[Yaml] do not remove "comments" in scalar blocks

xabbuh · xabbuh · commit 3c72fccc1b31 · 2015-12-05T14:17:16.000+01:00
Inside scalar blocks, lines starting with a `#` character must be
treated like every other strings and must not be ignored as comments.
diff --git a/src/Symfony/Component/Yaml/Parser.php b/src/Symfony/Component/Yaml/Parser.php
@@ -303,6 +303,7 @@ private function getCurrentLineIndentation()
     private function getNextEmbedBlock($indentation = null, $inSequence = false)
     {
         $oldLineIndentation = $this->getCurrentLineIndentation();
+        $insideBlockScalar = $this->isBlockScalarHeader();
 
         if (!$this->moveToNextLine()) {
             return;
@@ -339,17 +340,21 @@ private function getNextEmbedBlock($indentation = null, $inSequence = false)
 
         $isItUnindentedCollection = $this->isStringUnIndentedCollectionItem();
 
-        // Comments must not be removed inside a block scalar
-        $removeCommentsPattern = '~'.self::BLOCK_SCALAR_HEADER_PATTERN.'$~';
-        $removeComments = !preg_match($removeCommentsPattern, $this->currentLine);
+        if (!$insideBlockScalar) {
+            $insideBlockScalar = $this->isBlockScalarHeader();
+        }
+
+        $previousLineIndentation = $this->getCurrentLineIndentation();
 
         while ($this->moveToNextLine()) {
             $indent = $this->getCurrentLineIndentation();
 
-            if ($indent === $newIndent) {
-                $removeComments = !preg_match($removeCommentsPattern, $this->currentLine);
+            if (!$insideBlockScalar && $indent === $previousLineIndentation) {
+                $insideBlockScalar = $this->isBlockScalarHeader();
             }
 
+            $previousLineIndentation = $indent;
+
             if ($isItUnindentedCollection && !$this->isStringUnIndentedCollectionItem() && $newIndent === $indent) {
                 $this->moveToPreviousLine();
                 break;
@@ -360,7 +365,8 @@ private function getNextEmbedBlock($indentation = null, $inSequence = false)
                 continue;
             }
 
-            if ($removeComments && $this->isCurrentLineComment()) {
+            // we ignore "comment" lines only when we are not inside a scalar block
+            if (!$insideBlockScalar && $this->isCurrentLineComment()) {
                 continue;
             }
 
@@ -672,4 +678,14 @@ private function isStringUnIndentedCollectionItem()
     {
         return 0 === strpos($this->currentLine, '- ');
     }
+
+    /**
+     * Tests whether or not the current line is the header of a block scalar.
+     *
+     * @return bool
+     */
+    private function isBlockScalarHeader()
+    {
+        return (bool) preg_match('~'.self::BLOCK_SCALAR_HEADER_PATTERN.'$~', $this->currentLine);
+    }
 }
diff --git a/src/Symfony/Component/Yaml/Tests/ParserTest.php b/src/Symfony/Component/Yaml/Tests/ParserTest.php
@@ -726,6 +726,100 @@ public function testFloatKeys()
 
         $this->assertEquals($expected, $this->parser->parse($yaml));
     }
+
+    /**
+     * @dataProvider getCommentLikeStringInScalarBlockData
+     */
+    public function testCommentLikeStringsAreNotStrippedInBlockScalars($yaml, $expectedParserResult)
+    {
+        $this->assertSame($expectedParserResult, $this->parser->parse($yaml));
+    }
+
+    public function getCommentLikeStringInScalarBlockData()
+    {
+        $yaml1 = <<<EOT
+pages:
+    -
+        title: some title
+        content: |
+            # comment 1
+            header
+
+                # comment 2
+                <body>
+                    <h1>title</h1>
+                </body>
+
+            footer # comment3
+EOT;
+        $expected1 = array(
+            'pages' => array(
+                array(
+                    'title' => 'some title',
+                    'content' => <<<EOT
+# comment 1
+header
+
+    # comment 2
+    <body>
+        <h1>title</h1>
+    </body>
+
+footer # comment3
+EOT
+                    ,
+                ),
+            ),
+        );
+
+        $yaml2 = <<<EOT
+test: |
+    foo
+    # bar
+    baz
+collection:
+    - one: |
+        foo
+        # bar
+        baz
+    - two: |
+        foo
+        # bar
+        baz
+EOT;
+        $expected2 = array(
+            'test' => <<<EOT
+foo
+# bar
+baz
+
+EOT
+            ,
+            'collection' => array(
+                array(
+                    'one' => <<<EOT
+foo
+# bar
+baz
+EOT
+                    ,
+                ),
+                array(
+                    'two' => <<<EOT
+foo
+# bar
+baz
+EOT
+                    ,
+                ),
+            ),
+        );
+
+        return array(
+            array($yaml1, $expected1),
+            array($yaml2, $expected2),
+        );
+    }
 }
 
 class B
