fix(cards): parse HTML release-feed tables before generating Bluefin cards

scripts/generate-card-images.test.js

@@ -53,6 +53,70 @@ test("parseFeedItem extracts markdown package, diff, and commit data", async ()
   });
 });
 
+test("parseFeedItem extracts HTML release-table data (Atom feed fallback)", async () => {
+  const { parseFeedItem } = await import("./lib/card-feed-parser.mjs");
+
+  // Real shape produced when GitHub renders release Markdown to HTML, as
+  // returned by the Atom feed fallback (fetch-feeds.js) when GITHUB_TOKEN
+  // is unavailable and the REST API can't be used.
+  const item = {
+    title: "stable-20260401 (F43.20260401, #123)",
+    pubDate: "2026-04-01T00:00:00Z",
+    link: "https://example.com/release",
+    content: [
+      "<h3>Major packages</h3>",
+      "<table>",
+      "<thead><tr><th>Name</th><th>Version</th></tr></thead>",
+      "<tbody>",
+      "<tr><td><strong>Kernel</strong></td><td>6.14.0 ➡️ 6.15.0</td></tr>",
+      "<tr><td><strong>Mesa</strong></td><td>25.0</td></tr>",
+      "</tbody>",
+      "</table>",
+      "<h3>Major DX packages</h3>",
+      "<table>",
+      "<thead><tr><th>Name</th><th>Version</th></tr></thead>",
+      "<tbody>",
+      "<tr><td><strong>Devpod</strong></td><td>0.5 ➡️ 0.6</td></tr>",
+      "</tbody>",
+      "</table>",
+      "<h3>All Images</h3>",
+      "<table>",
+      "<thead><tr><th></th><th>Name</th><th>Previous</th><th>New</th></tr></thead>",
+      "<tbody>",
+      "<tr><td>✨</td><td>ghcr.io/example/new</td><td></td><td></td></tr>",
+      "<tr><td>🔄</td><td>ghcr.io/example/changed</td><td></td><td></td></tr>",
+      "<tr><td>❌</td><td>ghcr.io/example/removed</td><td></td><td></td></tr>",
+      "</tbody>",
+      "</table>",
+      "<h3>Commits</h3>",
+      "<table>",
+      "<thead><tr><th>Hash</th><th>Message</th></tr></thead>",
+      "<tbody>",
+      "<tr><td>abc123</td><td>Update kernel</td></tr>",
+      "<tr><td>def456</td><td>Update mesa</td></tr>",
+      "</tbody>",
+      "</table>",
+    ].join("\n"),
+  };
+
+  assert.deepEqual(parseFeedItem(item, "stable"), {
+    stream: "stable",
+    tag: "stable-20260401",
+    fedoraVersion: "43",
+    centosVersion: null,
+    majorPackages: [
+      { name: "Kernel", version: "6.15.0", prevVersion: "6.14.0" },
+      { name: "Mesa", version: "25.0", prevVersion: null },
+    ],
+    dxPackages: [{ name: "Devpod", version: "0.6", prevVersion: "0.5" }],
+    gdxPackages: [],
+    diffStats: { added: 1, changed: 1, removed: 1 },
+    commitCount: 2,
+    dateMs: Date.parse("2026-04-01T00:00:00Z"),
+    link: "https://example.com/release",
+  });
+});
+
 test("parseFeedItem returns null when markdown tables are missing major packages", async () => {
   const { parseFeedItem } = await import("./lib/card-feed-parser.mjs");
 

scripts/lib/card-feed-parser.mjs

@@ -72,12 +72,77 @@ export function parseCommitRows(rows) {
   return rows.filter((cells) => cells.length >= 2 && cells[0] && cells[0] !== "Hash").length;
 }
 
+/**
+ * Strip HTML tags and decode the handful of entities GitHub uses when
+ * rendering release notes (Atom feed content, or the release page itself).
+ */
+export function stripHtml(text) {
+  return text
+    .replace(/<[^>]+>/g, "")
+    .replace(/&amp;/g, "&")
+    .replace(/&lt;/g, "<")
+    .replace(/&gt;/g, ">")
+    .replace(/&quot;/g, '"')
+    .replace(/&#0?39;/g, "'")
+    .trim();
+}
+
+export function splitHtmlRow(rowHtml) {
+  const cells = [];
+  const cellRe = /<t[hd][^>]*>([\s\S]*?)<\/t[hd]>/gi;
+  let match;
+  while ((match = cellRe.exec(rowHtml))) {
+    cells.push(stripHtml(match[1]));
+  }
+  return cells;
+}
+
+/**
+ * Extract `### Heading` -> table-rows sections from an HTML release body
+ * (e.g. the GitHub Atom feed, which renders Markdown to HTML). Each heading
+ * is paired with the next <table> that appears before the following heading.
+ */
+export function extractSectionsHtml(content) {
+  const sections = new Map();
+  const headingRe = /<h[1-6][^>]*>([\s\S]*?)<\/h[1-6]>/gi;
+  const headings = [];
+  let hm;
+  while ((hm = headingRe.exec(content))) {
+    headings.push({ index: hm.index, end: headingRe.lastIndex, text: stripHtml(hm[1]) });
+  }
+
+  const tableRe = /<table[^>]*>([\s\S]*?)<\/table>/gi;
+  const tables = [];
+  let tm;
+  while ((tm = tableRe.exec(content))) {
+    tables.push({ index: tm.index, body: tm[1] });
+  }
+
+  for (let i = 0; i < headings.length; i++) {
+    const heading = headings[i];
+    const nextHeadingIndex = i + 1 < headings.length ? headings[i + 1].index : Infinity;
+    const table = tables.find((t) => t.index > heading.end && t.index < nextHeadingIndex);
+    if (!table) continue;
+
+    const rowRe = /<tr[^>]*>([\s\S]*?)<\/tr>/gi;
+    const rows = [];
+    let rm;
+    while ((rm = rowRe.exec(table.body))) {
+      const cells = splitHtmlRow(rm[1]);
+      if (cells.length > 0) rows.push(cells);
+    }
+    if (rows.length > 0) sections.set(heading.text, rows);
+  }
+  return sections;
+}
+
 export function parseFeedItem(item, streamHint) {
   const content = item.content ?? "";
   const isMarkdown = /^\|[\s|:-]*---[\s|:-]*\|/m.test(content);
-  if (!isMarkdown) return null;
+  const isHtml = !isMarkdown && /<table[^>]*>[\s\S]*<\/table>/i.test(content);
+  if (!isMarkdown && !isHtml) return null;
 
-  const sections = extractSectionsMd(content);
+  const sections = isHtml ? extractSectionsHtml(content) : extractSectionsMd(content);
   const majorPackages = parseTwoColTableMd(sections.get("Major packages") ?? []);
   const dxPackages = parseTwoColTableMd(sections.get("Major DX packages") ?? []);
   const gdxPackages = parseTwoColTableMd(sections.get("Major GDX packages") ?? []);