webpeel 0.21.85 → 0.21.86

package/dist/ee/challenge-re-export.d.ts

@@ -0,0 +1 @@
+export { solveChallenge } from './challenge-solver.js';

package/dist/ee/challenge-re-export.js

@@ -0,0 +1 @@
+export { solveChallenge } from './challenge-solver.js';

package/dist/ee/challenge-solver.d.ts

@@ -0,0 +1,72 @@
+/**
+ * Challenge / bot-protection solver.
+ *
+ * Attempts to bypass bot-protection challenges using free, in-process methods:
+ *  1. Cloudflare JS challenge — render in stealth Playwright, wait for auto-solve
+ *  2. hCaptcha — accessibility bypass (TODO: implement if API is confirmed available)
+ *
+ * Architecture note:
+ *  Browser-based solving is CPU/RAM intensive. When the env var BROWSER_WORKER_URL
+ *  is set, the solve request is proxied to an external worker (e.g. Hetzner 4GB VM)
+ *  instead of running locally. This keeps the main Render container (512 MB) lean.
+ *
+ * Usage:
+ *  const result = await solveChallenge(url, 'cloudflare', html);
+ *  if (result.solved) {
+ *    // result.html = real page content
+ *    // result.cookies = ["cf_clearance=...", ...]
+ *  }
+ */
+import type { ChallengeType } from '../core/challenge-detection.js';
+export interface ImageCaptchaResult {
+    solved: boolean;
+    rounds: number;
+    error?: string;
+}
+/**
+ * Ask the moondream vision model which grid cells contain the target object.
+ * Returns an array of 1-indexed grid positions (1–9), or null if the call fails.
+ */
+export declare function askVisionModel(base64Image: string, targetObject: string): Promise<number[] | null>;
+/**
+ * Detect if the page has an image grid CAPTCHA and extract the target object.
+ * Returns the object name (e.g. "traffic lights") or null if not detected.
+ */
+export declare function detectImageCaptchaTarget(page: import('playwright').Page): Promise<string | null>;
+/**
+ * Solve an image grid CAPTCHA using the moondream vision model.
+ *
+ * Flow per round:
+ *  1. Screenshot the CAPTCHA grid element
+ *  2. Send to moondream → get grid positions
+ *  3. Click identified cells
+ *  4. Click Verify button
+ *  5. Check if solved; if a new round appears, repeat (max 3 rounds)
+ */
+export declare function solveImageCaptcha(page: import('playwright').Page, targetObject: string): Promise<ImageCaptchaResult>;
+export interface SolveOptions {
+    /** Hard timeout in ms (default: 15 000) */
+    timeout?: number;
+    /** Optional proxy URL (http://user:pass@host:port) */
+    proxy?: string;
+}
+export interface SolveResult {
+    solved: boolean;
+    html: string;
+    /** Raw Set-Cookie header values extracted after solve */
+    cookies?: string[];
+    /** How the solve was performed */
+    method?: 'local-browser' | 'remote-worker' | 'accessibility';
+    /** Error details if solve failed */
+    error?: string;
+}
+/**
+ * Attempt to solve a bot-protection challenge.
+ *
+ * @param url            The page URL (used for proxy routing and cookie caching)
+ * @param challengeType  The type of challenge as detected by challenge-detection
+ * @param html           The raw challenge HTML (used for context / fallback)
+ * @param options        Optional timeout and proxy settings
+ * @returns              Solve result with real HTML content and cookies if successful
+ */
+export declare function solveChallenge(url: string, challengeType: ChallengeType, html: string, options?: SolveOptions): Promise<SolveResult>;