From 666c515c6e62bcc26396fc5211782907f4c8c2a8 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Johannes=20M=C3=BCller?= <straightshoota@gmail.com>
Date: Thu, 13 Apr 2023 10:32:33 +0200
Subject: [PATCH 1/3] Fix infinite loop with MATCH_INVALID_UTF8 in PCRE2 <10.36

---
 src/regex/pcre2.cr | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/src/regex/pcre2.cr b/src/regex/pcre2.cr
index 4bfa074f4289..6c64fa1d8e41 100644
--- a/src/regex/pcre2.cr
+++ b/src/regex/pcre2.cr
@@ -23,7 +23,9 @@ module Regex::PCRE2
   # :nodoc:
   def initialize(*, _source @source : String, _options @options)
     options = pcre2_compile_options(options) | LibPCRE2::UTF | LibPCRE2::DUPNAMES | LibPCRE2::UCP
-    if PCRE2.version_number >= {10, 34}
+    # MATCH_INVALID_UTF was introduced in 10.34 but a bug that can lead to an
+    # infinite loop is only fixed in 10.36 (https://github.com/PCRE2Project/pcre2/commit/e0c6029a62db9c2161941ecdf459205382d4d379).
+    if PCRE2.version_number >= {10, 36}
       options |= LibPCRE2::MATCH_INVALID_UTF
     end
     @re = PCRE2.compile(source, options) do |error_message|

From bfe9128ded69a55d7f0a983633c3e29dc2677e24 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Johannes=20M=C3=BCller?= <straightshoota@gmail.com>
Date: Thu, 13 Apr 2023 11:25:10 +0200
Subject: [PATCH 2/3] Adjust specs

---
 spec/std/regex_spec.cr | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/spec/std/regex_spec.cr b/spec/std/regex_spec.cr
index f71684ef66bb..25413b7db250 100644
--- a/spec/std/regex_spec.cr
+++ b/spec/std/regex_spec.cr
@@ -154,7 +154,7 @@ describe "Regex" do
     end
 
     it "multibyte index" do
-      if Regex::Engine.version_number < {10, 34}
+      if Regex::Engine.version_number < {10, 36}
         expect_raises(ArgumentError, "bad offset into UTF string") do
           /foo/.match_at_byte_index("öfoo", 1)
         end
@@ -246,7 +246,7 @@ describe "Regex" do
       end
 
       it "invalid codepoint" do
-        if Regex::Engine.version_number < {10, 34}
+        if Regex::Engine.version_number < {10, 36}
           expect_raises(ArgumentError, "UTF-8 error") do
             /foo/.matches?("f\x96o")
           end
@@ -310,7 +310,7 @@ describe "Regex" do
     end
 
     it "multibyte index" do
-      if Regex::Engine.version_number < {10, 34}
+      if Regex::Engine.version_number < {10, 36}
         expect_raises(ArgumentError, "bad offset into UTF string") do
           /foo/.matches_at_byte_index?("öfoo", 1)
         end

From 2bf6c5cd9c2aea6aae5c65337f2a4c9e792617e3 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Johannes=20M=C3=BCller?= <straightshoota@gmail.com>
Date: Thu, 13 Apr 2023 13:47:21 +0200
Subject: [PATCH 3/3] fixup

---
 spec/std/regex_spec.cr | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/spec/std/regex_spec.cr b/spec/std/regex_spec.cr
index 25413b7db250..698ca17c9a7d 100644
--- a/spec/std/regex_spec.cr
+++ b/spec/std/regex_spec.cr
@@ -115,8 +115,8 @@ describe "Regex" do
           /([\w_\.@#\/\*])+/.match("\xFF\xFE")
         end
       {% else %}
-        if Regex::PCRE2.version_number < {10, 35}
-          pending! "Error in libpcre2 < 10.35"
+        if Regex::PCRE2.version_number < {10, 36}
+          pending! "Error in libpcre2 < 10.36"
         else
           /([\w_\.@#\/\*])+/.match("\xFF\xFE").should be_nil
         end