        groups = match.groups()
-
-        for m in set(groups):
-            if not m:
-                continue
-            ip_str, ip = self.extract_ip(m)
-            if ip:
-                trunc_ip = self.process_ip(ip)
-                line = line.replace(ip_str, str(trunc_ip))
-            elif self.replace:
-                line = line.replace(m, self.replace)
-
-        return line
+        return re.sub(self.regex, self.process_regex_match, line)
 
     def process_line_column(self, line):
         """
diff --git a/tests.py b/tests.py
index d54fac1..4f73a4a 100755
--- a/tests.py
+++ b/tests.py
@@ -140,6 +140,12 @@ def test_column(line, columns, expected):
             '3.3.0.0 - - [20/May/2015:21:05:01 +0000] "GET /723.3.3.357 HTTP/1.1" 200 13358 "-" "useragent"',
             None,
         ),
+        (
+            '3.3.3.3 - - [20/May/2015:21:05:01 +0000] "GET /723.3.3.357 HTTP/1.1" 200 13358 "-" "useragent [ip:1.2.3.4]"',
+            re.compile(r"\b([0-9a-fA-F][0-9a-fA-F:\.]*|::[0-9a-fA-F:\.]+)\b"),
+            '3.3.0.0 - - [20/May/2015:21:05:01 +0000] "GET /723.3.3.357 HTTP/1.1" 200 13358 "-" "useragent [ip:1.2.0.0]"',
+            None,
+        ),
         (
             "blabla/ 3.3.3.3 /blublu",
             re.compile(r"^blabla/ ([^,]+) /blublu"),