Merge pull request #146 from tomek12311/fix/question-marks-are-not-pa…

…rsed-correctly FIX: Question marks are not parsed correctly
php-edifact · Jan 8, 2025 · d59afa4 · d59afa4
2 parents 385dd28 + 55eb38a
commit d59afa4
Showing 2 changed files with 6 additions and 4 deletions.
diff --git a/src/EDI/Parser.php b/src/EDI/Parser.php
@@ -522,10 +522,10 @@ private function unwrap(string &$string): array
             $this->errors[] = 'This file contains some segments without terminators';
         }
 
-        $terminatorRegex = '/(([^'.$this->symbRel.']'.$this->symbRel.'{2})+|[^'.$this->symbRel.'])'.$this->symbEnd.'|[\r\n]+/';
+        $terminatorRegex = '/((?<!'.$this->symbRel.')(?:'.$this->symbRel.$this->symbRel.')*)'.$this->symbEnd.'|[\r\n]+/';
 
         if ($this->strict) {
-            $terminatorRegex = '/(([^'.$this->symbRel.']'.$this->symbRel.'{2})+|[^'.$this->symbRel.'])'.$this->symbEnd.'/';
+            $terminatorRegex = '/((?<!'.$this->symbRel.')(?:'.$this->symbRel.$this->symbRel.')*)'.$this->symbEnd.'/';
         }
 
         $string = (string) \preg_replace(

diff --git a/tests/EDITest/ParserTest.php b/tests/EDITest/ParserTest.php
@@ -90,8 +90,10 @@ public static function multipleEscapedSegmentsProvider(): array
             ["EQD+CX??????DU12?+3456+2?:0'",  [['EQD', 'CX???DU12+3456', '2:0']]],
             ["EQD+CX????????DU12?+3456+2?:0'", [['EQD', 'CX????DU12+3456', '2:0']]],
             ["EQD+CX??DU12?+3456+2?:0??'",    [['EQD', 'CX?DU12+3456', '2:0?']]],
-            ["EQD+CX??DU12?+3456+2?:0????'",  [['EQD', 'CX?DU12+3456', "2:0??'"]]], // TODO? -> ' is correct here?
-            ["EQD+CX??DU12?+3456+2?:0??????'", [['EQD', 'CX?DU12+3456', "2:0???'"]]], // TODO? -> ' is correct here?
+            ["EQD+CX??DU12?+3456+2?:0???'",    [['EQD', 'CX?DU12+3456', "2:0?'"]]],
+            ["EQD+CX??DU12?+3456+2?:0????'",  [['EQD', 'CX?DU12+3456', "2:0??"]]],
+            ["EQD+CX??DU12?+3456+2?:0??????'", [['EQD', 'CX?DU12+3456', "2:0???"]]],
+            ["EQD+CX??DU12?+3456+2?:0???????'", [['EQD', 'CX?DU12+3456', "2:0???'"]]],
             ["??EQD+CX??DU12?+3456+2?:0'",    [['?EQD', 'CX?DU12+3456', '2:0']]],
             ["????EQD+CX??DU12?+3456+2?:0'",  [['??EQD', 'CX?DU12+3456', '2:0']]],
             ["??????EQD+CX??DU12?+3456+2?:0'", [['???EQD', 'CX?DU12+3456', '2:0']]],