mcp-scraper 0.65.1 → 0.66.1

package/dist/bin/mcp-stdio-server.js

@@ -13,20 +13,21 @@ import {
   registerScheduledResultsMcpTools,
   registerSerpIntelligenceCaptureTools,
   resolveDeploymentProfile
-} from "../chunk-PDAKCYHZ.js";
-import "../chunk-342OUCSG.js";
-import "../chunk-VIWT35S6.js";
-import "../chunk-4LINSJNI.js";
+} from "../chunk-E4LMY5OJ.js";
+import "../chunk-PGJQDMC2.js";
+import "../chunk-4OT7BAIE.js";
+import "../chunk-TNLGHNLX.js";
 import {
   renderInstallTerminal
-} from "../chunk-DCWXVAQT.js";
-import "../chunk-UVFB24RZ.js";
+} from "../chunk-WID3MFRS.js";
+import "../chunk-VXLU74YZ.js";
+import "../chunk-GGZEC22A.js";
 import {
   PACKAGE_VERSION
-} from "../chunk-CGKWNXJR.js";
-import "../chunk-EKJ65E3Z.js";
-import "../chunk-2XUXVKT4.js";
-import "../chunk-4VG3NUUY.js";
+} from "../chunk-VRJ7CO6L.js";
+import "../chunk-DNM65UCK.js";
+import "../chunk-AYJHI2FK.js";
+import "../chunk-GTGSD4YS.js";
 
 // bin/mcp-stdio-server.ts
 import { readFileSync } from "fs";
@@ -168,4 +169,3 @@ serveStdio(() => buildStdioServer(), {
 `);
   }
 });
-//# sourceMappingURL=mcp-stdio-server.js.map

package/dist/bin/paa-harvest.cjs

@@ -4018,6 +4018,9 @@ function classifyAttemptError(err) {
   if (err instanceof RequestAbortedError) return "request_aborted";
   if (err instanceof DOMException && (err.name === "TimeoutError" || err.name === "AbortError")) return "timeout";
   const message = err instanceof Error ? err.message : String(err);
+  if (/browser (?:has been )?closed|context (?:has been )?closed|target page, context or browser has been closed|session closed|page has been closed/i.test(message)) {
+    return "browser_session_interrupted";
+  }
   if (looksLikeProxyTunnelFailure(message)) return "proxy_tunnel_failed";
   if (looksLikeProxyUnavailable(message)) return "proxy_unavailable";
   return /timeout|timed out|Timeout \d+ms exceeded|deadline/i.test(message) ? "timeout" : "error";
@@ -4065,6 +4068,9 @@ function looksLikeProxyUnavailable(message) {
 function retryableLocationProxyError(outcome) {
   return outcome === "captcha" || outcome === "proxy_tunnel_failed" || outcome === "proxy_unavailable";
 }
+function retryableFreshBrowserError(outcome) {
+  return outcome === "browser_session_interrupted";
+}
 function locationMismatchMessage(result) {
   const evidence = result.diagnostics.debug?.locationEvidence;
   const expected = evidence?.expected?.canonicalLocation ?? result.location ?? "requested location";
@@ -4248,7 +4254,7 @@ async function harvest(rawOptions) {
           proxyFailures++;
           if (proxyFailures >= 2) effectiveProxyMode = "none";
         }
-        const willRetry2 = i < maxAttempts - 1 && (outcome2 === "captcha" || proxyMode === "location" && retryableLocationProxyError(outcome2));
+        const willRetry2 = i < maxAttempts - 1 && (outcome2 === "captcha" || retryableFreshBrowserError(outcome2) || proxyMode === "location" && retryableLocationProxyError(outcome2));
         if (outcome2 === "captcha") {
           console.warn(JSON.stringify({
             event: "harvest_attempt_captcha",
@@ -4366,7 +4372,7 @@ async function harvest(rawOptions) {
         proxyFailures++;
         if (proxyFailures >= 2) effectiveProxyMode = "none";
       }
-      const willRetry = i < maxAttempts - 1 && (outcome === "captcha" || proxyMode === "location" && retryableLocationProxyError(outcome));
+      const willRetry = i < maxAttempts - 1 && (outcome === "captcha" || retryableFreshBrowserError(outcome) || proxyMode === "location" && retryableLocationProxyError(outcome));
       if (outcome === "captcha") {
         console.warn(JSON.stringify({
           event: "harvest_attempt_captcha",
@@ -4451,4 +4457,3 @@ async function runCli() {
 
 // bin/paa-harvest.ts
 runCli();
-//# sourceMappingURL=paa-harvest.cjs.map

package/dist/bin/paa-harvest.js

@@ -1,12 +1,12 @@
 #!/usr/bin/env node
 import {
   harvest
-} from "../chunk-ORKXE7VX.js";
+} from "../chunk-4P3PHFPQ.js";
 import {
   browserServiceApiKey
-} from "../chunk-4LINSJNI.js";
-import "../chunk-FG6LJGJ7.js";
-import "../chunk-4VG3NUUY.js";
+} from "../chunk-TNLGHNLX.js";
+import "../chunk-77LQV73B.js";
+import "../chunk-GTGSD4YS.js";
 
 // src/cli.ts
 import { Command } from "commander";
@@ -39,4 +39,3 @@ async function runCli() {
 
 // bin/paa-harvest.ts
 runCli();
-//# sourceMappingURL=paa-harvest.js.map

package/dist/{chunk-5PIO7QBG.js → chunk-3PIXKLGZ.js}

@@ -4,10 +4,10 @@ import {
   privateArtifactOwnerId,
   readPrivateArtifactBuffer,
   renewPrivateArtifactDownload
-} from "./chunk-EKJ65E3Z.js";
+} from "./chunk-DNM65UCK.js";
 import {
   LedgerOperation
-} from "./chunk-2XUXVKT4.js";
+} from "./chunk-AYJHI2FK.js";
 import {
   CREDIT_LOT_TTL,
   getDb,
@@ -15,7 +15,7 @@ import {
   sanitizeVendorName,
   serializePublicErrorEnvelope,
   settleDebitMcIdempotent
-} from "./chunk-4VG3NUUY.js";
+} from "./chunk-GTGSD4YS.js";
 
 // src/api/site-extract-content-store.ts
 import { createHash } from "crypto";
@@ -797,4 +797,3 @@ export {
   setExtractJobPublicError,
   settleExtractJob
 };
-//# sourceMappingURL=chunk-5PIO7QBG.js.map

package/dist/{chunk-VIWT35S6.js → chunk-4OT7BAIE.js}

@@ -1,7 +1,7 @@
 import {
   publicErrorMessage,
   sanitizeVendorName
-} from "./chunk-4VG3NUUY.js";
+} from "./chunk-GTGSD4YS.js";
 
 // src/api/outbound-sanitize.ts
 var KEY_RENAMES = {
@@ -101,4 +101,3 @@ export {
   sanitizeAttempts,
   sanitizeHarvestResult
 };
-//# sourceMappingURL=chunk-VIWT35S6.js.map

package/dist/{chunk-ORKXE7VX.js → chunk-4P3PHFPQ.js}

@@ -6,14 +6,14 @@ import {
   currentCostContext,
   recordKernelSession,
   runWithCostContext
-} from "./chunk-4LINSJNI.js";
+} from "./chunk-TNLGHNLX.js";
 import {
   DEFAULT_PROXY_MODE,
   HarvestOptionsSchema,
   RawPAAItemSchema,
   postToMemoryLibrary,
   recencyToTbs
-} from "./chunk-FG6LJGJ7.js";
+} from "./chunk-77LQV73B.js";
 import {
   CaptchaError,
   ExtractionError,
@@ -21,7 +21,7 @@ import {
   RECAPTCHA_INSTRUCTIONS,
   RequestAbortedError,
   sanitizeVendorName
-} from "./chunk-4VG3NUUY.js";
+} from "./chunk-GTGSD4YS.js";
 
 // src/driver/BrowserDriver.ts
 import { chromium } from "playwright-extra";
@@ -3578,6 +3578,9 @@ function classifyAttemptError(err) {
   if (err instanceof RequestAbortedError) return "request_aborted";
   if (err instanceof DOMException && (err.name === "TimeoutError" || err.name === "AbortError")) return "timeout";
   const message = err instanceof Error ? err.message : String(err);
+  if (/browser (?:has been )?closed|context (?:has been )?closed|target page, context or browser has been closed|session closed|page has been closed/i.test(message)) {
+    return "browser_session_interrupted";
+  }
   if (looksLikeProxyTunnelFailure(message)) return "proxy_tunnel_failed";
   if (looksLikeProxyUnavailable(message)) return "proxy_unavailable";
   return /timeout|timed out|Timeout \d+ms exceeded|deadline/i.test(message) ? "timeout" : "error";
@@ -3625,6 +3628,9 @@ function looksLikeProxyUnavailable(message) {
 function retryableLocationProxyError(outcome) {
   return outcome === "captcha" || outcome === "proxy_tunnel_failed" || outcome === "proxy_unavailable";
 }
+function retryableFreshBrowserError(outcome) {
+  return outcome === "browser_session_interrupted";
+}
 function locationMismatchMessage(result) {
   const evidence = result.diagnostics.debug?.locationEvidence;
   const expected = evidence?.expected?.canonicalLocation ?? result.location ?? "requested location";
@@ -3808,7 +3814,7 @@ async function harvest(rawOptions) {
           proxyFailures++;
           if (proxyFailures >= 2) effectiveProxyMode = "none";
         }
-        const willRetry2 = i < maxAttempts - 1 && (outcome2 === "captcha" || proxyMode === "location" && retryableLocationProxyError(outcome2));
+        const willRetry2 = i < maxAttempts - 1 && (outcome2 === "captcha" || retryableFreshBrowserError(outcome2) || proxyMode === "location" && retryableLocationProxyError(outcome2));
         if (outcome2 === "captcha") {
           console.warn(JSON.stringify({
             event: "harvest_attempt_captcha",
@@ -3926,7 +3932,7 @@ async function harvest(rawOptions) {
         proxyFailures++;
         if (proxyFailures >= 2) effectiveProxyMode = "none";
       }
-      const willRetry = i < maxAttempts - 1 && (outcome === "captcha" || proxyMode === "location" && retryableLocationProxyError(outcome));
+      const willRetry = i < maxAttempts - 1 && (outcome === "captcha" || retryableFreshBrowserError(outcome) || proxyMode === "location" && retryableLocationProxyError(outcome));
       if (outcome === "captcha") {
         console.warn(JSON.stringify({
           event: "harvest_attempt_captcha",
@@ -4009,4 +4015,3 @@ export {
   resolveKernelProxyId,
   harvest
 };
-//# sourceMappingURL=chunk-ORKXE7VX.js.map

package/dist/{chunk-AMLFKPLL.js → chunk-75VW7CWE.js}

@@ -1,16 +1,14 @@
 import {
   isVendorUnavailableError
-} from "./chunk-UN4LZVCB.js";
-import {
-  loadHtmlDocument
-} from "./chunk-MQGBCWZD.js";
+} from "./chunk-OZJMVCDK.js";
 import {
+  loadHtmlDocument,
   validatePublicHttpUrl
-} from "./chunk-UVFB24RZ.js";
+} from "./chunk-GGZEC22A.js";
 import {
   buildPublicErrorEnvelope,
   getDb
-} from "./chunk-4VG3NUUY.js";
+} from "./chunk-GTGSD4YS.js";
 
 // src/lib/media-extractor.ts
 import { createWriteStream, mkdirSync, rmSync } from "fs";
@@ -732,4 +730,3 @@ export {
   queuedCommonsDocumentCount,
   semanticCommonsEntityScores
 };
-//# sourceMappingURL=chunk-AMLFKPLL.js.map

package/dist/{chunk-FG6LJGJ7.js → chunk-77LQV73B.js}

@@ -170,4 +170,3 @@ export {
   RawMapsAboutAttributeSchema,
   postToMemoryLibrary
 };
-//# sourceMappingURL=chunk-FG6LJGJ7.js.map

package/dist/{chunk-DKIG2TEE.js → chunk-AQKJOSIB.js}

@@ -1,9 +1,9 @@
 import {
   csvRecords
-} from "./chunk-G3P3ZDB4.js";
+} from "./chunk-RJ7JVYKU.js";
 import {
   getDb
-} from "./chunk-4VG3NUUY.js";
+} from "./chunk-GTGSD4YS.js";
 
 // src/api/location-data-repository.ts
 import { createHash, randomUUID } from "crypto";
@@ -681,4 +681,3 @@ export {
   importHostedCensusPlacesCsv,
   queryHostedLocationMarkets
 };
-//# sourceMappingURL=chunk-DKIG2TEE.js.map

package/dist/{chunk-2XUXVKT4.js → chunk-AYJHI2FK.js}

@@ -1011,4 +1011,3 @@ export {
   MEDIA_TRANSCRIBE_HOLD_MC,
   MC_PER_USD
 };
-//# sourceMappingURL=chunk-2XUXVKT4.js.map

package/dist/{chunk-5BEPOFKG.js → chunk-CNLLTD5B.js}

@@ -3007,4 +3007,3 @@ export {
   refreshAnalyticsDailyRollupsIfDue,
   createAnalyticsExport
 };
-//# sourceMappingURL=chunk-5BEPOFKG.js.map

package/dist/{chunk-EKJ65E3Z.js → chunk-DNM65UCK.js}

@@ -297,4 +297,3 @@ export {
   readPrivateArtifactWindow,
   readPrivateArtifactBuffer
 };
-//# sourceMappingURL=chunk-EKJ65E3Z.js.map