se-scraper/test/test_googleimage.js

80 lines
2.9 KiB
JavaScript
Raw Normal View History

2019-08-02 17:41:29 +02:00
'use strict';
const se_scraper = require('./../index.js');
2019-08-02 17:41:29 +02:00
const assert = require('chai').assert;
/*
* Use chai and mocha for tests.
* https://mochajs.org/#installation
*/
const normal_search_keywords = ['apple', 'rain'];
async function normal_image_search_test() {
let config = {
compress: false,
2019-06-12 18:14:49 +02:00
debug_level: 0,
headless: true,
block_assets: true,
user_agent: 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.110 Safari/537.36',
random_user_agent: false,
};
2019-06-12 18:14:49 +02:00
let scrape_config = {
search_engine: 'google_image',
keywords: normal_search_keywords,
num_pages: 2,
};
console.log('normal_image_search_test()');
2019-06-12 18:14:49 +02:00
normal_image_search_test_case( await se_scraper.scrape(config, scrape_config) );
}
// we test with a callback function to our handler
2019-06-12 18:14:49 +02:00
function normal_image_search_test_case(response) {
assert.equal(response.metadata.num_requests, 2);
2019-06-12 18:14:49 +02:00
for (let query in response.results) {
2019-06-12 18:14:49 +02:00
let total_rank = 1;
2019-06-12 18:14:49 +02:00
assert.containsAllKeys(response.results, normal_search_keywords, 'not all keywords were scraped.');
2019-06-12 18:14:49 +02:00
for (let page_number in response.results[query]) {
2019-06-12 18:14:49 +02:00
assert.isNumber(parseInt(page_number), 'page_number must be numeric');
2019-06-12 18:14:49 +02:00
let obj = response.results[query][page_number];
2019-06-12 18:14:49 +02:00
assert.containsAllKeys(obj, ['results', 'time', 'no_results', 'effective_query'], 'not all keys are in the object');
2019-06-12 18:14:49 +02:00
assert.isAtLeast(obj.results.length, 15, 'results must have at least 15 SERP objects');
assert.equal(obj.no_results, false, 'no results should be false');
assert.typeOf(Date.parse(obj.time), 'number', 'time should be a valid date');
2019-06-12 18:14:49 +02:00
for (let res of obj.results) {
2019-06-12 18:14:49 +02:00
assert.containsAllKeys(res, ['link', 'snippet', 'rank', 'clean_link'], 'not all keys are in the SERP object');
2019-06-12 18:14:49 +02:00
assert.isOk(res.link, 'link must be ok');
assert.typeOf(res.link, 'string', 'link must be string');
assert.isAtLeast(res.link.length, 5, 'link must have at least 5 chars');
2019-06-12 18:14:49 +02:00
assert.isOk(res.clean_link, 'clean_link must be ok');
assert.typeOf(res.clean_link, 'string', 'clean_link must be string');
assert.isAtLeast(res.clean_link.length, 5, 'clean_link must have at least 5 chars');
2019-06-12 18:14:49 +02:00
assert.isOk(res.snippet, 'snippet must be ok');
assert.typeOf(res.snippet, 'string', 'snippet must be string');
assert.isAtLeast(res.snippet.length, 10, 'snippet must have at least 10 chars');
2019-06-12 18:14:49 +02:00
assert.isNumber(res.rank, 'rank must be integer');
assert.equal(res.rank, total_rank++, 'rank ist wrong');
}
}
}
}
2019-08-02 17:41:29 +02:00
describe('Google Image', function(){
this.timeout(30000);
it('normal image search test', normal_image_search_test);
});