@anubhav-choubey/web-for-agents 0.1.0 → 0.2.0

package/dist/{chunk-ZYMNO5AO.js → chunk-PSNLS4GL.js}

@@ -2,11 +2,16 @@ import {
   collapseWhitespace
 } from "./chunk-FEAKVNMG.js";
 import {
+  ACTION_ATTR,
+  ACTION_DESCRIPTION_ATTR,
+  ACTION_DESTRUCTIVE_ATTR,
   CONTENT_ATTR,
   EXCLUDE_ATTR,
   HIDDEN_SELECTOR,
+  WEBMCP_TOOLDESCRIPTION_ATTR,
+  WEBMCP_TOOLNAME_ATTR,
   resolveUrl
-} from "./chunk-AZG5MLNG.js";
+} from "./chunk-F7PN5R25.js";
 
 // src/shared/shortcut.ts
 var DEFAULT_SHORTCUT = "alt+shift+a";
@@ -37,6 +42,60 @@ function matchesShortcut(e, s) {
   return e.code === s.code && e.altKey === s.alt && e.shiftKey === s.shift && e.ctrlKey === s.ctrl && e.metaKey === s.meta;
 }
 
+// src/shared/content.ts
+var DEFAULT_CONTENT_SELECTOR = "main";
+var STRIP = `script, style, noscript, template, nav, [${EXCLUDE_ATTR}], ${HIDDEN_SELECTOR}`;
+var BODY_STRIP = "header, nav, footer, aside";
+function selectContentRegion(doc, opts = {}) {
+  const element = doc.createElement("div");
+  const take = (n) => opts.inPlace ? n : n.cloneNode(true);
+  let source;
+  const markers = Array.from(doc.querySelectorAll(`[${CONTENT_ATTR}]`)).filter(
+    (el) => !el.parentElement?.closest(`[${CONTENT_ATTR}]`)
+  );
+  let selected = null;
+  if (markers.length > 0) {
+    source = "marker";
+    for (const m of markers) element.appendChild(take(m));
+  } else {
+    try {
+      selected = doc.querySelector(opts.contentSelector ?? DEFAULT_CONTENT_SELECTOR);
+    } catch {
+      selected = null;
+    }
+    if (selected) {
+      source = "selector";
+      element.appendChild(take(selected));
+    } else {
+      source = "body";
+      for (const child of Array.from(doc.body?.childNodes ?? [])) {
+        element.appendChild(take(child));
+      }
+      for (const el of Array.from(element.querySelectorAll(BODY_STRIP))) el.remove();
+    }
+  }
+  for (const el of Array.from(element.querySelectorAll(STRIP))) el.remove();
+  return { element, source };
+}
+
+// src/shared/pageMeta.ts
+function readPageMeta(doc) {
+  const titleTag = (doc.title ?? "").trim();
+  const title = titleTag || (doc.querySelector("h1")?.textContent ?? "").trim();
+  const description = doc.querySelector('meta[name="description"]')?.getAttribute("content")?.trim() || null;
+  const robots = doc.querySelector('meta[name="robots"]')?.getAttribute("content") ?? null;
+  const canonical = doc.querySelector('link[rel~="canonical"]')?.getAttribute("href")?.trim() || null;
+  return {
+    title: title || null,
+    titleFromDocument: title !== "",
+    titleTag: titleTag !== "",
+    description,
+    robots,
+    noindex: /noindex/i.test(robots ?? ""),
+    canonical
+  };
+}
+
 // src/shared/markdown.ts
 import TurndownService from "turndown";
 import { gfm } from "turndown-plugin-gfm";
@@ -108,68 +167,45 @@ function buildPageMarkdown(title, contentHtml, pageUrl) {
 ${body}`.replace(/\n+$/, "\n");
 }
 
-// src/shared/content.ts
-var DEFAULT_CONTENT_SELECTOR = "main";
-var STRIP = `script, style, noscript, template, nav, [${EXCLUDE_ATTR}], ${HIDDEN_SELECTOR}`;
-var BODY_STRIP = "header, nav, footer, aside";
-function selectContentRegion(doc, opts = {}) {
-  const element = doc.createElement("div");
-  const take = (n) => opts.inPlace ? n : n.cloneNode(true);
-  let source;
-  const markers = Array.from(doc.querySelectorAll(`[${CONTENT_ATTR}]`)).filter(
-    (el) => !el.parentElement?.closest(`[${CONTENT_ATTR}]`)
-  );
-  let selected = null;
-  if (markers.length > 0) {
-    source = "marker";
-    for (const m of markers) element.appendChild(take(m));
-  } else {
-    try {
-      selected = doc.querySelector(opts.contentSelector ?? DEFAULT_CONTENT_SELECTOR);
-    } catch {
-      selected = null;
-    }
-    if (selected) {
-      source = "selector";
-      element.appendChild(take(selected));
-    } else {
-      source = "body";
-      for (const child of Array.from(doc.body?.childNodes ?? [])) {
-        element.appendChild(take(child));
-      }
-      for (const el of Array.from(element.querySelectorAll(BODY_STRIP))) el.remove();
-    }
-  }
-  for (const el of Array.from(element.querySelectorAll(STRIP))) el.remove();
-  return { element, source };
-}
-
-// src/shared/pageMeta.ts
-function readPageMeta(doc) {
-  const titleTag = (doc.title ?? "").trim();
-  const title = titleTag || (doc.querySelector("h1")?.textContent ?? "").trim();
-  const description = doc.querySelector('meta[name="description"]')?.getAttribute("content")?.trim() || null;
-  const robots = doc.querySelector('meta[name="robots"]')?.getAttribute("content") ?? null;
-  const canonical = doc.querySelector('link[rel~="canonical"]')?.getAttribute("href")?.trim() || null;
-  return {
-    title: title || null,
-    titleFromDocument: title !== "",
-    titleTag: titleTag !== "",
-    description,
-    robots,
-    noindex: /noindex/i.test(robots ?? ""),
-    canonical
-  };
+// src/shared/markupActions.ts
+function markupActions(doc) {
+  const out = [];
+  doc.querySelectorAll(`[${ACTION_ATTR}]`).forEach((el) => {
+    const name = el.getAttribute(ACTION_ATTR)?.trim();
+    if (!name) return;
+    out.push({
+      element: el,
+      name,
+      description: el.getAttribute(ACTION_DESCRIPTION_ATTR) || null,
+      kind: el.tagName === "FORM" ? "form" : "click",
+      destructive: el.getAttribute(ACTION_DESTRUCTIVE_ATTR) === "true",
+      source: "markup"
+    });
+  });
+  doc.querySelectorAll(`form[${WEBMCP_TOOLNAME_ATTR}]`).forEach((el) => {
+    const name = el.getAttribute(WEBMCP_TOOLNAME_ATTR)?.trim();
+    if (!name) return;
+    out.push({
+      element: el,
+      name,
+      description: el.getAttribute(WEBMCP_TOOLDESCRIPTION_ATTR) || null,
+      kind: "form",
+      destructive: false,
+      source: "webmcp-declarative"
+    });
+  });
+  return out;
 }
 
 export {
-  escapeMarkdownInline,
-  escapeLinkTarget,
-  buildPageMarkdown,
   DEFAULT_CONTENT_SELECTOR,
   selectContentRegion,
   DEFAULT_SHORTCUT,
   parseShortcut,
   matchesShortcut,
-  readPageMeta
+  readPageMeta,
+  escapeMarkdownInline,
+  escapeLinkTarget,
+  buildPageMarkdown,
+  markupActions
 };

package/dist/{chunk-HONWT6GK.js → chunk-SMFRAZAA.js}

@@ -1,19 +1,23 @@
 import {
   READY_GLOBAL
-} from "./chunk-AZG5MLNG.js";
+} from "./chunk-F7PN5R25.js";
 import {
   POLL_MS,
   RenderError,
   SERIALIZE_EXPR,
   SETTLED_POLL_MS,
+  SNAPSHOT_GRACE_MS,
+  TimeoutError,
+  assertNoPageErrors,
   blockedHint,
   createDeadline,
   createOriginPolicy,
   readyTimeoutMessage,
   runInPage,
+  snapshotWarning,
   waitForReady,
   withoutBlockedLoads
-} from "./chunk-5TNG4QMP.js";
+} from "./chunk-PGZH33B5.js";
 import {
   LOG_PREFIX,
   errorMessage
@@ -56,8 +60,11 @@ var PROBE_SCRIPT = `(function () {
     };
   }
   if (w.MutationObserver) {
+    // Content changes only: attribute churn (class, style, aria-*, data-* animations) never settles
+    // on an animated page. Attributes are still saved, so a late attribute-only update (a meta tag or
+    // an img src changed alone, after the content is quiet) may be captured before or after it lands.
     new w.MutationObserver(touch).observe(document, {
-      childList: true, subtree: true, attributes: true, characterData: true
+      childList: true, subtree: true, characterData: true
     });
   }
   w.addEventListener("error", function (e) {
@@ -74,12 +81,25 @@ var PROBE_SCRIPT = `(function () {
     p.errors.push(String((r && r.message) || r || "unhandledrejection")); touch();
   });
 })();`;
-function readyCheck(idleMs = IDLE_MS) {
-  return `(function () {
+function readyCheck(idleMs = IDLE_MS, readySelector) {
+  if (readySelector === void 0) {
+    return `(function () {
   var p = window.${PROBE_GLOBAL};
   if (!p) return { ready: false, errors: [], path: location.pathname };
   var ready = document.readyState === "complete" && p.inflight === 0 && Date.now() - p.last >= ${idleMs} && window.${READY_GLOBAL} !== false;
   return { ready: ready, errors: p.errors.slice(), path: location.pathname };
+})()`;
+  }
+  const sel = JSON.stringify(readySelector);
+  const invalid = JSON.stringify(`prerender.readySelector ${sel} is not a valid CSS selector: `);
+  return `(function () {
+  var p = window.${PROBE_GLOBAL};
+  if (!p) return { ready: false, errors: [], path: location.pathname };
+  var found;
+  try { found = document.querySelector(${sel}) !== null; }
+  catch (e) { return { ready: false, errors: [${invalid} + String((e && e.message) || e)], path: location.pathname }; }
+  var ready = document.readyState === "complete" && p.inflight === 0 && found && window.${READY_GLOBAL} !== false;
+  return { ready: ready, errors: p.errors.slice(), path: location.pathname };
 })()`;
 }
 var READY_CHECK = readyCheck(IDLE_MS);
@@ -250,6 +270,24 @@ function createHappyDomRenderer(opts = {}) {
       }
     }
   });
+  const waitUntilReady = async (page, blocked, deadline, timeoutMs) => {
+    const poll = async (check2) => {
+      const r = page.evaluate(check2);
+      return { ready: r.ready, errors: withoutBlockedLoads([...r.errors], blocked) };
+    };
+    if (opts.readySelector !== void 0) {
+      const check2 = readyCheck(IDLE_MS, opts.readySelector);
+      return waitForReady(() => poll(check2), timeoutMs, POLL_MS);
+    }
+    let timer;
+    const settled = await Promise.race([
+      page.waitUntilComplete().then(() => true),
+      new Promise((r) => timer = setTimeout(() => r(false), Math.min(deadline.remaining(), 2e3)))
+    ]).finally(() => clearTimeout(timer));
+    const idleMs = opts.idleMs ?? (settled ? SETTLED_IDLE_MS : IDLE_MS);
+    const check = readyCheck(idleMs);
+    return waitForReady(() => poll(check), timeoutMs, Math.min(settled ? SETTLED_POLL_MS : POLL_MS, idleMs));
+  };
   return {
     name: "happy-dom",
     async render(url, timeoutMs, inPage) {
@@ -265,48 +303,42 @@ function createHappyDomRenderer(opts = {}) {
       };
       const sink = pageSink(page);
       attach(sink);
-      const step = (work) => deadline.race(Promise.race([work, sink.failed]), timeoutText);
+      const within = (d, text, work) => d.race(Promise.race([work, sink.failed]), text);
+      const step = (work) => within(deadline, timeoutText, work);
+      let timedOut = false;
       try {
         try {
           await step(page.goto(url));
-          let timer;
-          const settled = await step(
-            Promise.race([
-              page.waitUntilComplete().then(() => true),
-              new Promise((r) => timer = setTimeout(() => r(false), Math.min(deadline.remaining(), 2e3)))
-            ]).finally(() => clearTimeout(timer))
-          );
-          const idleMs = opts.idleMs ?? (settled ? SETTLED_IDLE_MS : IDLE_MS);
-          const check = readyCheck(idleMs);
-          await step(
-            waitForReady(
-              async () => {
-                const r = page.evaluate(check);
-                return { ready: r.ready, errors: withoutBlockedLoads([...r.errors], state.blocked) };
-              },
-              timeoutMs,
-              Math.min(settled ? SETTLED_POLL_MS : POLL_MS, idleMs)
-            )
-          );
+          try {
+            await step(waitUntilReady(page, state.blocked, deadline, timeoutMs));
+          } catch (e) {
+            if (!(opts.onTimeout === "snapshot" && e instanceof TimeoutError)) throw e;
+            timedOut = true;
+            await assertNoPageErrors(() => {
+              const r = page.evaluate(readyCheck(IDLE_MS, opts.readySelector));
+              return withoutBlockedLoads([...r.errors], state.blocked);
+            });
+          }
         } catch (e) {
           throw new RenderError("happy-dom", errorMessage(e) + blockedHint(state.blocked) + syncHint(state.syncRefused));
         }
         const html = page.evaluate(SERIALIZE_EXPR);
         const finalPath = new URL(page.url).pathname;
+        const after = timedOut ? createDeadline(SNAPSHOT_GRACE_MS) : deadline;
+        const afterMs = timedOut ? SNAPSHOT_GRACE_MS : timeoutMs;
         const results = await runInPage(
           "happy-dom",
           inPage,
-          (expr) => step(Promise.resolve().then(() => page.evaluate(expr))),
-          deadline,
-          timeoutMs
+          (expr) => within(after, readyTimeoutMessage(afterMs), Promise.resolve().then(() => page.evaluate(expr))),
+          after,
+          afterMs
         );
-        return {
-          html,
-          finalPath,
-          inPage: results,
+        const warnings = [
+          ...timedOut ? [snapshotWarning(timeoutMs)] : [],
           // The page rendered anyway, but without what the refused requests would have loaded.
-          ...state.syncRefused.length > 0 ? { warnings: [...new Map(state.syncRefused.map((r) => [r.kind + r.url, r])).values()].map(syncRefusalText) } : {}
-        };
+          ...[...new Map(state.syncRefused.map((r) => [r.kind + r.url, r])).values()].map(syncRefusalText)
+        ];
+        return { html, finalPath, inPage: results, ...warnings.length > 0 ? { warnings } : {} };
       } finally {
         active.delete(state);
         await page.close().catch(() => {

package/dist/cli.js

@@ -2,12 +2,12 @@ import {
   audit,
   formatText,
   runCli
-} from "./chunk-IF32TPGR.js";
-import "./chunk-CKXHTGGT.js";
-import "./chunk-LCVI2PRP.js";
+} from "./chunk-P6JSPAA2.js";
+import "./chunk-CGIIB6AF.js";
+import "./chunk-PHTV5NAY.js";
 import "./chunk-LCLDQ3A3.js";
 import "./chunk-FEAKVNMG.js";
-import "./chunk-AZG5MLNG.js";
+import "./chunk-F7PN5R25.js";
 import "./chunk-JJN6FNJD.js";
 export {
   audit,

package/dist/{happyDom-XVDHYJDM.js → happyDom-WTYQ5UAC.js}

@@ -2,9 +2,9 @@ import {
   createHappyDomRenderer,
   syncHint,
   syncRefusalText
-} from "./chunk-HONWT6GK.js";
-import "./chunk-AZG5MLNG.js";
-import "./chunk-5TNG4QMP.js";
+} from "./chunk-SMFRAZAA.js";
+import "./chunk-F7PN5R25.js";
+import "./chunk-PGZH33B5.js";
 import "./chunk-JJN6FNJD.js";
 export {
   createHappyDomRenderer,