|
|
@ -225,12 +225,27 @@ class Site {
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
timeout() {
|
|
|
|
/**
|
|
|
|
return new Promise((resolve, reject) =>
|
|
|
|
* @param {Promise} promise
|
|
|
|
setTimeout(() => {
|
|
|
|
* @param {string} [errorMessage]
|
|
|
|
reject(new Error('The website took too long to respond'))
|
|
|
|
*
|
|
|
|
}, this.options.maxWait)
|
|
|
|
* @return {Promise}
|
|
|
|
)
|
|
|
|
*/
|
|
|
|
|
|
|
|
async promiseTimeout(promise, errorMessage = 'The website took too long to respond') {
|
|
|
|
|
|
|
|
let timerId = null;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
return Promise.race([
|
|
|
|
|
|
|
|
new Promise((resolve, reject) => {
|
|
|
|
|
|
|
|
timerId = setTimeout(() => {
|
|
|
|
|
|
|
|
clearTimeout(timerId);
|
|
|
|
|
|
|
|
reject(new Error(errorMessage));
|
|
|
|
|
|
|
|
}, this.options.maxWait);
|
|
|
|
|
|
|
|
}),
|
|
|
|
|
|
|
|
promise.then((value) => {
|
|
|
|
|
|
|
|
clearTimeout(timerId);
|
|
|
|
|
|
|
|
return value;
|
|
|
|
|
|
|
|
}),
|
|
|
|
|
|
|
|
]);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
async goto(url) {
|
|
|
|
async goto(url) {
|
|
|
@ -337,17 +352,12 @@ class Site {
|
|
|
|
)
|
|
|
|
)
|
|
|
|
|
|
|
|
|
|
|
|
try {
|
|
|
|
try {
|
|
|
|
await Promise.race([
|
|
|
|
await this.promiseTimeout(page.goto(url.href, { waitUntil: 'domcontentloaded' }))
|
|
|
|
this.timeout(),
|
|
|
|
|
|
|
|
page.goto(url.href, { waitUntil: 'domcontentloaded' }),
|
|
|
|
|
|
|
|
])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
await sleep(1000)
|
|
|
|
await sleep(1000)
|
|
|
|
|
|
|
|
|
|
|
|
// Links
|
|
|
|
// Links
|
|
|
|
const links = await (
|
|
|
|
const links = await (
|
|
|
|
await Promise.race([
|
|
|
|
await this.promiseTimeout(
|
|
|
|
this.timeout(),
|
|
|
|
|
|
|
|
page.evaluateHandle(() =>
|
|
|
|
page.evaluateHandle(() =>
|
|
|
|
Array.from(document.getElementsByTagName('a')).map(
|
|
|
|
Array.from(document.getElementsByTagName('a')).map(
|
|
|
|
({ hash, hostname, href, pathname, protocol, rel }) => ({
|
|
|
|
({ hash, hostname, href, pathname, protocol, rel }) => ({
|
|
|
@ -360,13 +370,12 @@ class Site {
|
|
|
|
})
|
|
|
|
})
|
|
|
|
)
|
|
|
|
)
|
|
|
|
),
|
|
|
|
),
|
|
|
|
])
|
|
|
|
)
|
|
|
|
).jsonValue()
|
|
|
|
).jsonValue()
|
|
|
|
|
|
|
|
|
|
|
|
// CSS
|
|
|
|
// CSS
|
|
|
|
const css = await (
|
|
|
|
const css = await (
|
|
|
|
await Promise.race([
|
|
|
|
await this.promiseTimeout(
|
|
|
|
this.timeout(),
|
|
|
|
|
|
|
|
page.evaluateHandle((maxRows) => {
|
|
|
|
page.evaluateHandle((maxRows) => {
|
|
|
|
const css = []
|
|
|
|
const css = []
|
|
|
|
|
|
|
|
|
|
|
@ -390,25 +399,23 @@ class Site {
|
|
|
|
|
|
|
|
|
|
|
|
return css.join('\n')
|
|
|
|
return css.join('\n')
|
|
|
|
}, this.options.htmlMaxRows),
|
|
|
|
}, this.options.htmlMaxRows),
|
|
|
|
])
|
|
|
|
)
|
|
|
|
).jsonValue()
|
|
|
|
).jsonValue()
|
|
|
|
|
|
|
|
|
|
|
|
// Script tags
|
|
|
|
// Script tags
|
|
|
|
const scripts = await (
|
|
|
|
const scripts = await (
|
|
|
|
await Promise.race([
|
|
|
|
await this.promiseTimeout(
|
|
|
|
this.timeout(),
|
|
|
|
|
|
|
|
page.evaluateHandle(() =>
|
|
|
|
page.evaluateHandle(() =>
|
|
|
|
Array.from(document.getElementsByTagName('script'))
|
|
|
|
Array.from(document.getElementsByTagName('script'))
|
|
|
|
.map(({ src }) => src)
|
|
|
|
.map(({ src }) => src)
|
|
|
|
.filter((src) => src)
|
|
|
|
.filter((src) => src)
|
|
|
|
),
|
|
|
|
),
|
|
|
|
])
|
|
|
|
)
|
|
|
|
).jsonValue()
|
|
|
|
).jsonValue()
|
|
|
|
|
|
|
|
|
|
|
|
// Meta tags
|
|
|
|
// Meta tags
|
|
|
|
const meta = await (
|
|
|
|
const meta = await (
|
|
|
|
await Promise.race([
|
|
|
|
await this.promiseTimeout(
|
|
|
|
this.timeout(),
|
|
|
|
|
|
|
|
page.evaluateHandle(() =>
|
|
|
|
page.evaluateHandle(() =>
|
|
|
|
Array.from(document.querySelectorAll('meta')).reduce(
|
|
|
|
Array.from(document.querySelectorAll('meta')).reduce(
|
|
|
|
(metas, meta) => {
|
|
|
|
(metas, meta) => {
|
|
|
@ -424,12 +431,11 @@ class Site {
|
|
|
|
{}
|
|
|
|
{}
|
|
|
|
)
|
|
|
|
)
|
|
|
|
),
|
|
|
|
),
|
|
|
|
])
|
|
|
|
)
|
|
|
|
).jsonValue()
|
|
|
|
).jsonValue()
|
|
|
|
|
|
|
|
|
|
|
|
// JavaScript
|
|
|
|
// JavaScript
|
|
|
|
const js = await Promise.race([
|
|
|
|
const js = await this.promiseTimeout(
|
|
|
|
this.timeout(),
|
|
|
|
|
|
|
|
page.evaluate(
|
|
|
|
page.evaluate(
|
|
|
|
(technologies) => {
|
|
|
|
(technologies) => {
|
|
|
|
return technologies.reduce((technologies, { name, chains }) => {
|
|
|
|
return technologies.reduce((technologies, { name, chains }) => {
|
|
|
@ -462,7 +468,7 @@ class Site {
|
|
|
|
.filter(({ js }) => Object.keys(js).length)
|
|
|
|
.filter(({ js }) => Object.keys(js).length)
|
|
|
|
.map(({ name, js }) => ({ name, chains: Object.keys(js) }))
|
|
|
|
.map(({ name, js }) => ({ name, chains: Object.keys(js) }))
|
|
|
|
),
|
|
|
|
),
|
|
|
|
])
|
|
|
|
)
|
|
|
|
|
|
|
|
|
|
|
|
// Cookies
|
|
|
|
// Cookies
|
|
|
|
const cookies = (await page.cookies()).reduce(
|
|
|
|
const cookies = (await page.cookies()).reduce(
|
|
|
|