@bagelink/blox 1.15.228 → 1.15.232

package/dist/{prerender-FPRGhRSk.js → prerender-DnqZAE4x.js}

@@ -329,16 +329,23 @@ function extractPrimaryContext(contexts) {
   if (!contexts || typeof contexts !== "object") return empty;
   const ctx = Object.values(contexts).find((c) => c != null);
   if (!ctx) return empty;
-  const str = (key) => {
+  const str2 = (key) => {
     const v = ctx[key];
     return typeof v === "string" && v.trim() ? v.trim() : null;
   };
   return {
-    title: str("meta_title") || str("title"),
-    description: str("meta_description") || str("excerpt") || str("description") || str("blurb") || str("summary"),
-    image: str("og_image") || str("cover_image_url") || str("image_url") || str("image")
+    title: str2("meta_title") || str2("title"),
+    description: str2("meta_description") || str2("excerpt") || str2("description") || str2("blurb") || str2("summary"),
+    image: str2("og_image") || str2("cover_image_url") || str2("image_url") || str2("image")
   };
 }
+function primaryContextEntry(contexts) {
+  if (!contexts || typeof contexts !== "object") return null;
+  for (const [key, ctx] of Object.entries(contexts)) {
+    if (ctx != null && typeof ctx === "object") return { key, ctx };
+  }
+  return null;
+}
 function esc(s) {
   return s.replace(/&/g, "&amp;").replace(/"/g, "&quot;").replace(/</g, "&lt;").replace(/>/g, "&gt;");
 }
@@ -350,6 +357,196 @@ function absUrl(value) {
   if (/^(https?:)?\/\//i.test(v) || v.startsWith("/")) return v;
   return `${FILES_BASE_URL}/${v.replace(/^\/+/, "")}`;
 }
+function typeFromContextKey(key) {
+  const k = key.replace(/^\$/, "").toLowerCase();
+  if (/(post|article|blog|story|news)/.test(k)) return "Article";
+  if (/(product|listing|item|sku)/.test(k)) return "Product";
+  if (/(event|session|webinar)/.test(k)) return "Event";
+  return "WebPage";
+}
+function str(o, ...keys) {
+  for (const key of keys) {
+    const v = o[key];
+    if (typeof v === "string" && v.trim()) return v.trim();
+  }
+  return void 0;
+}
+function breadcrumbList(url, base, siteName) {
+  const segs = url.split("/").filter(Boolean);
+  const items = [
+    { "@type": "ListItem", "position": 1, "name": siteName || "Home", "item": base || "/" }
+  ];
+  let acc = "";
+  segs.forEach((seg, i) => {
+    acc += `/${seg}`;
+    items.push({
+      "@type": "ListItem",
+      "position": i + 2,
+      "name": decodeURIComponent(seg).replace(/[-_]/g, " ").replace(/\b\w/g, (c) => c.toUpperCase()),
+      "item": base ? base + acc : acc
+    });
+  });
+  return { "@context": "https://schema.org", "@type": "BreadcrumbList", "itemListElement": items };
+}
+function buildJsonLd(options) {
+  var _a;
+  const { url, page, website, contextKey, context, config } = options;
+  const seo = (website == null ? void 0 : website.seo) ?? {};
+  const base = (seo.canonical_base_url || (website == null ? void 0 : website.domain) || "").replace(/\/$/, "");
+  const siteName = seo.site_name || void 0;
+  const pageUrl = base ? base + url : url;
+  const title = (page == null ? void 0 : page.meta_title) || (page == null ? void 0 : page.title) || str(context ?? {}, "meta_title", "title") || seo.default_meta_title;
+  const description = (page == null ? void 0 : page.meta_description) || str(context ?? {}, "meta_description", "excerpt", "description", "summary") || seo.default_meta_description;
+  const graphs = [];
+  const webPage = { "@context": "https://schema.org", "@type": "WebPage", "url": pageUrl };
+  if (title) webPage.name = title;
+  if (description) webPage.description = description;
+  if (siteName) webPage.isPartOf = { "@type": "WebSite", "name": siteName, "url": base || void 0 };
+  graphs.push(webPage);
+  if (url && url !== "/") graphs.push(breadcrumbList(url, base, siteName));
+  if (contextKey && context) {
+    const type = ((_a = config == null ? void 0 : config.typeMap) == null ? void 0 : _a[contextKey]) || typeFromContextKey(contextKey);
+    const entity = { "@context": "https://schema.org", "@type": type, "url": pageUrl };
+    const name = str(context, "title", "name", "meta_title");
+    const desc = str(context, "meta_description", "excerpt", "description", "summary", "blurb");
+    const image = str(context, "og_image", "cover_image_url", "image_url", "image");
+    if (name) entity.name = name;
+    if (desc) entity.description = desc;
+    if (image) entity.image = image;
+    if (type === "Article") {
+      if (siteName) entity.publisher = { "@type": "Organization", "name": siteName };
+      const date = str(context, "published_at", "created_at", "date");
+      if (date) entity.datePublished = date;
+      const upd = str(context, "updated_at", "modified_at");
+      if (upd) entity.dateModified = upd;
+      const author = str(context, "author", "author_name");
+      if (author) entity.author = { "@type": "Person", "name": author };
+      if (name) {
+        entity.headline = name;
+        delete entity.name;
+      }
+    } else if (type === "Product") {
+      const price = str(context, "price", "amount");
+      const currency = str(context, "currency", "currency_code") || "USD";
+      if (price) {
+        entity.offers = { "@type": "Offer", "price": price, "priceCurrency": currency, "url": pageUrl };
+      }
+      const sku = str(context, "sku", "id");
+      if (sku) entity.sku = sku;
+    } else if (type === "Event") {
+      const start = str(context, "start_at", "starts_at", "start_date", "date");
+      const end = str(context, "end_at", "ends_at", "end_date");
+      if (start) entity.startDate = start;
+      if (end) entity.endDate = end;
+      const loc = str(context, "location", "venue", "address");
+      if (loc) entity.location = { "@type": "Place", "name": loc };
+    }
+    graphs.push(entity);
+  }
+  return graphs.map((g) => `<script type="application/ld+json">${JSON.stringify(g)}${"<"}/script>`).join("\n");
+}
+function generateLlmsTxt(options) {
+  var _a;
+  const { website, pages, config } = options;
+  const seo = (website == null ? void 0 : website.seo) ?? {};
+  const base = (seo.canonical_base_url || (website == null ? void 0 : website.domain) || "").replace(/\/$/, "");
+  const siteName = seo.site_name || (website == null ? void 0 : website.name) || "Website";
+  const summary = seo.default_meta_description;
+  const visible = pages.filter((p) => !isPathExcluded(p.url, config == null ? void 0 : config.excludePaths)).sort((a, b) => a.url.localeCompare(b.url));
+  const lines = [`# ${siteName}`, ""];
+  if (summary) lines.push(`> ${summary}`, "");
+  if (config == null ? void 0 : config.intro) lines.push(config.intro.trim(), "");
+  const abs = (u) => base ? base + u : u;
+  const link = (p) => {
+    const label = p.title || p.url;
+    const desc = p.description ? `: ${p.description}` : "";
+    return `- [${label}](${abs(p.url)})${desc}`;
+  };
+  const used = /* @__PURE__ */ new Set();
+  for (const section of (config == null ? void 0 : config.sections) ?? []) {
+    const inSection = visible.filter((p) => p.url.startsWith(section.prefix));
+    if (inSection.length === 0) continue;
+    lines.push(`## ${section.title}`, "");
+    for (const p of inSection) {
+      lines.push(link(p));
+      used.add(p.url);
+    }
+    lines.push("");
+  }
+  const rest = visible.filter((p) => !used.has(p.url));
+  if (rest.length > 0) {
+    lines.push(((_a = config == null ? void 0 : config.sections) == null ? void 0 : _a.length) ? "## Pages" : "## Pages", "");
+    for (const p of rest) lines.push(link(p));
+    lines.push("");
+  }
+  return `${lines.join("\n").trimEnd()}
+`;
+}
+function generateLlmsFullTxt(options) {
+  const { website, pages, config } = options;
+  const seo = (website == null ? void 0 : website.seo) ?? {};
+  const base = (seo.canonical_base_url || (website == null ? void 0 : website.domain) || "").replace(/\/$/, "");
+  const siteName = seo.site_name || (website == null ? void 0 : website.name) || "Website";
+  const visible = pages.filter((p) => !isPathExcluded(p.url, config == null ? void 0 : config.excludePaths)).filter((p) => p.text && p.text.trim()).sort((a, b) => a.url.localeCompare(b.url));
+  const blocks = [`# ${siteName}`, ""];
+  for (const p of visible) {
+    const abs = base ? base + p.url : p.url;
+    blocks.push(`## ${p.title || p.url}`, "", `Source: ${abs}`, "");
+    if (p.description) blocks.push(`> ${p.description}`, "");
+    blocks.push(p.text.trim(), "", "---", "");
+  }
+  return `${blocks.join("\n").trimEnd()}
+`;
+}
+function extractPageText(html) {
+  if (!html) return "";
+  let s = html;
+  s = s.replace(/<(script|style|noscript|svg|template)\b[^>]*>[\s\S]*?<\/\1>/gi, " ");
+  s = s.replace(/<(nav|header|footer)\b[^>]*>[\s\S]*?<\/\1>/gi, " ");
+  s = s.replace(/<h1\b[^>]*>/gi, "\n\n# ").replace(/<h2\b[^>]*>/gi, "\n\n## ").replace(/<h3\b[^>]*>/gi, "\n\n### ").replace(/<(h4|h5|h6)\b[^>]*>/gi, "\n\n#### ");
+  s = s.replace(/<li\b[^>]*>/gi, "\n- ");
+  s = s.replace(/<(\/p|\/div|\/section|\/article|br\s*\/?|\/li|\/h[1-6])>/gi, "\n");
+  s = s.replace(/<[^>]+>/g, " ");
+  s = s.replace(/&nbsp;/g, " ").replace(/&amp;/g, "&").replace(/&lt;/g, "<").replace(/&gt;/g, ">").replace(/&quot;/g, '"').replace(/&#39;/g, "'");
+  s = s.split("\n").map((l) => l.replace(/[ \t]+/g, " ").trim()).join("\n");
+  s = s.replace(/\n{3,}/g, "\n\n").trim();
+  return s;
+}
+function isPathExcluded(path2, patterns) {
+  if (!patterns || patterns.length === 0) return false;
+  return patterns.some((p) => {
+    if (p.endsWith("*")) return path2.startsWith(p.slice(0, -1));
+    return path2 === p;
+  });
+}
+async function mergeLlmParts(options) {
+  const fs2 = await import("node:fs/promises");
+  const path2 = await import("node:path");
+  const byUrl = /* @__PURE__ */ new Map();
+  for (const file of options.partFiles) {
+    try {
+      const raw = await fs2.readFile(file, "utf8");
+      const arr = JSON.parse(raw);
+      for (const e of arr) if (e == null ? void 0 : e.url) byUrl.set(e.url, e);
+    } catch {
+    }
+  }
+  const pages = [...byUrl.values()];
+  const { website, config } = options;
+  const llmsTxt = generateLlmsTxt({ website, pages, config });
+  await fs2.writeFile(path2.join(options.outDir, "llms.txt"), llmsTxt, "utf8");
+  if ((config == null ? void 0 : config.fullText) !== false) {
+    const full = generateLlmsFullTxt({ website, pages, config });
+    await fs2.writeFile(path2.join(options.outDir, "llms-full.txt"), full, "utf8");
+  }
+  for (const file of options.partFiles) {
+    try {
+      await fs2.unlink(file);
+    } catch {
+    }
+  }
+  return { pages: pages.length };
+}
 async function prerender({
   root = process.cwd(),
   clientOutDir = "dist/client",
@@ -364,15 +561,20 @@ async function prerender({
   /**
    * How many `/resolve-path` prefetches to run in parallel, ahead of the render
    * pool. Decouples the (I/O-bound) resolve latency from the (CPU-bound) render
-   * so pages render from an in-memory cache with no per-page round-trip. 0
-   * disables prefetch (render resolves inline). Default 24.
+   * so pages render from an in-memory cache with no per-page round-trip.
+   *
+   * NOTE: this is PER PROCESS. When running N shards, the *total* load on the
+   * API is `N × this`, so keep it modest (a small API saturates ~5-wide). The
+   * sharded runner divides a global budget by the shard count. 0 disables
+   * prefetch (render resolves inline). Default 8.
    */
-  resolvePrefetchConcurrency = 24,
+  resolvePrefetchConcurrency = 8,
   mode = "dir",
   website = null,
   websiteId = "",
   redirects = [],
-  collections = {}
+  collections = {},
+  llm = {}
 } = {}) {
   var _a, _b;
   const absRoot = path.resolve(root);
@@ -407,6 +609,8 @@ async function prerender({
   const rendered = [];
   const noindexPaths = /* @__PURE__ */ new Set();
   const lastmod = {};
+  const llmPages = [];
+  const llmEnabled = llm.enabled !== false;
   const failures = [];
   const resolveCache = /* @__PURE__ */ new Map();
   const prefetch = serverMod.prefetchResolve;
@@ -434,14 +638,16 @@ async function prerender({
       html,
       head = "",
       htmlAttrs = "",
-      lastmod: pageLastmod
+      lastmod: pageLastmod,
+      llm: pageLlm
     } = await serverMod.render(urlPath, {
       manifest,
       template,
       website,
       websiteId,
       collections,
-      resolveCache
+      resolveCache,
+      llm
     });
     const outHtml = injectIntoTemplate(template, head, html, fontPreloads, htmlAttrs);
     const outfile = outFilePath(absClient, urlPath, mode);
@@ -449,6 +655,7 @@ async function prerender({
     await fs.writeFile(outfile, outHtml, "utf8");
     rendered.push(urlPath);
     if (pageLastmod) lastmod[urlPath] = pageLastmod;
+    if (llmEnabled && pageLlm) llmPages.push(pageLlm);
     if (/<meta[^>]+name=["']robots["'][^>]*noindex/i.test(head)) {
       noindexPaths.add(urlPath);
     }
@@ -478,7 +685,8 @@ async function prerender({
       }
     }
   }
-  const prefetchDone = runResolvePrefetch();
+  await runResolvePrefetch().catch(() => {
+  });
   while (queue.length && !stopped && rendered.length < maxPages) {
     const workers = Array.from(
       { length: Math.min(concurrency, queue.length) },
@@ -486,8 +694,6 @@ async function prerender({
     );
     await Promise.all(workers);
   }
-  await prefetchDone.catch(() => {
-  });
   const canonicalBase = (((_a = website == null ? void 0 : website.seo) == null ? void 0 : _a.canonical_base_url) || (website == null ? void 0 : website.domain) || "").replace(/\/$/, "");
   if (writeSharedAssets) {
     const sitemapPaths = rendered.filter((p) => !noindexPaths.has(p));
@@ -515,6 +721,24 @@ async function prerender({
       await fs.writeFile(redirectsPath, redirectsContent + existing, "utf8");
       console.log(`  Generated _redirects (${redirects.length} rules)`);
     }
+    if (llmEnabled && llm.writeIndex !== false && llmPages.length > 0) {
+      const indexable = llmPages.filter((p) => !noindexPaths.has(p.url));
+      const llmsTxt = generateLlmsTxt({ website, pages: indexable, config: llm });
+      await fs.writeFile(path.join(absClient, "llms.txt"), llmsTxt, "utf8");
+      console.log(`  Generated llms.txt (${indexable.length} pages)`);
+      if (llm.fullText !== false) {
+        const full = generateLlmsFullTxt({ website, pages: indexable, config: llm });
+        await fs.writeFile(path.join(absClient, "llms-full.txt"), full, "utf8");
+        console.log("  Generated llms-full.txt");
+      }
+    }
+  }
+  if (llmEnabled && llm.partsFile && llmPages.length > 0) {
+    const indexable = llmPages.filter((p) => !noindexPaths.has(p.url));
+    const partPath = path.resolve(absRoot, llm.partsFile);
+    await fs.mkdir(path.dirname(partPath), { recursive: true });
+    await fs.writeFile(partPath, JSON.stringify(indexable), "utf8");
+    console.log(`  Wrote ${indexable.length} llm page extract(s) → ${llm.partsFile}`);
   }
   return {
     rendered,
@@ -650,12 +874,18 @@ function discoverInternalLinks(html) {
 export {
   prerender as a,
   buildPageHead as b,
-  buildSiteHead as c,
-  fetchCmsPrerenderPaths as d,
-  generateRobotsTxt as e,
+  primaryContextEntry as c,
+  buildJsonLd as d,
+  extractPageText as e,
   fetchCmsSiteData as f,
-  generateNetlifyRedirects as g,
-  generateSitemapXml as h,
-  getSiteIconFlags as i,
+  buildSiteHead as g,
+  fetchCmsPrerenderPaths as h,
+  generateLlmsFullTxt as i,
+  generateLlmsTxt as j,
+  generateNetlifyRedirects as k,
+  generateRobotsTxt as l,
+  generateSitemapXml as m,
+  getSiteIconFlags as n,
+  mergeLlmParts as o,
   polyfillBloxSsgGlobals as p
 };