browsertrix-crawler/tests/mult_url_crawl_with_favicon.test.js

26 lines
1.3 KiB
JavaScript
Raw Normal View History

import child_process from "child_process";
import fs from "fs";
test("ensure multi url crawl run with docker run passes", async () => {
2023-11-08 16:40:49 -05:00
child_process.execSync("docker run -v $PWD/test-crawls:/crawls webrecorder/browsertrix-crawler crawl --url https://www.iana.org/ --url https://webrecorder.net/ --generateWACZ --text --collection advanced --combineWARC --rolloverSize 10000 --workers 2 --title \"test title\" --description \"test description\" --pages 2 --limit 2");
child_process.execSync("docker run -v $PWD/test-crawls:/crawls webrecorder/browsertrix-crawler wacz validate --file collections/advanced/advanced.wacz");
});
test("check that the favicon made it into the pages jsonl file", () => {
2023-11-08 16:40:49 -05:00
expect(fs.existsSync("test-crawls/collections/advanced/pages/pages.jsonl")).toBe(true);
2023-11-08 16:40:49 -05:00
const data1 = JSON.parse(fs.readFileSync("test-crawls/collections/advanced/pages/pages.jsonl", "utf8").split("\n")[1]);
const data2 = JSON.parse(fs.readFileSync("test-crawls/collections/advanced/pages/pages.jsonl", "utf8").split("\n")[2]);
const data = [ data1, data2 ];
for (const d of data) {
if (d.url === "https://webrecorder.net/") {
2023-11-08 16:40:49 -05:00
expect(d.favIconUrl).toEqual("https://webrecorder.net/assets/favicon.ico");
}
if (d.url === "https://iana.org/") {
2023-11-08 16:40:49 -05:00
expect(d.favIconUrl).toEqual("https://www.iana.org/_img/bookmark_icon.ico");
}
}
});