Web Scraper

1import { ApifyClient } from 'apify-client';
2
3// Initialize the ApifyClient with API token
4const client = new ApifyClient({
5    token: '<YOUR_API_TOKEN>',
6});
7
8// Prepare Actor input
9const input = {
10    "startUrls": [
11        {
12            "url": "https://apify.com"
13        }
14    ],
15    "pseudoUrls": [
16        {
17            "purl": "https://apify.com[(/[\\w-]+)?]"
18        }
19    ],
20    "linkSelector": "a",
21    "pageFunction": async function pageFunction(context) {
22        // See README for context properties. If the syntax is unfamiliar see the link
23        // https://javascript.info/destructuring-assignment#object-destructuring
24        const { request, log, jQuery } = context;
25    
26        // To be able to use jQuery as $, one needs save it into a variable
27        // and select the inject jQuery option. We've selected it for you.
28        const $ = jQuery;
29        const title = $('title').text();
30    
31        // This is yet another new feature of Javascript called template strings.
32        // https://javascript.info/string#quotes
33        log.info(`URL: ${request.url} TITLE: ${title}`);
34    
35        // To save data just return an object with the requested properties.
36        return {
37            url: request.url,
38            title
39        };
40    },
41    "proxyConfiguration": {
42        "useApifyProxy": false
43    },
44    "initialCookies": [],
45    "waitUntil": [
46        "networkidle2"
47    ],
48    "customData": {}
49};
50
51(async () => {
52    // Run the Actor and wait for it to finish
53    const run = await client.actor("apify/web-scraper").call(input);
54
55    // Fetch and print Actor results from the run's dataset (if any)
56    console.log('Results from dataset');
57    const { items } = await client.dataset(run.defaultDatasetId).listItems();
58    items.forEach((item) => {
59        console.dir(item);
60    });
61})();