solver.js

import { spawnSync } from 'child_process';
import fs from 'fs';
import path from 'path';
import os from 'os';
import { Readable } from 'stream';
import vosk from 'vosk-koffi';
import wav from 'wav';
import { fileURLToPath } from 'url';

const __filename = fileURLToPath(import.meta.url);
const __dirname = path.dirname(__filename);

vosk.setLogLevel(-1);
const MODEL_DIR = path.resolve(__dirname, "model");
const model = new vosk.Model(MODEL_DIR);

const SOURCE_FILE = "sound.mp3";
const OUT_FILE = "out.wav";
const MAIN_FRAME = "iframe[title='reCAPTCHA']";
const BFRAME = "iframe[src*='google.com/recaptcha'][src*='bframe']";
const CHALLENGE = "body > div > div";

class Mutex {
    constructor(name = "Mutex") {
        this.name = name;
        this._locked = false;
        this._queue = [];
    }

    async lock(tag) {
        if (this._locked) {
            if (tag) {
                console.log(`[${this.name}] ${tag} waiting`);
            }

            return new Promise((resolve) => {
                this._queue.push(() => {
                    this._lock(tag);
                    resolve();
                });
            });
        }

        this._lock(tag);
    }

    unlock(tag) {
        if (this._locked) {
            if (tag) {
                console.log(`[${this.name}] ${tag} unlocked`);
            }

            this._locked = false;
            this._queue.shift()?.();
        }
    }

    _lock(tag) {
        this._locked = true;

        if (tag) {
            console.log(`[${this.name}] ${tag} locked`);
        }
    }
}

function sleep(ms) {
    return new Promise((resolve) => setTimeout(resolve, ms));
}

function createDir() {
    const dir = path.resolve(os.tmpdir(), "reSOLVER-" + Math.random().toString().slice(2));
    if (fs.existsSync(dir)) {
        fs.rmSync(dir, { recursive: true });
    }
    fs.mkdirSync(dir, { recursive: true });
    return dir;
}

function convert(dir, ffmpeg = "ffmpeg") {
    const args = [
        "-loglevel",
        "error",
        "-i",
        SOURCE_FILE,
        "-acodec",
        "pcm_s16le",
        "-ac",
        "1",
        "-ar",
        "16000",
        OUT_FILE,
    ];

    spawnSync(ffmpeg, args, { cwd: dir });
}

function recognize(dir) {
    return new Promise((resolve) => {
        const stream = fs.createReadStream(path.resolve(dir, OUT_FILE), { highWaterMark: 4096 });

        const reader = new wav.Reader();
        const readable = new Readable().wrap(reader);
        reader.on("format", async ({ audioFormat, sampleRate, channels }) => {
            if (audioFormat != 1 || channels != 1) {
                throw new Error("Audio file must be WAV with mono PCM.");
            }

            const rec = new vosk.Recognizer({ model, sampleRate });
            rec.setMaxAlternatives(10);
            rec.setWords(true);
            rec.setPartialWords(true);

            for await (const data of readable) {
                const end_of_speech = rec.acceptWaveform(data);
                if (end_of_speech) {
                    const result = rec
                        .result()
                        .alternatives.sort((a, b) => b.confidence - a.confidence)[0].text;
                    stream.close(() => resolve(result));
                }
            }

            rec.free();
        });

        stream.pipe(reader);
    });
}

async function getText(res, ffmpeg = "ffmpeg") {
    const tempDir = createDir();

    fs.writeFileSync(path.resolve(tempDir, SOURCE_FILE), await res.body());
    convert(tempDir, ffmpeg);
    const result = await recognize(tempDir);

    fs.rmSync(tempDir, { recursive: true });
    return result;
}

export async function solve(page, { delay = 64, wait = 5000, retry = 3, ffmpeg = "ffmpeg" } = {}) {
    console.log("Starting reCAPTCHA solver...");
    
    try {
        await page.waitForSelector('iframe', { state: "attached", timeout: wait });
        console.log("Found iframes on page");
        
        const allIframes = await page.$$('iframe');
        console.log(`Total iframes found: ${allIframes.length}`);
        
        for (let i = 0; i < allIframes.length; i++) {
            const src = await allIframes[i].getAttribute('src');
            const title = await allIframes[i].getAttribute('title');
            console.log(`Iframe ${i}: src=${src}, title=${title}`);
        }
    } catch (error) {
        console.error("Error finding iframes:", error.message);
        throw new Error("No reCAPTCHA detected");
    }

    let invisible = false;

    let b_iframe = null;
    const bframeSelectors = [
        "iframe[src*='/recaptcha/api2/bframe']",
        "iframe[src*='/recaptcha/enterprise/bframe']",
        "iframe[src*='bframe']"
    ];
    
    for (const selector of bframeSelectors) {
        try {
            await page.waitForSelector(selector, { state: "attached", timeout: 2000 });
            b_iframe = await page.$(selector);
            if (b_iframe) {
                console.log(`Found bframe with selector: ${selector}`);
                break;
            }
        } catch {
            continue;
        }
    }

    if (b_iframe === null) {
        console.log("Bframe not found yet, checking main frame...");
        
        try {
            await page.waitForSelector(MAIN_FRAME, { state: "attached", timeout: wait });
            const iframe = await page.$(MAIN_FRAME);
            
            if (iframe === null) {
                throw new Error("Could not find reCAPTCHA iframe");
            }

            const box_page = await iframe.contentFrame();
            if (box_page === null) {
                throw new Error("Could not find reCAPTCHA iframe content");
            }

            invisible = (await box_page.$("div.rc-anchor-invisible")) ? true : false;
            console.log("invisible:", invisible);

            if (invisible === true) {
                return false;
            } else {
                const label = await box_page.$("#recaptcha-anchor-label");
                if (label === null) {
                    throw new Error("Could not find reCAPTCHA label");
                }

                console.log("Clicking reCAPTCHA checkbox...");
                await label.click();
                
                await sleep(2000);
                
                for (const selector of bframeSelectors) {
                    b_iframe = await page.$(selector);
                    if (b_iframe) {
                        console.log(`Found bframe after click: ${selector}`);
                        break;
                    }
                }
            }
        } catch (error) {
            console.error("Error handling main frame:", error.message);
            throw new Error("Could not find reCAPTCHA popup iframe");
        }
    }

    if (b_iframe === null) {
        throw new Error("Could not find reCAPTCHA popup iframe after all attempts");
    }

    const bframe = await b_iframe.contentFrame();
    if (bframe === null) {
        throw new Error("Could not find reCAPTCHA popup iframe content");
    }

    await sleep(1000);
    
    const challenge = await bframe.$(CHALLENGE);
    if (challenge === null) {
        console.log("No challenge found yet, this might be OK");
        return false;
    }

    const required = await challenge.evaluate(
        (elm) => !elm.classList.contains("rc-footer"),
    );
    console.log("action required:", required);

    if (required === false) {
        return false;
    }

    await bframe.waitForSelector("#recaptcha-audio-button", { timeout: wait });
    const audio_button = await bframe.$("#recaptcha-audio-button");
    if (audio_button === null) {
        throw new Error("Could not find reCAPTCHA audio button");
    }

    const mutex = new Mutex();
    await mutex.lock("init");
    let passed = false;
    let answer = Promise.resolve("");
    const listener = async (res) => {
        if (res.headers()["content-type"] === "audio/mp3") {
            console.log(`got audio from ${res.url()}`);
            answer = new Promise((resolve) => {
                getText(res, ffmpeg)
                    .then(resolve)
                    .catch(() => undefined);
            });
            mutex.unlock("get sound");
        } else if (
            res.url().startsWith("https://www.google.com/recaptcha/api2/userverify") ||
            res.url().startsWith("https://www.google.com/recaptcha/enterprise/userverify")
        ) {
            const raw = (await res.body()).toString().replace(")]}'\n", "");
            const json = JSON.parse(raw);
            passed = json[2] === 1;
            mutex.unlock("verified");
        }
    };
    page.on("response", listener);

    await audio_button.click();

    let tried = 0;
    while (passed === false) {
        if (tried++ >= retry) {
            throw new Error("Could not solve reCAPTCHA");
        }

        await Promise.race([
            mutex.lock("ready"),
            sleep(wait).then(() => {
                throw new Error("No Audio Found");
            }),
        ]);
        await bframe.waitForSelector("#audio-source", { state: "attached", timeout: wait });
        await bframe.waitForSelector("#audio-response", { timeout: wait });

        console.log("recognized:", await answer);

        const input = await bframe.$("#audio-response");
        if (input === null) {
            throw new Error("Could not find reCAPTCHA audio input");
        }

        await input.type(await answer, { delay });

        const button = await bframe.$("#recaptcha-verify-button");
        if (button === null) {
            throw new Error("Could not find reCAPTCHA verify button");
        }

        await button.click();
        await mutex.lock("done");
        console.log("passed:", passed);
    }

    page.off("response", listener);

    await sleep(1000);

    let token = null;
    
    try {
        const iframe = await page.$(MAIN_FRAME);
        if (iframe) {
            const box_page = await iframe.contentFrame();
            if (box_page) {
                const textarea = await box_page.$('#g-recaptcha-response');
                if (textarea) {
                    token = await textarea.evaluate(el => el.value);
                    console.log("Token found in textarea:", token?.substring(0, 50) + "...");
                }
            }
        }
    } catch (error) {
        console.log("Error extracting token from iframe:", error.message);
    }

    if (!token) {
        try {
            const textarea = await page.$('#g-recaptcha-response');
            if (textarea) {
                token = await textarea.evaluate(el => el.value);
                console.log("Token found in main page:", token?.substring(0, 50) + "...");
            }
        } catch (error) {
            console.log("Error extracting token from main page:", error.message);
        }
    }

    if (!token) {
        try {
            const allTextareas = await page.$('textarea');
            for (const textarea of allTextareas) {
                const name = await textarea.getAttribute('name');
                const id = await textarea.getAttribute('id');
                if (name === 'g-recaptcha-response' || id === 'g-recaptcha-response') {
                    token = await textarea.evaluate(el => el.value);
                    console.log("Token found in textarea by attribute:", token?.substring(0, 50) + "...");
                    break;
                }
            }
        } catch (error) {
            console.log("Error searching all textareas:", error.message);
        }
    }

    console.log("Final token:", token ? `${token.substring(0, 50)}... (length: ${token.length})` : "not found");

    return { success: true, token };
}