1import type { WaitForInteractionRequest } from "../adapters/types";
2import type { BrowserSession } from "./session";
3 
4interface GateSnapshot {
5  title: string;
6  currentUrl: string;
7  bodyText: string;
8  hasCloudflareTurnstile: boolean;
9  hasCloudflareChallenge: boolean;
10  hasRecaptcha: boolean;
11  hasRecaptchaIframe: boolean;
12  hasHcaptcha: boolean;
13  hasHcaptchaIframe: boolean;
14}
15 
16export function detectInteractionGateFromSnapshot(snapshot: GateSnapshot): WaitForInteractionRequest | null {
17  const text = snapshot.bodyText.toLowerCase();
18  const title = snapshot.title.toLowerCase();
19  const url = snapshot.currentUrl.toLowerCase();
20 
21  if (
22    snapshot.hasCloudflareTurnstile ||
23    snapshot.hasCloudflareChallenge ||
24    title.includes("just a moment") ||
25    text.includes("verify you are human") ||
26    text.includes("checking your browser before accessing") ||
27    text.includes("enable javascript and cookies to continue") ||
28    url.includes("/cdn-cgi/challenge-platform/")
29  ) {
30    return {
31      type: "wait_for_interaction",
32      kind: "cloudflare",
33      provider: "cloudflare",
34      reason: "Cloudflare human verification detected",
35      prompt: "Please complete the Cloudflare verification in the opened Chrome window. Extraction will continue automatically once the challenge disappears.",
36      requiresVisibleBrowser: true,
37    };
38  }
39 
40  if (
41    snapshot.hasRecaptcha ||
42    snapshot.hasRecaptchaIframe ||
43    text.includes("i'm not a robot") ||
44    text.includes("recaptcha")
45  ) {
46    return {
47      type: "wait_for_interaction",
48      kind: "recaptcha",
49      provider: "google_recaptcha",
50      reason: "Google reCAPTCHA detected",
51      prompt: "Please complete the reCAPTCHA verification in the opened Chrome window. Extraction will continue automatically once the challenge disappears.",
52      requiresVisibleBrowser: true,
53    };
54  }
55 
56  if (
57    snapshot.hasHcaptcha ||
58    snapshot.hasHcaptchaIframe ||
59    text.includes("hcaptcha")
60  ) {
61    return {
62      type: "wait_for_interaction",
63      kind: "hcaptcha",
64      provider: "hcaptcha",
65      reason: "hCaptcha verification detected",
66      prompt: "Please complete the hCaptcha verification in the opened Chrome window. Extraction will continue automatically once the challenge disappears.",
67      requiresVisibleBrowser: true,
68    };
69  }
70 
71  return null;
72}
73 
74export async function detectInteractionGate(browser: BrowserSession): Promise<WaitForInteractionRequest | null> {
75  const snapshot = await browser.evaluate<GateSnapshot>(`
76    (() => {
77      const bodyText = (document.body?.innerText ?? "").slice(0, 4000);
78      return {
79        title: document.title ?? "",
80        currentUrl: window.location.href,
81        bodyText,
82        hasCloudflareTurnstile: Boolean(
83          document.querySelector(
84            '.cf-turnstile, [name="cf-turnstile-response"], iframe[src*="challenges.cloudflare.com"]'
85          )
86        ),
87        hasCloudflareChallenge: Boolean(
88          document.querySelector(
89            '#challenge-running, #cf-challenge-running, .challenge-platform, [data-ray], [data-translate="checking_browser"]'
90          )
91        ),
92        hasRecaptcha: Boolean(
93          document.querySelector(
94            '.g-recaptcha, textarea[name="g-recaptcha-response"], iframe[title*="reCAPTCHA"]'
95          )
96        ),
97        hasRecaptchaIframe: Boolean(
98          document.querySelector('iframe[src*="google.com/recaptcha"], iframe[src*="recaptcha/api2"]')
99        ),
100        hasHcaptcha: Boolean(
101          document.querySelector(
102            '.h-captcha, textarea[name="h-captcha-response"], iframe[title*="hCaptcha"]'
103          )
104        ),
105        hasHcaptchaIframe: Boolean(
106          document.querySelector('iframe[src*="hcaptcha.com"]')
107        ),
108      };
109    })()
110  `).catch(() => ({
111    title: "",
112    currentUrl: "",
113    bodyText: "",
114    hasCloudflareTurnstile: false,
115    hasCloudflareChallenge: false,
116    hasRecaptcha: false,
117    hasRecaptchaIframe: false,
118    hasHcaptcha: false,
119    hasHcaptchaIframe: false,
120  }));
121 
122  return detectInteractionGateFromSnapshot(snapshot);
123}
124