@m13v/seo-components 0.8.9 → 0.8.11

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@m13v/seo-components",
-  "version": "0.8.9",
+  "version": "0.8.11",
   "scripts": {
     "build:css": "tailwind -i src/_build.css -o dist/styles.css --minify",
     "prepublishOnly": "npm run build:css"

package/src/lib/discover-guides.ts

@@ -1,3 +1,4 @@
+import fs from "node:fs";
 import path from "node:path";
 import { walkPages, type PageEntry } from "./walk-pages";
 
@@ -25,31 +26,55 @@ export function discoverGuides(contentDir?: string): GuideEntry[] {
 
   if (cachedDir === dir && cachedGuides) return cachedGuides;
 
-  // Determine the Next.js app directory root from contentDir.
-  // e.g. "src/app/t" -> "src/app", "app" -> "app", "src/app/(content)/t" -> "src/app"
-  const relDir = path.relative(process.cwd(), dir);
-  const relParts = relDir.split(path.sep);
-  const appIdx = relParts.indexOf("app");
-  const appDir =
-    appIdx >= 0
-      ? path.join(process.cwd(), ...relParts.slice(0, appIdx + 1))
-      : path.join(process.cwd(), "src/app");
-  const allPages = walkPages({ appDir });
+  // Try to load the build-time manifest first (generated by withSeoContent).
+  // This is required on Vercel where fs.readdirSync can't walk source dirs.
+  // Check both project root and .next/ for backwards compat.
+  const candidates = [
+    path.join(process.cwd(), "seo-guides-manifest.json"),
+    path.join(process.cwd(), ".next", "seo-guides-manifest.json"),
+  ];
+  let manifestPages: PageEntry[] | null = null;
+  for (const manifestPath of candidates) {
+    try {
+      const raw = fs.readFileSync(manifestPath, "utf-8");
+      const manifest = JSON.parse(raw);
+      if (Array.isArray(manifest.pages)) {
+        manifestPages = manifest.pages;
+        break;
+      }
+    } catch {
+      // Try next candidate
+    }
+  }
 
-  // Figure out the href prefix from contentDir.
-  // e.g. "src/app/(content)/t" -> "/t", "src/app/t" -> "/t"
-  const relative = path.relative(appDir, dir);
-  const segments = relative
-    ? relative
-        .split(path.sep)
-        .filter((s) => !(s.startsWith("(") && s.endsWith(")")))
-    : [];
-  const hrefPrefix = "/" + segments.join("/");
+  let guides: GuideEntry[];
 
-  const guides: GuideEntry[] = allPages
-    .filter((p) => p.href.startsWith(hrefPrefix + "/") || p.href === hrefPrefix)
-    .filter((p) => p.href !== hrefPrefix) // exclude the index page
-    .map(pageToGuide);
+  if (manifestPages && manifestPages.length > 0) {
+    guides = manifestPages.map(pageToGuide);
+  } else {
+    // Fallback: walk the filesystem (works locally, may fail on Vercel)
+    const relDir = path.relative(process.cwd(), dir);
+    const relParts = relDir.split(path.sep);
+    const appIdx = relParts.indexOf("app");
+    const appDir =
+      appIdx >= 0
+        ? path.join(process.cwd(), ...relParts.slice(0, appIdx + 1))
+        : path.join(process.cwd(), "src/app");
+    const allPages = walkPages({ appDir });
+
+    const relative = path.relative(appDir, dir);
+    const segments = relative
+      ? relative
+          .split(path.sep)
+          .filter((s) => !(s.startsWith("(") && s.endsWith(")")))
+      : [];
+    const hrefPrefix = "/" + segments.join("/");
+
+    guides = allPages
+      .filter((p) => p.href.startsWith(hrefPrefix + "/") || p.href === hrefPrefix)
+      .filter((p) => p.href !== hrefPrefix)
+      .map(pageToGuide);
+  }
 
   guides.sort(
     (a, b) =>

package/src/next.js

@@ -1,8 +1,18 @@
+import { readdirSync, readFileSync, writeFileSync, mkdirSync } from "node:fs";
+import { join, sep, relative } from "node:path";
+
 /**
  * Wrap a Next.js config so the guide-chat API route can read content files
- * at runtime on Vercel. Adds `outputFileTracingIncludes` for the guide-chat
- * endpoint so Next.js's file tracer includes every page.tsx under `contentDir`
- * in the serverless function bundle.
+ * at runtime on Vercel. Does two things:
+ *
+ * 1. Walks `contentDir` at config evaluation time (during `next build`) and
+ *    writes a JSON manifest of all discovered pages to
+ *    `.next/seo-guides-manifest.json`. The serverless function reads this
+ *    manifest instead of walking the filesystem at runtime.
+ *
+ * 2. Adds `outputFileTracingIncludes` so Next.js bundles every page.tsx
+ *    under `contentDir` into the guide-chat serverless function (needed for
+ *    reading article text at runtime).
  *
  * Usage (next.config.ts or next.config.mjs):
  *
@@ -16,14 +26,147 @@
  */
 export function withSeoContent(config = {}, opts = {}) {
   const contentDir = opts.contentDir ?? "src/app/t";
-  const glob = `./${contentDir}/**/*`;
+  const contentGlob = `./${contentDir}/**/*`;
   const existing = config.outputFileTracingIncludes ?? {};
+
+  // Build a manifest at config evaluation time (runs during `next build`).
+  // Write to project root so outputFileTracingIncludes can find it reliably.
+  const manifestFile = "seo-guides-manifest.json";
+  try {
+    const manifest = buildManifest(contentDir);
+    writeFileSync(join(process.cwd(), manifestFile), JSON.stringify(manifest));
+  } catch (e) {
+    // Non-fatal: the runtime will fall back to filesystem walking
+  }
+
   return {
     ...config,
     outputFileTracingIncludes: {
       ...existing,
-      "/api/guide-chat": [glob],
-      "/api/guide-chat/route": [glob],
+      "/api/guide-chat": [contentGlob, `./${manifestFile}`],
+      "/api/guide-chat/route": [contentGlob, `./${manifestFile}`],
     },
   };
 }
+
+/* ------------------------------------------------------------------ */
+/*  Build-time manifest generation (mirrors walk-pages.ts logic)       */
+/* ------------------------------------------------------------------ */
+
+const CONST_TITLE_RE = /const\s+TITLE\s*=\s*["'`](.+?)["'`]/;
+const CONST_DESC_RE = /const\s+DESCRIPTION\s*=\s*\n?\s*["'`](.+?)["'`]/;
+const CONST_DATE_RE = /const\s+DATE_PUBLISHED\s*=\s*["'`](.+?)["'`]/;
+const META_TITLE_RE = /title:\s*["'`](.+?)["'`]/;
+const META_DESC_RE = /description:\s*["'`](.+?)["'`]/;
+const H2_RE = /<h2\b[^>]*>([\s\S]*?)<\/h2>/g;
+const FAQ_SECTION_RE = /<FaqSection\b/;
+
+function slugify(text) {
+  return text
+    .toLowerCase()
+    .replace(/[^a-z0-9]+/g, "-")
+    .replace(/^-+|-+$/g, "")
+    .slice(0, 80);
+}
+
+function extractMeta(src, fallbackSlug) {
+  const title =
+    src.match(CONST_TITLE_RE)?.[1] ??
+    src.match(META_TITLE_RE)?.[1] ??
+    fallbackSlug.split("-").map(w => w.charAt(0).toUpperCase() + w.slice(1)).join(" ");
+  const description =
+    src.match(CONST_DESC_RE)?.[1] ?? src.match(META_DESC_RE)?.[1] ?? "";
+  const datePublished = src.match(CONST_DATE_RE)?.[1] ?? undefined;
+  return { title, description, datePublished };
+}
+
+function extractSections(src) {
+  const seen = new Set();
+  const out = [];
+  let m;
+  H2_RE.lastIndex = 0;
+  while ((m = H2_RE.exec(src)) !== null) {
+    const text = m[1]
+      .replace(/\{[^}]*\}/g, " ")
+      .replace(/<[^>]+>/g, " ")
+      .replace(/\s+/g, " ")
+      .trim();
+    if (!text) continue;
+    const base = slugify(text);
+    if (!base) continue;
+    let id = base;
+    let i = 2;
+    while (seen.has(id)) { id = `${base}-${i++}`; }
+    seen.add(id);
+    out.push({ id, title: text });
+  }
+  if (FAQ_SECTION_RE.test(src)) {
+    const faqId = "frequently-asked-questions";
+    if (!seen.has(faqId)) out.push({ id: faqId, title: "Frequently asked questions" });
+  }
+  return out;
+}
+
+function buildManifest(contentDir) {
+  const cwd = process.cwd();
+  const relParts = contentDir.split(/[/\\]/);
+  const appIdx = relParts.indexOf("app");
+  const appDirRel = appIdx >= 0 ? relParts.slice(0, appIdx + 1).join(sep) : contentDir;
+  const appDir = join(cwd, appDirRel);
+  const contentAbs = join(cwd, contentDir);
+
+  const pages = [];
+
+  function walk(dir, urlSegments) {
+    let entries;
+    try { entries = readdirSync(dir, { withFileTypes: true }); }
+    catch { return; }
+
+    for (const entry of entries) {
+      if (entry.isFile() && entry.name === "page.tsx") {
+        const href = urlSegments.length === 0 ? "/" : "/" + urlSegments.join("/");
+        if (href === "/") continue;
+        const filePath = join(dir, entry.name);
+        let src;
+        try { src = readFileSync(filePath, "utf-8"); }
+        catch { continue; }
+        const lastSeg = urlSegments[urlSegments.length - 1] ?? "";
+        const meta = extractMeta(src, lastSeg);
+        const sections = extractSections(src);
+        pages.push({
+          href,
+          title: meta.title,
+          description: meta.description,
+          datePublished: meta.datePublished,
+          sections,
+          category: urlSegments[0] ?? "",
+        });
+        continue;
+      }
+      if (!entry.isDirectory()) continue;
+      const name = entry.name;
+      if (name.startsWith("_") || name === "api" || (name.startsWith("[") && name.endsWith("]"))) continue;
+      const isRouteGroup = name.startsWith("(") && name.endsWith(")");
+      walk(join(dir, name), isRouteGroup ? urlSegments : [...urlSegments, name]);
+    }
+  }
+
+  walk(appDir, []);
+
+  // Filter to contentDir prefix
+  const relContent = relative(appDir, contentAbs);
+  const segments = relContent
+    ? relContent.split(sep).filter(s => !(s.startsWith("(") && s.endsWith(")")))
+    : [];
+  const hrefPrefix = "/" + segments.join("/");
+
+  const filtered = pages.filter(p =>
+    (p.href.startsWith(hrefPrefix + "/") || p.href === hrefPrefix) && p.href !== hrefPrefix
+  );
+
+  filtered.sort((a, b) =>
+    (b.datePublished ?? "").localeCompare(a.datePublished ?? "") || a.title.localeCompare(b.title)
+  );
+
+  return { contentDir, appDir: appDirRel, pages: filtered };
+}