From 54ca1ab2dec779327d32b5806741f49885cb9da9 Mon Sep 17 00:00:00 2001
From: Markus Makela <markus.makela@mariadb.com>
Date: Mon, 11 Jan 2016 12:25:33 +0200
Subject: [PATCH] Canonicalized queries remove non-executable comments

The comments which do not alter the functionality of a query are now removed
from canonicalized queries.

Also fixed missing semicolon in the comment removal regex and added tests
for comment removal.
---
 query_classifier/test/canonical_tests/CMakeLists.txt  |  7 +++++++
 .../test/canonical_tests/comment.expected             | 11 +++++++++++
 query_classifier/test/canonical_tests/comment.sql     | 11 +++++++++++
 utils/skygw_utils.cc                                  |  6 +++---
 4 files changed, 32 insertions(+), 3 deletions(-)
 create mode 100644 query_classifier/test/canonical_tests/comment.expected
 create mode 100644 query_classifier/test/canonical_tests/comment.sql

diff --git a/query_classifier/test/canonical_tests/CMakeLists.txt b/query_classifier/test/canonical_tests/CMakeLists.txt
index 0092b7841..8295893ef 100644
--- a/query_classifier/test/canonical_tests/CMakeLists.txt
+++ b/query_classifier/test/canonical_tests/CMakeLists.txt
@@ -29,3 +29,10 @@ add_test(NAME Internal-CanonicalQueryAlter COMMAND ${CMAKE_CURRENT_SOURCE_DIR}/c
   ${CMAKE_CURRENT_BINARY_DIR}/alter.output
   ${CMAKE_CURRENT_SOURCE_DIR}/alter.expected
   $<TARGET_FILE:canonizer>)
+
+add_test(NAME Internal-CanonicalQueryComment COMMAND ${CMAKE_CURRENT_SOURCE_DIR}/canontest.sh
+  ${CMAKE_CURRENT_BINARY_DIR}/test.log
+  ${CMAKE_CURRENT_SOURCE_DIR}/comment.sql
+  ${CMAKE_CURRENT_BINARY_DIR}/comment.output
+  ${CMAKE_CURRENT_SOURCE_DIR}/comment.expected
+  $<TARGET_FILE:canonizer>)
diff --git a/query_classifier/test/canonical_tests/comment.expected b/query_classifier/test/canonical_tests/comment.expected
new file mode 100644
index 000000000..f88c4ffde
--- /dev/null
+++ b/query_classifier/test/canonical_tests/comment.expected
@@ -0,0 +1,11 @@
+select ?;
+select ?;
+select  ?;
+select /*! ? + */ ?;
+select /*!? ? + */ ?;
+select /*!? ? + */ ?;
+SELECT ? ;
+SELECT ? /*! +? */;
+SELECT ? /*!? +? */;
+SELECT ? /*M! +? */;
+SELECT ? /*M!? +? */;
diff --git a/query_classifier/test/canonical_tests/comment.sql b/query_classifier/test/canonical_tests/comment.sql
new file mode 100644
index 000000000..904752017
--- /dev/null
+++ b/query_classifier/test/canonical_tests/comment.sql
@@ -0,0 +1,11 @@
+select 1;-- comment after statement
+select 1;# comment after statement
+select /* inline comment */ 1;
+select /*! 1 + */ 1;
+select /*!300000 1 + */ 1;
+select /*!300000 1 + */ 1;
+SELECT 2 /* +1 */;
+SELECT 1 /*! +1 */;
+SELECT 1 /*!50101 +1 */;
+SELECT 2 /*M! +1 */;
+SELECT 2 /*M!50101 +1 */;
diff --git a/utils/skygw_utils.cc b/utils/skygw_utils.cc
index 77770982a..e4a7506a2 100644
--- a/utils/skygw_utils.cc
+++ b/utils/skygw_utils.cc
@@ -2034,13 +2034,13 @@ void skygw_file_close(
 #define BUFFER_GROWTH_RATE 1.2
 static pcre2_code* remove_comments_re = NULL;
 static const PCRE2_SPTR remove_comments_pattern = (PCRE2_SPTR)
-"(?:`[^`]*`\\K)|(?:#.*|--[[:space]].*)";
+"(?:`[^`]*`\\K)|(\\/[*](?!(M?!)).*?[*]\\/)|(?:#.*|--[[:space:]].*)";
 
 /**
  * Remove SQL comments from the end of a string
  *
- * The inline comments are not removed due to the fact that they can alter the
- * behavior of the query.
+ * The inline executable comments are not removed due to the fact that they can
+ * alter the behavior of the query.
  * @param src Pointer to the string to modify.
  * @param srcsize Pointer to a size_t variable which holds the length of the string to
  * be modified.