From a7a805ce6cf7f54c992e6086d7b9efd371e1988f Mon Sep 17 00:00:00 2001
From: Yura Sorokin <yura.sorokin@percona.com>
Date: Wed, 11 Mar 2026 17:50:09 +0100
Subject: [PATCH] PS-10679 fix: cannot parse GTID set with multiple UUIDs if no
 space after a coma before start of next UUID

https://perconadev.atlassian.net/browse/PS-10679

Fixed an issue with "binsrv::gtids::gtid_set' parser not being able to accept GTIDs without
spaces between UUID groups.

'gtid_set_test' unit test (BOOST_TEST_MODULE GtidSetTests) extended with additional test
case 'GtidSetWhitespaces' that checks for
'binsrv::gtids::gtid_set' -> 'std::string' -> 'binsrv::gtids::gtid_set roundtrip' with removed
whitespaces.
---
 src/binsrv/gtids/gtid_set.cpp |  6 ++--
 tests/gtid_set_test.cpp       | 56 +++++++++++++++++++++++++++++++++++
 2 files changed, 58 insertions(+), 4 deletions(-)

diff --git a/src/binsrv/gtids/gtid_set.cpp b/src/binsrv/gtids/gtid_set.cpp
index e95de5a..2ba9171 100644
--- a/src/binsrv/gtids/gtid_set.cpp
+++ b/src/binsrv/gtids/gtid_set.cpp
@@ -80,7 +80,7 @@ class gtid_set_parser {
     <optionally_tagged_interval> ::= (<tag> ':')? <gno> ('-' <gno>)?
     <tag> ::= [a-zA-Z_][a-zA-Z0-9_]{0,31}
     <gno> ::= [0-9]+
-    <space> ::= [ \t\r\n\f\v]+
+    <space> ::= [ \t\r\n\f\v]*
   */
 
   gtid_set *result_gtids_;
@@ -252,10 +252,8 @@ class gtid_set_parser {
     parse_optionally_tagged_intervals(remainder);
   }
 
-  // <space> ::= [ \t\r\n\f\v]+
+  // <space> ::= [ \t\r\n\f\v]*
   static void parse_space(std::string_view &remainder) {
-    util::parse_character_predicate(remainder, space_predicate);
-
     char character{};
     while (util::parse_character_predicate_ex(remainder, space_predicate,
                                               character)) {
diff --git a/tests/gtid_set_test.cpp b/tests/gtid_set_test.cpp
index b628c37..68933be 100644
--- a/tests/gtid_set_test.cpp
+++ b/tests/gtid_set_test.cpp
@@ -665,3 +665,59 @@ BOOST_AUTO_TEST_CASE(GtidSetStreamOperatorMixed) {
       boost::lexical_cast<binsrv::gtids::gtid_set>(gtids_str)};
   BOOST_CHECK_EQUAL(gtids, restored_gtids);
 }
+
+BOOST_AUTO_TEST_CASE(GtidSetWhitespaces) {
+  const binsrv::gtids::uuid first_uuid{first_uuid_sv};
+  const binsrv::gtids::uuid second_uuid{second_uuid_sv};
+
+  const binsrv::gtids::tag first_tag{first_tag_sv};
+  const binsrv::gtids::tag second_tag{second_tag_sv};
+
+  binsrv::gtids::gtid_set gtids{};
+  // NOLINTBEGIN(cppcoreguidelines-avoid-magic-numbers,readability-magic-numbers)
+  gtids += binsrv::gtids::gtid{first_uuid, 101ULL};
+  gtids += binsrv::gtids::gtid{first_uuid, 102ULL};
+  gtids += binsrv::gtids::gtid{first_uuid, 103ULL};
+  gtids += binsrv::gtids::gtid{first_uuid, 105ULL};
+
+  gtids += binsrv::gtids::gtid{first_uuid, first_tag, 111ULL};
+  gtids += binsrv::gtids::gtid{first_uuid, first_tag, 112ULL};
+  gtids += binsrv::gtids::gtid{first_uuid, first_tag, 113ULL};
+  gtids += binsrv::gtids::gtid{first_uuid, first_tag, 115ULL};
+
+  gtids += binsrv::gtids::gtid{first_uuid, second_tag, 121ULL};
+  gtids += binsrv::gtids::gtid{first_uuid, second_tag, 122ULL};
+  gtids += binsrv::gtids::gtid{first_uuid, second_tag, 123ULL};
+  gtids += binsrv::gtids::gtid{first_uuid, second_tag, 125ULL};
+
+  gtids += binsrv::gtids::gtid{second_uuid, 201ULL};
+  gtids += binsrv::gtids::gtid{second_uuid, 202ULL};
+  gtids += binsrv::gtids::gtid{second_uuid, 203ULL};
+  gtids += binsrv::gtids::gtid{second_uuid, 205ULL};
+
+  gtids += binsrv::gtids::gtid{second_uuid, first_tag, 211ULL};
+  gtids += binsrv::gtids::gtid{second_uuid, first_tag, 212ULL};
+  gtids += binsrv::gtids::gtid{second_uuid, first_tag, 213ULL};
+  gtids += binsrv::gtids::gtid{second_uuid, first_tag, 215ULL};
+
+  gtids += binsrv::gtids::gtid{second_uuid, second_tag, 221ULL};
+  gtids += binsrv::gtids::gtid{second_uuid, second_tag, 222ULL};
+  gtids += binsrv::gtids::gtid{second_uuid, second_tag, 223ULL};
+  gtids += binsrv::gtids::gtid{second_uuid, second_tag, 225ULL};
+  // NOLINTEND(cppcoreguidelines-avoid-magic-numbers,readability-magic-numbers)
+
+  auto gtids_str{boost::lexical_cast<std::string>(gtids)};
+  // erasing all space characters from the GTIDs string
+
+  // looks like a misconfiguration in clang-tyidy-19 that doesn't know that
+  // std::erase() for std::string is located in the <string> system header
+
+  // TODO: re-check this when switching to clang-20
+
+  // NOLINTNEXTLINE(misc-include-cleaner)
+  std::erase(gtids_str, ' ');
+  BOOST_CHECK(gtids_str.find(' ') == std::string::npos);
+  const auto restored_gtids{
+      boost::lexical_cast<binsrv::gtids::gtid_set>(gtids_str)};
+  BOOST_CHECK_EQUAL(gtids, restored_gtids);
+}