[markdown] Preserve metadata passed to fenced code blocks

pkgs/markdown/CHANGELOG.md

@@ -1,5 +1,7 @@
 ## 7.3.1-wip
 
+* Preserve metadata passed to fenced code blocks as
+  `data-metadata` on the created `pre` element.
 * Update the README link to the markdown playground
   (https://dart-lang.github.io/tools).
 * Update `package:web` API references in the example.

pkgs/markdown/lib/src/block_syntaxes/fenced_code_block_syntax.dart

@@ -21,9 +21,9 @@ class FencedCodeBlockSyntax extends BlockSyntax {
 
   @override
   Node parse(BlockParser parser) {
-    final openingFence = _FenceMatch.fromMatch(pattern.firstMatch(
-      escapePunctuation(parser.current.content),
-    )!);
+    final openingFence = _FenceMatch.fromMatch(
+      pattern.firstMatch(escapePunctuation(parser.current.content))!,
+    );
 
     var text = parseChildLines(
       parser,
@@ -38,16 +38,27 @@ class FencedCodeBlockSyntax extends BlockSyntax {
       text = '$text\n';
     }
 
+    final (languageString, metadataString) = openingFence.languageAndMetadata;
+
     final code = Element.text('code', text);
-    if (openingFence.hasLanguage) {
-      var language = decodeHtmlCharacters(openingFence.language);
+    if (languageString != null) {
+      var language = decodeHtmlCharacters(languageString);
       if (parser.document.encodeHtml) {
         language = escapeHtmlAttribute(language);
       }
       code.attributes['class'] = 'language-$language';
     }
 
-    return Element('pre', [code]);
+    final pre = Element('pre', [code]);
+    if (metadataString != null) {
+      var metadata = decodeHtmlCharacters(metadataString);
+      if (parser.document.encodeHtml) {
+        metadata = escapeHtmlAttribute(metadata);
+      }
+      pre.attributes['data-metadata'] = metadata;
+    }
+
+    return pre;
   }
 
   String _removeIndentation(String content, int length) {
@@ -130,12 +141,30 @@ class _FenceMatch {
   // https://spec.commonmark.org/0.30/#info-string.
   final String info;
 
-  // The first word of the info string is typically used to specify the language
-  // of the code sample,
-  // https://spec.commonmark.org/0.30/#example-143.
-  String get language => info.split(' ').first;
+  /// Returns the language and remaining metadata from the [info] string.
+  ///
+  /// The language is the first word of the info string,
+  /// to match the (unspecified, but typical) behavior of CommonMark parsers,
+  /// as suggested in https://spec.commonmark.org/0.30/#example-143.
+  ///
+  /// The metadata is any remaining part of the info string after the language.
+  (String? language, String? metadata) get languageAndMetadata {
+    if (info.isEmpty) {
+      return (null, null);
+    }
 
-  bool get hasInfo => info.isNotEmpty;
+    // We assume the info string is trimmed already.
+    final firstSpaceIndex = info.indexOf(' ');
+    if (firstSpaceIndex == -1) {
+      // If there is no space, the whole string is the language.
+      return (info, null);
+    }
+
+    return (
+      info.substring(0, firstSpaceIndex),
+      info.substring(firstSpaceIndex + 1),
+    );
+  }
 
-  bool get hasLanguage => language.isNotEmpty;
+  bool get hasInfo => info.isNotEmpty;
 }

pkgs/markdown/test/common_mark/fenced_code_blocks.unit

@@ -214,7 +214,7 @@ def foo(x)
 end
 ~~~~~~~
 <<<
-<pre><code class="language-ruby">def foo(x)
+<pre data-metadata="startline=3 $%@#$"><code class="language-ruby">def foo(x)
   return 3
 end
 </code></pre>
@@ -234,7 +234,7 @@ foo</p>
 foo
 ~~~
 <<<
-<pre><code class="language-aa">foo
+<pre data-metadata="``` ~~~"><code class="language-aa">foo
 </code></pre>
 >>> Fenced code blocks - 147
 ```

pkgs/markdown/test/gfm/fenced_code_blocks.unit

@@ -214,7 +214,7 @@ def foo(x)
 end
 ~~~~~~~
 <<<
-<pre><code class="language-ruby">def foo(x)
+<pre data-metadata="startline=3 $%@#$"><code class="language-ruby">def foo(x)
   return 3
 end
 </code></pre>
@@ -234,7 +234,7 @@ foo</p>
 foo
 ~~~
 <<<
-<pre><code class="language-aa">foo
+<pre data-metadata="``` ~~~"><code class="language-aa">foo
 </code></pre>
 >>> Fenced code blocks - 117
 ```

pkgs/markdown/test/original/fenced_code_block.unit

@@ -5,3 +5,27 @@
 <<<
 <pre><code>'foo'
 </code></pre>
+>>> with basic metadata string
+```dart meta
+code
+```
+
+<<<
+<pre data-metadata="meta"><code class="language-dart">code
+</code></pre>
+>>> with characters to escape
+```dart title="main.dart"
+code
+```
+
+<<<
+<pre data-metadata="title=&quot;main.dart&quot;"><code class="language-dart">code
+</code></pre>
+>>> with HTML character reference
+```dart &#124;
+code
+```
+
+<<<
+<pre data-metadata="|"><code class="language-dart">code
+</code></pre>