.. | ||
abort_controller_test.ts | ||
blob_test.ts | ||
body_test.ts | ||
buffer_test.ts | ||
build_test.ts | ||
chmod_test.ts | ||
chown_test.ts | ||
console_test.ts | ||
copy_file_test.ts | ||
custom_event_test.ts | ||
dir_test.ts | ||
dispatch_json_test.ts | ||
dispatch_minimal_test.ts | ||
dom_exception_test.ts | ||
dom_iterable_test.ts | ||
error_stack_test.ts | ||
event_target_test.ts | ||
event_test.ts | ||
fetch_test.ts | ||
file_test.ts | ||
files_test.ts | ||
form_data_test.ts | ||
format_error_test.ts | ||
fs_events_test.ts | ||
get_random_values_test.ts | ||
globals_test.ts | ||
headers_test.ts | ||
internals_test.ts | ||
link_test.ts | ||
location_test.ts | ||
make_temp_test.ts | ||
metrics_test.ts | ||
mkdir_test.ts | ||
net_test.ts | ||
os_test.ts | ||
performance_test.ts | ||
permissions_test.ts | ||
process_test.ts | ||
read_dir_test.ts | ||
read_file_test.ts | ||
read_link_test.ts | ||
README.md | ||
realpath_test.ts | ||
remove_test.ts | ||
rename_test.ts | ||
request_test.ts | ||
resources_test.ts | ||
signal_test.ts | ||
stat_test.ts | ||
symbols_test.ts | ||
symlink_test.ts | ||
test_util.ts | ||
testing_test.ts | ||
text_encoding_test.ts | ||
timers_test.ts | ||
tls_test.ts | ||
truncate_test.ts | ||
tty_test.ts | ||
umask_test.ts | ||
unit_test_runner.ts | ||
unit_tests.ts | ||
url_search_params_test.ts | ||
url_test.ts | ||
utime_test.ts | ||
version_test.ts | ||
write_file_test.ts |
Deno runtime tests
Files in this directory are unit tests for Deno runtime.
They are run under compiled Deno binary as opposed to files in cli/js/
which
are bundled and snapshotted using deno_typescript
crate.
Testing Deno runtime code requires checking API under different runtime
permissions (ie. running with different --allow-*
flags). To accomplish this
all tests exercised are created using unitTest()
function.
import { unitTest } from "./test_util.ts";
unitTest(function simpleTestFn(): void {
// test code here
});
unitTest({
ignore: Deno.build.os === "win",
perms: { read: true, write: true },
},
function complexTestFn(): void {
// test code here
}
);
unitTest
is is a wrapper function that enhances Deno.test()
API in several
ways:
- ability to conditionally skip tests using
UnitTestOptions.skip
- ability to register required set of permissions for given test case using
UnitTestOptions.perms
- sanitization of resources - ensuring that tests close all opened resources preventing interference between tests
- sanitization of async ops - ensuring that tests don't leak async ops by ensuring that all started async ops are done before test finishes
Running tests
unit_test_runner.ts
is the main script used to run unit tests.
Runner discoveres required permissions combinations by loading
cli/js/tests/unit_tests.ts
and going through all registered instances of
unitTest
.
There are three ways to run unit_test_runner.ts
:
# Run all tests. Spawns worker processes for each discovered permission
# combination:
target/debug/deno -A cli/js/tests/unit_test_runner.ts --master
# By default all output of worker processes is discarded; for debug purposes
# the --verbose flag preserves output from the worker
target/debug/deno -A cli/js/tests/unit_test_runner.ts --master --verbose
# Run subset of tests that don't require any permissions
target/debug/deno cli/js/tests/unit_test_runner.ts
# Run subset tests that require "net" and "read" permissions
target/debug/deno --allow-net --allow-read cli/js/tests/unit_test_runner.ts
# "worker" mode communicates with parent using TCP socket on provided address;
# after initial setup drops permissions to specified set. It shouldn't be used
# directly, only be "master" process.
target/debug/deno -A cli/js/tests/unit_test_runner.ts --worker --addr=127.0.0.1:4500 --perms=net,write,run
# Run specific tests
target/debug/deno --allow-net cli/js/tests/unit_test_runner.ts -- netTcpListenClose
RUST_BACKTRACE=1 cargo run -- --allow-read --allow-write cli/js/tests/unit_test_runner.ts -- netUnixDialListen
Http server
tools/http_server.py
is required to run when one's running unit tests. During
CI it's spawned automatically, but if you want to run tests manually make sure
that server is spawned otherwise there'll be cascade of test failures.