1+ import ScrapersFactory from '../ScrapersFactory' ;
2+ import Https from '../ScrapersFactory/products/Https' ;
3+ import https from 'https' ;
4+ import Puppetter from '../ScrapersFactory/products/Puppetter' ;
5+ import { ScraperType } from '../configs/types' ;
6+
7+ jest . mock ( '../db/MongoDBConnection' , ( ) => ( {
8+ getInstance : jest . fn ( ) . mockResolvedValue ( {
9+ getDb : jest . fn ( ) . mockReturnValue ( {
10+ collection : jest . fn ( ) . mockReturnValue ( {
11+ insertOne : jest . fn ( ) . mockResolvedValue ( { } ) ,
12+ } ) ,
13+ } ) ,
14+ } ) ,
15+ } ) ) ;
16+
17+ jest . mock ( '../db/MongoCRUD' , ( ) => {
18+ return jest . fn ( ) . mockImplementation ( ( ) => ( {
19+ create : jest . fn ( ) . mockResolvedValue ( { } ) ,
20+ } ) ) ;
21+ } ) ;
22+
23+ describe ( 'ScrapersFactory' , ( ) => {
24+ it ( 'should create an instance of Https scraper' , ( ) => {
25+ const scraper = ScrapersFactory . createScraper ( ScraperType . HTTPS ) ;
26+ expect ( scraper ) . toBeInstanceOf ( Https ) ;
27+ } ) ;
28+
29+ it ( 'should create an instance of Puppetter scraper' , ( ) => {
30+ const scraper = ScrapersFactory . createScraper ( ScraperType . PUPPETTER ) ;
31+ expect ( scraper ) . toBeInstanceOf ( Puppetter ) ;
32+ } ) ;
33+
34+ it ( 'should throw an error for unsupported scraper types' , ( ) => {
35+ expect ( ( ) => ScrapersFactory . createScraper ( 'UNSUPPORTED_TYPE' as ScraperType ) )
36+ . toThrow ( 'Unsupported scrape type: UNSUPPORTED_TYPE' ) ;
37+ } ) ;
38+ } ) ;
39+
40+ describe ( 'Https Scraper' , ( ) => {
41+ let httpsScraper : Https ;
42+
43+ beforeAll ( ( ) => {
44+ httpsScraper = new Https ( ) ;
45+ } ) ;
46+
47+ afterAll ( ( ) => {
48+ // Clean up any open handles, especially https requests
49+ https . globalAgent . destroy ( ) ;
50+ } ) ;
51+
52+ it ( 'should get HTML content from a given URL' , async ( ) => {
53+ const mockHost = 'example.com' ;
54+ const mockPath = '/' ;
55+ const htmlContent = await httpsScraper . get ( mockHost , mockPath ) ;
56+ expect ( htmlContent ) . toContain ( '<html>' ) ;
57+ } ) ;
58+
59+ it ( 'should parse HTML content correctly' , ( ) => {
60+ const mockHtml = '<html><body><h1>Test</h1></body></html>' ;
61+ const rootElement = httpsScraper . parse ( mockHtml ) ;
62+ expect ( rootElement . querySelector ( 'h1' ) ?. textContent ) . toBe ( 'Test' ) ;
63+ } ) ;
64+
65+ it ( 'should save data correctly in MongoDB' , async ( ) => {
66+ const mockData = { key : 'value' } ;
67+ const spySave = jest . spyOn ( httpsScraper , 'save' ) . mockImplementation ( async ( ) => { } ) ;
68+
69+ await httpsScraper . save ( mockData ) ;
70+ expect ( spySave ) . toHaveBeenCalledWith ( mockData ) ;
71+ } ) ;
72+ } ) ;
73+
74+ describe ( 'Puppetter Scraper' , ( ) => {
75+ let puppetterScraper : Puppetter ;
76+
77+ beforeAll ( ( ) => {
78+ puppetterScraper = new Puppetter ( ) ;
79+ } ) ;
80+
81+ it ( 'should get HTML content from a given URL using Puppeteer' , async ( ) => {
82+ const mockHost = 'https://example.com' ;
83+ const mockPath = '/' ;
84+ const htmlContent = await puppetterScraper . get ( mockHost , mockPath ) ;
85+ expect ( htmlContent ) . toContain ( '<html>' ) ;
86+ } ) ;
87+
88+ it ( 'should parse HTML content correctly' , ( ) => {
89+ const mockHtml = '<html><body><h1>Test</h1></body></html>' ;
90+ const rootElement = puppetterScraper . parse ( mockHtml ) ;
91+ expect ( rootElement . querySelector ( 'h1' ) ?. textContent ) . toBe ( 'Test' ) ;
92+ } ) ;
93+
94+ it ( 'should save data correctly in MongoDB' , async ( ) => {
95+ const mockData = { key : 'value' } ;
96+ const spySave = jest . spyOn ( puppetterScraper , 'save' ) . mockImplementation ( async ( ) => { } ) ;
97+
98+ await puppetterScraper . save ( mockData ) ;
99+ expect ( spySave ) . toHaveBeenCalledWith ( mockData ) ;
100+ } ) ;
101+ } ) ;
0 commit comments