Spaces:
Build error
Build error
fix: stop using pool
Browse files
backend/functions/package-lock.json
CHANGED
|
@@ -20,7 +20,6 @@
|
|
| 20 |
"express": "^4.19.2",
|
| 21 |
"firebase-admin": "^12.1.0",
|
| 22 |
"firebase-functions": "^4.9.0",
|
| 23 |
-
"generic-pool": "^3.9.0",
|
| 24 |
"htmlparser2": "^9.0.0",
|
| 25 |
"jose": "^5.1.0",
|
| 26 |
"jsdom": "^24.0.0",
|
|
@@ -5796,6 +5795,7 @@
|
|
| 5796 |
"version": "3.9.0",
|
| 5797 |
"resolved": "https://registry.npmjs.org/generic-pool/-/generic-pool-3.9.0.tgz",
|
| 5798 |
"integrity": "sha512-hymDOu5B53XvN4QT9dBmZxPX4CWhBPPLguTZ9MMFeFa/Kg0xWVfylOVNlJji/E7yTZWFd/q9GO5TxDLq156D7g==",
|
|
|
|
| 5799 |
"engines": {
|
| 5800 |
"node": ">= 4"
|
| 5801 |
}
|
|
|
|
| 20 |
"express": "^4.19.2",
|
| 21 |
"firebase-admin": "^12.1.0",
|
| 22 |
"firebase-functions": "^4.9.0",
|
|
|
|
| 23 |
"htmlparser2": "^9.0.0",
|
| 24 |
"jose": "^5.1.0",
|
| 25 |
"jsdom": "^24.0.0",
|
|
|
|
| 5795 |
"version": "3.9.0",
|
| 5796 |
"resolved": "https://registry.npmjs.org/generic-pool/-/generic-pool-3.9.0.tgz",
|
| 5797 |
"integrity": "sha512-hymDOu5B53XvN4QT9dBmZxPX4CWhBPPLguTZ9MMFeFa/Kg0xWVfylOVNlJji/E7yTZWFd/q9GO5TxDLq156D7g==",
|
| 5798 |
+
"devOptional": true,
|
| 5799 |
"engines": {
|
| 5800 |
"node": ">= 4"
|
| 5801 |
}
|
backend/functions/package.json
CHANGED
|
@@ -40,7 +40,6 @@
|
|
| 40 |
"express": "^4.19.2",
|
| 41 |
"firebase-admin": "^12.1.0",
|
| 42 |
"firebase-functions": "^4.9.0",
|
| 43 |
-
"generic-pool": "^3.9.0",
|
| 44 |
"htmlparser2": "^9.0.0",
|
| 45 |
"jose": "^5.1.0",
|
| 46 |
"jsdom": "^24.0.0",
|
|
|
|
| 40 |
"express": "^4.19.2",
|
| 41 |
"firebase-admin": "^12.1.0",
|
| 42 |
"firebase-functions": "^4.9.0",
|
|
|
|
| 43 |
"htmlparser2": "^9.0.0",
|
| 44 |
"jose": "^5.1.0",
|
| 45 |
"jsdom": "^24.0.0",
|
backend/functions/src/services/puppeteer.ts
CHANGED
|
@@ -1,7 +1,6 @@
|
|
| 1 |
import os from 'os';
|
| 2 |
import fs from 'fs';
|
| 3 |
import { container, singleton } from 'tsyringe';
|
| 4 |
-
import genericPool from 'generic-pool';
|
| 5 |
import { AsyncService, Defer, marshalErrorLike, AssertionFailureError, delay, maxConcurrency } from 'civkit';
|
| 6 |
import { Logger } from '../shared/services/logger';
|
| 7 |
import { JSDOM } from 'jsdom';
|
|
@@ -76,43 +75,32 @@ puppeteer.use(puppeteerPageProxy({
|
|
| 76 |
@singleton()
|
| 77 |
export class PuppeteerControl extends AsyncService {
|
| 78 |
|
|
|
|
| 79 |
browser!: Browser;
|
| 80 |
logger = this.globalLogger.child({ service: this.constructor.name });
|
| 81 |
|
| 82 |
-
pagePool = genericPool.createPool({
|
| 83 |
-
create: async () => {
|
| 84 |
-
const page = await this.newPage();
|
| 85 |
-
return page;
|
| 86 |
-
},
|
| 87 |
-
destroy: async (page) => {
|
| 88 |
-
await Promise.race([
|
| 89 |
-
(async () => {
|
| 90 |
-
const ctx = page.browserContext();
|
| 91 |
-
await page.close();
|
| 92 |
-
await ctx.close();
|
| 93 |
-
})(), delay(5000)
|
| 94 |
-
]).catch((err) => {
|
| 95 |
-
this.logger.error(`Failed to destroy page`, { err: marshalErrorLike(err) });
|
| 96 |
-
});
|
| 97 |
-
},
|
| 98 |
-
validate: async (page) => {
|
| 99 |
-
return page.browser().connected && !page.isClosed();
|
| 100 |
-
}
|
| 101 |
-
}, {
|
| 102 |
-
max: Math.max(1 + Math.floor(os.totalmem() / (256 * 1024 * 1024)), 16),
|
| 103 |
-
min: 1,
|
| 104 |
-
acquireTimeoutMillis: 60_000,
|
| 105 |
-
testOnBorrow: true,
|
| 106 |
-
testOnReturn: true,
|
| 107 |
-
autostart: false,
|
| 108 |
-
priorityRange: 3
|
| 109 |
-
});
|
| 110 |
-
|
| 111 |
private __healthCheckInterval?: NodeJS.Timeout;
|
| 112 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 113 |
constructor(protected globalLogger: Logger) {
|
| 114 |
super(...arguments);
|
| 115 |
-
this.setMaxListeners(2 *
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 116 |
}
|
| 117 |
|
| 118 |
override async init() {
|
|
@@ -121,8 +109,6 @@ export class PuppeteerControl extends AsyncService {
|
|
| 121 |
this.__healthCheckInterval = undefined;
|
| 122 |
}
|
| 123 |
await this.dependencyReady();
|
| 124 |
-
this.logger.info(`PuppeteerControl initializing with pool size ${this.pagePool.max}`, { poolSize: this.pagePool.max });
|
| 125 |
-
this.pagePool.start();
|
| 126 |
|
| 127 |
if (this.browser) {
|
| 128 |
if (this.browser.connected) {
|
|
@@ -151,24 +137,33 @@ export class PuppeteerControl extends AsyncService {
|
|
| 151 |
this.emit('ready');
|
| 152 |
|
| 153 |
this.__healthCheckInterval = setInterval(() => this.healthCheck(), 30_000);
|
|
|
|
| 154 |
}
|
| 155 |
|
| 156 |
@maxConcurrency(1)
|
| 157 |
async healthCheck() {
|
| 158 |
-
this.
|
| 159 |
-
|
|
|
|
|
|
|
|
|
|
| 160 |
this.logger.warn(`Health check failed`, { err: marshalErrorLike(err) });
|
| 161 |
return null;
|
| 162 |
});
|
| 163 |
-
this.pagePool.max -= 1;
|
| 164 |
|
| 165 |
if (healthyPage) {
|
| 166 |
-
this.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 167 |
return;
|
| 168 |
}
|
| 169 |
|
| 170 |
this.logger.warn(`Trying to clean up...`);
|
| 171 |
-
await this.pagePool.clear();
|
| 172 |
this.browser.process()?.kill('SIGKILL');
|
| 173 |
Reflect.deleteProperty(this, 'browser');
|
| 174 |
this.emit('crippled');
|
|
@@ -178,7 +173,7 @@ export class PuppeteerControl extends AsyncService {
|
|
| 178 |
async newPage() {
|
| 179 |
await this.serviceReady();
|
| 180 |
const dedicatedContext = await this.browser.createBrowserContext();
|
| 181 |
-
|
| 182 |
const page = await dedicatedContext.newPage();
|
| 183 |
const preparations = [];
|
| 184 |
|
|
@@ -300,18 +295,72 @@ document.addEventListener('readystatechange', handlePageLoad);
|
|
| 300 |
document.addEventListener('load', handlePageLoad);
|
| 301 |
`);
|
| 302 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 303 |
return page;
|
| 304 |
}
|
| 305 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 306 |
async *scrap(parsedUrl: URL, options?: ScrappingOptions): AsyncGenerator<PageSnapshot | undefined> {
|
| 307 |
// parsedUrl.search = '';
|
| 308 |
const url = parsedUrl.toString();
|
| 309 |
|
| 310 |
-
this.logger.info(`Scraping ${url}`, { url });
|
| 311 |
let snapshot: PageSnapshot | undefined;
|
| 312 |
let screenshot: Buffer | undefined;
|
| 313 |
-
|
| 314 |
-
const
|
|
|
|
| 315 |
if (options?.proxyUrl) {
|
| 316 |
await page.useProxy(options.proxyUrl);
|
| 317 |
}
|
|
@@ -342,7 +391,7 @@ document.addEventListener('load', handlePageLoad);
|
|
| 342 |
|
| 343 |
const gotoPromise = page.goto(url, { waitUntil: ['load', 'domcontentloaded', 'networkidle0'], timeout: 30_000 })
|
| 344 |
.catch((err) => {
|
| 345 |
-
this.logger.warn(`Browsing of ${url} did not fully succeed`, { err: marshalErrorLike(err) });
|
| 346 |
return Promise.reject(new AssertionFailureError({
|
| 347 |
message: `Failed to goto ${url}: ${err}`,
|
| 348 |
cause: err,
|
|
@@ -362,7 +411,7 @@ document.addEventListener('load', handlePageLoad);
|
|
| 362 |
}
|
| 363 |
}
|
| 364 |
finalized = true;
|
| 365 |
-
this.logger.info(`Snapshot of ${url} done`, { url, title: snapshot?.title, href: snapshot?.href });
|
| 366 |
this.emit(
|
| 367 |
'crawled',
|
| 368 |
{ ...snapshot, screenshot },
|
|
@@ -378,7 +427,7 @@ document.addEventListener('load', handlePageLoad);
|
|
| 378 |
nextSnapshotDeferred.resolve(snapshot);
|
| 379 |
})
|
| 380 |
.catch((err) => {
|
| 381 |
-
this.logger.warn(`Failed to wait for selector ${options.waitForSelector}`, { err: marshalErrorLike(err) });
|
| 382 |
});
|
| 383 |
}
|
| 384 |
|
|
@@ -401,9 +450,7 @@ document.addEventListener('load', handlePageLoad);
|
|
| 401 |
} finally {
|
| 402 |
gotoPromise.finally(() => {
|
| 403 |
page.off('snapshot', hdl);
|
| 404 |
-
this.
|
| 405 |
-
this.logger.warn(`Failed to destroy page`, { err: marshalErrorLike(err) });
|
| 406 |
-
});
|
| 407 |
});
|
| 408 |
nextSnapshotDeferred.resolve();
|
| 409 |
}
|
|
|
|
| 1 |
import os from 'os';
|
| 2 |
import fs from 'fs';
|
| 3 |
import { container, singleton } from 'tsyringe';
|
|
|
|
| 4 |
import { AsyncService, Defer, marshalErrorLike, AssertionFailureError, delay, maxConcurrency } from 'civkit';
|
| 5 |
import { Logger } from '../shared/services/logger';
|
| 6 |
import { JSDOM } from 'jsdom';
|
|
|
|
| 75 |
@singleton()
|
| 76 |
export class PuppeteerControl extends AsyncService {
|
| 77 |
|
| 78 |
+
_sn = 0;
|
| 79 |
browser!: Browser;
|
| 80 |
logger = this.globalLogger.child({ service: this.constructor.name });
|
| 81 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 82 |
private __healthCheckInterval?: NodeJS.Timeout;
|
| 83 |
|
| 84 |
+
__loadedPage: Page[] = [];
|
| 85 |
+
|
| 86 |
+
finalizerMap = new WeakMap<Page, ReturnType<typeof setTimeout>>();
|
| 87 |
+
snMap = new WeakMap<Page, number>();
|
| 88 |
+
livePages = new Set<Page>();
|
| 89 |
+
lastPageCratedAt: number = 0;
|
| 90 |
+
|
| 91 |
constructor(protected globalLogger: Logger) {
|
| 92 |
super(...arguments);
|
| 93 |
+
this.setMaxListeners(2 * Math.floor(os.totalmem() / (256 * 1024 * 1024)) + 1); 148 - 95;
|
| 94 |
+
|
| 95 |
+
this.on('crippled', () => {
|
| 96 |
+
this.__loadedPage.length = 0;
|
| 97 |
+
this.livePages.clear();
|
| 98 |
+
});
|
| 99 |
+
}
|
| 100 |
+
|
| 101 |
+
briefPages() {
|
| 102 |
+
this.logger.info(`Status: ${this.livePages.size} pages alive: ${Array.from(this.livePages).map((x) => this.snMap.get(x)).sort().join(', ')}; ${this.__loadedPage.length} idle pages: ${this.__loadedPage.map((x) => this.snMap.get(x)).sort().join(', ')}`);
|
| 103 |
+
this.logger.info(``);
|
| 104 |
}
|
| 105 |
|
| 106 |
override async init() {
|
|
|
|
| 109 |
this.__healthCheckInterval = undefined;
|
| 110 |
}
|
| 111 |
await this.dependencyReady();
|
|
|
|
|
|
|
| 112 |
|
| 113 |
if (this.browser) {
|
| 114 |
if (this.browser.connected) {
|
|
|
|
| 137 |
this.emit('ready');
|
| 138 |
|
| 139 |
this.__healthCheckInterval = setInterval(() => this.healthCheck(), 30_000);
|
| 140 |
+
this.newPage().then((r) => this.__loadedPage.push(r));
|
| 141 |
}
|
| 142 |
|
| 143 |
@maxConcurrency(1)
|
| 144 |
async healthCheck() {
|
| 145 |
+
if (Date.now() - this.lastPageCratedAt <= 10_000) {
|
| 146 |
+
this.briefPages();
|
| 147 |
+
return;
|
| 148 |
+
}
|
| 149 |
+
const healthyPage = await this.newPage().catch((err) => {
|
| 150 |
this.logger.warn(`Health check failed`, { err: marshalErrorLike(err) });
|
| 151 |
return null;
|
| 152 |
});
|
|
|
|
| 153 |
|
| 154 |
if (healthyPage) {
|
| 155 |
+
this.__loadedPage.push(healthyPage);
|
| 156 |
+
|
| 157 |
+
if (this.__loadedPage.length > 3) {
|
| 158 |
+
this.ditchPage(this.__loadedPage.shift()!);
|
| 159 |
+
}
|
| 160 |
+
|
| 161 |
+
this.briefPages();
|
| 162 |
+
|
| 163 |
return;
|
| 164 |
}
|
| 165 |
|
| 166 |
this.logger.warn(`Trying to clean up...`);
|
|
|
|
| 167 |
this.browser.process()?.kill('SIGKILL');
|
| 168 |
Reflect.deleteProperty(this, 'browser');
|
| 169 |
this.emit('crippled');
|
|
|
|
| 173 |
async newPage() {
|
| 174 |
await this.serviceReady();
|
| 175 |
const dedicatedContext = await this.browser.createBrowserContext();
|
| 176 |
+
const sn = this._sn++;
|
| 177 |
const page = await dedicatedContext.newPage();
|
| 178 |
const preparations = [];
|
| 179 |
|
|
|
|
| 295 |
document.addEventListener('load', handlePageLoad);
|
| 296 |
`);
|
| 297 |
|
| 298 |
+
this.snMap.set(page, sn);
|
| 299 |
+
this.logger.warn(`Page ${sn} created.`);
|
| 300 |
+
this.lastPageCratedAt = Date.now();
|
| 301 |
+
this.livePages.add(page);
|
| 302 |
+
|
| 303 |
return page;
|
| 304 |
}
|
| 305 |
|
| 306 |
+
async getNextPage() {
|
| 307 |
+
let thePage;
|
| 308 |
+
if (this.__loadedPage.length) {
|
| 309 |
+
thePage = this.__loadedPage.shift();
|
| 310 |
+
if (this.__loadedPage.length <= 1) {
|
| 311 |
+
this.newPage()
|
| 312 |
+
.then((r) => this.__loadedPage.push(r))
|
| 313 |
+
.catch((err) => {
|
| 314 |
+
this.logger.warn(`Failed to load new page ahead of time`, { err: marshalErrorLike(err) });
|
| 315 |
+
});
|
| 316 |
+
}
|
| 317 |
+
}
|
| 318 |
+
|
| 319 |
+
if (!thePage) {
|
| 320 |
+
thePage = await this.newPage();
|
| 321 |
+
}
|
| 322 |
+
|
| 323 |
+
const timer = setTimeout(() => {
|
| 324 |
+
this.logger.warn(`Page is not allowed to live past 5 minutes, ditching page ${this.snMap.get(thePage)}...`);
|
| 325 |
+
this.ditchPage(thePage);
|
| 326 |
+
}, 300 * 1000);
|
| 327 |
+
|
| 328 |
+
this.finalizerMap.set(thePage, timer);
|
| 329 |
+
|
| 330 |
+
return thePage;
|
| 331 |
+
}
|
| 332 |
+
|
| 333 |
+
async ditchPage(page: Page) {
|
| 334 |
+
if (this.finalizerMap.has(page)) {
|
| 335 |
+
clearTimeout(this.finalizerMap.get(page)!);
|
| 336 |
+
this.finalizerMap.delete(page);
|
| 337 |
+
}
|
| 338 |
+
if (page.isClosed()) {
|
| 339 |
+
return;
|
| 340 |
+
}
|
| 341 |
+
const sn = this.snMap.get(page);
|
| 342 |
+
this.logger.info(`Closing page ${sn}`);
|
| 343 |
+
this.livePages.delete(page);
|
| 344 |
+
await Promise.race([
|
| 345 |
+
(async () => {
|
| 346 |
+
const ctx = page.browserContext();
|
| 347 |
+
await page.close();
|
| 348 |
+
await ctx.close();
|
| 349 |
+
})(), delay(5000)
|
| 350 |
+
]).catch((err) => {
|
| 351 |
+
this.logger.error(`Failed to destroy page ${sn}`, { err: marshalErrorLike(err) });
|
| 352 |
+
});
|
| 353 |
+
}
|
| 354 |
+
|
| 355 |
async *scrap(parsedUrl: URL, options?: ScrappingOptions): AsyncGenerator<PageSnapshot | undefined> {
|
| 356 |
// parsedUrl.search = '';
|
| 357 |
const url = parsedUrl.toString();
|
| 358 |
|
|
|
|
| 359 |
let snapshot: PageSnapshot | undefined;
|
| 360 |
let screenshot: Buffer | undefined;
|
| 361 |
+
const page = await this.getNextPage();
|
| 362 |
+
const sn = this.snMap.get(page);
|
| 363 |
+
this.logger.info(`Page ${sn}: Scraping ${url}`, { url });
|
| 364 |
if (options?.proxyUrl) {
|
| 365 |
await page.useProxy(options.proxyUrl);
|
| 366 |
}
|
|
|
|
| 391 |
|
| 392 |
const gotoPromise = page.goto(url, { waitUntil: ['load', 'domcontentloaded', 'networkidle0'], timeout: 30_000 })
|
| 393 |
.catch((err) => {
|
| 394 |
+
this.logger.warn(`Page ${sn}: Browsing of ${url} did not fully succeed`, { err: marshalErrorLike(err) });
|
| 395 |
return Promise.reject(new AssertionFailureError({
|
| 396 |
message: `Failed to goto ${url}: ${err}`,
|
| 397 |
cause: err,
|
|
|
|
| 411 |
}
|
| 412 |
}
|
| 413 |
finalized = true;
|
| 414 |
+
this.logger.info(`Page ${sn}: Snapshot of ${url} done`, { url, title: snapshot?.title, href: snapshot?.href });
|
| 415 |
this.emit(
|
| 416 |
'crawled',
|
| 417 |
{ ...snapshot, screenshot },
|
|
|
|
| 427 |
nextSnapshotDeferred.resolve(snapshot);
|
| 428 |
})
|
| 429 |
.catch((err) => {
|
| 430 |
+
this.logger.warn(`Page ${sn}: Failed to wait for selector ${options.waitForSelector}`, { err: marshalErrorLike(err) });
|
| 431 |
});
|
| 432 |
}
|
| 433 |
|
|
|
|
| 450 |
} finally {
|
| 451 |
gotoPromise.finally(() => {
|
| 452 |
page.off('snapshot', hdl);
|
| 453 |
+
this.ditchPage(page);
|
|
|
|
|
|
|
| 454 |
});
|
| 455 |
nextSnapshotDeferred.resolve();
|
| 456 |
}
|