102 lines
3.9 KiB
JavaScript
102 lines
3.9 KiB
JavaScript
const logger = require("logops")
|
|
, config = require('../../config')
|
|
, { By, until } = require('selenium-webdriver')
|
|
, url = require('url')
|
|
, poolingService = require('../service/PoolingService')
|
|
, fetch = require('node-fetch')
|
|
|
|
|
|
class SeleniumBrowser {
|
|
constructor() {
|
|
this.sleep = m => new Promise(r => setTimeout(r, m))
|
|
this.pool = poolingService.pool
|
|
}
|
|
|
|
async request(params) {
|
|
let uri = params.uri
|
|
logger.info('URL %s', uri)
|
|
if (uri.indexOf('://') > 3) {
|
|
return this.pool.acquire()
|
|
.then(async driver => {
|
|
let html = await driver.get(uri)
|
|
.then(() => driver.findElement(By.css('body')))
|
|
.then(body => body.getAttribute('innerHTML'))
|
|
let htmlPromise = this.isCaptchaPage(html) ? this.resolveCaptcha(driver) : this.getHtml(driver, html, params)
|
|
html = await htmlPromise
|
|
this.pool.destroy(driver)
|
|
return html
|
|
})
|
|
} else {
|
|
throw new Error('Invalid URI - ' + uri)
|
|
}
|
|
}
|
|
|
|
isCaptchaPage(html) {
|
|
return html.indexOf('grecaptcha') > -1 || html.indexOf('www.google.com/recaptcha/api2/anchor') > -1
|
|
}
|
|
|
|
getCaptchaRequestUri(id) {
|
|
return 'https://2captcha.com/res.php?key=62fc5ef52fd4befe266d224796a7ea6f&json=1&action=get&id=' + id
|
|
}
|
|
|
|
getHtml(driver, html, params) {
|
|
if (params.waitElements) {
|
|
return driver.get(params.uri)
|
|
.then(() => Promise.all(params.waitElements.map(s => driver.wait(until.elementLocated(By.css(s)), 10000))))
|
|
.then(() => driver.findElement(By.css('body')))
|
|
.then(body => body.getAttribute('innerHTML'))
|
|
.then(html => {
|
|
return html
|
|
})
|
|
} else {
|
|
return html
|
|
}
|
|
}
|
|
|
|
resolveCaptcha(driver) {
|
|
return driver.findElement(By.css('#grecaptcha iframe')).getAttribute('src')
|
|
.then(async src => {
|
|
return {
|
|
"key": "62fc5ef52fd4befe266d224796a7ea6f",
|
|
"method": "userrecaptcha",
|
|
"googlekey": url.parse(src, true).query.k,
|
|
"pageurl": await driver.getCurrentUrl(),
|
|
"json": 1
|
|
}
|
|
})
|
|
.then(params => {
|
|
let requestUrl = new url.URL('https://2captcha.com/in.php')
|
|
requestUrl.search = new url.URLSearchParams(params).toString();
|
|
return fetch(requestUrl).then(res => res.json()).then(res => {
|
|
logger.debug(res, '2captcha IN-request response')
|
|
if (!res.hasOwnProperty('request')) {
|
|
throw new Error('2captcha IN-request response has no data request id : ' + JSON.stringify(res))
|
|
}
|
|
return res.request
|
|
})
|
|
})
|
|
.then(async id => {
|
|
logger.info('Request 2captcha solving result after 20 senconds')
|
|
await this.sleep(1000 * 20)
|
|
let res = await fetch(this.getCaptchaRequestUri(id)).then(res => res.json())
|
|
let maxRetry = 10
|
|
while (res.request == 'CAPCHA_NOT_READY') {
|
|
await this.sleep(1000 * 5)
|
|
maxRetry = maxRetry - 1
|
|
logger.info('2captcha still working so request solving result after 5 senconds')
|
|
res = await fetch(this.getCaptchaRequestUri(id)).then(res => res.json())
|
|
if (maxRetry == 0) break
|
|
}
|
|
return res
|
|
})
|
|
.then(res => {
|
|
return driver.executeScript('captchaCallback("' + res.request + '")').then(() => this.sleep(1000 * 5))
|
|
})
|
|
}
|
|
|
|
close() {
|
|
return this.pool.drain()
|
|
}
|
|
}
|
|
|
|
module.exports = new SeleniumBrowser |