Add HTML data and ARIA attribute symbols

changelog.d/unreleased/2106.added.md

@@ -0,0 +1,16 @@
+---
+category: added
+issues:
+  - 2106
+affected:
+  - src/CodeIndex/Indexer/Symbols/SymbolExtractor.Markup.cs
+  - tests/CodeIndex.Tests/SymbolExtractorTests.cs
+---
+
+## English
+
+- **HTML `data-*` and `aria-*` attributes are now indexed as property symbols (#2106)** — HTML symbol extraction now records semantic attribute names such as `data-testid` and `aria-label`, including boolean and multiline attributes, without reading attribute-like text inside quoted values.
+
+## 日本語
+
+- **HTML の `data-*` / `aria-*` 属性を property シンボルとして索引するようになりました (#2106)** — HTML のシンボル抽出が `data-testid` や `aria-label` などの意味を持つ属性名を記録するようになり、boolean 属性や複数行属性にも対応しつつ、引用符付き値の中にある属性風テキストは拾いません。

src/CodeIndex/Indexer/Symbols/SymbolExtractor.Markup.cs

@@ -186,6 +186,22 @@ private static List<SymbolRecord> ExtractHtmlSymbols(long fileId, string[] lines
                     }
                 }
 
+                if (IsHtmlSemanticStateAttributeName(attrNameLower))
+                {
+                    var attrStartLine = FindHtmlLineNumber(lineStarts, attrNameStart);
+                    var attrSignatureIndex = Math.Clamp(attrStartLine - 1, 0, lines.Length - 1);
+                    symbols.Add(new SymbolRecord
+                    {
+                        FileId = fileId,
+                        Kind = "property",
+                        Name = attrNameLower,
+                        Line = attrStartLine,
+                        StartLine = attrStartLine,
+                        EndLine = attrStartLine,
+                        Signature = lines[attrSignatureIndex].Trim(),
+                    });
+                }
+
                 if (attrValue == null || attrValue.Length == 0)
                     continue;
 
@@ -256,6 +272,13 @@ private static List<SymbolRecord> ExtractHtmlSymbols(long fileId, string[] lines
         return symbols;
     }
 
+    private static bool IsHtmlSemanticStateAttributeName(string attrNameLower)
+    {
+        return (attrNameLower.StartsWith("data-", StringComparison.Ordinal) ||
+                attrNameLower.StartsWith("aria-", StringComparison.Ordinal)) &&
+               attrNameLower.Length > 5;
+    }
+
     private static List<SymbolRecord> ExtractMarkdownSymbols(long fileId, string[] lines)
     {
         // Markdown headings are the closest thing to navigable symbols in docs files.

tests/CodeIndex.Tests/SymbolExtractorTests.cs

@@ -22226,6 +22226,30 @@ public void Extract_Html_IgnoresDataIdAndAriaIdAndXmlIdAttributes()
         Assert.DoesNotContain(symbols, s => s.Kind == "property" && s.Name == "ns");
     }
 
+    [Fact]
+    public void Extract_Html_CapturesDataAndAriaAttributeNamesAsProperties()
+    {
+        var content = """
+            <button DATA-TestId="save-button" data-user-id=42 aria-label="Save" aria-expanded></button>
+            <section
+              data-panel-state="open"
+              aria-labelledby='panel-title'></section>
+            <div title="data-fake=&quot;nope&quot; aria-hidden=&quot;true&quot;" id="real"></div>
+            """;
+
+        var symbols = SymbolExtractor.Extract(1, "html", content);
+
+        Assert.Contains(symbols, s => s.Kind == "property" && s.Name == "data-testid" && s.Line == 1);
+        Assert.Contains(symbols, s => s.Kind == "property" && s.Name == "data-user-id" && s.Line == 1);
+        Assert.Contains(symbols, s => s.Kind == "property" && s.Name == "aria-label" && s.Line == 1);
+        Assert.Contains(symbols, s => s.Kind == "property" && s.Name == "aria-expanded" && s.Line == 1);
+        Assert.Contains(symbols, s => s.Kind == "property" && s.Name == "data-panel-state" && s.Line == 3);
+        Assert.Contains(symbols, s => s.Kind == "property" && s.Name == "aria-labelledby" && s.Line == 4);
+        Assert.DoesNotContain(symbols, s => s.Kind == "property" && s.Name == "data-fake");
+        Assert.DoesNotContain(symbols, s => s.Kind == "property" && s.Name == "aria-hidden");
+        Assert.Contains(symbols, s => s.Kind == "property" && s.Name == "real");
+    }
+
     [Fact]
     public void Extract_Html_CapturesExternalScriptAndLinkAsImports()
     {