From c042630ed8f94c32106d92a6a8deb192dabe558d Mon Sep 17 00:00:00 2001
From: Paul Wankadia <junyer@google.com>
Date: Fri, 22 Dec 2023 13:27:19 +0000
Subject: [PATCH] Report `kRegexpBadPerlOp` for look-behind assertions.

Fixes #468.

Change-Id: I9a72db0bbb9a04e5081cc1f0f94399476b63d1c3
Reviewed-on: https://code-review.googlesource.com/c/re2/+/62330
Reviewed-by: Perry Lorier <perryl@google.com>
Reviewed-by: Paul Wankadia <junyer@google.com>
---
 re2/parse.cc              | 11 +++++++++++
 re2/testing/parse_test.cc | 26 ++++++++++++++++++++++++++
 2 files changed, 37 insertions(+)

diff --git a/re2/parse.cc b/re2/parse.cc
index 46335bdb1..904599280 100644
--- a/re2/parse.cc
+++ b/re2/parse.cc
@@ -2070,6 +2070,17 @@ bool Regexp::ParseState::ParsePerlFlags(absl::string_view* s) {
     return false;
   }
 
+  // Check for look-around assertions. This is NOT because we support them! ;)
+  // As per https://github.com/google/re2/issues/468, we really want to report
+  // kRegexpBadPerlOp (not kRegexpBadNamedCapture) for look-behind assertions.
+  // Additionally, it would be nice to report not "(?<", but "(?<=" or "(?<!".
+  if ((t.size() > 3 && (t[2] == '=' || t[2] == '!')) ||
+      (t.size() > 4 && t[2] == '<' && (t[3] == '=' || t[3] == '!'))) {
+    status_->set_code(kRegexpBadPerlOp);
+    status_->set_error_arg(absl::string_view(t.data(), t[2] == '<' ? 4 : 3));
+    return false;
+  }
+
   // Check for named captures, first introduced in Python's regexp library.
   // As usual, there are three slightly different syntaxes:
   //
diff --git a/re2/testing/parse_test.cc b/re2/testing/parse_test.cc
index acb949c4f..7684b62a4 100644
--- a/re2/testing/parse_test.cc
+++ b/re2/testing/parse_test.cc
@@ -532,4 +532,30 @@ TEST(NamedCaptures, ErrorArgs) {
   EXPECT_EQ(status.error_arg(), "(?<space bar>");
 }
 
+// Test that look-around error args are correct.
+TEST(LookAround, ErrorArgs) {
+  RegexpStatus status;
+  Regexp* re;
+
+  re = Regexp::Parse("(?=foo).*", Regexp::LikePerl, &status);
+  EXPECT_TRUE(re == NULL);
+  EXPECT_EQ(status.code(), kRegexpBadPerlOp);
+  EXPECT_EQ(status.error_arg(), "(?=");
+
+  re = Regexp::Parse("(?!foo).*", Regexp::LikePerl, &status);
+  EXPECT_TRUE(re == NULL);
+  EXPECT_EQ(status.code(), kRegexpBadPerlOp);
+  EXPECT_EQ(status.error_arg(), "(?!");
+
+  re = Regexp::Parse("(?<=foo).*", Regexp::LikePerl, &status);
+  EXPECT_TRUE(re == NULL);
+  EXPECT_EQ(status.code(), kRegexpBadPerlOp);
+  EXPECT_EQ(status.error_arg(), "(?<=");
+
+  re = Regexp::Parse("(?<!foo).*", Regexp::LikePerl, &status);
+  EXPECT_TRUE(re == NULL);
+  EXPECT_EQ(status.code(), kRegexpBadPerlOp);
+  EXPECT_EQ(status.error_arg(), "(?<!");
+}
+
 }  // namespace re2