From 346cdaad72aa9a71ad96b444139220c2508ce1f3 Mon Sep 17 00:00:00 2001
From: Clemens Schwaighofer <clemens.schwaighofer@egplusww.com>
Date: Wed, 11 Dec 2024 11:23:19 +0900
Subject: [PATCH] Fix for params regex comment update

---
 src/DB/IO.php                         |  1 -
 src/DB/Support/ConvertPlaceholder.php | 17 +++++++++-----
 test/phpunit/DB/CoreLibsDBIOTest.php  | 32 +++++++++++++++++++++++++--
 3 files changed, 42 insertions(+), 8 deletions(-)

diff --git a/src/DB/IO.php b/src/DB/IO.php
index d579fd1..4daad5b 100644
--- a/src/DB/IO.php
+++ b/src/DB/IO.php
@@ -1332,7 +1332,6 @@ class IO
 	 */
 	private function __dbCheckQueryParams(string $query, array $params): bool
 	{
-		// $this->log->debug('DB QUERY PARAMS REGEX', ConvertPlaceholder::REGEX_LOOKUP_PLACEHOLDERS);
 		$placeholder_count = $this->__dbCountQueryParams($query);
 		$params_count = count($params);
 		if ($params_count != $placeholder_count) {
diff --git a/src/DB/Support/ConvertPlaceholder.php b/src/DB/Support/ConvertPlaceholder.php
index fd8eea6..dff0378 100644
--- a/src/DB/Support/ConvertPlaceholder.php
+++ b/src/DB/Support/ConvertPlaceholder.php
@@ -18,17 +18,20 @@ class ConvertPlaceholder
 	// NOTE some combinations are allowed, but the query will fail before this
 	/** @var string split regex, entries before $ group */
 	private const PATTERN_QUERY_SPLIT =
-		',|' // for ',' mostly in INSERT
-		. '[(<>=]|' // general set for (, <, >, = in any query with any combination
-		. '(?:[\(,]\s*\-\-[\s\w]*)\r?\n|' // a comment that starts after a ( or ,
+		'\?\?|' // UNKNOWN: double ??, is this to avoid something?
+		. '[\(,]|' // for ',' and '(' mostly in INSERT or ANY()
+		. '[<>=]|' // general set for <, >, = in any query with any combination
 		. '\^@|' // text search for start from text with ^@
 		. '\|\||' // concats two elements
 		. '&&|' // array overlap
-		. '\-\|\-|' // range overlap
+		. '\-\|\-|' // range overlap for array
 		. '[^-]-{1}|' // single -, used in JSON too
 		. '->|->>|#>|#>>|@>|<@|@@|@\?|\?{1}|\?\||\?&|#-'; //JSON searches, Array searchs, etc
 	/** @var string the main regex including  the pattern query split */
-	private const PATTERN_ELEMENT = '(?:\'.*?\')?\s*(?:\?\?|' . self::PATTERN_QUERY_SPLIT . ')\s*';
+	private const PATTERN_ELEMENT = '(?:\'.*?\')?\s*(?:' . self::PATTERN_QUERY_SPLIT . ')\s*';
+	/** @var string comment regex
+	 * anything that starts with -- and ends with a line break but any character that is not line break inbetween */
+	private const PATTERN_COMMENT = '(?:\-\-[^\r\n]*?\r?\n)*\s*';
 	/** @var string parts to ignore in the SQL */
 	private const PATTERN_IGNORE =
 		// digit -> ignore
@@ -45,6 +48,7 @@ class ConvertPlaceholder
 	/** @var string replace regex for named (:...) entries */
 	public const REGEX_REPLACE_NAMED = '/'
 		. '(' . self::PATTERN_ELEMENT . ')'
+		. self::PATTERN_COMMENT
 		. '('
 		. self::PATTERN_IGNORE
 		. self::PATTERN_NAMED
@@ -53,6 +57,7 @@ class ConvertPlaceholder
 	/** @var string replace regex for question mark (?) entries */
 	public const REGEX_REPLACE_QUESTION_MARK = '/'
 		. '(' . self::PATTERN_ELEMENT . ')'
+		. self::PATTERN_COMMENT
 		. '('
 		. self::PATTERN_IGNORE
 		. self::PATTERN_QUESTION_MARK
@@ -61,6 +66,7 @@ class ConvertPlaceholder
 	/** @var string replace regex for numbered ($n) entries */
 	public const REGEX_REPLACE_NUMBERED = '/'
 		. '(' . self::PATTERN_ELEMENT . ')'
+		. self::PATTERN_COMMENT
 		. '('
 		. self::PATTERN_IGNORE
 		. self::PATTERN_NUMBERED
@@ -71,6 +77,7 @@ class ConvertPlaceholder
 		// prefix string part, must match towards
 		// seperator for ( = , ? - [and json/jsonb in pg doc section 9.15]
 		. self::PATTERN_ELEMENT
+		. self::PATTERN_COMMENT
 		// match for replace part
 		. '(?:'
 		// ignore parts
diff --git a/test/phpunit/DB/CoreLibsDBIOTest.php b/test/phpunit/DB/CoreLibsDBIOTest.php
index faadf6e..1476729 100644
--- a/test/phpunit/DB/CoreLibsDBIOTest.php
+++ b/test/phpunit/DB/CoreLibsDBIOTest.php
@@ -5141,9 +5141,9 @@ final class CoreLibsDBIOTest extends TestCase
 				INSERT INTO table_with_primary_key (
 					row_int, row_numeric, row_varchar, row_varchar_literal
 				) VALUES (
-					-- comment 1
+					-- comment 1 かな
 					$1, $2,
-					-- comment 2
+					-- comment 2 -
 					$3
 					-- comment 3
 					, $4
@@ -5152,6 +5152,23 @@ final class CoreLibsDBIOTest extends TestCase
 				'count' => 4,
 				'convert' => false
 			],
+			'comment in update' => [
+				'query' => <<<SQL
+				UPDATE table_with_primary_key SET
+					row_int =
+					-- COMMENT 1
+					$1,
+					row_numeric =
+					$2 -- COMMENT 2
+					,
+					row_varchar -- COMMENT 3
+					= $3
+				WHERE
+					row_varchar = $4
+				SQL,
+				'count' => 4,
+				'convert' => false,
+			],
 			// Note some are not set
 			'a complete set of possible' => [
 				'query' => <<<SQL
@@ -5168,6 +5185,17 @@ final class CoreLibsDBIOTest extends TestCase
 				SQL,
 				'count' => 12,
 				'convert' => false,
+			],
+			// all the same
+			'all the same numbered' => [
+				'query' => <<<SQL
+				UPDATE table_with_primary_key SET
+					row_int = $1::INT, row_numeric = $1::NUMERIC, row_varchar = $1
+				WHERE
+					row_varchar = $1
+				SQL,
+				'count' => 1,
+				'convert' => false,
 			]
 		];
 	}