extract-webpage 1.2.333 → 1.2.335

package/dist/html-to-content/prism-global.d.ts

@@ -1,2 +1,12 @@
 import { default as Prism } from 'prismjs';
+/**
+ * Registers the grammars above, once per runtime. Cheap to call repeatedly —
+ * callers are meant to invoke it wherever they are about to highlight, so that
+ * the work does not hinge on a module-level statement surviving tree-shaking.
+ *
+ * Never rejects: a grammar that fails to load just leaves its language out of
+ * `Prism.languages`, which every caller already treats as "render this block
+ * unhighlighted".
+ */
+export declare function loadPrismGrammars(): Promise<void>;
 export default Prism;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "extract-webpage",
-  "version": "1.2.333",
+  "version": "1.2.335",
   "module": "./dist/extract-webpage.es.js",
   "description": "Search, extract, cite, and outline the web for a topic with AI Research Agent.",
   "author": "vtempest <grokthiscontact@gmail.com>",
@@ -74,11 +74,11 @@
   "dependencies": {
     "@huggingface/transformers": "^4.2.0",
     "ai": "^7.0.94",
-    "chat-agent-toolkit": "^1.2.339",
+    "chat-agent-toolkit": "^1.2.341",
     "chrono-node": "^2.10.1",
     "drizzle-orm": "^0.45.2",
-    "extract-pdf": "^0.1.323",
-    "extract-youtube": "^1.0.327",
+    "extract-pdf": "^0.1.325",
+    "extract-youtube": "^1.0.329",
     "html-entities": "^2.6.0",
     "js-yaml": "^5.4.1",
     "jsdom": "^30.0.1",

package/src/html-to-content/html-utils.ts

@@ -139,26 +139,15 @@ export function convertURLToAbsoluteURL(base, relative) {
 }
 
 import { marked } from "marked";
-// Must precede every `prismjs/components/*` import: the grammar scripts read
-// `Prism` off the global object, which this module publishes (see its docs).
-import Prism from "./prism-global";
-import "prismjs/components/prism-markup.js";
-import "prismjs/components/prism-css.js";
-import "prismjs/components/prism-javascript.js";
-import "prismjs/components/prism-typescript.js";
-import "prismjs/components/prism-jsx.js";
-import "prismjs/components/prism-tsx.js";
-import "prismjs/components/prism-python.js";
-import "prismjs/components/prism-bash.js";
-import "prismjs/components/prism-json.js";
-import "prismjs/components/prism-yaml.js";
-import "prismjs/components/prism-markdown.js";
-import "prismjs/components/prism-sql.js";
-import "prismjs/components/prism-rust.js";
-import "prismjs/components/prism-go.js";
-import "prismjs/components/prism-java.js";
-import "prismjs/components/prism-c.js";
-import "prismjs/components/prism-cpp.js";
+// `prism-global` owns both the Prism instance and the grammars beyond the ones
+// prismjs' entry point bundles; importing `prismjs/components/*` here directly
+// would reintroduce the load-order crash its docs describe.
+import Prism, { loadPrismGrammars } from "./prism-global";
+
+// Start the grammars loading now, and again from the renderer below — the
+// second call is what guarantees they are requested at all, since a bundler is
+// free to drop this one as a side effect of a module it thinks is pure.
+void loadPrismGrammars();
 
 // Configure marked once at module load with Prism.js syntax highlighting.
 // marked v17 removed the `highlight` option from setOptions, so highlighting
@@ -166,6 +155,7 @@ import "prismjs/components/prism-cpp.js";
 marked.use({
   renderer: {
     code({ text, lang }) {
+      void loadPrismGrammars();
       const language = lang && Prism.languages[lang] ? lang : null;
       const highlighted = language
         ? Prism.highlight(text, Prism.languages[language], language)

package/src/html-to-content/prism-global.ts

@@ -1,22 +1,82 @@
 /**
  * @module html-to-content/prism-global
- * @description Publishes Prism on `globalThis` so the `prismjs/components/*`
- * grammar scripts can find it.
+ * @description Owns the Prism instance: publishes it on `globalThis`, then
+ * registers the language grammars that prismjs' own entry point leaves out —
+ * without ever depending on module evaluation order.
  *
- * Those grammar files are plain browser scripts, not modules — `prism-markup.js`
- * literally opens with `Prism.languages.markup = {...}`, resolving `Prism` as a
- * free variable off the global object. Prism's core only publishes that global
- * when it can see `window` (browser), a `WorkerGlobalScope` `self`, or Node's
- * `global`. On Cloudflare Workers / edge runtimes none of the three exist, so
- * the first grammar file throws `ReferenceError: Prism is not defined` while the
- * module graph is still evaluating, taking the whole SSR render down with it.
+ * The `prismjs/components/*` files are plain browser scripts, not modules:
+ * `prism-markup.js` literally opens with `Prism.languages.markup = {...}`,
+ * resolving `Prism` as a free variable off the global object. Someone has to
+ * put it there first. prismjs' own entry point does that for `window`
+ * (browser) and `global` (Node); on Cloudflare Workers / edge runtimes it sees
+ * neither, which is what the assignment below is for.
  *
- * Import this module *before* any `prismjs/components/*` import. ES modules are
- * evaluated in import order, so this file's body runs — and the global lands —
- * before the grammars reference it.
+ * Publishing it is only half the job, though: it has to happen *before* the
+ * grammar scripts run, and a static `import "prismjs/components/..."` cannot
+ * promise that once a bundler is in the loop. Those files declare no
+ * dependency on Prism — reading a global is invisible to the module graph — so
+ * nothing pins them after whoever publishes it, and a bundler is free to hoist
+ * them, split them into another chunk, or drop the publishing statement as a
+ * dead `globalThis` write. When that happens the chunk dies on load with
+ * `ReferenceError: Prism is not defined`, which takes down the whole route
+ * that lazily imported it rather than just the syntax highlighting.
+ *
+ * So the grammars are loaded with `import()` from inside `loadPrismGrammars()`
+ * instead. The global is published by a statement earlier in that same
+ * function body, which makes the ordering a runtime fact rather than a promise
+ * the bundler has to keep.
  */
 import Prism from "prismjs";
 
-(globalThis as typeof globalThis & { Prism?: unknown }).Prism ??= Prism;
+/**
+ * The grammars to register, in order — several extend an earlier one, so this
+ * list is loaded sequentially rather than in parallel (`tsx` needs `jsx` and
+ * `typescript`; `cpp` needs `c`). Markup, CSS, C-like and JavaScript are
+ * absent because prismjs' entry point already bundles them.
+ */
+const GRAMMARS: ReadonlyArray<() => Promise<unknown>> = [
+  () => import("prismjs/components/prism-typescript.js"),
+  () => import("prismjs/components/prism-jsx.js"),
+  () => import("prismjs/components/prism-tsx.js"),
+  () => import("prismjs/components/prism-python.js"),
+  () => import("prismjs/components/prism-bash.js"),
+  () => import("prismjs/components/prism-json.js"),
+  () => import("prismjs/components/prism-yaml.js"),
+  () => import("prismjs/components/prism-markdown.js"),
+  () => import("prismjs/components/prism-sql.js"),
+  () => import("prismjs/components/prism-rust.js"),
+  () => import("prismjs/components/prism-go.js"),
+  () => import("prismjs/components/prism-java.js"),
+  () => import("prismjs/components/prism-c.js"),
+  () => import("prismjs/components/prism-cpp.js"),
+];
+
+/** Puts Prism where the grammar scripts look for it. Idempotent. */
+function publishPrismGlobal(): void {
+  (globalThis as typeof globalThis & { Prism?: typeof Prism }).Prism ??= Prism;
+}
+
+publishPrismGlobal();
+
+let loading: Promise<void> | undefined;
+
+/**
+ * Registers the grammars above, once per runtime. Cheap to call repeatedly —
+ * callers are meant to invoke it wherever they are about to highlight, so that
+ * the work does not hinge on a module-level statement surviving tree-shaking.
+ *
+ * Never rejects: a grammar that fails to load just leaves its language out of
+ * `Prism.languages`, which every caller already treats as "render this block
+ * unhighlighted".
+ */
+export function loadPrismGrammars(): Promise<void> {
+  loading ??= (async () => {
+    publishPrismGlobal();
+    for (const load of GRAMMARS) {
+      await load();
+    }
+  })().catch(() => {});
+  return loading;
+}
 
 export default Prism;