Merge pull request #5593 from kamil-holubicki/PS-9777-8.0

PS-9777: binlog_utils_udf plugin should handle binlog.index entries the same as the server code does

Author: kamil-holubicki

mysql-test/r/percona_binlog_utils_udf.result

@@ -1655,7 +1655,5 @@ DELETE FROM loaded_ts;
 LOAD DATA LOCAL INFILE 'MYSQLTEST_VARDIR/filtered_binlog_file' INTO TABLE loaded_ts;
 include/assert.inc ['last record timestamp extracted via get_last_record_timestamp_by_binlog() for stage 16 should match the value extracted via mysqlbinlog']
 DROP FUNCTION get_last_record_timestamp_by_binlog;
-DROP TABLE loaded_ts;
 UNINSTALL PLUGIN binlog_utils_udf;
-DROP TABLE captured_gtid;
 DROP TABLE t1;

mysql-test/t/percona_binlog_utils_udf.test

@@ -338,11 +338,59 @@ while($stage <= $number_of_stages)
 }
 DROP FUNCTION get_last_record_timestamp_by_binlog;
 
-# temporary table cleanup
-DROP TABLE loaded_ts;
+
+#
+# PS-9777 - binlog_utils_udf plugin should handle binlog.index entries the same as the server code does
+# Check that binlog_utils_udf plugin handles binlog.index file entries
+# that are filenames (no relative or absolute path) correctly.
+#
+--source include/have_util_sed.inc
+
+# We only expect that the following won't fail at any stage. Actual results don't matter.
+--disable_query_log
+--disable_result_log
+eval CREATE FUNCTION get_gtid_set_by_binlog RETURNS STRING SONAME "$BINLOG_UTILS_UDF_LIB";
+eval CREATE FUNCTION get_binlog_by_gtid_set RETURNS STRING SONAME "$BINLOG_UTILS_UDF_LIB";
+eval CREATE FUNCTION get_binlog_by_gtid RETURNS STRING SONAME "$BINLOG_UTILS_UDF_LIB";
+eval CREATE FUNCTION get_last_gtid_from_binlog RETURNS STRING SONAME "$BINLOG_UTILS_UDF_LIB";
+eval CREATE FUNCTION get_first_record_timestamp_by_binlog RETURNS INTEGER SONAME "$BINLOG_UTILS_UDF_LIB";
+eval CREATE FUNCTION get_last_record_timestamp_by_binlog RETURNS INTEGER SONAME "$BINLOG_UTILS_UDF_LIB";
+
+# Modify the index file (remove leading ./ leaving only binlog file names)
+--let $index_file = `SELECT @@log_bin_index`
+--exec $SED -i 's|^\./binlog\.|binlog.|' $index_file
+
+# There will be no temp tables after restart
+--let $binlog_file_name = `SELECT binlog_name FROM captured_gtid LIMIT 1`
+--let $gtid_value = `SELECT gtid_value FROM captured_gtid LIMIT 1`
+
+# mysqld uses a cache over index file. Restart the server to fetch the new index file content.
+--source include/restart_mysqld.inc
+
+# restart re-enables query and resutl logs
+--disable_query_log
+--disable_result_log
+eval SELECT get_gtid_set_by_binlog('$binlog_file_name');
+--let $gtidset = `SELECT get_gtid_set_by_binlog('$binlog_file_name')`
+eval SELECT get_binlog_by_gtid_set('$gtidset');
+eval SELECT get_binlog_by_gtid('$gtid_value');
+eval SELECT get_last_gtid_from_binlog('$binlog_file_name');
+eval SELECT get_first_record_timestamp_by_binlog('$binlog_file_name');
+eval SELECT get_last_record_timestamp_by_binlog('$binlog_file_name');
+
+DROP FUNCTION get_gtid_set_by_binlog;
+DROP FUNCTION get_binlog_by_gtid_set;
+DROP FUNCTION get_binlog_by_gtid;
+DROP FUNCTION get_last_gtid_from_binlog;
+DROP FUNCTION get_first_record_timestamp_by_binlog;
+DROP FUNCTION get_last_record_timestamp_by_binlog;
+--enable_query_log
+--enable_result_log
+
+
+# temporary table cleanup not needed as we restarted the server
 
 --replace_result $BINLOG_UTILS_UDF_LIB BINLOG_UTILS_UDF_LIB
 UNINSTALL PLUGIN binlog_utils_udf;
 
-DROP TABLE captured_gtid;
 DROP TABLE t1;

plugin/binlog_utils_udf/binlog_utils_udf.cc

@@ -168,6 +168,25 @@ const char *get_short_binlog_name(const std::string &binlog_name) noexcept {
   return binlog_name.c_str() + dirname_length(binlog_name.c_str());
 }
 
+//
+// Retrieves the contents of the index file. Then normalizes its entries (each
+// entry is prefixed with relative or absolute path)
+//
+std::pair<int, std::list<std::string>> get_normalized_log_index() {
+  auto log_index = mysql_bin_log.get_log_index(true /* need_lock_index */);
+
+  auto &binlog_names = log_index.second;
+  auto bg = std::begin(binlog_names);
+  auto en = std::end(binlog_names);
+  std::for_each(bg, en, [](std::string &s) {
+    fn_reflen_buffer buf;
+    check_and_normalize_binlog_name(get_short_binlog_name(s), buf);
+    s.assign(buf);
+  });
+
+  return log_index;
+}
+
 log_event_ptr find_first_event(std::string_view binlog_name) {
   DBUG_TRACE;
 
@@ -378,7 +397,7 @@ mysqlpp::udf_result_t<STRING_RESULT> get_binlog_by_gtid_impl::calculate(
       throw std::runtime_error("Cannot parse 'gtid_executed'");
   }
 
-  auto log_index = mysql_bin_log.get_log_index(true /* need_lock_index */);
+  auto log_index = get_normalized_log_index();
   if (log_index.first != LOG_INFO_EOF)
     throw std::runtime_error("Cannot read binary log index'");
   if (log_index.second.empty())
@@ -480,7 +499,7 @@ mysqlpp::udf_result_t<STRING_RESULT> get_gtid_set_by_binlog_impl::calculate(
     const mysqlpp::udf_context &ctx) {
   DBUG_TRACE;
 
-  auto log_index = mysql_bin_log.get_log_index(true /* need_lock_index */);
+  auto log_index = get_normalized_log_index();
   if (log_index.first != LOG_INFO_EOF)
     throw std::runtime_error("Cannot read binary log index");
   if (log_index.second.empty())
@@ -584,7 +603,7 @@ mysqlpp::udf_result_t<STRING_RESULT> get_binlog_by_gtid_set_impl::calculate(
       throw std::runtime_error("Cannot parse 'gtid_executed'");
   }
 
-  auto log_index = mysql_bin_log.get_log_index(true /* need_lock_index */);
+  auto log_index = get_normalized_log_index();
   if (log_index.first != LOG_INFO_EOF)
     throw std::runtime_error("Cannot read binary log index'");
   if (log_index.second.empty())