browsertrix-crawler/tests/custom_selector.test.js
Ilya Kreymer fd49041f63
flow behaviors: add scrolling into view (#892)
Some page elements don't quite respond correctly if the element is not
in view, so should add the setEnsureElementIsInTheViewport() to click,
doubleclick, hover and change step locators.
2025-10-07 08:17:56 -07:00

99 lines
2.5 KiB
JavaScript

import child_process from "child_process";
import fs from "fs";
test("test custom selector crawls JS files as pages", async () => {
try {
child_process.execSync(
"docker run -v $PWD/test-crawls:/crawls webrecorder/browsertrix-crawler crawl --url https://www.iana.org/ --collection custom-sel-1 --selectLinks \"script[src]->src\"",
);
} catch (error) {
console.log(error);
}
const crawledPages = fs.readFileSync(
"test-crawls/collections/custom-sel-1/pages/pages.jsonl",
"utf8",
);
const pages = new Set();
for (const line of crawledPages.trim().split("\n")) {
const url = JSON.parse(line).url;
if (!url) {
continue;
}
pages.add(url);
}
const crawledExtraPages = fs.readFileSync(
"test-crawls/collections/custom-sel-1/pages/extraPages.jsonl",
"utf8",
);
const extraPages = new Set();
for (const line of crawledExtraPages.trim().split("\n")) {
const url = JSON.parse(line).url;
if (!url) {
continue;
}
extraPages.add(url);
}
const expectedPages = new Set([
"https://www.iana.org/",
]);
const expectedExtraPages = new Set([
"https://www.iana.org/static/_js/jquery.js",
"https://www.iana.org/static/_js/iana.js",
]);
expect(pages).toEqual(expectedPages);
expect(extraPages).toEqual(expectedExtraPages);
});
test("test invalid selector, crawl fails", async () => {
let status = 0;
try {
child_process.execSync(
"docker run -v $PWD/test-crawls:/crawls webrecorder/browsertrix-crawler crawl --url https://www.iana.org/ --collection custom-sel-invalid --selectLinks \"script[\"",
);
} catch (e) {
status = e.status;
}
// logger fatal exit code
expect(status).toBe(17);
});
test("test valid autoclick selector passes validation", async () => {
let failed = false;
try {
child_process.execSync(
"docker run -v $PWD/test-crawls:/crawls webrecorder/browsertrix-crawler crawl --url https://example-com.webrecorder.net/ --clickSelector button --scopeType page",
);
} catch (e) {
failed = true;
}
// valid clickSelector
expect(failed).toBe(false);
});
test("test invalid autoclick selector fails validation, crawl fails", async () => {
let status = 0;
try {
child_process.execSync(
"docker run -v $PWD/test-crawls:/crawls webrecorder/browsertrix-crawler crawl --url https://example-com.webrecorder.net/ --clickSelector \",\" --scopeType page",
);
} catch (e) {
status = e.status;
}
// logger fatal exit code
expect(status).toBe(17);
});