drop comments while lexing unquoted strings

Author: xabbuh

Parser.php

@@ -1158,7 +1158,18 @@ private function lexInlineQuotedString(int &$cursor = 0): string
     private function lexUnquotedString(int &$cursor): string
     {
         $offset = $cursor;
-        $cursor += strcspn($this->currentLine, '[]{},:', $cursor);
+
+        while ($cursor < strlen($this->currentLine)) {
+            if (in_array($this->currentLine[$cursor], ['[', ']', '{', '}', ',', ':'], true)) {
+                break;
+            }
+
+            if (\in_array($this->currentLine[$cursor], [' ', "\t"], true) && '#' === ($this->currentLine[$cursor + 1] ?? '')) {
+                break;
+            }
+
+            ++$cursor;
+        }
 
         if ($cursor === $offset) {
             throw new ParseException('Malformed unquoted YAML string.');
@@ -1235,7 +1246,7 @@ private function consumeWhitespaces(int &$cursor): bool
         $whitespacesConsumed = 0;
 
         do {
-            $whitespaceOnlyTokenLength = strspn($this->currentLine, ' ', $cursor);
+            $whitespaceOnlyTokenLength = strspn($this->currentLine, " \t", $cursor);
             $whitespacesConsumed += $whitespaceOnlyTokenLength;
             $cursor += $whitespaceOnlyTokenLength;
 

Tests/ParserTest.php

@@ -1751,6 +1751,69 @@ public function testParseMultiLineUnquotedString()
         $this->assertSame(['foo' => 'bar baz foobar foo', 'bar' => 'baz'], $this->parser->parse($yaml));
     }
 
+    /**
+     * @dataProvider unquotedStringWithTrailingComment
+     */
+    public function testParseMultiLineUnquotedStringWithTrailingComment(string $yaml, array $expected)
+    {
+        $this->assertSame($expected, $this->parser->parse($yaml));
+    }
+
+    public function unquotedStringWithTrailingComment()
+    {
+        return [
+            'comment after comma' => [
+                <<<'YAML'
+                {
+                    foo: 3, # comment
+                    bar: 3
+                }
+                YAML,
+                ['foo' => 3, 'bar' => 3],
+            ],
+            'comment after space' => [
+                <<<'YAML'
+                {
+                    foo: 3 # comment
+                }
+                YAML,
+                ['foo' => 3],
+            ],
+            'comment after space, but missing space after #' => [
+                <<<'YAML'
+                {
+                    foo: 3 #comment
+                }
+                YAML,
+                ['foo' => 3],
+            ],
+            'comment after tab' => [
+                <<<YAML
+                {
+                    foo: 3\t# comment
+                }
+                YAML,
+                ['foo' => 3],
+            ],
+            'comment after tab, but missing space after #' => [
+                <<<YAML
+                {
+                    foo: 3\t#comment
+                }
+                YAML,
+                ['foo' => 3],
+            ],
+            '# in mapping value' => [
+                <<<'YAML'
+                {
+                    foo: example.com/#about
+                }
+                YAML,
+                ['foo' => 'example.com/#about'],
+            ],
+        ];
+    }
+
     /**
      * @dataProvider escapedQuotationCharactersInQuotedStrings
      */