mcp-scraper 0.32.1 → 0.32.2

package/dist/index.js

@@ -1,11 +1,11 @@
 import {
   harvest
-} from "./chunk-RDCCU7CQ.js";
+} from "./chunk-AEMABLEL.js";
 import {
   recordVendorUsage
 } from "./chunk-MTSBI7ZH.js";
 import "./chunk-K443GQY5.js";
-import "./chunk-PTHY7NYD.js";
+import "./chunk-V73MPRU6.js";
 import "./chunk-M2S27J6Z.js";
 import "./chunk-ICT7DDHL.js";
 

package/dist/{server-GXTYKIAX.js → server-LTZDJHFG.js}

@@ -15,8 +15,9 @@ import {
   encodeUule,
   harvest,
   normalizeLocation,
-  resolveKernelProxyId
-} from "./chunk-RDCCU7CQ.js";
+  resolveKernelProxyId,
+  serpArrivalUrl
+} from "./chunk-AEMABLEL.js";
 import {
   HttpMcpToolExecutor,
   assessTranscriptSignal,
@@ -36,7 +37,7 @@ import {
   sanitizeAttempts,
   sanitizeHarvestResult,
   transcribeMediaUrl
-} from "./chunk-2FY4VBAY.js";
+} from "./chunk-XEED325T.js";
 import {
   auditImages,
   buildLinkReport,
@@ -65,7 +66,7 @@ import {
   workflowOutputBaseDir,
   workflowStepCount,
   workflowSupportsSteps
-} from "./chunk-55M5CE5L.js";
+} from "./chunk-MKZIL4EA.js";
 import "./chunk-K443GQY5.js";
 import {
   DEFAULT_MAPS_PROXY_MODE,
@@ -76,8 +77,8 @@ import {
   RawMapsHoursRowSchema,
   RawMapsOverviewSchema,
   RawMapsReviewStatsSchema
-} from "./chunk-PTHY7NYD.js";
-import "./chunk-VJBS6RPB.js";
+} from "./chunk-V73MPRU6.js";
+import "./chunk-XLO5CZ6T.js";
 import {
   completeExtractJob,
   countSuccessfulPages,
@@ -13083,6 +13084,7 @@ import { Hono as Hono11 } from "hono";
 // src/extractor/MapsSearchExtractor.ts
 var LOCAL_RESULTS_PAGE_LIMIT = 8;
 var LOCAL_RESULTS_WAIT_MS = 1e3;
+var MAPS_REDIRECT_BASE = "https://serp-redirector.vercel.app";
 function buildGoogleLocalResultsUrl(input) {
   const query = [input.query, input.location].filter(Boolean).join(" ");
   const params = new URLSearchParams({ q: query, udm: "1", gl: input.gl, hl: input.hl, pws: "0" });
@@ -13104,21 +13106,23 @@ var MapsSearchExtractor = class {
     const startMs = Date.now();
     const searchQuery = [options.query, options.location].filter(Boolean).join(" ");
     const searchUrl = buildGoogleOrganicSearchUrl(options);
+    const keepDefaultProxy = !options.forceDirectEgress && options.proxyMode !== "location" && options.proxyMode !== "configured" && !options.kernelProxyId;
     const config = {
       headless: options.headless,
       kernelApiKey: options.kernelApiKey,
       kernelProxyId: options.kernelProxyId,
       kernelProxyResolution: options.kernelProxyResolution,
       proxyMode: options.proxyMode,
-      keepDefaultProxy: options.proxyMode !== "location" && options.proxyMode !== "configured" && !options.kernelProxyId,
+      keepDefaultProxy,
       viewport: { width: 1280, height: 900 },
       locale: `${options.hl}-${options.gl.toUpperCase()}`,
       debug: options.debug
     };
+    const navUrl = options.serpRedirect ? serpArrivalUrl(searchUrl, process.env.SERP_REDIRECT_BASE?.trim() || MAPS_REDIRECT_BASE) : searchUrl;
     try {
       await this.driver.launch(config);
       const page = this.driver.getPage();
-      await page.goto(searchUrl, { waitUntil: "domcontentloaded", timeout: 6e4 });
+      await page.goto(navUrl, { waitUntil: "domcontentloaded", timeout: 6e4 });
       await page.waitForTimeout(LOCAL_RESULTS_WAIT_MS);
       if (await this.detectBlock(page)) throw new CaptchaError(RECAPTCHA_INSTRUCTIONS);
       const reachedFinder = await this.clickIntoLocalFinder(page);
@@ -13978,7 +13982,9 @@ async function runMapsSearchWithRotation(options) {
         ...options,
         proxyZip,
         kernelProxyId: options.proxyMode === "none" ? void 0 : resolution.kernelProxyId,
-        kernelProxyResolution: resolution.resolution
+        kernelProxyResolution: resolution.resolution,
+        serpRedirect: options.proxyMode === "none" && attemptIndex >= 1 && attemptIndex < maxAttempts - 1,
+        forceDirectEgress: options.proxyMode === "none" && attemptIndex >= 1
       };
       const isHeadfulEscalation = attemptIndex === maxAttempts - 1;
       const baseCtx = currentCostContext();
@@ -27424,7 +27430,7 @@ app.get("/cron/tick", async (c) => {
   if (!process.env.CRON_SECRET || secret2 !== `Bearer ${process.env.CRON_SECRET}`) {
     return c.json({ error: "Unauthorized" }, 401);
   }
-  const { drainQueue } = await import("./worker-7DJYUACV.js");
+  const { drainQueue } = await import("./worker-ZZCZNY3R.js");
   const budget = { maxJobs: 10, deadlineMs: Date.now() + 28e4 };
   const workflowDispatchResult = await dispatchDueWorkflowSchedules(`${new URL(c.req.url).protocol}//${new URL(c.req.url).host}`);
   const [results, sweepResult, reapResult, expiredResult, blobCleanup, connectedDataArtifactCleanup] = await Promise.all([
@@ -27627,4 +27633,4 @@ app.get("/blog/:slug/", (c) => {
 export {
   app
 };
-//# sourceMappingURL=server-GXTYKIAX.js.map
+//# sourceMappingURL=server-LTZDJHFG.js.map