try to use speechtext server

This commit is contained in:
2022-09-26 18:33:11 +02:00
parent 08e6330c78
commit cf5a97c844
+54 -13
View File
@@ -1,6 +1,8 @@
const axios = require("axios");
const PLAY_AUDIO_BTN_ID = ".audio-captcha-play-button"
const wget = require('node-wget');
const crypto = require('crypto');
const fs = require("fs");
const homedir = require('os').homedir();
const dest_dir = homedir + "/wavs/"
// iframe.query_selector('.audio-captcha-track').inner_html()
@@ -17,6 +19,32 @@ function delay(delayInMs) {
});
}
function getRandom() {
return Math.floor(Math.random() * 3);
}
function getRandomWaitTime() {
return getRandom() * 1000
}
async function sendRequest(fileName, callback) {
const fileStream = fs.createReadStream(fileName);
let response = await axios.post('http://127.0.0.1:8000', fileStream, {
headers: {
'Content-Type': 'wav'
}
})
if (response.status === 200) {
let result = response.data;
console.log(result);
callback(result)
} else {
console.log("error")
callback([])
}
}
class GeoCaptchaSolver {
constructor(page) {
this.page = page
@@ -29,36 +57,49 @@ class GeoCaptchaSolver {
let iframeHandler = await this.page.frameLocator("body > iframe");
let playAudioBtn = await iframeHandler.locator(PLAY_AUDIO_BTN_ID);
playAudioBtn.click();
await delay(5000);
// let captcha_track = await iframeHandler.locator(AUDIO_CAPTCHA_TRACK);
// const handle = await this.page.frameLocator("iframe");
// const contentFrame = await handle.contentFrame();
// console.log(contentFrame)
await delay(5000 + getRandomWaitTime());
let captcha_container = await iframeHandler.locator(CAPTCHA_CONTAINER)
// let result = await iframeHandler.evaluate(() => {
// document.getElementsByClassName(AUDIO_CAPTCHA_TRACK)
// })
let html = await captcha_container.innerHTML()
console.log("audio_tag: " + html);
// find wav from html
this.findWavFile(html)
await this.findTextFromWavFile(html, (number_list) => {
for (i; i < number_list.length; i++) {
this.page.evaluate((info) => {
let input_fileds = document.getElementsByClassName("audio-captcha-inputs");
input_fileds[info.index].focus();
input_fileds[info.index].value = info.number;
}, {'index': i, 'number': number_list[i]})
delay(getRandomWaitTime());
}
}
)
}
findWavFile(html) {
async findTextFromWavFile(html) {
let result = re.exec(html);
//create the directory if not exist
await fs.promises.mkdir(dest_dir, {recursive: true});
if (result) {
let audioUrl = result[0]
let fileName = crypto.randomUUID() + ".wav"
let audioUrl = result[0];
let fileName = crypto.randomUUID() + ".wav";
wget({url: audioUrl, dest: dest_dir + fileName}, function (error, response, body) {
if (error) {
console.log('--- error:');
console.log(error); // error encountered
} else {
console.log('download the file successfully');
// send request to get numbers
sendRequest(fileName);
}
})
}
}
}
module.exports = GeoCaptchaSolver
module.exports = GeoCaptchaSolver
// sendRequest("/Users/lpan/wavs/42f90d53-17c0-4a9b-a585-59eaa737a80d.wav").then(r => {
// console.log(r)
// })