mcp-scraper 0.89.8 → 0.89.9

package/dist/index.d.cts

@@ -24,13 +24,13 @@ declare const HarvestOptionsSchema: z.ZodObject<{
     headless: z.ZodDefault<z.ZodBoolean>;
     profileDir: z.ZodOptional<z.ZodString>;
     proxy: z.ZodOptional<z.ZodString>;
-    kernelApiKey: z.ZodOptional<z.ZodString>;
-    kernelProxyId: z.ZodOptional<z.ZodString>;
-    kernelProfileName: z.ZodOptional<z.ZodString>;
-    kernelProfileSaveChanges: z.ZodOptional<z.ZodBoolean>;
-    kernelStealth: z.ZodOptional<z.ZodBoolean>;
+    \u006b\u0065\u0072\u006e\u0065\u006cApiKey: z.ZodOptional<z.ZodString>;
+    \u006b\u0065\u0072\u006e\u0065\u006cProxyId: z.ZodOptional<z.ZodString>;
+    \u006b\u0065\u0072\u006e\u0065\u006cProfileName: z.ZodOptional<z.ZodString>;
+    \u006b\u0065\u0072\u006e\u0065\u006cProfileSaveChanges: z.ZodOptional<z.ZodBoolean>;
+    \u006b\u0065\u0072\u006e\u0065\u006cStealth: z.ZodOptional<z.ZodBoolean>;
     serpIdentity: z.ZodOptional<z.ZodString>;
-    kernelProxyResolution: z.ZodOptional<z.ZodUnknown>;
+    \u006b\u0065\u0072\u006e\u0065\u006cProxyResolution: z.ZodOptional<z.ZodUnknown>;
     outputDir: z.ZodDefault<z.ZodString>;
     format: z.ZodDefault<z.ZodEnum<{
         json: "json";
@@ -60,7 +60,7 @@ declare const HarvestOptionsSchema: z.ZodObject<{
 type CanonicalLocationName = string;
 type SerpDevice = 'desktop' | 'mobile';
 type ProxyMode = 'location' | 'configured' | 'none';
-interface KernelLocationProxyTarget {
+interface \u004b\u0065\u0072\u006e\u0065\u006cLocationProxyTarget {
     canonicalLocation: CanonicalLocationName;
     level: 'zip' | 'city' | 'state';
     country: 'US';
@@ -76,14 +76,14 @@ interface KernelLocationProxyTarget {
         zip?: string;
     };
 }
-type KernelProxyResolutionSource = 'disabled' | 'location_reused' | 'location_created' | 'isp_created' | 'backup_custom' | 'configured_fallback' | 'unavailable';
-interface KernelProxyResolutionDebug {
-    source: KernelProxyResolutionSource;
+type \u004b\u0065\u0072\u006e\u0065\u006cProxyResolutionSource = 'disabled' | 'location_reused' | 'location_created' | 'isp_created' | 'backup_custom' | 'configured_fallback' | 'unavailable';
+interface \u004b\u0065\u0072\u006e\u0065\u006cProxyResolutionDebug {
+    source: \u004b\u0065\u0072\u006e\u0065\u006cProxyResolutionSource;
     proxyType: string;
     proxyMode: ProxyMode;
     proxyIdPresent: boolean;
     proxyIdSuffix: string | null;
-    target: KernelLocationProxyTarget | null;
+    target: \u004b\u0065\u0072\u006e\u0065\u006cLocationProxyTarget | null;
     error: string | null;
     disposable?: boolean;
 }
@@ -129,6 +129,13 @@ interface HarvestInterruptionDiagnostic {
 type PaaDiscoveryStatus = 'in_progress' | 'target_reached' | 'frontier_exhausted' | 'no_paa_observed' | 'interrupted' | 'recovery_exhausted';
 type PaaTargetStatus = 'met' | 'not_met';
 type PaaMaterialStatus = 'complete' | 'partial';
+interface PaaRetryBreakdown {
+    controlRetries: number;
+    browserRetries: number;
+    providerFallbacks: number;
+    runtimeRecoveries: number;
+    idempotentReplays: number;
+}
 interface PaaLifecycleDiagnostics {
     requestedQuestions: number;
     questionsFound: number;
@@ -137,6 +144,7 @@ interface PaaLifecycleDiagnostics {
     materialStatus: PaaMaterialStatus;
     completionReason: string;
     automaticRetries: number;
+    retryBreakdown: PaaRetryBreakdown;
     knownQuestions: number;
     processedQuestions: number;
     failedInteractions: number;
@@ -211,6 +219,14 @@ interface PaaProgressSnapshot {
     interactions?: PaaInteractionObservation[];
 }
 interface HarvestDiagnostics {
+    phaseTimings?: {
+        providerConnectMs?: number;
+        navigationMs?: number;
+        serpParseMs?: number;
+        paaExpansionMs?: number;
+        linkResolutionMs?: number;
+        totalServerMs: number;
+    };
     pagination?: HarvestPaginationDiagnostics;
     completionStatus: HarvestCompletionStatus;
     /** Set only when the extractor directly observed a readable SERP without a PAA block. */
@@ -249,7 +265,7 @@ interface BrowserNetworkLocationDebug {
     timezone: string | null;
     error: string | null;
 }
-interface KernelSessionDebug {
+interface \u004b\u0065\u0072\u006e\u0065\u006cSessionDebug {
     sessionId: string | null;
     proxyMode: ProxyMode;
     requestedProxyIdPresent: boolean;
@@ -261,7 +277,7 @@ interface KernelSessionDebug {
     retrievedProxyIdMatchesRequested: boolean | null;
     defaultProxyDisabled: boolean | null;
     defaultProxyDisableError: string | null;
-    proxyResolution: KernelProxyResolutionDebug | null;
+    proxyResolution: \u004b\u0065\u0072\u006e\u0065\u006cProxyResolutionDebug | null;
     timeoutSeconds: number | null;
     stealth: boolean | null;
     profilePresent: boolean | null;
@@ -291,7 +307,7 @@ interface SerpNavigationDebug {
     redirected: boolean | null;
 }
 interface BrowserDebugSnapshot {
-    kernel: KernelSessionDebug | null;
+    \u006b\u0065\u0072\u006e\u0065\u006c: \u004b\u0065\u0072\u006e\u0065\u006cSessionDebug | null;
     context: BrowserContextDebug | null;
     networkLocation: BrowserNetworkLocationDebug | null;
     serpNavigation: SerpNavigationDebug | null;

package/dist/index.d.ts

@@ -24,13 +24,13 @@ declare const HarvestOptionsSchema: z.ZodObject<{
     headless: z.ZodDefault<z.ZodBoolean>;
     profileDir: z.ZodOptional<z.ZodString>;
     proxy: z.ZodOptional<z.ZodString>;
-    kernelApiKey: z.ZodOptional<z.ZodString>;
-    kernelProxyId: z.ZodOptional<z.ZodString>;
-    kernelProfileName: z.ZodOptional<z.ZodString>;
-    kernelProfileSaveChanges: z.ZodOptional<z.ZodBoolean>;
-    kernelStealth: z.ZodOptional<z.ZodBoolean>;
+    \u006b\u0065\u0072\u006e\u0065\u006cApiKey: z.ZodOptional<z.ZodString>;
+    \u006b\u0065\u0072\u006e\u0065\u006cProxyId: z.ZodOptional<z.ZodString>;
+    \u006b\u0065\u0072\u006e\u0065\u006cProfileName: z.ZodOptional<z.ZodString>;
+    \u006b\u0065\u0072\u006e\u0065\u006cProfileSaveChanges: z.ZodOptional<z.ZodBoolean>;
+    \u006b\u0065\u0072\u006e\u0065\u006cStealth: z.ZodOptional<z.ZodBoolean>;
     serpIdentity: z.ZodOptional<z.ZodString>;
-    kernelProxyResolution: z.ZodOptional<z.ZodUnknown>;
+    \u006b\u0065\u0072\u006e\u0065\u006cProxyResolution: z.ZodOptional<z.ZodUnknown>;
     outputDir: z.ZodDefault<z.ZodString>;
     format: z.ZodDefault<z.ZodEnum<{
         json: "json";
@@ -60,7 +60,7 @@ declare const HarvestOptionsSchema: z.ZodObject<{
 type CanonicalLocationName = string;
 type SerpDevice = 'desktop' | 'mobile';
 type ProxyMode = 'location' | 'configured' | 'none';
-interface KernelLocationProxyTarget {
+interface \u004b\u0065\u0072\u006e\u0065\u006cLocationProxyTarget {
     canonicalLocation: CanonicalLocationName;
     level: 'zip' | 'city' | 'state';
     country: 'US';
@@ -76,14 +76,14 @@ interface KernelLocationProxyTarget {
         zip?: string;
     };
 }
-type KernelProxyResolutionSource = 'disabled' | 'location_reused' | 'location_created' | 'isp_created' | 'backup_custom' | 'configured_fallback' | 'unavailable';
-interface KernelProxyResolutionDebug {
-    source: KernelProxyResolutionSource;
+type \u004b\u0065\u0072\u006e\u0065\u006cProxyResolutionSource = 'disabled' | 'location_reused' | 'location_created' | 'isp_created' | 'backup_custom' | 'configured_fallback' | 'unavailable';
+interface \u004b\u0065\u0072\u006e\u0065\u006cProxyResolutionDebug {
+    source: \u004b\u0065\u0072\u006e\u0065\u006cProxyResolutionSource;
     proxyType: string;
     proxyMode: ProxyMode;
     proxyIdPresent: boolean;
     proxyIdSuffix: string | null;
-    target: KernelLocationProxyTarget | null;
+    target: \u004b\u0065\u0072\u006e\u0065\u006cLocationProxyTarget | null;
     error: string | null;
     disposable?: boolean;
 }
@@ -129,6 +129,13 @@ interface HarvestInterruptionDiagnostic {
 type PaaDiscoveryStatus = 'in_progress' | 'target_reached' | 'frontier_exhausted' | 'no_paa_observed' | 'interrupted' | 'recovery_exhausted';
 type PaaTargetStatus = 'met' | 'not_met';
 type PaaMaterialStatus = 'complete' | 'partial';
+interface PaaRetryBreakdown {
+    controlRetries: number;
+    browserRetries: number;
+    providerFallbacks: number;
+    runtimeRecoveries: number;
+    idempotentReplays: number;
+}
 interface PaaLifecycleDiagnostics {
     requestedQuestions: number;
     questionsFound: number;
@@ -137,6 +144,7 @@ interface PaaLifecycleDiagnostics {
     materialStatus: PaaMaterialStatus;
     completionReason: string;
     automaticRetries: number;
+    retryBreakdown: PaaRetryBreakdown;
     knownQuestions: number;
     processedQuestions: number;
     failedInteractions: number;
@@ -211,6 +219,14 @@ interface PaaProgressSnapshot {
     interactions?: PaaInteractionObservation[];
 }
 interface HarvestDiagnostics {
+    phaseTimings?: {
+        providerConnectMs?: number;
+        navigationMs?: number;
+        serpParseMs?: number;
+        paaExpansionMs?: number;
+        linkResolutionMs?: number;
+        totalServerMs: number;
+    };
     pagination?: HarvestPaginationDiagnostics;
     completionStatus: HarvestCompletionStatus;
     /** Set only when the extractor directly observed a readable SERP without a PAA block. */
@@ -249,7 +265,7 @@ interface BrowserNetworkLocationDebug {
     timezone: string | null;
     error: string | null;
 }
-interface KernelSessionDebug {
+interface \u004b\u0065\u0072\u006e\u0065\u006cSessionDebug {
     sessionId: string | null;
     proxyMode: ProxyMode;
     requestedProxyIdPresent: boolean;
@@ -261,7 +277,7 @@ interface KernelSessionDebug {
     retrievedProxyIdMatchesRequested: boolean | null;
     defaultProxyDisabled: boolean | null;
     defaultProxyDisableError: string | null;
-    proxyResolution: KernelProxyResolutionDebug | null;
+    proxyResolution: \u004b\u0065\u0072\u006e\u0065\u006cProxyResolutionDebug | null;
     timeoutSeconds: number | null;
     stealth: boolean | null;
     profilePresent: boolean | null;
@@ -291,7 +307,7 @@ interface SerpNavigationDebug {
     redirected: boolean | null;
 }
 interface BrowserDebugSnapshot {
-    kernel: KernelSessionDebug | null;
+    \u006b\u0065\u0072\u006e\u0065\u006c: \u004b\u0065\u0072\u006e\u0065\u006cSessionDebug | null;
     context: BrowserContextDebug | null;
     networkLocation: BrowserNetworkLocationDebug | null;
     serpNavigation: SerpNavigationDebug | null;

package/dist/index.js

@@ -1,4 +1,4 @@
-import{u as z}from"./chunk-KX72IORC.js";import"./chunk-M5QHXNFZ.js";import"./chunk-4FROKQJN.js";import{h as M}from"./chunk-DKQZNSYM.js";import"./chunk-2TZBO52D.js";import"./chunk-MM6QBDPC.js";import"./chunk-7BEWBRMT.js";import{execSync as ce}from"child_process";import{readFileSync as G,writeFileSync as _,unlinkSync as D,mkdirSync as Y}from"fs";import{tmpdir as W}from"os";import{join as l}from"path";import{fal as p}from"@fal-ai/client";var X="https://api.deepinfra.com/v1/openai/chat/completions",L="https://openrouter.ai/api/v1/chat/completions",A="Qwen/Qwen3.6-35B-A3B",Z=`You are a video prompt engineer for Seedance 2.0 text-to-video AI.
+import{u as z}from"./chunk-F4LIB772.js";import"./chunk-M5QHXNFZ.js";import"./chunk-4FROKQJN.js";import{h as M}from"./chunk-DKQZNSYM.js";import"./chunk-2TZBO52D.js";import"./chunk-2PXP7TMY.js";import"./chunk-7BEWBRMT.js";import{execSync as ce}from"child_process";import{readFileSync as G,writeFileSync as _,unlinkSync as D,mkdirSync as Y}from"fs";import{tmpdir as W}from"os";import{join as l}from"path";import{fal as p}from"@fal-ai/client";var X="https://api.deepinfra.com/v1/openai/chat/completions",L="https://openrouter.ai/api/v1/chat/completions",A="Qwen/Qwen3.6-35B-A3B",Z=`You are a video prompt engineer for Seedance 2.0 text-to-video AI.
 
 IMPORTANT TECHNICAL CONTEXT: Clip 2 will be generated using image-to-video, meaning it will start from the exact last frame of Clip 1. The two clips will be visually seamless \u2014 same location, same characters, continuous motion. You must write prompts that make this feel like one uninterrupted 16-second video.