Doctolib

1import { ApifyClient } from 'apify-client';
2
3// Initialize the ApifyClient with your Apify 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://www.doctolib.fr/infectiologue/75001-paris"
13        }
14    ],
15    "pageFunction": async function pageFunction(context) {
16        let data = {}
17        let userData = context.request.userData
18        data.url = context.request.url
19        data.label = userData.label
20        // product here is a reference to a doctor page. Naming is mislieading, my apologies. 
21        // it's here because usually I work with marketplaces.
22        if(userData && userData.label === 'product'){   
23            context.log.info('label product.');     
24            // data.img = await context.page.locator('[data-qa-id=adview_spotlight_container] img >> nth=0').getAttribute('src')
25            data.nom = await context.page.locator('#main-content h1').innerText({timeout:6000})
26            try{
27                data.tarif = await context.page.locator('#payment_means').first().innerText({timeout:3000})
28                data.horaire_contact = await context.page.locator('#openings_and_contact').first().innerText({timeout:3000})
29                data.description = await context.page.locator('.dl-profile-bio').first().innerText({timeout:3000})
30                data.specialite = await context.page.locator('.dl-profile-header-speciality').first().innerText({timeout:3000})
31                data.expertise = await context.page.locator('#skills').first().innerText({timeout:3000})        
32                data.phones = await context.getPhones(data.horaire_contact)
33            }catch(e){
34                context.log.info('not found',e);     
35            }        
36            
37        }else{
38            context.log.info('we are not on a doctor page, so a search or pagination page.');
39            // we are looking for "doctors" (called "product" here) to be queued, let's write it down
40            userData.label = 'product';
41            const elements = context.page.locator('.search-result-card a[href]');
42            const links = await elements.evaluateAll(elems => elems.map(elem => elem.getAttribute('href')));
43            links.forEach(async link => {
44                if(link.startsWith('/')){ link = 'https://www.doctolib.fr' + link }
45                await context.enqueueRequest(link, userData , false);
46            })
47        }
48        context.log.info(`function ended`);
49        delete data.label
50        return data;
51    },
52    "pseudoUrls": [
53        {
54            "purl": "https://www.doctolib.fr/[.*]"
55        }
56    ]
57};
58
59(async () => {
60    // Run the Actor and wait for it to finish
61    const run = await client.actor("anchor/doctolib").call(input);
62
63    // Fetch and print Actor results from the run's dataset (if any)
64    console.log('Results from dataset');
65    console.log(`💾 Check your data here: https://console.apify.com/storage/datasets/${run.defaultDatasetId}`);
66    const { items } = await client.dataset(run.defaultDatasetId).listItems();
67    items.forEach((item) => {
68        console.dir(item);
69    });
70})();
71
72// 📚 Want to learn more 📖? Go to → https://docs.apify.com/api/client/js/docs