unbrowse 6.1.3 → 6.2.0

package/dist/cli.js

@@ -31,7 +31,7 @@ var __promiseAll = (args) => Promise.all(args);
 var __require = /* @__PURE__ */ createRequire(import.meta.url);
 
 // ../../src/build-info.generated.ts
-var BUILD_RELEASE_VERSION = "6.1.3", BUILD_GIT_SHA = "2efa82e8f6a7", BUILD_CODE_HASH = "5d9ebf619c61", BUILD_RELEASE_MANIFEST_BASE64 = "eyJzY2hlbWFfdmVyc2lvbiI6MSwicmVsZWFzZV92ZXJzaW9uIjoiNi4xLjMiLCJnaXRfc2hhIjoiMmVmYTgyZThmNmE3IiwiY29kZV9oYXNoIjoiNWQ5ZWJmNjE5YzYxIiwidHJhY2VfdmVyc2lvbiI6IjVkOWViZjYxOWM2MUAyZWZhODJlOGY2YTciLCJpc3N1ZWRfYXQiOiIyMDI2LTA1LTAxVDAxOjA3OjIxLjM5NFoifQ", BUILD_RELEASE_MANIFEST_SIGNATURE = "bzwIl25aiAhrxXJfAHiTDAFUYKkOUntait0ZAKFE2Ys", BUILD_DEFAULT_BACKEND_URL = "https://beta-api.unbrowse.ai";
+var BUILD_RELEASE_VERSION = "6.2.0", BUILD_GIT_SHA = "d39e0773367a", BUILD_CODE_HASH = "5d9ebf619c61", BUILD_RELEASE_MANIFEST_BASE64 = "eyJzY2hlbWFfdmVyc2lvbiI6MSwicmVsZWFzZV92ZXJzaW9uIjoiNi4yLjAiLCJnaXRfc2hhIjoiZDM5ZTA3NzMzNjdhIiwiY29kZV9oYXNoIjoiNWQ5ZWJmNjE5YzYxIiwidHJhY2VfdmVyc2lvbiI6IjVkOWViZjYxOWM2MUBkMzllMDc3MzM2N2EiLCJpc3N1ZWRfYXQiOiIyMDI2LTA1LTAxVDAyOjA5OjIwLjQ1N1oifQ", BUILD_RELEASE_MANIFEST_SIGNATURE = "R7RjaEXQD0TV-UY0ZOZwLAJ7iFqZc3n-ZIfmrZHKtec", BUILD_DEFAULT_BACKEND_URL = "https://beta-api.unbrowse.ai";
 
 // ../../src/version.ts
 import { createHash } from "crypto";

package/dist/mcp.js

@@ -226,11 +226,11 @@ import { dirname, join, parse } from "path";
 import { fileURLToPath as fileURLToPath2 } from "url";
 
 // ../../src/build-info.generated.ts
-var BUILD_RELEASE_VERSION = "6.1.3";
-var BUILD_GIT_SHA = "2efa82e8f6a7";
+var BUILD_RELEASE_VERSION = "6.2.0";
+var BUILD_GIT_SHA = "d39e0773367a";
 var BUILD_CODE_HASH = "5d9ebf619c61";
-var BUILD_RELEASE_MANIFEST_BASE64 = "eyJzY2hlbWFfdmVyc2lvbiI6MSwicmVsZWFzZV92ZXJzaW9uIjoiNi4xLjMiLCJnaXRfc2hhIjoiMmVmYTgyZThmNmE3IiwiY29kZV9oYXNoIjoiNWQ5ZWJmNjE5YzYxIiwidHJhY2VfdmVyc2lvbiI6IjVkOWViZjYxOWM2MUAyZWZhODJlOGY2YTciLCJpc3N1ZWRfYXQiOiIyMDI2LTA1LTAxVDAxOjA3OjIxLjM5NFoifQ";
-var BUILD_RELEASE_MANIFEST_SIGNATURE = "bzwIl25aiAhrxXJfAHiTDAFUYKkOUntait0ZAKFE2Ys";
+var BUILD_RELEASE_MANIFEST_BASE64 = "eyJzY2hlbWFfdmVyc2lvbiI6MSwicmVsZWFzZV92ZXJzaW9uIjoiNi4yLjAiLCJnaXRfc2hhIjoiZDM5ZTA3NzMzNjdhIiwiY29kZV9oYXNoIjoiNWQ5ZWJmNjE5YzYxIiwidHJhY2VfdmVyc2lvbiI6IjVkOWViZjYxOWM2MUBkMzllMDc3MzM2N2EiLCJpc3N1ZWRfYXQiOiIyMDI2LTA1LTAxVDAyOjA5OjIwLjQ1N1oifQ";
+var BUILD_RELEASE_MANIFEST_SIGNATURE = "R7RjaEXQD0TV-UY0ZOZwLAJ7iFqZc3n-ZIfmrZHKtec";
 var BUILD_DEFAULT_BACKEND_URL = "https://beta-api.unbrowse.ai";
 
 // ../../src/version.ts

package/dist/server.js

@@ -7285,7 +7285,7 @@ var init_capture = __esm(async () => {
 });
 
 // ../../src/build-info.generated.ts
-var BUILD_RELEASE_VERSION = "6.1.3", BUILD_GIT_SHA = "2efa82e8f6a7", BUILD_CODE_HASH = "5d9ebf619c61", BUILD_RELEASE_MANIFEST_BASE64 = "eyJzY2hlbWFfdmVyc2lvbiI6MSwicmVsZWFzZV92ZXJzaW9uIjoiNi4xLjMiLCJnaXRfc2hhIjoiMmVmYTgyZThmNmE3IiwiY29kZV9oYXNoIjoiNWQ5ZWJmNjE5YzYxIiwidHJhY2VfdmVyc2lvbiI6IjVkOWViZjYxOWM2MUAyZWZhODJlOGY2YTciLCJpc3N1ZWRfYXQiOiIyMDI2LTA1LTAxVDAxOjA3OjIxLjM5NFoifQ", BUILD_RELEASE_MANIFEST_SIGNATURE = "bzwIl25aiAhrxXJfAHiTDAFUYKkOUntait0ZAKFE2Ys", BUILD_DEFAULT_BACKEND_URL = "https://beta-api.unbrowse.ai";
+var BUILD_RELEASE_VERSION = "6.2.0", BUILD_GIT_SHA = "d39e0773367a", BUILD_CODE_HASH = "5d9ebf619c61", BUILD_RELEASE_MANIFEST_BASE64 = "eyJzY2hlbWFfdmVyc2lvbiI6MSwicmVsZWFzZV92ZXJzaW9uIjoiNi4yLjAiLCJnaXRfc2hhIjoiZDM5ZTA3NzMzNjdhIiwiY29kZV9oYXNoIjoiNWQ5ZWJmNjE5YzYxIiwidHJhY2VfdmVyc2lvbiI6IjVkOWViZjYxOWM2MUBkMzllMDc3MzM2N2EiLCJpc3N1ZWRfYXQiOiIyMDI2LTA1LTAxVDAyOjA5OjIwLjQ1N1oifQ", BUILD_RELEASE_MANIFEST_SIGNATURE = "R7RjaEXQD0TV-UY0ZOZwLAJ7iFqZc3n-ZIfmrZHKtec", BUILD_DEFAULT_BACKEND_URL = "https://beta-api.unbrowse.ai";
 
 // ../../src/version.ts
 import { createHash as createHash2 } from "crypto";
@@ -12880,6 +12880,67 @@ function extractDefinitionSpecial(html, intent) {
     element_count: 1
   }];
 }
+function extractArticleBodySpecial(html, intent) {
+  const intentLower = intent.toLowerCase();
+  const articleIntent = /(wikipedia|article|wiki page|page on|read|content of|body of|summary of|about )/i.test(intentLower);
+  const looksWikipedia = /id="mw-content-text"|class="mw-parser-output"/i.test(html);
+  if (!articleIntent && !looksWikipedia)
+    return [];
+  const $ = cheerio.load(html);
+  const root = looksWikipedia ? $(".mw-parser-output").first() : $("article, [role='article']").first().length > 0 ? $("article, [role='article']").first() : $("main").first();
+  if (!root.length)
+    return [];
+  const title = cleanText($("h1").first().text());
+  if (!title)
+    return [];
+  root.find(".mw-editsection, .reference, .references, .navbox, .infobox, .reflist, .hatnote, .ambox, .toc, sup.reference, style, script, .sidebar").remove();
+  let summary = "";
+  root.find("> p, .mw-parser-output > p, p").each((_, el) => {
+    if (summary.length >= 80)
+      return false;
+    const t = cleanText($(el).text());
+    if (t.length >= 80)
+      summary = t;
+    return;
+  });
+  const sections = [];
+  let current = null;
+  root.find("h2, h3, p, li").each((_, el) => {
+    const tag = el.tagName?.toLowerCase?.() ?? el.name ?? "";
+    const txt = cleanText($(el).text());
+    if (!txt)
+      return;
+    if (tag === "h2" || tag === "h3") {
+      if (current && current.parts.length)
+        sections.push({ heading: current.heading, text: current.parts.join(`
+`).slice(0, 1500) });
+      if (/^(references?|external links?|see also|notes?|further reading|bibliography|sources?)$/i.test(txt)) {
+        current = null;
+      } else {
+        current = { heading: txt, parts: [] };
+      }
+    } else if (current && (tag === "p" || tag === "li") && txt.length > 20) {
+      if (current.parts.join(`
+`).length < 1500)
+        current.parts.push(txt);
+    }
+  });
+  if (current && current.parts.length)
+    sections.push({ heading: current.heading, text: current.parts.join(`
+`).slice(0, 1500) });
+  if (!summary && sections.length === 0)
+    return [];
+  return [{
+    type: "article",
+    data: {
+      title,
+      summary: summary || undefined,
+      sections: sections.slice(0, 12),
+      section_count: sections.length
+    },
+    element_count: 1 + sections.length
+  }];
+}
 function extractCourseSearchSpecial(html, intent) {
   if (!/\b(course|courses)\b/i.test(intent))
     return [];
@@ -13440,7 +13501,8 @@ function extractFromDOM(html, intent) {
   const trendStructures = extractTrendSpecial(workingHtml, intent);
   const definitionStructures = extractDefinitionSpecial(workingHtml, intent);
   const courseStructures = extractCourseSearchSpecial(workingHtml, intent);
-  const structures = [...flashStructures, ...githubStructures, ...linkedInStructures, ...packageSearchStructures, ...xProfileStructures, ...postStructures, ...trendStructures, ...definitionStructures, ...courseStructures, ...spaStructures, ...parseStructured(cleaned)].map((structure) => normalizeStructureForIntent(structure, intent));
+  const articleStructures = extractArticleBodySpecial(workingHtml, intent);
+  const structures = [...flashStructures, ...githubStructures, ...linkedInStructures, ...packageSearchStructures, ...xProfileStructures, ...postStructures, ...trendStructures, ...definitionStructures, ...courseStructures, ...articleStructures, ...spaStructures, ...parseStructured(cleaned)].map((structure) => normalizeStructureForIntent(structure, intent));
   if (structures.length === 0) {
     return { data: null, extraction_method: "none", confidence: 0 };
   }
@@ -17425,6 +17487,13 @@ async function executeEndpoint(skill, endpoint, params = {}, projection, options
     if (mergedParams.features == null || mergedParams.features === "{features}") {
       mergedParams.features = encodeURIComponent(__gqlEnc.features);
     }
+    if (endpoint.body && typeof endpoint.body === "object" && !Array.isArray(endpoint.body)) {
+      const b = endpoint.body;
+      if ("variables" in b)
+        b.variables = __gqlEnc.variables;
+      if ("features" in b)
+        b.features = __gqlEnc.features;
+    }
   }
   let url = interpolate(urlTemplate, mergedParams);
   const __callerUrl = typeof mergedParams.url === "string" && mergedParams.url ? mergedParams.url : endpoint.trigger_url ?? "";
@@ -17871,7 +17940,8 @@ async function executeEndpoint(skill, endpoint, params = {}, projection, options
   });
   if (!trace.success) {
     trace.error = status === 0 ? `HTTP 0 — network failure or browser fetch was blocked (DNS, TLS, CORS, anti-bot, or kuri tab error). Try \`unbrowse go\` to open a live session, then re-run.` : status === 404 ? `HTTP 404 — endpoint may be stale. Re-run via POST /v1/intent/resolve to get fresh endpoints.` : `HTTP ${status}`;
-    if (data == null) {
+    const isEmptyData = data == null || typeof data === "object" && !Array.isArray(data) && Object.keys(data).length === 0;
+    if (isEmptyData) {
       data = {
         error: status === 0 ? "network_failure" : `http_${status}`,
         message: trace.error,
@@ -21904,7 +21974,7 @@ async function resolveAndExecute(intent, params = {}, context, projection, optio
       requiredRecovery: false
     });
     const isSameHostResolve = !!context?.url && !!endpointScopedSkill.domain;
-    const hostMatches = isSameHostResolve && new URL(context.url).hostname === endpointScopedSkill.domain;
+    const hostMatches = isSameHostResolve && (new URL(context.url).hostname === endpointScopedSkill.domain || getRegistrableDomain(new URL(context.url).hostname) === getRegistrableDomain(endpointScopedSkill.domain));
     const allNegative = epRanked.length > 0 && epRanked.every((r) => r.score < 0);
     if ((epRanked.length === 0 || allNegative) && hostMatches) {
       return {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "unbrowse",
-  "version": "6.1.3",
+  "version": "6.2.0",
   "description": "Reverse-engineer any website into reusable API skills. Zero-dep single binary with embedded browser engine.",
   "type": "module",
   "bin": {