@vakra-dev/reader 0.0.3 → 0.1.1

package/dist/index.d.ts

@@ -90,6 +90,20 @@ interface IBrowserPool {
     healthCheck?(): Promise<HealthStatus>;
 }
 
+/**
+ * Engine types for multi-engine scraping architecture
+ *
+ * Engine stack (in order of preference):
+ * 1. http - Native fetch, fastest, no browser
+ * 2. tlsclient - TLS fingerprinting via got-scraping
+ * 3. hero - Full browser with JavaScript execution
+ */
+
+/**
+ * Available engine names
+ */
+type EngineName = "http" | "tlsclient" | "hero";
+
 /**
  * Proxy configuration for Hero
  */
@@ -143,6 +157,8 @@ interface ScrapeOptions {
     formats?: Array<"markdown" | "html">;
     /** Custom user agent string */
     userAgent?: string;
+    /** Custom headers for requests */
+    headers?: Record<string, string>;
     /** Request timeout in milliseconds (default: 30000) */
     timeoutMs?: number;
     /** URL patterns to include (regex strings) */
@@ -187,6 +203,12 @@ interface ScrapeOptions {
     browserPool?: BrowserPoolConfig;
     /** Browser pool instance (internal, provided by ReaderClient) */
     pool?: IBrowserPool;
+    /** Engines to use in order (default: ['http', 'tlsclient', 'hero']) */
+    engines?: EngineName[];
+    /** Skip specific engines (e.g., ['http'] to skip native fetch) */
+    skipEngines?: EngineName[];
+    /** Force a specific engine, skipping the cascade */
+    forceEngine?: EngineName;
 }
 /**
  * Website metadata extracted from the base page
@@ -306,13 +328,17 @@ interface ScrapeResult {
 /**
  * Default scrape options
  */
-declare const DEFAULT_OPTIONS: Omit<Required<ScrapeOptions>, "proxy" | "waitForSelector" | "connectionToCore" | "userAgent" | "browserPool" | "pool"> & {
+declare const DEFAULT_OPTIONS: Omit<Required<ScrapeOptions>, "proxy" | "waitForSelector" | "connectionToCore" | "userAgent" | "headers" | "browserPool" | "pool" | "engines" | "skipEngines" | "forceEngine"> & {
     proxy?: ProxyConfig;
     waitForSelector?: string;
     connectionToCore?: any;
     userAgent?: string;
+    headers?: Record<string, string>;
     browserPool?: BrowserPoolConfig;
     pool?: IBrowserPool;
+    engines?: EngineName[];
+    skipEngines?: EngineName[];
+    forceEngine?: EngineName;
 };
 /**
  * Format type guard
@@ -547,7 +573,6 @@ declare class ReaderClient {
  */
 declare class Scraper {
     private options;
-    private pool;
     private logger;
     private robotsCache;
     constructor(options: ScrapeOptions);
@@ -570,13 +595,7 @@ declare class Scraper {
      */
     private scrapeSingleUrlWithRetry;
     /**
-     * Wait for the final page to load after any Cloudflare redirects
-     * Cloudflare often does silent redirects even when bypassed, we need to ensure
-     * we're on the actual content page before scraping.
-     */
-    private waitForFinalPage;
-    /**
-     * Scrape a single URL
+     * Scrape a single URL using the engine orchestrator
      */
     private scrapeSingleUrl;
     /**
@@ -832,6 +851,8 @@ declare function isDaemonRunning(port?: number): Promise<boolean>;
  *
  * Simple conversion without any headers, metadata, or formatting wrappers.
  * Returns clean markdown content ready for LLM consumption.
+ *
+ * Uses supermarkdown (Rust-based) for high-performance conversion.
  */
 declare function htmlToMarkdown(html: string): string;
 /**