Reset Crawler

import { CheerioCrawler, RequestQueue } from 'crawlee';
import type { RequestOptions } from 'crawlee';

const startRequests: RequestOptions[] = [
    {
        url: 'https://google.com',
    },
    {
        url: 'https://crawlee.dev',
    },
    {
        url: 'https://yahoo.com',
    },
    {
        url: 'https://apify.com',
        label: 'apify',
    },
];

const crawler = new CheerioCrawler({
    requestHandler: ({ log, request }) => {
        // if it's going to apify.com, throw an error
        if (request.label === 'apify') throw new Error('403');

        // otherwise, just log something
        log.info(`${request.url} handled`);
    },
    errorHandler: async ({ crawler, log }, error) => {
        // store the number of 403 errors hit in the crawler's state
        const state = await crawler.useState<{ count403: number }>({ count403: 0 });

        // if the error was a 403 error, increment the error count
        if (error.message.includes('403')) state.count403++;

        // if the total number of 403s hit is 2 (or greater),
        if (state.count403 >= 2) {
            log.warning('Restarting crawler.');
            // reset the count
            state.count403 = 0;

            // stop the crawler
            await crawler.teardown();
            // drop the existing request queue so requests can be handled again
            await crawler.requestQueue?.drop();

            // create a new request queue with a random name
            crawler.requestQueue = await RequestQueue.open(Math.round(Math.random() * 100).toString());

            // re-run the crawler
            await crawler.run(startRequests);
        }
    },
});

await crawler.run(startRequests);

import { CheerioCrawler, RequestQueue } from 'crawlee';
import type { RequestOptions } from 'crawlee';

const startRequests: RequestOptions[] = [
    {
        url: 'https://google.com',
    },
    {
        url: 'https://crawlee.dev',
    },
    {
        url: 'https://yahoo.com',
    },
    {
        url: 'https://apify.com',
        label: 'apify',
    },
];

const crawler = new CheerioCrawler({
    requestHandler: ({ log, request }) => {
        // if it's going to apify.com, throw an error
        if (request.label === 'apify') throw new Error('403');

        // otherwise, just log something
        log.info(`${request.url} handled`);
    },
    errorHandler: async ({ crawler, log }, error) => {
        // store the number of 403 errors hit in the crawler's state
        const state = await crawler.useState<{ count403: number }>({ count403: 0 });

        // if the error was a 403 error, increment the error count
        if (error.message.includes('403')) state.count403++;

        // if the total number of 403s hit is 2 (or greater),
        if (state.count403 >= 2) {
            log.warning('Restarting crawler.');
            // reset the count
            state.count403 = 0;

            // stop the crawler
            await crawler.teardown();
            // drop the existing request queue so requests can be handled again
            await crawler.requestQueue?.drop();

            // create a new request queue with a random name
            crawler.requestQueue = await RequestQueue.open(Math.round(Math.random() * 100).toString());

            // re-run the crawler
            await crawler.run(startRequests);
        }
    },
});

await crawler.run(startRequests);