se-scraper/test/test_baidu.js

85 lines
3.2 KiB
JavaScript
Raw Normal View History

const se_scraper = require('./../index.js');
var assert = require('chai').assert;
/*
* Use chai and mocha for tests.
* https://mochajs.org/#installation
*/
const normal_search_keywords = ['mouse', 'cat'];
async function normal_search_test() {
let config = {
compress: false,
2019-06-11 18:16:59 +02:00
debug_level: 1,
keyword_file: '',
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',
};
2019-06-11 18:16:59 +02:00
let scrape_config = {
search_engine: 'baidu',
keywords: normal_search_keywords,
num_pages: 2,
};
console.log('normal_search_test()');
2019-06-11 18:16:59 +02:00
normal_search_test_case( await se_scraper.scrape(config, scrape_config) );
}
// we test with a callback function to our handler
2019-06-11 18:16:59 +02:00
function normal_search_test_case(response) {
assert.equal(response.metadata.num_requests, 4);
2019-06-11 18:16:59 +02:00
for (let query in response.results) {
let total_rank = 1;
2019-06-11 18:16:59 +02:00
assert.containsAllKeys(response.results, normal_search_keywords, 'not all keywords were scraped.');
2019-06-11 18:16:59 +02:00
for (let page_number in response.results[query]) {
2019-06-11 18:16:59 +02:00
assert.isNumber(parseInt(page_number), 'page_number must be numeric');
2019-06-11 18:16:59 +02:00
let obj = response.results[query][page_number];
2019-06-11 18:16:59 +02:00
assert.containsAllKeys(obj, ['results', 'time', 'num_results', 'no_results'], 'not all keys are in the object');
2019-06-11 18:16:59 +02:00
assert.isAtLeast(obj.results.length, 7, 'results must have at least 7 SERP objects');
assert.typeOf(Date.parse(obj.time), 'number', 'time should be a valid date');
2019-06-11 18:16:59 +02:00
assert.typeOf(obj.num_results, 'string', 'num_results must be a string');
assert.isAtLeast(obj.num_results.length, 5, 'num_results should be a string of at least 5 chars');
2019-06-11 18:16:59 +02:00
assert.equal(obj.no_results, false, 'no results should be false');
2019-06-11 18:16:59 +02:00
for (let res of obj.results) {
2019-06-11 18:16:59 +02:00
assert.containsAllKeys(res, ['link', 'title', 'rank', 'visible_link', 'snippet'], 'not all keys are in the SERP object');
2019-06-11 18:16:59 +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');
2019-06-11 18:16:59 +02:00
assert.isOk(res.visible_link, 'visible_link must be ok');
assert.typeOf(res.visible_link, 'string', 'visible_link must be string');
assert.isAtLeast(res.visible_link.length, 5, 'visible_link must have at least 5 chars');
2019-06-11 18:16:59 +02:00
assert.isOk(res.title, 'title must be ok');
assert.typeOf(res.title, 'string', 'title must be string');
assert.isAtLeast(res.title.length, 5, 'title must have at least 5 chars');
2019-06-11 18:16:59 +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');
2019-06-11 18:16:59 +02:00
assert.isNumber(res.rank, 'rank must be integer');
assert.equal(res.rank, total_rank++, 'rank ist wrong');
}
}
}
}
(async () => {
await normal_search_test();
})();