Fix basic binary LIKE/REGEXP for PostgreSQL

atk4 · Apr 29, 2024 · e741a8c · e741a8c
1 parent 77fc8e2
commit e741a8c
Show file tree

Hide file tree

Showing 6 changed files with 39 additions and 43 deletions.
diff --git a/src/Persistence/Sql/Expression.php b/src/Persistence/Sql/Expression.php
@@ -384,7 +384,7 @@ public function getDebugQuery(): string
 
         $i = 0;
         $sql = preg_replace_callback(
-            '~' . self::QUOTED_TOKEN_REGEX . '\K|(?:\?|:\w+)~',
+            '~' . self::QUOTED_TOKEN_REGEX . '\K|(?:\?|(?<!:):\w+)~',
             function ($matches) use ($params, &$i) {
                 if ($matches[0] === '') {
                     return '';

diff --git a/src/Persistence/Sql/Postgresql/ExpressionTrait.php b/src/Persistence/Sql/Postgresql/ExpressionTrait.php
@@ -62,7 +62,7 @@ protected function updateRenderBeforeExecute(array $render): array
         [$sql, $params] = parent::updateRenderBeforeExecute($render);
 
         $sql = preg_replace_callback(
-            '~' . self::QUOTED_TOKEN_REGEX . '\K|:\w+~',
+            '~' . self::QUOTED_TOKEN_REGEX . '\K|(?<!:):\w+~',
             static function ($matches) use ($params) {
                 if ($matches[0] === '') {
                     return '';

diff --git a/src/Persistence/Sql/Postgresql/Query.php b/src/Persistence/Sql/Postgresql/Query.php
@@ -19,39 +19,45 @@ class Query extends BaseQuery
     protected string $templateUpdate = 'update [table][join] set [set] [where]';
     protected string $templateReplace;
 
-    #[\Override]
-    protected function _renderConditionLikeOperator(bool $negated, string $sqlLeft, string $sqlRight): string
+    /**
+     * @param \Closure(string, string): string $makeSqlFx
+     */
+    private function _renderConditionConditionalCastToText(string $sqlLeft, string $sqlRight, \Closure $makeSqlFx): string
     {
-        $sqlRightEscaped = 'regexp_replace(' . $sqlRight . ', '
-            . $this->escapeStringLiteral('(\\\[\\\_%])|(\\\)') . ', '
-            . $this->escapeStringLiteral('\1\2\2') . ', '
-            . $this->escapeStringLiteral('g') . ')';
-
-        return $sqlLeft . ($negated ? ' not' : '') . ' like ' . $sqlRightEscaped
-            . ' escape ' . $this->escapeStringLiteral('\\');
+        return $this->_renderConditionBinaryReuse(
+            $sqlLeft,
+            $sqlRight,
+            function ($sqlLeft, $sqlRight) use ($makeSqlFx) {
+                return 'case when pg_typeof(' . $sqlLeft . ') = ' . $this->escapeStringLiteral('bytea') . '::regtype'
+                    . ' then ' . $makeSqlFx('convert_from(cast(cast(' . $sqlLeft . ' as text) as bytea), '
+                    . $this->escapeStringLiteral('UTF8') . ')', $sqlRight) // will raise SQL error for 0x00 or 0x80+ bytes
+                    . ' else ' . $makeSqlFx('cast(' . $sqlLeft . ' as citext)', $sqlRight)
+                    . ' end';
+            }
+        );
     }
 
-    // needed for PostgreSQL v14 and lower
     #[\Override]
-    protected function _renderConditionRegexpOperator(bool $negated, string $sqlLeft, string $sqlRight, bool $binary = false): string
+    protected function _renderConditionLikeOperator(bool $negated, string $sqlLeft, string $sqlRight): string
     {
-        return $sqlLeft . ' ' . ($negated ? '!' : '') . '~ ' . $sqlRight;
+        return $this->_renderConditionConditionalCastToText($sqlLeft, $sqlRight, function ($sqlLeft, $sqlRight) use ($negated) {
+            $sqlRightEscaped = 'regexp_replace(' . $sqlRight . ', '
+                . $this->escapeStringLiteral('(\\\[\\\_%])|(\\\)') . ', '
+                . $this->escapeStringLiteral('\1\2\2') . ', '
+                . $this->escapeStringLiteral('g') . ')';
+
+            return $sqlLeft . ($negated ? ' not' : '') . ' like ' . $sqlRightEscaped
+                . ' escape ' . $this->escapeStringLiteral('\\');
+        });
     }
 
+    // needed for PostgreSQL v14 and lower
     #[\Override]
-    protected function _subrenderCondition(array $row): string
+    protected function _renderConditionRegexpOperator(bool $negated, string $sqlLeft, string $sqlRight, bool $binary = false): string
     {
-        if (count($row) !== 1) {
-            [$field, $operator, $value] = $row;
-
-            if (in_array(strtolower($operator ?? '='), ['like', 'not like', 'regexp', 'not regexp'], true)) {
-                $field = new Expression('CAST([] AS citext)', [$field]);
-
-                $row = [$field, $operator, $value];
-            }
-        }
-
-        return parent::_subrenderCondition($row);
+        return $this->_renderConditionConditionalCastToText($sqlLeft, $sqlRight, static function ($sqlLeft, $sqlRight) use ($negated) {
+            return $sqlLeft . ' ' . ($negated ? '!' : '') . '~ ' . $sqlRight;
+        });
     }
 
     #[\Override]

diff --git a/tests/ConditionSqlTest.php b/tests/ConditionSqlTest.php
@@ -544,12 +544,6 @@ public function testLikeCondition(string $type, bool $isBinary): void
             return $res;
         };
 
-        if ($this->getDatabasePlatform() instanceof PostgreSQLPlatform && $isBinary) {
-            self::assertTrue(true); // @phpstan-ignore-line
-
-            return; // TODO
-        }
-
         if ($this->getDatabasePlatform() instanceof SQLServerPlatform && $isBinary) {
             self::assertTrue(true); // @phpstan-ignore-line
 
@@ -683,12 +677,6 @@ public function testRegexpCondition(string $type, bool $isBinary): void
             self::markTestIncomplete('MSSQL has no REGEXP support yet');
         }
 
-        if ($this->getDatabasePlatform() instanceof PostgreSQLPlatform && $isBinary) {
-            self::assertTrue(true); // @phpstan-ignore-line
-
-            return; // TODO
-        }
-
         if ($this->getDatabasePlatform() instanceof OraclePlatform && $isBinary) {
             $this->expectException(Exception::class);
             $this->expectExceptionMessage('Unsupported binary field operator');
@@ -703,7 +691,8 @@ public function testRegexpCondition(string $type, bool $isBinary): void
 
         self::assertSame([1], $findIdsRegexFx('name', 'John'));
         self::assertSame($isBinary ? [] : [1], $findIdsRegexFx('name', 'john'));
-        self::assertSame($this->getDatabasePlatform() instanceof MySQLPlatform && $isBinary && !$isMysql5x && !$isMariadb ? [] : [13], $findIdsRegexFx('name', 'heiß')); // TODO investigate/report MySQL 8.x bug
+        // TODO investigate/report MySQL 8.x bug
+        self::assertSame($this->getDatabasePlatform() instanceof MySQLPlatform && $isBinary && !$isMysql5x && !$isMariadb ? [] : [13], $findIdsRegexFx('name', 'heiß'));
         self::assertSame($isBinary ? [] : [13], $findIdsRegexFx('name', 'Heiß'));
         self::assertSame([1], $findIdsRegexFx('name', 'Joh'));
         self::assertSame([1], $findIdsRegexFx('name', 'ohn'));
@@ -776,7 +765,8 @@ public function testRegexpCondition(string $type, bool $isBinary): void
             self::assertSame([5, 6], $findIdsRegexFx('name', 'Sa\s'));
             self::assertSame([7, 8, 9, 10, 11, 12], $findIdsRegexFx('name', 'Sa\S'));
             self::assertSame([1, 3], $findIdsRegexFx('name', '\wo'));
-            self::assertSame($this->getDatabasePlatform() instanceof MySQLPlatform && $isBinary ? [] : [13], $findIdsRegexFx('name', 'hei\w$')); // TODO align SQLite with MySQL
+            // TODO align SQLite binary behaviour with MySQL
+            self::assertSame($isBinary && ($this->getDatabasePlatform() instanceof MySQLPlatform || $this->getDatabasePlatform() instanceof PostgreSQLPlatform) ? [] : [13], $findIdsRegexFx('name', 'hei\w$'));
             self::assertSame([10], $findIdsRegexFx('name', '\W\\\\'));
             if ($type !== 'string' && !$this->getDatabasePlatform() instanceof OraclePlatform) {
                 self::assertSame([5], $findIdsRegexFx('name', '\x20'));

diff --git a/tests/Persistence/Sql/QueryTest.php b/tests/Persistence/Sql/QueryTest.php
@@ -881,7 +881,7 @@ public function testWhereSpecialValues(): void
         }
         self::assertSame(
             <<<'EOF'
-                where CAST(:a AS citext) like regexp_replace(:b, '(\\[\\_%])|(\\)', '\1\2\2', 'g') escape chr(92)
+                where case when pg_typeof("name") = 'bytea'::regtype then convert_from(cast(cast("name" as text) as bytea), 'UTF8') like regexp_replace(:a, '(\\[\\_%])|(\\)', '\1\2\2', 'g') escape chr(92) else cast("name" as citext) like regexp_replace(:a, '(\\[\\_%])|(\\)', '\1\2\2', 'g') escape chr(92) end
                 EOF,
             (new PostgresqlQuery('[where]'))->where('name', 'like', 'foo')->render()[0]
         );
@@ -927,7 +927,7 @@ public function testWhereSpecialValues(): void
         }
         self::assertSame(
             <<<'EOF'
-                where CAST(:a AS citext) ~ :b
+                where case when pg_typeof("name") = 'bytea'::regtype then convert_from(cast(cast("name" as text) as bytea), 'UTF8') ~ :a else cast("name" as citext) ~ :a end
                 EOF,
             (new PostgresqlQuery('[where]'))->where('name', 'regexp', 'foo')->render()[0]
         );

diff --git a/tests/Schema/MigratorTest.php b/tests/Schema/MigratorTest.php
@@ -119,7 +119,7 @@ public function testCharacterTypeFieldCaseSensitivity(string $type, bool $isBina
             $model->addCondition('v', 'in', ['MixedCase', 'foo']);
             self::assertSameExportUnordered($expectedExport, $model->export(['id']));
 
-            if (!$this->getDatabasePlatform() instanceof PostgreSQLPlatform && !$this->getDatabasePlatform() instanceof SQLServerPlatform && !$this->getDatabasePlatform() instanceof OraclePlatform) {
+            if (!$this->getDatabasePlatform() instanceof SQLServerPlatform && !$this->getDatabasePlatform() instanceof OraclePlatform) {
                 $model->scope()->clear();
                 $model->addCondition('v', 'like', 'MixedCase');
                 self::assertSameExportUnordered($expectedExport, $model->export(['id']));