From 34a5841b019c8c38d55aad3981e3fb919d9ee60c Mon Sep 17 00:00:00 2001
From: Alex Kenion <alexk@wordfence.com>
Date: Tue, 24 Oct 2023 11:35:34 -0400
Subject: [PATCH] Improved messaging around file inclusion/exclusion patterns

---
 wordfence/cli/malwarescan/definition.py  |  8 ++++----
 wordfence/cli/malwarescan/malwarescan.py | 16 ++++++++++++++--
 wordfence/scanning/filtering.py          | 13 +++++++++++--
 3 files changed, 29 insertions(+), 8 deletions(-)

diff --git a/wordfence/cli/malwarescan/definition.py b/wordfence/cli/malwarescan/definition.py
index 41b4dcd3..6c60246b 100644
--- a/wordfence/cli/malwarescan/definition.py
+++ b/wordfence/cli/malwarescan/definition.py
@@ -89,8 +89,8 @@
     },
     "include-files-pattern": {
         "short_name": "N",
-        "description": "PCRE regex allow pattern. Only matching filenames will"
-                       " be scanned.",
+        "description": "Python regex allow pattern. Only matching filenames "
+                       "will be scanned.",
         "context": "ALL",
         "argument_type": "OPTION_REPEATABLE",
         "default": None,
@@ -100,8 +100,8 @@
     },
     "exclude-files-pattern": {
         "short_name": "X",
-        "description": "PCRE regex deny pattern. Matching filenames will not "
-                       "be scanned.",
+        "description": "Python regex deny pattern. Matching filenames will "
+                       "not be scanned.",
         "context": "ALL",
         "argument_type": "OPTION_REPEATABLE",
         "default": None,
diff --git a/wordfence/cli/malwarescan/malwarescan.py b/wordfence/cli/malwarescan/malwarescan.py
index 577db943..41f02da4 100644
--- a/wordfence/cli/malwarescan/malwarescan.py
+++ b/wordfence/cli/malwarescan/malwarescan.py
@@ -86,13 +86,25 @@ def _initialize_file_filter(self) -> filtering.FileFilter:
         if self.config.include_files_pattern is not None:
             has_include_overrides = True
             for pattern in self.config.include_files_pattern:
-                filter.add(filtering.filter_pattern(pattern))
+                try:
+                    filter.add(filtering.filter_pattern(pattern))
+                except filtering.InvalidPatternException:
+                    raise Exception(
+                            'File inclusion pattern must be a Python regex, '
+                            f'received: "{pattern}"'
+                        )
         if self.config.exclude_files is not None:
             for name in self.config.exclude_files:
                 filter.add(filtering.filter_filename(name), False)
         if self.config.exclude_files_pattern is not None:
             for pattern in self.config.exclude_files_pattern:
-                filter.add(filtering.filter_pattern(pattern), False)
+                try:
+                    filter.add(filtering.filter_pattern(pattern), False)
+                except filtering.InvalidPatternException:
+                    raise Exception(
+                            'File exclusion pattern must be a Python regex, '
+                            f'received: "{pattern}"'
+                        )
         if not has_include_overrides:
             filter.add(filtering.filter_php)
             filter.add(filtering.filter_html)
diff --git a/wordfence/scanning/filtering.py b/wordfence/scanning/filtering.py
index dcf5d093..2c7b2ed1 100644
--- a/wordfence/scanning/filtering.py
+++ b/wordfence/scanning/filtering.py
@@ -98,6 +98,15 @@ def __call__(self, path: str) -> bool:
         return matches_regex(self.pattern, path)
 
 
+class InvalidPatternException(Exception):
+
+    def __init__(self, pattern: str):
+        self.pattern = pattern
+
+
 def filter_pattern(regex: str) -> Callable[[str], bool]:
-    pattern = re.compile(regex)
-    return Filter(pattern)
+    try:
+        pattern = re.compile(regex)
+        return Filter(pattern)
+    except re.error:
+        raise InvalidPatternException(regex)