unbrowse 9.6.2 → 9.8.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "unbrowse",
-  "version": "9.6.2",
+  "version": "9.8.0",
   "repository": {
     "type": "git",
     "url": "git+https://github.com/unbrowse-ai/unbrowse.git"

package/runtime/cli.js

@@ -2350,7 +2350,7 @@ var init_telemetry = __esm(() => {
 });
 
 // .tmp-runtime-src/build-info.generated.ts
-var BUILD_RELEASE_VERSION = "9.6.2", BUILD_GIT_SHA = "d2d14a6629a0", BUILD_CODE_HASH = "5d9ebf619c61", BUILD_RELEASE_MANIFEST_BASE64 = "eyJzY2hlbWFfdmVyc2lvbiI6MSwicmVsZWFzZV92ZXJzaW9uIjoiOS42LjIiLCJnaXRfc2hhIjoiZDJkMTRhNjYyOWEwIiwiY29kZV9oYXNoIjoiNWQ5ZWJmNjE5YzYxIiwidHJhY2VfdmVyc2lvbiI6IjVkOWViZjYxOWM2MUBkMmQxNGE2NjI5YTAiLCJpc3N1ZWRfYXQiOiIyMDI2LTA2LTE4VDA0OjE3OjAzLjg1M1oifQ", BUILD_RELEASE_MANIFEST_SIGNATURE = "OefvyW2iLVVPQ-0HMg9Mz-bciCmP8LE5u7fVEss_98E", BUILD_DEFAULT_BACKEND_URL = "https://beta-api.unbrowse.ai", BUILD_DEFAULT_PROFILE = "";
+var BUILD_RELEASE_VERSION = "9.8.0", BUILD_GIT_SHA = "255142bb4c25", BUILD_CODE_HASH = "5d9ebf619c61", BUILD_RELEASE_MANIFEST_BASE64 = "eyJzY2hlbWFfdmVyc2lvbiI6MSwicmVsZWFzZV92ZXJzaW9uIjoiOS44LjAiLCJnaXRfc2hhIjoiMjU1MTQyYmI0YzI1IiwiY29kZV9oYXNoIjoiNWQ5ZWJmNjE5YzYxIiwidHJhY2VfdmVyc2lvbiI6IjVkOWViZjYxOWM2MUAyNTUxNDJiYjRjMjUiLCJpc3N1ZWRfYXQiOiIyMDI2LTA2LTE4VDA2OjA4OjA2LjY3MVoifQ", BUILD_RELEASE_MANIFEST_SIGNATURE = "Tw3ScHlFYGaEtPwKLhcPI_lgQUgjAZmhWKSi4fDFMw4", BUILD_DEFAULT_BACKEND_URL = "https://beta-api.unbrowse.ai", BUILD_DEFAULT_PROFILE = "";
 
 // .tmp-runtime-src/version.ts
 import { createHash as createHash7 } from "crypto";
@@ -46122,6 +46122,42 @@ function urlPathLooksListLike2(contextUrl) {
     return false;
   }
 }
+function entityPointerTemplate(href) {
+  let path7 = href;
+  try {
+    path7 = new URL(href, "https://_").pathname;
+  } catch {
+    path7 = href.split("?")[0];
+  }
+  const segs = path7.split("/").filter(Boolean);
+  if (segs.length === 0)
+    return null;
+  const shape = [];
+  let hasId = false;
+  for (const s of segs) {
+    const low = s.toLowerCase();
+    if (/\d{3,}/.test(low) || low.length > 30 || /^[0-9a-f-]{8,}$/.test(low) || /-\d{2,}$/.test(low)) {
+      shape.push("{id}");
+      hasId = true;
+    } else {
+      shape.push(low);
+    }
+  }
+  return hasId ? shape.slice(0, 3).join("/") : null;
+}
+function linksFormEntityCollection(hrefs, min = 4) {
+  const groups = new Map;
+  for (const href of hrefs) {
+    const t = entityPointerTemplate(href);
+    if (!t)
+      continue;
+    const n = (groups.get(t) ?? 0) + 1;
+    if (n >= min)
+      return true;
+    groups.set(t, n);
+  }
+  return false;
+}
 function cardinalityMatches2(intent, subject, opts) {
   const wantsMany = isListLikeIntent2(intent) || urlPathLooksListLike2(opts?.contextUrl);
   if (!wantsMany)
@@ -46694,6 +46730,7 @@ __export(exports_capture, {
   tagRequestProvenance: () => tagRequestProvenance,
   shutdownAllBrowsers: () => shutdownAllBrowsers,
   shouldStopHydrationWait: () => shouldStopHydrationWait,
+  shouldScrollStimulate: () => shouldScrollStimulate,
   selectPerformanceReplayCandidates: () => selectPerformanceReplayCandidates,
   registerDocumentStartScript: () => registerDocumentStartScript,
   navigatePageForCapture: () => navigatePageForCapture,
@@ -46956,6 +46993,9 @@ function extractRouteHint(url) {
   } catch {}
   return null;
 }
+function shouldScrollStimulate(captureUrl, intent) {
+  return isListLikeIntent2(intent) || urlPathLooksListLike2(captureUrl);
+}
 function deriveIntentHints(captureUrl, intent) {
   const derivedHints = new Set;
   if (captureUrl) {
@@ -47583,8 +47623,7 @@ async function waitForContentReady(tabId, captureUrl, intent, responseBodies) {
       log("capture", `intent-aware wait: already captured API matching one of [${[...derivedHints].join(", ")}], skipping`);
     }
   }
-  const lowerIntent = intent?.toLowerCase() ?? "";
-  if (captureUrl && responseBodies && (/search|explore|trending|tabs|discover/i.test(captureUrl) || /\b(person|people|profile|profiles|user|users|member|members|company|companies|organization|organisations|business|post|posts|tweet|tweets|status|statuses)\b/.test(lowerIntent))) {
+  if (captureUrl && responseBodies && shouldScrollStimulate(captureUrl, intent)) {
     try {
       const before = responseBodies.size;
       await evaluate(tabId, "window.scrollTo(0, Math.max(window.innerHeight, Math.min(document.body.scrollHeight, window.innerHeight * 2)))");
@@ -48658,6 +48697,7 @@ var init_capture = __esm(async () => {
   init_domain();
   init_logger();
   init_header_classify();
+  init_cardinality2();
   init_browser_access();
   await init_vault();
   waitQueue = [];
@@ -56326,6 +56366,215 @@ var init_curl_impersonate_fallback = __esm(() => {
   };
 });
 
+// .tmp-runtime-src/execution/search-forms.ts
+var exports_search_forms = {};
+__export(exports_search_forms, {
+  isStructuredSearchForm: () => isStructuredSearchForm,
+  fillSearchRoute: () => fillSearchRoute,
+  detectSearchForms: () => detectSearchForms,
+  deriveSearchRouteTemplates: () => deriveSearchRouteTemplates
+});
+function deriveSearchRouteTemplates(html, minDistinct = 4) {
+  const hrefs = new Set;
+  for (const m of html.matchAll(/href\s*=\s*["'](\/[^"'?#\s]+)["']/gi))
+    hrefs.add(m[1]);
+  const groups = new Map;
+  for (const h of hrefs) {
+    const segs = h.split("/").filter(Boolean);
+    if (segs.length < 1 || segs.length > 4)
+      continue;
+    for (let i = 0;i < segs.length; i++) {
+      const val = segs[i];
+      if (!/^[a-z][a-z0-9-]{1,40}$/i.test(val) || /\d{3,}/.test(val) || /\.[a-z0-9]{1,5}$/i.test(val))
+        continue;
+      const shape = segs.map((s, j) => j === i ? "{query}" : s).join("/");
+      const trailing = h.endsWith("/") ? "/" : "";
+      const key = `/${shape}${trailing}`;
+      if (!groups.has(key))
+        groups.set(key, new Set);
+      groups.get(key).add(val.toLowerCase());
+    }
+  }
+  const out = [];
+  for (const [template, vals] of groups) {
+    if (vals.size >= minDistinct)
+      out.push({ template, samples: [...vals].slice(0, 5), count: vals.size });
+  }
+  return out.sort((a, b) => b.count - a.count);
+}
+function fillSearchRoute(origin, template, query) {
+  const slug = encodeURIComponent(query.trim().toLowerCase());
+  return origin.replace(/\/+$/, "") + template.replace("{query}", slug);
+}
+function isStructuredSearchForm(spec) {
+  return spec.fields.length > 0 && !!spec.submit_selector;
+}
+function formSelectorFromElement(attribs, index) {
+  const id = attribs.id;
+  if (id)
+    return `form#${id}`;
+  const name = attribs.name;
+  if (name)
+    return `form[name="${name}"]`;
+  const action2 = attribs.action;
+  if (action2)
+    return `form[action="${action2}"]`;
+  return `form:nth-of-type(${index + 1})`;
+}
+function inputSelectorFromElement(attribs, tagName) {
+  const id = attribs.id;
+  if (id)
+    return `#${id}`;
+  const name = attribs.name;
+  if (name)
+    return `${tagName}[name="${name}"]`;
+  return tagName;
+}
+function mapInputType(typeAttr, tagName) {
+  if (tagName === "select")
+    return "select";
+  if (tagName === "textarea")
+    return "text";
+  const t = (typeAttr ?? "text").toLowerCase();
+  if (t === "radio")
+    return "radio";
+  if (t === "checkbox")
+    return "checkbox";
+  if (t === "date")
+    return "date";
+  if (t === "hidden")
+    return "hidden";
+  if (t === "submit" || t === "button" || t === "image" || t === "reset")
+    return null;
+  if (t === "password" || t === "file")
+    return null;
+  if (SUPPORTED_INPUT_TYPES.has(t))
+    return "text";
+  return "text";
+}
+function parseAttrs(attrStr) {
+  const attrs = {};
+  const attrRegex = /(\w[\w-]*)(?:\s*=\s*(?:"([^"]*)"|'([^']*)'|(\S+)))?/g;
+  let m;
+  while ((m = attrRegex.exec(attrStr)) !== null) {
+    attrs[m[1]] = m[2] ?? m[3] ?? m[4] ?? "";
+  }
+  return attrs;
+}
+function detectSearchForms(html) {
+  const results = [];
+  const formRegex = /<form([^>]*)>([\s\S]*?)<\/form>/gi;
+  let formMatch;
+  let formIndex = 0;
+  while ((formMatch = formRegex.exec(html)) !== null) {
+    const formAttrs = formMatch[1];
+    const formBody = formMatch[2];
+    const formElAttrs = parseAttrs(formAttrs);
+    const fieldRegex = /<(input|select|textarea)([^>]*)\/?>/gi;
+    let fieldMatch;
+    const fields = [];
+    const seenNames = new Set;
+    let hasLoginField = false;
+    let hasSearchLikeField = false;
+    while ((fieldMatch = fieldRegex.exec(formBody)) !== null) {
+      const tagName = fieldMatch[1].toLowerCase();
+      const fieldAttrs = parseAttrs(fieldMatch[2]);
+      const name = fieldAttrs.name ?? "";
+      const typeAttr = fieldAttrs.type;
+      if (LOGIN_FIELD_NAMES.has(name.toLowerCase()) || typeAttr === "password") {
+        hasLoginField = true;
+      }
+      if (SEARCH_FIELD_NAMES.has(name.toLowerCase())) {
+        hasSearchLikeField = true;
+      }
+      const mappedType = mapInputType(typeAttr, tagName);
+      if (!mappedType)
+        continue;
+      if (!name && mappedType !== "text")
+        continue;
+      if (seenNames.has(name) && mappedType !== "radio")
+        continue;
+      if (name)
+        seenNames.add(name);
+      let options;
+      if (tagName === "select") {
+        const optRegex = /<option[^>]*value="([^"]*)"[^>]*>/gi;
+        let optMatch;
+        options = [];
+        while ((optMatch = optRegex.exec(formBody)) !== null) {
+          options.push(optMatch[1]);
+        }
+        if (options.length === 0)
+          options = undefined;
+      }
+      fields.push({
+        name: name || `unnamed_${fields.length}`,
+        type: mappedType,
+        selector: inputSelectorFromElement(fieldAttrs, tagName),
+        ...options ? { options } : {},
+        required: fieldAttrs.required !== undefined
+      });
+    }
+    let submitSelector = "";
+    if (/<button[^>]*type\s*=\s*"submit"/i.test(formBody)) {
+      submitSelector = "button[type=submit]";
+    } else if (/<input[^>]*type\s*=\s*"submit"/i.test(formBody)) {
+      submitSelector = 'input[type="submit"]';
+    } else if (/<button/i.test(formBody)) {
+      submitSelector = "button";
+    }
+    const nonHiddenFields = fields.filter((f) => f.type !== "hidden");
+    if (!hasLoginField && nonHiddenFields.length > 0 && submitSelector && (hasSearchLikeField || nonHiddenFields.length >= 1)) {
+      const formSelector = formSelectorFromElement(formElAttrs, formIndex);
+      results.push({
+        form_selector: formSelector,
+        submit_selector: submitSelector,
+        fields
+      });
+    }
+    formIndex++;
+  }
+  return results;
+}
+var SEARCH_FIELD_NAMES, LOGIN_FIELD_NAMES, SUPPORTED_INPUT_TYPES;
+var init_search_forms = __esm(() => {
+  SEARCH_FIELD_NAMES = new Set([
+    "q",
+    "query",
+    "search",
+    "keyword",
+    "keywords",
+    "term",
+    "terms",
+    "find",
+    "lookup",
+    "filter",
+    "s",
+    "text",
+    "input"
+  ]);
+  LOGIN_FIELD_NAMES = new Set([
+    "password",
+    "passwd",
+    "pass",
+    "pwd",
+    "confirm_password",
+    "username",
+    "email",
+    "login",
+    "user"
+  ]);
+  SUPPORTED_INPUT_TYPES = new Set([
+    "text",
+    "search",
+    "hidden",
+    "date",
+    "number",
+    "tel",
+    "email"
+  ]);
+});
+
 // node_modules/.bun/@mixmark-io+domino@2.2.0/node_modules/@mixmark-io/domino/lib/Event.js
 var require_Event = __commonJS((exports, module) => {
   module.exports = Event2;
@@ -72977,20 +73226,24 @@ function buildDirectDocumentResult(url, html, contentType, intent) {
       const hits = intentTokens.filter((tok) => haystack.includes(tok));
       const hitRate = hits.length / intentTokens.length;
       if (hitRate < 0.34) {
-        return {
-          rejected: true,
-          reason: "intent_mismatch",
-          evidence: {
-            intent_tokens: intentTokens,
-            response_token_hits: hits,
-            response_token_hit_rate: hitRate,
-            html_bytes: html.length
-          }
-        };
+        const isCollection = isListLikeIntent2(intent) && linksFormEntityCollection(Array.from(html.matchAll(/href\s*=\s*["']([^"']+)["']/gi), (m) => m[1]));
+        if (!isCollection) {
+          return {
+            rejected: true,
+            reason: "intent_mismatch",
+            evidence: {
+              intent_tokens: intentTokens,
+              response_token_hits: hits,
+              response_token_hit_rate: hitRate,
+              html_bytes: html.length
+            }
+          };
+        }
       }
     }
   }
   const { url_template, input_params, path_params, query } = extractHtmlHoles(url);
+  const routing_candidates = buildSearchRouteCandidates(html, url, intent);
   return {
     rejected: false,
     title,
@@ -73004,12 +73257,40 @@ function buildDirectDocumentResult(url, html, contentType, intent) {
     text_excerpt: bodyText.slice(0, MARKDOWN_BUDGET),
     markdown: htmlToMarkdownSafe(html, bodyText),
     tables: extractTables(html),
+    ...routing_candidates.length > 0 ? { routing_candidates } : {},
     extraction: {
       source: "direct-document",
       rejected: false
     }
   };
 }
+function intentQueryTerm(intent, url) {
+  let domTokens = new Set;
+  try {
+    domTokens = new Set(new URL(url).hostname.toLowerCase().split(/[.-]/));
+  } catch {}
+  const toks = (intent.toLowerCase().match(/[a-z][a-z0-9]{2,}/g) ?? []).filter((t) => !QUERY_STOPWORDS.has(t) && !domTokens.has(t));
+  return [...new Set(toks)].join(" ").trim();
+}
+function buildSearchRouteCandidates(html, url, intent) {
+  if (!intent || !isListLikeIntent2(intent))
+    return [];
+  const queryTerm = intentQueryTerm(intent, url);
+  if (!queryTerm)
+    return [];
+  let origin = "";
+  try {
+    origin = new URL(url).origin;
+  } catch {
+    return [];
+  }
+  return deriveSearchRouteTemplates(html).slice(0, 3).map((t) => ({
+    url: fillSearchRoute(origin, t.template, queryTerm),
+    template: t.template,
+    query: queryTerm,
+    samples: t.samples
+  }));
+}
 async function fetchDirectDocument(url) {
   if (!isDirectDocumentEligibleUrl(url))
     return null;
@@ -73197,10 +73478,12 @@ function cellText(html) {
 function decodeHtmlEntityText(input) {
   return input.replace(/&nbsp;/g, " ").replace(/&amp;/g, "&").replace(/&lt;/g, "<").replace(/&gt;/g, ">").replace(/&quot;/g, '"').replace(/&#39;/g, "'");
 }
-var HTML_RE, MIN_DIRECT_DOCUMENT_HTML_BYTES = 5000, CHALLENGE_RE, INTERSTITIAL_RE, MIN_DIRECT_DOCUMENT_BODY_TEXT = 500, SPA_HYDRATION_RE, SPA_HYDRATION_BODY_TEXT_FLOOR = 2000, SPA_ROOT_CONTAINER_RE, PARKED_RE, INTENT_STOPWORDS, MARKDOWN_BUDGET, MAX_TABLES = 10, MAX_TABLE_ROWS = 50, buildBloombergDirectDocumentResult;
+var HTML_RE, MIN_DIRECT_DOCUMENT_HTML_BYTES = 5000, CHALLENGE_RE, INTERSTITIAL_RE, MIN_DIRECT_DOCUMENT_BODY_TEXT = 500, SPA_HYDRATION_RE, SPA_HYDRATION_BODY_TEXT_FLOOR = 2000, SPA_ROOT_CONTAINER_RE, PARKED_RE, INTENT_STOPWORDS, MARKDOWN_BUDGET, MAX_TABLES = 10, MAX_TABLE_ROWS = 50, QUERY_STOPWORDS, buildBloombergDirectDocumentResult;
 var init_direct_document = __esm(() => {
   init_curl_impersonate_fallback();
   init_proxy_fetch();
+  init_cardinality2();
+  init_search_forms();
   HTML_RE = /text\/html|application\/xhtml\+xml/i;
   CHALLENGE_RE = /\b(access denied|are you a robot|captcha|just a moment|pardon our interruption|robot check|unusual traffic|verify you are human)\b/i;
   INTERSTITIAL_RE = /\b(please wait for verification|just a moment|cf-mitigated|datadome|akamai bot|perimeterx|sign in to continue|log in to (?:continue|access)|javascript is not available)\b/i;
@@ -73283,6 +73566,7 @@ var init_direct_document = __esm(() => {
     "look"
   ]);
   MARKDOWN_BUDGET = Math.max(1000, Number(process.env.UNBROWSE_MARKDOWN_BUDGET ?? "12000") || 12000);
+  QUERY_STOPWORDS = new Set(("resolve unbrowse execute run walk go fetch open view want need please " + "find search browse list lookup discover show get me a an the on of for in to " + "with and or all my your this that some good best top new latest cheap near").split(" "));
   buildBloombergDirectDocumentResult = buildDirectDocumentResult;
 });
 
@@ -121955,181 +122239,6 @@ function clampToFloor(score, demotion, floor) {
 }
 var HARD_NEGATIVE_FLOOR = -2000, WEAK_NEGATIVE_FLOOR = -400, PAGE_ARTIFACT_DEMOTION = 800, EMPTY_ENTITY_BAG_DEMOTION = 650, EMPTY_ENTITY_BAG_FLOOR = -700;
 
-// .tmp-runtime-src/execution/search-forms.ts
-var exports_search_forms = {};
-__export(exports_search_forms, {
-  isStructuredSearchForm: () => isStructuredSearchForm,
-  detectSearchForms: () => detectSearchForms
-});
-function isStructuredSearchForm(spec) {
-  return spec.fields.length > 0 && !!spec.submit_selector;
-}
-function formSelectorFromElement(attribs, index2) {
-  const id = attribs.id;
-  if (id)
-    return `form#${id}`;
-  const name = attribs.name;
-  if (name)
-    return `form[name="${name}"]`;
-  const action2 = attribs.action;
-  if (action2)
-    return `form[action="${action2}"]`;
-  return `form:nth-of-type(${index2 + 1})`;
-}
-function inputSelectorFromElement(attribs, tagName) {
-  const id = attribs.id;
-  if (id)
-    return `#${id}`;
-  const name = attribs.name;
-  if (name)
-    return `${tagName}[name="${name}"]`;
-  return tagName;
-}
-function mapInputType(typeAttr, tagName) {
-  if (tagName === "select")
-    return "select";
-  if (tagName === "textarea")
-    return "text";
-  const t = (typeAttr ?? "text").toLowerCase();
-  if (t === "radio")
-    return "radio";
-  if (t === "checkbox")
-    return "checkbox";
-  if (t === "date")
-    return "date";
-  if (t === "hidden")
-    return "hidden";
-  if (t === "submit" || t === "button" || t === "image" || t === "reset")
-    return null;
-  if (t === "password" || t === "file")
-    return null;
-  if (SUPPORTED_INPUT_TYPES.has(t))
-    return "text";
-  return "text";
-}
-function parseAttrs(attrStr) {
-  const attrs = {};
-  const attrRegex = /(\w[\w-]*)(?:\s*=\s*(?:"([^"]*)"|'([^']*)'|(\S+)))?/g;
-  let m;
-  while ((m = attrRegex.exec(attrStr)) !== null) {
-    attrs[m[1]] = m[2] ?? m[3] ?? m[4] ?? "";
-  }
-  return attrs;
-}
-function detectSearchForms(html3) {
-  const results = [];
-  const formRegex = /<form([^>]*)>([\s\S]*?)<\/form>/gi;
-  let formMatch;
-  let formIndex = 0;
-  while ((formMatch = formRegex.exec(html3)) !== null) {
-    const formAttrs = formMatch[1];
-    const formBody = formMatch[2];
-    const formElAttrs = parseAttrs(formAttrs);
-    const fieldRegex = /<(input|select|textarea)([^>]*)\/?>/gi;
-    let fieldMatch;
-    const fields = [];
-    const seenNames = new Set;
-    let hasLoginField = false;
-    let hasSearchLikeField = false;
-    while ((fieldMatch = fieldRegex.exec(formBody)) !== null) {
-      const tagName = fieldMatch[1].toLowerCase();
-      const fieldAttrs = parseAttrs(fieldMatch[2]);
-      const name = fieldAttrs.name ?? "";
-      const typeAttr = fieldAttrs.type;
-      if (LOGIN_FIELD_NAMES.has(name.toLowerCase()) || typeAttr === "password") {
-        hasLoginField = true;
-      }
-      if (SEARCH_FIELD_NAMES.has(name.toLowerCase())) {
-        hasSearchLikeField = true;
-      }
-      const mappedType = mapInputType(typeAttr, tagName);
-      if (!mappedType)
-        continue;
-      if (!name && mappedType !== "text")
-        continue;
-      if (seenNames.has(name) && mappedType !== "radio")
-        continue;
-      if (name)
-        seenNames.add(name);
-      let options;
-      if (tagName === "select") {
-        const optRegex = /<option[^>]*value="([^"]*)"[^>]*>/gi;
-        let optMatch;
-        options = [];
-        while ((optMatch = optRegex.exec(formBody)) !== null) {
-          options.push(optMatch[1]);
-        }
-        if (options.length === 0)
-          options = undefined;
-      }
-      fields.push({
-        name: name || `unnamed_${fields.length}`,
-        type: mappedType,
-        selector: inputSelectorFromElement(fieldAttrs, tagName),
-        ...options ? { options } : {},
-        required: fieldAttrs.required !== undefined
-      });
-    }
-    let submitSelector = "";
-    if (/<button[^>]*type\s*=\s*"submit"/i.test(formBody)) {
-      submitSelector = "button[type=submit]";
-    } else if (/<input[^>]*type\s*=\s*"submit"/i.test(formBody)) {
-      submitSelector = 'input[type="submit"]';
-    } else if (/<button/i.test(formBody)) {
-      submitSelector = "button";
-    }
-    const nonHiddenFields = fields.filter((f) => f.type !== "hidden");
-    if (!hasLoginField && nonHiddenFields.length > 0 && submitSelector && (hasSearchLikeField || nonHiddenFields.length >= 1)) {
-      const formSelector = formSelectorFromElement(formElAttrs, formIndex);
-      results.push({
-        form_selector: formSelector,
-        submit_selector: submitSelector,
-        fields
-      });
-    }
-    formIndex++;
-  }
-  return results;
-}
-var SEARCH_FIELD_NAMES, LOGIN_FIELD_NAMES, SUPPORTED_INPUT_TYPES;
-var init_search_forms = __esm(() => {
-  SEARCH_FIELD_NAMES = new Set([
-    "q",
-    "query",
-    "search",
-    "keyword",
-    "keywords",
-    "term",
-    "terms",
-    "find",
-    "lookup",
-    "filter",
-    "s",
-    "text",
-    "input"
-  ]);
-  LOGIN_FIELD_NAMES = new Set([
-    "password",
-    "passwd",
-    "pass",
-    "pwd",
-    "confirm_password",
-    "username",
-    "email",
-    "login",
-    "user"
-  ]);
-  SUPPORTED_INPUT_TYPES = new Set([
-    "text",
-    "search",
-    "hidden",
-    "date",
-    "number",
-    "tel",
-    "email"
-  ]);
-});
-
 // .tmp-runtime-src/state/stateless.ts
 function isStateless() {
   const v = process.env.UNBROWSE_STATELESS;
@@ -123558,7 +123667,7 @@ function isProtobufContentType(contentType) {
 function isProtobufLikeEndpoint(url, contentType) {
   if (isProtobufContentType(contentType))
     return true;
-  return /\/(field-data-proto|proto|protobuf)(\/|$|-)/i.test(url);
+  return /[-/](proto|protobuf)(\/|$|-)/i.test(url);
 }
 function decodeProtobufBytes(bytes) {
   return decodeBytes(bytes, "bytes");

package/runtime/mcp.js

@@ -36310,7 +36310,7 @@ var init_cached_resolution = __esm(() => {
 });
 
 // .tmp-runtime-src/build-info.generated.ts
-var BUILD_RELEASE_VERSION = "9.6.2", BUILD_GIT_SHA = "d2d14a6629a0", BUILD_CODE_HASH = "5d9ebf619c61", BUILD_RELEASE_MANIFEST_BASE64 = "eyJzY2hlbWFfdmVyc2lvbiI6MSwicmVsZWFzZV92ZXJzaW9uIjoiOS42LjIiLCJnaXRfc2hhIjoiZDJkMTRhNjYyOWEwIiwiY29kZV9oYXNoIjoiNWQ5ZWJmNjE5YzYxIiwidHJhY2VfdmVyc2lvbiI6IjVkOWViZjYxOWM2MUBkMmQxNGE2NjI5YTAiLCJpc3N1ZWRfYXQiOiIyMDI2LTA2LTE4VDA0OjE3OjAzLjg1M1oifQ", BUILD_RELEASE_MANIFEST_SIGNATURE = "OefvyW2iLVVPQ-0HMg9Mz-bciCmP8LE5u7fVEss_98E", BUILD_DEFAULT_BACKEND_URL = "https://beta-api.unbrowse.ai", BUILD_DEFAULT_PROFILE = "";
+var BUILD_RELEASE_VERSION = "9.8.0", BUILD_GIT_SHA = "255142bb4c25", BUILD_CODE_HASH = "5d9ebf619c61", BUILD_RELEASE_MANIFEST_BASE64 = "eyJzY2hlbWFfdmVyc2lvbiI6MSwicmVsZWFzZV92ZXJzaW9uIjoiOS44LjAiLCJnaXRfc2hhIjoiMjU1MTQyYmI0YzI1IiwiY29kZV9oYXNoIjoiNWQ5ZWJmNjE5YzYxIiwidHJhY2VfdmVyc2lvbiI6IjVkOWViZjYxOWM2MUAyNTUxNDJiYjRjMjUiLCJpc3N1ZWRfYXQiOiIyMDI2LTA2LTE4VDA2OjA4OjA2LjY3MVoifQ", BUILD_RELEASE_MANIFEST_SIGNATURE = "Tw3ScHlFYGaEtPwKLhcPI_lgQUgjAZmhWKSi4fDFMw4", BUILD_DEFAULT_BACKEND_URL = "https://beta-api.unbrowse.ai", BUILD_DEFAULT_PROFILE = "";
 
 // .tmp-runtime-src/version.ts
 import { createHash as createHash4 } from "crypto";
@@ -43252,6 +43252,42 @@ function urlPathLooksListLike(contextUrl) {
     return false;
   }
 }
+function entityPointerTemplate(href) {
+  let path5 = href;
+  try {
+    path5 = new URL(href, "https://_").pathname;
+  } catch {
+    path5 = href.split("?")[0];
+  }
+  const segs = path5.split("/").filter(Boolean);
+  if (segs.length === 0)
+    return null;
+  const shape = [];
+  let hasId = false;
+  for (const s of segs) {
+    const low = s.toLowerCase();
+    if (/\d{3,}/.test(low) || low.length > 30 || /^[0-9a-f-]{8,}$/.test(low) || /-\d{2,}$/.test(low)) {
+      shape.push("{id}");
+      hasId = true;
+    } else {
+      shape.push(low);
+    }
+  }
+  return hasId ? shape.slice(0, 3).join("/") : null;
+}
+function linksFormEntityCollection(hrefs, min = 4) {
+  const groups = new Map;
+  for (const href of hrefs) {
+    const t = entityPointerTemplate(href);
+    if (!t)
+      continue;
+    const n = (groups.get(t) ?? 0) + 1;
+    if (n >= min)
+      return true;
+    groups.set(t, n);
+  }
+  return false;
+}
 function cardinalityMatches(intent, subject, opts) {
   const wantsMany = isListLikeIntent(intent) || urlPathLooksListLike(opts?.contextUrl);
   if (!wantsMany)
@@ -43827,6 +43863,7 @@ __export(exports_capture, {
   tagRequestProvenance: () => tagRequestProvenance,
   shutdownAllBrowsers: () => shutdownAllBrowsers,
   shouldStopHydrationWait: () => shouldStopHydrationWait,
+  shouldScrollStimulate: () => shouldScrollStimulate,
   selectPerformanceReplayCandidates: () => selectPerformanceReplayCandidates,
   registerDocumentStartScript: () => registerDocumentStartScript,
   navigatePageForCapture: () => navigatePageForCapture,
@@ -44089,6 +44126,9 @@ function extractRouteHint(url) {
   } catch {}
   return null;
 }
+function shouldScrollStimulate(captureUrl, intent) {
+  return isListLikeIntent(intent) || urlPathLooksListLike(captureUrl);
+}
 function deriveIntentHints(captureUrl, intent) {
   const derivedHints = new Set;
   if (captureUrl) {
@@ -44716,8 +44756,7 @@ async function waitForContentReady(tabId, captureUrl, intent, responseBodies) {
       log("capture", `intent-aware wait: already captured API matching one of [${[...derivedHints].join(", ")}], skipping`);
     }
   }
-  const lowerIntent = intent?.toLowerCase() ?? "";
-  if (captureUrl && responseBodies && (/search|explore|trending|tabs|discover/i.test(captureUrl) || /\b(person|people|profile|profiles|user|users|member|members|company|companies|organization|organisations|business|post|posts|tweet|tweets|status|statuses)\b/.test(lowerIntent))) {
+  if (captureUrl && responseBodies && shouldScrollStimulate(captureUrl, intent)) {
     try {
       const before = responseBodies.size;
       await evaluate(tabId, "window.scrollTo(0, Math.max(window.innerHeight, Math.min(document.body.scrollHeight, window.innerHeight * 2)))");
@@ -45791,6 +45830,7 @@ var init_capture = __esm(async () => {
   init_domain();
   init_logger();
   init_header_classify();
+  init_cardinality();
   init_browser_access();
   await init_vault();
   waitQueue = [];
@@ -54383,6 +54423,215 @@ var init_curl_impersonate_fallback = __esm(() => {
   };
 });
 
+// .tmp-runtime-src/execution/search-forms.ts
+var exports_search_forms = {};
+__export(exports_search_forms, {
+  isStructuredSearchForm: () => isStructuredSearchForm,
+  fillSearchRoute: () => fillSearchRoute,
+  detectSearchForms: () => detectSearchForms,
+  deriveSearchRouteTemplates: () => deriveSearchRouteTemplates
+});
+function deriveSearchRouteTemplates(html, minDistinct = 4) {
+  const hrefs = new Set;
+  for (const m of html.matchAll(/href\s*=\s*["'](\/[^"'?#\s]+)["']/gi))
+    hrefs.add(m[1]);
+  const groups = new Map;
+  for (const h of hrefs) {
+    const segs = h.split("/").filter(Boolean);
+    if (segs.length < 1 || segs.length > 4)
+      continue;
+    for (let i = 0;i < segs.length; i++) {
+      const val = segs[i];
+      if (!/^[a-z][a-z0-9-]{1,40}$/i.test(val) || /\d{3,}/.test(val) || /\.[a-z0-9]{1,5}$/i.test(val))
+        continue;
+      const shape = segs.map((s, j) => j === i ? "{query}" : s).join("/");
+      const trailing = h.endsWith("/") ? "/" : "";
+      const key = `/${shape}${trailing}`;
+      if (!groups.has(key))
+        groups.set(key, new Set);
+      groups.get(key).add(val.toLowerCase());
+    }
+  }
+  const out = [];
+  for (const [template, vals] of groups) {
+    if (vals.size >= minDistinct)
+      out.push({ template, samples: [...vals].slice(0, 5), count: vals.size });
+  }
+  return out.sort((a, b) => b.count - a.count);
+}
+function fillSearchRoute(origin, template, query) {
+  const slug = encodeURIComponent(query.trim().toLowerCase());
+  return origin.replace(/\/+$/, "") + template.replace("{query}", slug);
+}
+function isStructuredSearchForm(spec) {
+  return spec.fields.length > 0 && !!spec.submit_selector;
+}
+function formSelectorFromElement(attribs, index) {
+  const id = attribs.id;
+  if (id)
+    return `form#${id}`;
+  const name = attribs.name;
+  if (name)
+    return `form[name="${name}"]`;
+  const action2 = attribs.action;
+  if (action2)
+    return `form[action="${action2}"]`;
+  return `form:nth-of-type(${index + 1})`;
+}
+function inputSelectorFromElement(attribs, tagName) {
+  const id = attribs.id;
+  if (id)
+    return `#${id}`;
+  const name = attribs.name;
+  if (name)
+    return `${tagName}[name="${name}"]`;
+  return tagName;
+}
+function mapInputType(typeAttr, tagName) {
+  if (tagName === "select")
+    return "select";
+  if (tagName === "textarea")
+    return "text";
+  const t = (typeAttr ?? "text").toLowerCase();
+  if (t === "radio")
+    return "radio";
+  if (t === "checkbox")
+    return "checkbox";
+  if (t === "date")
+    return "date";
+  if (t === "hidden")
+    return "hidden";
+  if (t === "submit" || t === "button" || t === "image" || t === "reset")
+    return null;
+  if (t === "password" || t === "file")
+    return null;
+  if (SUPPORTED_INPUT_TYPES.has(t))
+    return "text";
+  return "text";
+}
+function parseAttrs(attrStr) {
+  const attrs = {};
+  const attrRegex = /(\w[\w-]*)(?:\s*=\s*(?:"([^"]*)"|'([^']*)'|(\S+)))?/g;
+  let m;
+  while ((m = attrRegex.exec(attrStr)) !== null) {
+    attrs[m[1]] = m[2] ?? m[3] ?? m[4] ?? "";
+  }
+  return attrs;
+}
+function detectSearchForms(html) {
+  const results = [];
+  const formRegex = /<form([^>]*)>([\s\S]*?)<\/form>/gi;
+  let formMatch;
+  let formIndex = 0;
+  while ((formMatch = formRegex.exec(html)) !== null) {
+    const formAttrs = formMatch[1];
+    const formBody = formMatch[2];
+    const formElAttrs = parseAttrs(formAttrs);
+    const fieldRegex = /<(input|select|textarea)([^>]*)\/?>/gi;
+    let fieldMatch;
+    const fields = [];
+    const seenNames = new Set;
+    let hasLoginField = false;
+    let hasSearchLikeField = false;
+    while ((fieldMatch = fieldRegex.exec(formBody)) !== null) {
+      const tagName = fieldMatch[1].toLowerCase();
+      const fieldAttrs = parseAttrs(fieldMatch[2]);
+      const name = fieldAttrs.name ?? "";
+      const typeAttr = fieldAttrs.type;
+      if (LOGIN_FIELD_NAMES.has(name.toLowerCase()) || typeAttr === "password") {
+        hasLoginField = true;
+      }
+      if (SEARCH_FIELD_NAMES.has(name.toLowerCase())) {
+        hasSearchLikeField = true;
+      }
+      const mappedType = mapInputType(typeAttr, tagName);
+      if (!mappedType)
+        continue;
+      if (!name && mappedType !== "text")
+        continue;
+      if (seenNames.has(name) && mappedType !== "radio")
+        continue;
+      if (name)
+        seenNames.add(name);
+      let options;
+      if (tagName === "select") {
+        const optRegex = /<option[^>]*value="([^"]*)"[^>]*>/gi;
+        let optMatch;
+        options = [];
+        while ((optMatch = optRegex.exec(formBody)) !== null) {
+          options.push(optMatch[1]);
+        }
+        if (options.length === 0)
+          options = undefined;
+      }
+      fields.push({
+        name: name || `unnamed_${fields.length}`,
+        type: mappedType,
+        selector: inputSelectorFromElement(fieldAttrs, tagName),
+        ...options ? { options } : {},
+        required: fieldAttrs.required !== undefined
+      });
+    }
+    let submitSelector = "";
+    if (/<button[^>]*type\s*=\s*"submit"/i.test(formBody)) {
+      submitSelector = "button[type=submit]";
+    } else if (/<input[^>]*type\s*=\s*"submit"/i.test(formBody)) {
+      submitSelector = 'input[type="submit"]';
+    } else if (/<button/i.test(formBody)) {
+      submitSelector = "button";
+    }
+    const nonHiddenFields = fields.filter((f) => f.type !== "hidden");
+    if (!hasLoginField && nonHiddenFields.length > 0 && submitSelector && (hasSearchLikeField || nonHiddenFields.length >= 1)) {
+      const formSelector = formSelectorFromElement(formElAttrs, formIndex);
+      results.push({
+        form_selector: formSelector,
+        submit_selector: submitSelector,
+        fields
+      });
+    }
+    formIndex++;
+  }
+  return results;
+}
+var SEARCH_FIELD_NAMES, LOGIN_FIELD_NAMES, SUPPORTED_INPUT_TYPES;
+var init_search_forms = __esm(() => {
+  SEARCH_FIELD_NAMES = new Set([
+    "q",
+    "query",
+    "search",
+    "keyword",
+    "keywords",
+    "term",
+    "terms",
+    "find",
+    "lookup",
+    "filter",
+    "s",
+    "text",
+    "input"
+  ]);
+  LOGIN_FIELD_NAMES = new Set([
+    "password",
+    "passwd",
+    "pass",
+    "pwd",
+    "confirm_password",
+    "username",
+    "email",
+    "login",
+    "user"
+  ]);
+  SUPPORTED_INPUT_TYPES = new Set([
+    "text",
+    "search",
+    "hidden",
+    "date",
+    "number",
+    "tel",
+    "email"
+  ]);
+});
+
 // node_modules/.bun/@mixmark-io+domino@2.2.0/node_modules/@mixmark-io/domino/lib/Event.js
 var require_Event = __commonJS((exports, module) => {
   module.exports = Event2;
@@ -71034,20 +71283,24 @@ function buildDirectDocumentResult(url, html, contentType, intent) {
       const hits = intentTokens.filter((tok) => haystack.includes(tok));
       const hitRate = hits.length / intentTokens.length;
       if (hitRate < 0.34) {
-        return {
-          rejected: true,
-          reason: "intent_mismatch",
-          evidence: {
-            intent_tokens: intentTokens,
-            response_token_hits: hits,
-            response_token_hit_rate: hitRate,
-            html_bytes: html.length
-          }
-        };
+        const isCollection = isListLikeIntent(intent) && linksFormEntityCollection(Array.from(html.matchAll(/href\s*=\s*["']([^"']+)["']/gi), (m) => m[1]));
+        if (!isCollection) {
+          return {
+            rejected: true,
+            reason: "intent_mismatch",
+            evidence: {
+              intent_tokens: intentTokens,
+              response_token_hits: hits,
+              response_token_hit_rate: hitRate,
+              html_bytes: html.length
+            }
+          };
+        }
       }
     }
   }
   const { url_template, input_params, path_params, query } = extractHtmlHoles(url);
+  const routing_candidates = buildSearchRouteCandidates(html, url, intent);
   return {
     rejected: false,
     title,
@@ -71061,12 +71314,40 @@ function buildDirectDocumentResult(url, html, contentType, intent) {
     text_excerpt: bodyText.slice(0, MARKDOWN_BUDGET),
     markdown: htmlToMarkdownSafe(html, bodyText),
     tables: extractTables(html),
+    ...routing_candidates.length > 0 ? { routing_candidates } : {},
     extraction: {
       source: "direct-document",
       rejected: false
     }
   };
 }
+function intentQueryTerm(intent, url) {
+  let domTokens = new Set;
+  try {
+    domTokens = new Set(new URL(url).hostname.toLowerCase().split(/[.-]/));
+  } catch {}
+  const toks = (intent.toLowerCase().match(/[a-z][a-z0-9]{2,}/g) ?? []).filter((t) => !QUERY_STOPWORDS.has(t) && !domTokens.has(t));
+  return [...new Set(toks)].join(" ").trim();
+}
+function buildSearchRouteCandidates(html, url, intent) {
+  if (!intent || !isListLikeIntent(intent))
+    return [];
+  const queryTerm = intentQueryTerm(intent, url);
+  if (!queryTerm)
+    return [];
+  let origin = "";
+  try {
+    origin = new URL(url).origin;
+  } catch {
+    return [];
+  }
+  return deriveSearchRouteTemplates(html).slice(0, 3).map((t) => ({
+    url: fillSearchRoute(origin, t.template, queryTerm),
+    template: t.template,
+    query: queryTerm,
+    samples: t.samples
+  }));
+}
 async function fetchDirectDocument(url) {
   if (!isDirectDocumentEligibleUrl(url))
     return null;
@@ -71254,10 +71535,12 @@ function cellText(html) {
 function decodeHtmlEntityText(input) {
   return input.replace(/&nbsp;/g, " ").replace(/&amp;/g, "&").replace(/&lt;/g, "<").replace(/&gt;/g, ">").replace(/&quot;/g, '"').replace(/&#39;/g, "'");
 }
-var HTML_RE, MIN_DIRECT_DOCUMENT_HTML_BYTES = 5000, CHALLENGE_RE, INTERSTITIAL_RE, MIN_DIRECT_DOCUMENT_BODY_TEXT = 500, SPA_HYDRATION_RE, SPA_HYDRATION_BODY_TEXT_FLOOR = 2000, SPA_ROOT_CONTAINER_RE, PARKED_RE, INTENT_STOPWORDS, MARKDOWN_BUDGET, MAX_TABLES = 10, MAX_TABLE_ROWS = 50, buildBloombergDirectDocumentResult;
+var HTML_RE, MIN_DIRECT_DOCUMENT_HTML_BYTES = 5000, CHALLENGE_RE, INTERSTITIAL_RE, MIN_DIRECT_DOCUMENT_BODY_TEXT = 500, SPA_HYDRATION_RE, SPA_HYDRATION_BODY_TEXT_FLOOR = 2000, SPA_ROOT_CONTAINER_RE, PARKED_RE, INTENT_STOPWORDS, MARKDOWN_BUDGET, MAX_TABLES = 10, MAX_TABLE_ROWS = 50, QUERY_STOPWORDS, buildBloombergDirectDocumentResult;
 var init_direct_document = __esm(() => {
   init_curl_impersonate_fallback();
   init_proxy_fetch();
+  init_cardinality();
+  init_search_forms();
   HTML_RE = /text\/html|application\/xhtml\+xml/i;
   CHALLENGE_RE = /\b(access denied|are you a robot|captcha|just a moment|pardon our interruption|robot check|unusual traffic|verify you are human)\b/i;
   INTERSTITIAL_RE = /\b(please wait for verification|just a moment|cf-mitigated|datadome|akamai bot|perimeterx|sign in to continue|log in to (?:continue|access)|javascript is not available)\b/i;
@@ -71340,6 +71623,7 @@ var init_direct_document = __esm(() => {
     "look"
   ]);
   MARKDOWN_BUDGET = Math.max(1000, Number(process.env.UNBROWSE_MARKDOWN_BUDGET ?? "12000") || 12000);
+  QUERY_STOPWORDS = new Set(("resolve unbrowse execute run walk go fetch open view want need please " + "find search browse list lookup discover show get me a an the on of for in to " + "with and or all my your this that some good best top new latest cheap near").split(" "));
   buildBloombergDirectDocumentResult = buildDirectDocumentResult;
 });
 
@@ -120052,181 +120336,6 @@ function clampToFloor(score, demotion, floor) {
 }
 var HARD_NEGATIVE_FLOOR = -2000, WEAK_NEGATIVE_FLOOR = -400, PAGE_ARTIFACT_DEMOTION = 800, EMPTY_ENTITY_BAG_DEMOTION = 650, EMPTY_ENTITY_BAG_FLOOR = -700;
 
-// .tmp-runtime-src/execution/search-forms.ts
-var exports_search_forms = {};
-__export(exports_search_forms, {
-  isStructuredSearchForm: () => isStructuredSearchForm,
-  detectSearchForms: () => detectSearchForms
-});
-function isStructuredSearchForm(spec) {
-  return spec.fields.length > 0 && !!spec.submit_selector;
-}
-function formSelectorFromElement(attribs, index2) {
-  const id = attribs.id;
-  if (id)
-    return `form#${id}`;
-  const name = attribs.name;
-  if (name)
-    return `form[name="${name}"]`;
-  const action2 = attribs.action;
-  if (action2)
-    return `form[action="${action2}"]`;
-  return `form:nth-of-type(${index2 + 1})`;
-}
-function inputSelectorFromElement(attribs, tagName) {
-  const id = attribs.id;
-  if (id)
-    return `#${id}`;
-  const name = attribs.name;
-  if (name)
-    return `${tagName}[name="${name}"]`;
-  return tagName;
-}
-function mapInputType(typeAttr, tagName) {
-  if (tagName === "select")
-    return "select";
-  if (tagName === "textarea")
-    return "text";
-  const t = (typeAttr ?? "text").toLowerCase();
-  if (t === "radio")
-    return "radio";
-  if (t === "checkbox")
-    return "checkbox";
-  if (t === "date")
-    return "date";
-  if (t === "hidden")
-    return "hidden";
-  if (t === "submit" || t === "button" || t === "image" || t === "reset")
-    return null;
-  if (t === "password" || t === "file")
-    return null;
-  if (SUPPORTED_INPUT_TYPES.has(t))
-    return "text";
-  return "text";
-}
-function parseAttrs(attrStr) {
-  const attrs = {};
-  const attrRegex = /(\w[\w-]*)(?:\s*=\s*(?:"([^"]*)"|'([^']*)'|(\S+)))?/g;
-  let m;
-  while ((m = attrRegex.exec(attrStr)) !== null) {
-    attrs[m[1]] = m[2] ?? m[3] ?? m[4] ?? "";
-  }
-  return attrs;
-}
-function detectSearchForms(html3) {
-  const results = [];
-  const formRegex = /<form([^>]*)>([\s\S]*?)<\/form>/gi;
-  let formMatch;
-  let formIndex = 0;
-  while ((formMatch = formRegex.exec(html3)) !== null) {
-    const formAttrs = formMatch[1];
-    const formBody = formMatch[2];
-    const formElAttrs = parseAttrs(formAttrs);
-    const fieldRegex = /<(input|select|textarea)([^>]*)\/?>/gi;
-    let fieldMatch;
-    const fields = [];
-    const seenNames = new Set;
-    let hasLoginField = false;
-    let hasSearchLikeField = false;
-    while ((fieldMatch = fieldRegex.exec(formBody)) !== null) {
-      const tagName = fieldMatch[1].toLowerCase();
-      const fieldAttrs = parseAttrs(fieldMatch[2]);
-      const name = fieldAttrs.name ?? "";
-      const typeAttr = fieldAttrs.type;
-      if (LOGIN_FIELD_NAMES.has(name.toLowerCase()) || typeAttr === "password") {
-        hasLoginField = true;
-      }
-      if (SEARCH_FIELD_NAMES.has(name.toLowerCase())) {
-        hasSearchLikeField = true;
-      }
-      const mappedType = mapInputType(typeAttr, tagName);
-      if (!mappedType)
-        continue;
-      if (!name && mappedType !== "text")
-        continue;
-      if (seenNames.has(name) && mappedType !== "radio")
-        continue;
-      if (name)
-        seenNames.add(name);
-      let options;
-      if (tagName === "select") {
-        const optRegex = /<option[^>]*value="([^"]*)"[^>]*>/gi;
-        let optMatch;
-        options = [];
-        while ((optMatch = optRegex.exec(formBody)) !== null) {
-          options.push(optMatch[1]);
-        }
-        if (options.length === 0)
-          options = undefined;
-      }
-      fields.push({
-        name: name || `unnamed_${fields.length}`,
-        type: mappedType,
-        selector: inputSelectorFromElement(fieldAttrs, tagName),
-        ...options ? { options } : {},
-        required: fieldAttrs.required !== undefined
-      });
-    }
-    let submitSelector = "";
-    if (/<button[^>]*type\s*=\s*"submit"/i.test(formBody)) {
-      submitSelector = "button[type=submit]";
-    } else if (/<input[^>]*type\s*=\s*"submit"/i.test(formBody)) {
-      submitSelector = 'input[type="submit"]';
-    } else if (/<button/i.test(formBody)) {
-      submitSelector = "button";
-    }
-    const nonHiddenFields = fields.filter((f) => f.type !== "hidden");
-    if (!hasLoginField && nonHiddenFields.length > 0 && submitSelector && (hasSearchLikeField || nonHiddenFields.length >= 1)) {
-      const formSelector = formSelectorFromElement(formElAttrs, formIndex);
-      results.push({
-        form_selector: formSelector,
-        submit_selector: submitSelector,
-        fields
-      });
-    }
-    formIndex++;
-  }
-  return results;
-}
-var SEARCH_FIELD_NAMES, LOGIN_FIELD_NAMES, SUPPORTED_INPUT_TYPES;
-var init_search_forms = __esm(() => {
-  SEARCH_FIELD_NAMES = new Set([
-    "q",
-    "query",
-    "search",
-    "keyword",
-    "keywords",
-    "term",
-    "terms",
-    "find",
-    "lookup",
-    "filter",
-    "s",
-    "text",
-    "input"
-  ]);
-  LOGIN_FIELD_NAMES = new Set([
-    "password",
-    "passwd",
-    "pass",
-    "pwd",
-    "confirm_password",
-    "username",
-    "email",
-    "login",
-    "user"
-  ]);
-  SUPPORTED_INPUT_TYPES = new Set([
-    "text",
-    "search",
-    "hidden",
-    "date",
-    "number",
-    "tel",
-    "email"
-  ]);
-});
-
 // .tmp-runtime-src/state/stateless.ts
 function isStateless() {
   const v = process.env.UNBROWSE_STATELESS;
@@ -121655,7 +121764,7 @@ function isProtobufContentType(contentType) {
 function isProtobufLikeEndpoint(url, contentType) {
   if (isProtobufContentType(contentType))
     return true;
-  return /\/(field-data-proto|proto|protobuf)(\/|$|-)/i.test(url);
+  return /[-/](proto|protobuf)(\/|$|-)/i.test(url);
 }
 function decodeProtobufBytes(bytes) {
   return decodeBytes(bytes, "bytes");

package/vendor/kuri/manifest.json

@@ -2,7 +2,7 @@
   "repo_url": "https://github.com/justrach/kuri.git",
   "branch": "adding-extensions",
   "source_sha": "149881254046a20778f642b69f20f0c6468f6fb4",
-  "built_at": "2026-06-18T03:58:10.362Z",
+  "built_at": "2026-06-18T05:53:02.585Z",
   "binaries": {
     "darwin-arm64": {
       "zig_target": "aarch64-macos",
@@ -21,11 +21,11 @@
     },
     "linux-x64": {
       "zig_target": "x86_64-linux",
-      "sha256": "e73aecfbf07001ba0be5032118790eb253ad5d8d12caca6a1dd5ad3ccab44b9e"
+      "sha256": "f39955d73d86150fba2a4bec6393e7745feb42f5152870b6c27fd68a5cff3a6e"
     },
     "win-x64": {
       "zig_target": "x86_64-windows-gnu",
-      "sha256": "9ecbc82be646e755e4664051cf345d54dde3c6610e457d763deff67895047963",
+      "sha256": "376a34f508ea6a4e140150f9f6ddc00519f8bb0894ee2dd7a60bc0e7613d89b0",
       "source": "pre-staged"
     }
   },
@@ -33,22 +33,22 @@
     "darwin-arm64": {
       "zig_target": "aarch64-macos",
       "lib": "libkuri_ffi.dylib",
-      "sha256": "6c72cf383df4fa3f870b745da43d64eae8f67e58c6f971214ac29602fb649939"
+      "sha256": "2ca1be4d477f28c4a4ab1dd993cfe6766b9f6858c42befda3f503a3e2940bf6f"
     },
     "darwin-x64": {
       "zig_target": "x86_64-macos",
       "lib": "libkuri_ffi.dylib",
-      "sha256": "82480772ddc8e44c8e34e70b80d7dc0969004942f77276587af450b62d3d2750"
+      "sha256": "80e27865e521b4bc6a79dfcb4fd481535d70e03226125ce10d9625ba02320f47"
     },
     "linux-arm64": {
       "zig_target": "aarch64-linux",
       "lib": "libkuri_ffi.so",
-      "sha256": "ef8dfa2b634f04294f93a94472d9856ba777681afaab2d4213f0e29821882e07"
+      "sha256": "6fa04fc6b505212e5ae9cfdc0d21eb2f9b2eb97dc41850e02ba3f4112e252e9d"
     },
     "linux-x64": {
       "zig_target": "x86_64-linux",
       "lib": "libkuri_ffi.so",
-      "sha256": "fb29ad2b71186d176306321d17e88074a67fea139991faef9aa4862333942c9e"
+      "sha256": "fa520df6cca6eab9260bd43ed7a60d665c746645f12f6fb4e3ecddbb026bdb05"
     }
   }
 }