dogsbay 0.2.0-beta.44 → 0.2.0-beta.46

package/dist/audit/rules/seo/sitemap.js

@@ -22,7 +22,7 @@
  * `_resetSitemapCache()`.
  */
 import { existsSync, readFileSync } from "node:fs";
-import { join } from "node:path";
+import { basename, dirname, join } from "node:path";
 const STATE_CACHE = new Map();
 export function _resetSitemapCache() {
     STATE_CACHE.clear();
@@ -50,11 +50,26 @@ function getState(distRoot) {
             // sitemap integration which produces well-formed XML, so
             // the only failure mode is "missing entirely" or
             // "doesn't parse at all."
-            if (/<(urlset|sitemapindex)[\s>]/.test(content)) {
+            const isIndex = /<sitemapindex[\s>]/.test(content);
+            const isUrlset = /<urlset[\s>]/.test(content);
+            if (isIndex || isUrlset) {
                 sitemapValid = true;
-                const matches = content.matchAll(/<loc>([^<]+)<\/loc>/g);
-                for (const m of matches) {
-                    sitemapLocs.add(m[1].trim());
+                if (isIndex) {
+                    // sitemap-index.xml — each <loc> points at a child
+                    // sitemap file (not a page). Read each child and union
+                    // its <loc> entries (those ARE the page URLs). Without
+                    // this recursion, sitemapLocs would carry sitemap file
+                    // URLs only and every page would be flagged "missing
+                    // from sitemap" — the entire audit class is unusable.
+                    collectChildSitemapLocs(content, sitemapPath, sitemapLocs);
+                }
+                else {
+                    // Plain <urlset> — extract <loc>s directly. Same as
+                    // before.
+                    const matches = content.matchAll(/<loc>([^<]+)<\/loc>/g);
+                    for (const m of matches) {
+                        sitemapLocs.add(m[1].trim());
+                    }
                 }
             }
         }
@@ -146,8 +161,19 @@ export const sitemapComplete = {
         // Map each html file to its likely "page URL" form. Astro
         // emits `docs/intro/index.html` for the URL `/docs/intro/`,
         // so we strip `/index.html` and ensure a leading slash.
+        //
+        // Skip pages with `<meta name="robots" content="noindex">`.
+        // The sitemap emitter intentionally excludes those (per the
+        // platform contract documented in reference/frontmatter.md
+        // and `packages/format-astro/src/sitemap.ts` isExcluded), so
+        // flagging them as "missing" would tell users to fix something
+        // that's working as intended. The auto-emitted taxonomy index
+        // pages (/tags/, /by-audience/, …) all ship with noindex.
         const issues = [];
         for (const f of allFiles) {
+            const robots = f.$('meta[name="robots"]').attr("content") ?? "";
+            if (/\bnoindex\b/i.test(robots))
+                continue;
             const expectedPath = htmlPathToUrlPath(f.path);
             const found = pathListedInSitemap(expectedPath, state.sitemapLocs, urlBase);
             if (!found) {
@@ -203,6 +229,51 @@ export const sitemapRobotsCoherence = {
         return issues;
     },
 };
+/**
+ * Recurse into a `<sitemapindex>` document: for each child
+ * sitemap referenced by `<loc>`, read the file from disk and
+ * union its page-level `<loc>` entries into the supplied set.
+ *
+ * Resolution strategy: take the basename of the child URL and
+ * resolve it against the directory that contains the index file.
+ * That covers both layouts in use today — host-root
+ * (`dist/sitemap-index.xml` + `dist/sitemap-0.xml`) and
+ * per-mount (`dist/<basePath>/sitemap-index.xml` +
+ * `dist/<basePath>/sitemap-0.xml`). Children that don't resolve
+ * to a local file are silently skipped — better to miss a few
+ * URLs than to crash the audit when a deploy ships a partial
+ * sitemap. The page-level rule still surfaces orphan pages, so
+ * an incomplete recursion just downgrades to the previous
+ * misbehaviour (all pages flagged as missing) — never worse.
+ */
+function collectChildSitemapLocs(indexContent, indexPath, out) {
+    const indexDir = dirname(indexPath);
+    for (const m of indexContent.matchAll(/<loc>([^<]+)<\/loc>/g)) {
+        const childUrl = m[1].trim();
+        let childName;
+        try {
+            const u = new URL(childUrl);
+            childName = basename(u.pathname);
+        }
+        catch {
+            childName = basename(childUrl);
+        }
+        if (!childName || childName === "/")
+            continue;
+        const childPath = join(indexDir, childName);
+        if (!existsSync(childPath))
+            continue;
+        try {
+            const childContent = readFileSync(childPath, "utf-8");
+            for (const cm of childContent.matchAll(/<loc>([^<]+)<\/loc>/g)) {
+                out.add(cm[1].trim());
+            }
+        }
+        catch {
+            // Skip unreadable child — see function-level comment.
+        }
+    }
+}
 /**
  * Convert an HTML file path within `dist/` into the public URL
  * path the sitemap is likely to list. Drops `/index.html` and

package/dist/commands/site-build.js

@@ -32,6 +32,24 @@ export async function siteBuild(cwd, options) {
     const siteRoot = dirname(configPath);
     // 2. Load it
     let config = loadConfig(configPath);
+    // 2b. Migration warning: the default basePath changed from "/docs"
+    // to "" in v0.3 (see plans/default-basepath-root.md). Surface the
+    // change to users whose config doesn't pin a value — they get to
+    // accept the new default (basePath: "") or restore the previous
+    // shape (basePath: "/docs"), but can't be silently moved. Fires
+    // once per build until the user sets the field to anything.
+    if (config.site.basePath === undefined) {
+        console.warn(pc.yellow(`\nWarning: site.basePath is not set in dogsbay.config.yml.`));
+        console.warn(`  The default changed in v0.3: it was "/docs", it is now "" (host root).`);
+        console.warn(`  Your site will mount at the host root instead of under /docs/.`);
+        console.warn(``);
+        console.warn(`  ${pc.cyan('To accept the new default and silence this warning, add:')}`);
+        console.warn(`      ${pc.cyan('basePath: ""')}        # under site:`);
+        console.warn(``);
+        console.warn(`  ${pc.cyan('To restore the previous behaviour, add:')}`);
+        console.warn(`      ${pc.cyan('basePath: "/docs"')}   # under site:`);
+        console.warn(``);
+    }
     // 3. CLI overrides
     config = mergeOverrides(config, options);
     // 4. Resolve where the generated Astro project lives. Default

package/dist/commands/site-init.js

@@ -142,7 +142,11 @@ function seedStarterContent(absTarget, config) {
     mkdirSync(contentDir, { recursive: true });
     const written = [];
     const siteName = config.site.name?.trim() || "your documentation site";
-    const basePath = config.site.basePath ?? "/docs";
+    // `??` fallback used to be "/docs"; matched the framework default
+    // pre-v0.3. The default flipped to "" (host root); align this
+    // fallback so the scaffold's intra-link text matches the URL
+    // shape a new site actually serves. See plans/default-basepath-root.md.
+    const basePath = config.site.basePath ?? "";
     const indexPath = join(contentDir, "index.md");
     writeFileSync(indexPath, `---
 title: Welcome
@@ -483,8 +487,15 @@ async function resolveInteractive(opts) {
 }
 // ─── SiteInitOptions → DogsbayConfig ─────────────────────────────────────
 function buildConfig(opts) {
+    // Always write basePath explicitly. The v0.3 default is "" (host
+    // root); scaffolding it visibly stops the migration warning from
+    // firing on the user's first build AND makes the URL shape
+    // obvious without forcing the reader to know the framework
+    // default. Authors who want /docs (or any other prefix) just edit
+    // this line. See plans/default-basepath-root.md.
     const site = {
         name: opts.siteName.trim(),
+        basePath: "",
         ...(opts.siteUrl?.trim() ? { url: opts.siteUrl.trim() } : {}),
         ...(opts.description?.trim()
             ? { description: opts.description.trim() }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dogsbay",
-  "version": "0.2.0-beta.44",
+  "version": "0.2.0-beta.46",
   "description": "CLI for Dogsbay — scaffold, build, and serve documentation sites with markdown / MkDocs / Obsidian / OpenAPI sources",
   "type": "module",
   "bin": {
@@ -32,15 +32,15 @@
     "picocolors": "^1.1.0",
     "prompts": "^2.4.2",
     "yaml": "^2.8.3",
-    "@dogsbay/autodoc-python": "0.2.0-beta.44",
-    "@dogsbay/format-mkdocs": "0.2.0-beta.44",
-    "@dogsbay/format-astro": "0.2.0-beta.44",
-    "@dogsbay/format-obsidian": "0.2.0-beta.44",
-    "@dogsbay/format-mdx": "0.2.0-beta.44",
-    "@dogsbay/format-starlight": "0.2.0-beta.44",
-    "@dogsbay/format-dogsbay-md": "0.2.0-beta.44",
-    "@dogsbay/format-openapi": "0.2.0-beta.44",
-    "@dogsbay/types": "0.2.0-beta.44"
+    "@dogsbay/autodoc-python": "0.2.0-beta.46",
+    "@dogsbay/format-mkdocs": "0.2.0-beta.46",
+    "@dogsbay/format-mdx": "0.2.0-beta.46",
+    "@dogsbay/format-astro": "0.2.0-beta.46",
+    "@dogsbay/format-starlight": "0.2.0-beta.46",
+    "@dogsbay/format-dogsbay-md": "0.2.0-beta.46",
+    "@dogsbay/format-obsidian": "0.2.0-beta.46",
+    "@dogsbay/format-openapi": "0.2.0-beta.46",
+    "@dogsbay/types": "0.2.0-beta.46"
   },
   "devDependencies": {
     "@types/markdown-it": "^14.1.0",