@ibalzam/codejitsu-core 0.3.3 → 0.5.0

package/modules/llms/src/generate.mjs

@@ -198,6 +198,12 @@ async function generateContentScan({ config, cwd }) {
   const llms = config.llms;
   const scan = llms.contentScan ?? {};
 
+  // Date + draft field names come from the blog module config; CC schemas like
+  // pearl's use `pubDate` + `draft`, while simpler sites use `date` (+ no draft).
+  const blogCfg = config.blog && typeof config.blog === 'object' ? config.blog : {};
+  const dateField = blogCfg.dateField ?? 'date';
+  const draftField = blogCfg.draftField ?? null;
+
   const servicesDir = scan.servicesDir ? path.resolve(cwd, scan.servicesDir) : null;
   const locationsDir = scan.locationsDir ? path.resolve(cwd, scan.locationsDir) : null;
   const pagesDir = scan.pagesDir ? path.resolve(cwd, scan.pagesDir) : null;
@@ -205,10 +211,7 @@ async function generateContentScan({ config, cwd }) {
 
   const services = readContentDir(servicesDir);
   const locations = readContentDir(locationsDir);
-  const blogPosts = readBlogPosts(blogDir, 'pubDate', 'draft').concat(
-    // Also try 'date' field for fallback
-    blogDir && readBlogPosts(blogDir, 'date', 'draft').filter((p) => !p.pubDate) || []
-  );
+  const blogPosts = readBlogPosts(blogDir, dateField, draftField);
   const pages = pagesDir ? collectStaticPages(pagesDir) : [];
 
   const dynamicRoutes = scan.dynamicRoutes ?? [];
@@ -239,6 +242,7 @@ async function generateContentScan({ config, cwd }) {
     business: site.business,
     services,
     locations,
+    blogPosts: blogPosts.slice(0, llms.blogFullLimit ?? 20),
     aiGuidance: llms.aiGuidance,
     today: isoDate(),
   });
@@ -372,7 +376,7 @@ function renderContentScanConcise({ siteUrl, siteName, tagline, about, business,
   return lines.join('\n') + '\n';
 }
 
-function renderContentScanFull({ siteUrl, siteName, tagline, about, business, services, locations, aiGuidance, today }) {
+function renderContentScanFull({ siteUrl, siteName, tagline, about, business, services, locations, blogPosts, aiGuidance, today }) {
   const lines = [];
   lines.push(`# ${siteName} — Full Reference`);
   lines.push(`Last Updated: ${today}`, '');
@@ -431,6 +435,19 @@ function renderContentScanFull({ siteUrl, siteName, tagline, about, business, se
     lines.push('', '---', '');
   }
 
+  if (blogPosts && blogPosts.length) {
+    lines.push('## Blog Posts', '');
+    for (const post of blogPosts) {
+      lines.push(`### ${post.title}`, '');
+      if (post.date) lines.push(`**Published**: ${post.date}`);
+      if (post.author) lines.push(`**Author**: ${post.author}`);
+      if (post.tags?.length) lines.push(`**Tags**: ${post.tags.join(', ')}`);
+      lines.push(`**URL**: ${siteUrl}/blog/${post.slug}/`, '');
+      if (post.description) lines.push(post.description, '');
+      lines.push('---', '');
+    }
+  }
+
   lines.push(`## Optional`, '', `- Sitemap: ${siteUrl}/sitemap-index.xml`, '');
   if (aiGuidance) lines.push('## For AI Assistants', '', aiGuidance, '');
   return lines.join('\n') + '\n';

package/modules/rehype/CLAUDE.md

@@ -0,0 +1,64 @@
+# Rehype module — instructions for Claude
+
+When the user asks to **fix trailing-slash bugs in markdown content** (or pre-empt them across a Codejitsu site), wire up `rehypeTrailingSlash`.
+
+## What this module provides
+
+A single rehype plugin: `trailingSlash`. Runs during Astro's markdown→HTML conversion. Walks the HTML AST and rewrites internal `<a href="/foo">` to `<a href="/foo/">` (or vice versa with `policy: 'never'`).
+
+**Why this exists:** Astro's `trailingSlash: 'always'` config covers route resolution and `Astro.url.pathname` but does NOT touch href strings written by humans in markdown or `.astro` files. This plugin closes that gap for markdown-rendered HTML.
+
+It does **not** affect href strings inside `.astro` component source (Astro doesn't run rehype on those). Use the audit to catch those.
+
+## Wiring it into a site
+
+```ts
+// astro.config.mjs
+import { defineConfig } from 'astro/config';
+import trailingSlash from '@ibalzam/codejitsu-core/rehype/trailing-slash';
+
+export default defineConfig({
+  markdown: {
+    rehypePlugins: [trailingSlash],
+  },
+});
+```
+
+With explicit options:
+
+```ts
+rehypePlugins: [
+  [trailingSlash, { policy: 'always' }],
+],
+```
+
+## What it does NOT touch
+
+- External URLs (`http://`, `https://`, `//`)
+- `mailto:`, `tel:`, `javascript:`
+- Anchor-only links (`#section`)
+- Paths ending in a file extension (`.pdf`, `.html`, `.webp`, etc.)
+- Root path (`/`)
+
+## What it DOES touch
+
+- `<a href="/foo">` → `<a href="/foo/">`
+- `<a href="/foo?bar=1">` → `<a href="/foo/?bar=1">`
+- `<a href="/foo#section">` → `<a href="/foo/#section">`
+
+Preserves query strings and fragments. Path-only modification.
+
+## What must NOT be done
+
+- **Don't apply this to `.astro` component files** — the plugin runs on markdown rehype, not Astro components. If a `<a href="/foo">` lives in a `.astro` file, the plugin can't see it.
+- **Don't set `policy: 'never'` if `astro.config` has `trailingSlash: 'always'`** — they'd contradict each other. The audit will flag the inconsistency.
+- **Don't run this with `policy: 'preserve'` and expect anything to change** — that mode is a no-op (registered as a placeholder for symmetry).
+
+## Verify after wiring
+
+```bash
+npm run build
+npx codejitsu audit
+```
+
+The audit's "All internal links end with /" check should now report 0 markdown-level offenders. Component-level offenders (in `.astro` files) still surface — those must be fixed by hand.

package/modules/rehype/src/trailing-slash.mjs

@@ -0,0 +1,88 @@
+/**
+ * Rehype plugin that enforces a trailing-slash policy on internal `<a href>`
+ * values produced by markdown content. Astro's `trailingSlash: 'always'`
+ * controls page routing but does NOT rewrite hand-written hrefs in markdown
+ * or component templates. This plugin fills that gap for markdown content.
+ *
+ * Usage in astro.config.mjs:
+ *
+ *   import trailingSlash from '@ibalzam/codejitsu-core/rehype/trailing-slash';
+ *
+ *   export default defineConfig({
+ *     markdown: {
+ *       rehypePlugins: [trailingSlash],
+ *     },
+ *   });
+ *
+ * Or with options:
+ *
+ *   rehypePlugins: [[trailingSlash, { policy: 'always' }]]
+ *
+ * What it skips (leaves untouched):
+ *  - External URLs (http://, https://, //, mailto:, tel:, etc.)
+ *  - Anchor-only links (#section)
+ *  - Paths ending in a file extension (.pdf, .html, .webp, ...)
+ *  - Root path `/`
+ *
+ * @param {object} [opts]
+ * @param {'always' | 'never' | 'preserve'} [opts.policy='always']
+ */
+export default function rehypeTrailingSlash(opts = {}) {
+  const policy = opts.policy ?? 'always';
+  if (policy === 'preserve') return () => {};
+
+  return (tree) => {
+    walk(tree, (node) => {
+      if (node.tagName !== 'a') return;
+      const href = node.properties?.href;
+      if (typeof href !== 'string') return;
+
+      const normalized = normalize(href, policy);
+      if (normalized !== href) {
+        node.properties.href = normalized;
+      }
+    });
+  };
+}
+
+function walk(node, fn) {
+  if (node?.type === 'element') fn(node);
+  if (Array.isArray(node?.children)) {
+    for (const child of node.children) walk(child, fn);
+  }
+}
+
+/**
+ * Apply policy to a single href. Pure, no side-effects — exported for tests.
+ *
+ * @param {string} href
+ * @param {'always' | 'never'} policy
+ */
+export function normalize(href, policy) {
+  if (!href.startsWith('/')) return href;        // external, anchor, relative — skip
+  if (href.startsWith('//')) return href;        // protocol-relative external
+  if (href === '/') return href;                  // root is its own canonical
+
+  // Split path / query / fragment so we don't break /foo?bar=baz or /foo#anchor.
+  const m = href.match(/^([^?#]*)(\?[^#]*)?(#.*)?$/);
+  if (!m) return href;
+  let path = m[1];
+  const query = m[2] ?? '';
+  const fragment = m[3] ?? '';
+
+  if (!path || path === '/') return href;
+
+  // Last segment with a `.` is likely a file (e.g. /robots.txt, /og-image.webp).
+  const lastSeg = path.split('/').filter(Boolean).pop() ?? '';
+  if (lastSeg.includes('.')) return href;
+
+  const endsWithSlash = path.endsWith('/');
+
+  if (policy === 'always' && !endsWithSlash) {
+    path = `${path}/`;
+  } else if (policy === 'never' && endsWithSlash) {
+    path = path.replace(/\/+$/, '');
+  }
+
+  return `${path}${query}${fragment}`;
+}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ibalzam/codejitsu-core",
-  "version": "0.3.3",
+  "version": "0.5.0",
   "type": "module",
   "description": "Shared core for Codejitsu Astro sites — reusable code and Claude-facing instructions for blog, SEO, images, deploy, and llms.txt.",
   "keywords": [
@@ -60,6 +60,7 @@
       "default": "./modules/seo/src/sitemap.js"
     },
     "./seo/Head.astro": "./modules/seo/templates/Head.astro",
+    "./rehype/trailing-slash": "./modules/rehype/src/trailing-slash.mjs",
     "./images": {
       "types": "./modules/images/src/index.d.ts",
       "default": "./modules/images/src/index.js"
@@ -68,6 +69,7 @@
     "./package.json": "./package.json"
   },
   "bin": {
+    "codejitsu": "./bin/codejitsu.mjs",
     "codejitsu-llms": "./modules/llms/bin/generate.mjs",
     "codejitsu-optimize-images": "./modules/images/bin/optimize.mjs",
     "codejitsu-check": "./checklist/bin/run.mjs"