Merge pull request #17 from teichert/table-scope-col

(ADA) markdown table headers cells are column scoped

src/services/htmlMarkdownUtils.test.ts

@@ -0,0 +1,86 @@
+import { describe, it, expect } from 'vitest';
+import { markdownToHtmlNoImages } from './htmlMarkdownUtils';
+
+describe('markdownToHtmlNoImages', () => {
+  it('moves caption into table when caption immediately precedes table', () => {
+    const markdown = `<caption>My Table</caption>
+
+| Header |
+| --- |
+| Cell |`;
+    const html = markdownToHtmlNoImages(markdown);
+    // We expect the caption to be inside the table
+    expect(html).toMatch(/<table>\s*<caption>My Table<\/caption>/);
+  });
+
+  it('renders table correctly without caption', () => {
+    const markdown = `
+| Header |
+| --- |
+| Cell |`;
+    const html = markdownToHtmlNoImages(markdown);
+    expect(html).toMatch(/<table>/);
+    expect(html).not.toMatch(/<caption>/);
+    expect(html).toContain('Header');
+    expect(html).toContain('Cell');
+  });
+
+  it('moves caption with attributes into table', () => {
+    const markdown = `<caption style="color:red">My Table</caption>
+
+| Header |
+| --- |
+| Cell |`;
+    const html = markdownToHtmlNoImages(markdown);
+    expect(html).toMatch(/<table>\s*<caption style="color:red">My Table<\/caption>/);
+  });
+
+  it('adds scope="col" to table headers', () => {
+    const markdown = `
+| Header 1 | Header 2 |
+| --- | --- |
+| Cell 1 | Cell 2 |`;
+    const html = markdownToHtmlNoImages(markdown);
+    expect(html).toContain('<th scope="col">Header 1</th>');
+    expect(html).toContain('<th scope="col">Header 2</th>');
+  });
+
+  it('does not add an extra empty header row', () => {
+    const markdown = `
+| Header |
+| --- |
+| Cell |`;
+    const html = markdownToHtmlNoImages(markdown);
+    expect(html).not.toContain('<th scope="col"></th>');
+    const thCount = (html.match(/<th scope="col"/g) || []).length;
+    expect(thCount).toBe(1);
+  });
+
+  it('does not add scope="col" to raw HTML tables', () => {
+    const markdown = `
+<table>
+  <thead>
+    <tr>
+      <th>Raw Header</th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+      <td>Raw Cell</td>
+    </tr>
+  </tbody>
+</table>
+
+| MD Header |
+| --- |
+| MD Cell |`;
+    const html = markdownToHtmlNoImages(markdown);
+    
+    // Raw table should be untouched (or at least not have scope="col" added if it wasn't there)
+    expect(html).toContain('<th>Raw Header</th>');
+    expect(html).not.toContain('<th scope="col">Raw Header</th>');
+
+    // Markdown table should have scope="col"
+    expect(html).toContain('<th scope="col">MD Header</th>');
+  });
+});

src/services/htmlMarkdownUtils.ts

@@ -47,6 +47,20 @@ marked.use(
 
 marked.use({ extensions: [mermaidExtension] });
 
+// We use a custom renderer instead of a regex replace because regex is too aggressive.
+// It would add scope="col" to raw HTML tables (which we want to leave alone).
+// The renderer only applies to markdown tables.
+marked.use({
+  renderer: {
+    tablecell({ text, header, align }) {
+      const type = header ? "th" : "td";
+      const alignAttr = align ? ` align="${align}"` : "";
+      const scopeAttr = header ? ' scope="col"' : "";
+      return `<${type}${scopeAttr}${alignAttr}>${text}</${type}>\n`;
+    },
+  },
+});
+
 export function extractImageSources(htmlString: string) {
   const srcUrls = [];
   const regex = /<img[^>]+src=["']?([^"'>]+)["']?/g;
@@ -125,8 +139,21 @@ export function markdownToHTMLSafe({
 }
 
 export function markdownToHtmlNoImages(markdownString: string) {
-  const clean = DOMPurify.sanitize(
+  const parsedHtml = marked.parse(markdownString, {
-    marked.parse(markdownString, { async: false, pedantic: false, gfm: true })
+    async: false,
-  ).replaceAll(/>[^<>]*<\/math>/g, "></math>");
+    pedantic: false,
+    gfm: true,
+  }) as string;
+
+  // Move caption inside table
+  const htmlWithCaptionInTable = parsedHtml.replace(
+    /(<caption[^>]*>[\s\S]*?<\/caption>)\s*(<table[^>]*>)/g,
+    "$2$1"
+  );
+
+  const clean = DOMPurify.sanitize(htmlWithCaptionInTable).replaceAll(
+    />[^<>]*<\/math>/g,
+    "></math>"
+  );
   return clean;
 }