From c457abd5f097dd13fb21543381e7cfafe7d31cfb Mon Sep 17 00:00:00 2001
From: Andi Albrecht <albrecht.andi@gmail.com>
Date: Mon, 20 Mar 2023 08:33:46 +0100
Subject: [PATCH] Remove unnecessary parts in regex for bad escaping.

The regex tried to deal with situations where escaping in the
SQL to be parsed was suspicious.
---
 CHANGELOG            | 10 ++++++++++
 sqlparse/keywords.py |  4 ++--
 tests/test_split.py  |  4 ++--
 3 files changed, 14 insertions(+), 4 deletions(-)

diff --git a/CHANGELOG b/CHANGELOG
index 94864138..880a9ca9 100644
--- a/CHANGELOG
+++ b/CHANGELOG
@@ -1,12 +1,22 @@
 Development Version
 -------------------
 
+Notable Changes
+
+* IMPORTANT: This release fixes a security vulnerability in the
+  parser where a regular expression vulnerable to ReDOS (Regular
+  Expression Denial of Service) was used. See the security advisory
+  for details: https://github.com/andialbrecht/sqlparse/security/advisories/GHSA-rrm6-wvj7-cwh2
+  The vulnerability was discovered by @erik-krogh from GitHub
+  Security Lab (GHSL). Thanks for reporting!
+
 Bug Fixes
 
 * Revert a change from 0.4.0 that changed IN to be a comparison (issue694).
   The primary expectation is that IN is treated as a keyword and not as a
   comparison operator. That also follows the definition of reserved keywords
   for the major SQL syntax definitions.
+* Fix regular expressions for string parsing.
 
 Other
 
diff --git a/sqlparse/keywords.py b/sqlparse/keywords.py
index f85d4688..b45f3e0f 100644
--- a/sqlparse/keywords.py
+++ b/sqlparse/keywords.py
@@ -59,9 +59,9 @@
     (r'(?![_A-ZÀ-Ü])-?(\d+(\.\d*)|\.\d+)(?![_A-ZÀ-Ü])',
      tokens.Number.Float),
     (r'(?![_A-ZÀ-Ü])-?\d+(?![_A-ZÀ-Ü])', tokens.Number.Integer),
-    (r"'(''|\\\\|\\'|[^'])*'", tokens.String.Single),
+    (r"'(''|\\'|[^'])*'", tokens.String.Single),
     # not a real string literal in ANSI SQL:
-    (r'"(""|\\\\|\\"|[^"])*"', tokens.String.Symbol),
+    (r'"(""|\\"|[^"])*"', tokens.String.Symbol),
     (r'(""|".*?[^\\]")', tokens.String.Symbol),
     # sqlite names can be escaped with [square brackets]. left bracket
     # cannot be preceded by word character or a right bracket --
diff --git a/tests/test_split.py b/tests/test_split.py
index a9d75765..e79750e8 100644
--- a/tests/test_split.py
+++ b/tests/test_split.py
@@ -18,8 +18,8 @@ def test_split_semicolon():
 
 
 def test_split_backslash():
-    stmts = sqlparse.parse(r"select '\\'; select '\''; select '\\\'';")
-    assert len(stmts) == 3
+    stmts = sqlparse.parse("select '\'; select '\'';")
+    assert len(stmts) == 2
 
 
 @pytest.mark.parametrize('fn', ['function.sql',