Spaces:
Sleeping
API Tester
A test framework for testing the puter HTTP API and puterjs API.
Table of Contents
How to use
Workflow
All commands below should be run from the root directory of puter.
(Optional) Start a backend server:
npm startCopy
example_config.ymland add the correct values:cp ./tools/api-tester/example_config.yml ./tools/api-tester/config.ymlFields:
- url: The endpoint of the backend server. (default: http://api.puter.localhost:4100/)
- username: The username of the user to test. (e.g.
admin) - token: The token of the user. (can be obtained by logging in on the webpage and typing
puter.authTokenin Developer Tools's console) - mountpoints: The mountpoints to test. (default config includes 2 mountpoints:
/for "puter fs provider" and/admin/tmpfor "memory fs provider")
Run tests against the HTTP API (unit tests and benchmarks):
node ./tools/api-tester/apitest.js(experimental) Run tests against the puter-js client:
node ./tools/api-tester/apitest.js --puterjs
Shorthands
Run tests against the HTTP API (unit tests and benchmarks):
node ./tools/api-tester/apitest.jsRun unit tests against the HTTP API:
node ./tools/api-tester/apitest.js --unitRun benchmarks against the HTTP API:
node ./tools/api-tester/apitest.js --benchFilter tests by suite name:
node ./tools/api-tester/apitest.js --unit --suite=mkdirFilter benchmarks by name:
node ./tools/api-tester/apitest.js --bench --suite=stat_intensive_1Stop on first failure:
node ./tools/api-tester/apitest.js --unit --stop-on-failure(unimplemented) Filter tests by test name:
# (wildcard matching) Run tests containing "memoryfs" in the name node ./tools/api-tester/apitest.js --unit --test='*memoryfs*' # (exact matching) Run the test "mkdir in memoryfs" node ./tools/api-tester/apitest.js --unit --test='mkdir in memoryfs'(unimplemented) Rerun failed tests in the last run:
node ./tools/api-tester/apitest.js --rerun-failed
Basic Concepts
A test case is a function that tests a specific behavior of the backend API. Test cases can be nested:
await t.case('normal mkdir', async () => {
const result = await t.mkdir_v2('foo');
expect(result.name).equal('foo');
await t.case('can stat the created directory', async () => {
const stat = await t.stat('foo');
expect(stat.name).equal('foo');
});
});
A test suite is a collection of test cases. A .js file should contain exactly one test suite.
module.exports = {
name: 'mkdir',
do: async t => {
await t.case('normal mkdir', async () => {
...
});
await t.case('recursive mkdir', async () => {
...
});
}
};
Behaviors
Working directory (t.cwd)
- The working directory is stored in
t.cwd. - All filesystem operations are performed relative to the working directory, if the given path is not absolute. (e.g.,
t.mkdir('foo'),t.cd('foo'),t.stat('foo'), etc.) - Tests will be run under all mountpoints. The default working directory for a mountpoint is
${mountpoint.path}/{username}/api_test. (This is subject to change in the future, the reason we includeadminin the path is to ensure the test useradminhas write access, see Permission Documentation for details.) - The working directory is reset at the beginning of each test suite, since a test suite usually doesn't want to be affected by other test suites.
- The working directory will be inherited from the cases in the same test suite, since a leaf case might want to share the context with its parent/sibling cases.
module.exports = {
name: 'readdir',
do: async t => {
// t.cwd is reset to /admin/api_test
await t.case('normal mkdir', async () => {
// inherits cwd from parent/sibling cases
await t.case('mkdir in subdir', async () => {
// inherits cwd from parent/sibling cases
});
});
}
};
Implementation
- Test suites are registered in
tools/api-tester/tests/__entry__.js.
TODO
- Reset
t.cwdif a test case fails. Currently,t.cwdis not reset if a test case fails. - Integrate apitest into CI, optionally running it only in specific scenarios (e.g., when backend code changes).