Closes https://github.com/microsoft/playwright/issues/32070. We were
applying `additionalFileMatcher` not just to `filteredProjectSuites`,
but also to `projectSuites`. `projectSuites` is where we take dependency
projects from, though - so `--only-changed` led to empty dependency
projects, resulting in the reported bug.
The fix is to only apply `additionalFileMatcher` on
`filteredProjectSuites`.
Supercedes https://github.com/microsoft/playwright/pull/31915, closes
https://github.com/microsoft/playwright/issues/31811.
When TypeScript resolves a specifier via path mapping, it does not
interpret `package.json`. If path mapping resolves to a directory, it
only looks at the `index.js` file in that directory if it's in CommonJS
mode.
We need to mirror this in our `esmLoader.ts`.
---------
Co-authored-by: Dmitry Gozman <dgozman@gmail.com>
Also considered an alternative to not perform the locator handler check
during one-shot, but that would be somewhat against the promise of the
locator handler that is supposed to run **before** every expect check.
Fixes#32089.
Introduce `--tsconfig` to specify a single config to be used for all
imported files, instead of looking up tsconfig for each file separately.
Fixes#12829.
Ideally we generate the timestamp when the Event gets created. This
patch adds a best-effort logic, since we can't override the constructor
of natively created events, e.g. `MouseEvent`.
Fixes https://github.com/microsoft/playwright/issues/31924
This allows a dynamic import of a TS file to be processed by Babel.
For some reason, Playwright used to revert the CJS transforms. However,
ESM loader and transforms are always active, so CJS should be too.
When two attachments have the same content sha1, we used the first one's
name for the downloaded file, no matter which one the user clicked to
download. Now we pass the name explicitly.
References #31912.
Closes https://github.com/microsoft/playwright/issues/22211
Currently, when the server notifies the UI about changed files, the UI
determines what files to re-run based on an old test list. By listing
tests before that, we make sure that the test list is up-to-date, and
that added tests are included in the next run.
I've also removed the `listChanged` event as discussed in the team sync.
The event isn't used anywhere and fires in exactly the same cases where
`testFilesChanged` fired, so i've folded them into one another. This allowed simplifying `Watcher`.
- Update copy to clipboard button.
- Reveal test source in the Source tab instead of external editor.
- New button to reveal in the external editor in the Source tab.
- Move the Pick Locator button next to snapshot tabs.
Upon calling `browser.close()` or dropping remote connection, make sure
to reject api calls before resolving `browser.close()` and firing a
`disconnected` event.
This change aligns the order guarantee with non-remote case.
Resolves https://github.com/microsoft/playwright/issues/31847 by adding
playwright config's `baseURL` value to the `context-options` trace
event, and showing that in the Trace Viewer.
Because the added property is optional, I didn't increment the trace
format version.
I've also considered pulling the `baseURL` from the existing
`browser.newContext` step to get around modifying the trace format, but
that felt pretty hacky.
https://github.com/user-attachments/assets/ecaef747-727d-4937-9ca3-1605ca9907b9
---------
Signed-off-by: Simon Knott <info@simonknott.de>
Co-authored-by: Dmitry Gozman <dgozman@gmail.com>
Introduces an `--only-changed [base ref]` option.
`playwright test --only-changed` filters the test run to only run test
suites that have uncommitted changes.
`playwright test --only-changed=foo` runs only tests that were changed
since commit `foo`.
In pull request CI, this can be used to run changed tests first and fail
fast: `--only-changed=$GITHUB_BASE_REF`.
During local development, it can be used to quickly filter down to the
touched set of tests suites.
In some rare usecases, this can also help to cut down on CI usage for
pull requests. Tread with caution though.
File dependencies are taken into account to ensure that if you touched a
utility file, all relevant tests are still executed.
Closes https://github.com/microsoft/playwright/issues/15075
Broken out of https://github.com/microsoft/playwright/pull/31727 as per
@dgozman's
[request](https://github.com/microsoft/playwright/pull/31727#discussion_r1685793229).
The PR goal is to remove the `suite` argument from the Component
testing's Vite Plugin. `suite` is used to enrich Vite's dependency graph
with information about dependencies between test suites and helper
files. It essentially merges the Vite graph with the
`compilationCache.ts > fileDependencies` graph, and then writes the
result back into `compilationCache.ts > externalDependencies`.
By refactoring this to make the connection on the reading end in
`collectAffectedTestFiles`, we can drop the `suite` parameter.
We didn't yet have a test that depended on the dependency graph being
connected correctly between `fileDependencies` and
`externalDepedencies`, so I've [extended an existing
test](53a539938b)
to capture that.
This reverts commit 0aa2f06f68.
Discussed the new permission in the API review and decided not to
proceed with the feature as we are not ready to commit to supporting it
yet:
* the API is Chromium specific
* the API is still experimental
* there is no clarity to what extend the screen manipulation APIs will
work in old headless which is our main test environment
We'll keep an eye on the demand for the feature and may get back to
implementing it in the future.
Reference: https://github.com/microsoft/playwright/issues/27198
Previously, only the "actual" attachment was created, pointing to the
file in `test-results`. Now, the "expected" attachment pointing to the
file in `__screenshots__` is also created. This will help any reporters
that would like to know the "expected" path, for example to do a manual
accept/decline of the baseline.
Fixes#30693.
The following actions keep `noWaitAfter` option: `click`, `selectOption`
and `press`.
All other actions that used to have `noWaitAfter` now behave like it was
set to true, not waiting for follow-up navigations. In the docs, this
option is marked as completely ignored.
A small logic change was made to compensate for this behavior: when
waiting for the `hitTargetInterceptor`, we now race it against
navigations to avoid stalling when navigation stalls. Previously,
waiting for the interceptor was disabled when `noWaitAfter` was passed,
and since it's impossible to pass this option now, we mitigate by never
stalling instead.
Fixes#31469.
... unless an array of file-system-friendly parts is provided.
Motivation: attachment name is used as a file system path when
downloading attachments, so we keep them fs-friendly.
References #30693.
When used in a terminal, the `list` reporter prints out information
about test steps to help debugging. In non-TTY environments like GitHub
Actions, currently it doesn't.
This PR changes that, so that in non-TTY environments you'll see the
"step end" messages appearing, but not the "step begin" messages. This
is a good middleground, because it helps the user understand test
progress, without being too verbose.
Closes https://github.com/microsoft/playwright/issues/31674
Retaining traces in the following scenarios:
- browser crash;
- manual `browser.close()`;
- implicit `browser.close()` from the `browser` fixture upon test end.
This does not affect the library, where `browser.close()` will not
retain the trace and will close the browser as fast as possible.
References #31541, #31535, #31537.
* Revert harTracer change from
aeba083da0
to make sure that har.Entry._monotonicTime always represents request
start time. The issue from the corresponding report was due to HEAD and
GET request sent for the same URL, that use case is still addressed as
we match by url + method
* Adjust resources monotonic time as well when several contexts are
shown in the trace viewer.
Fixes https://github.com/microsoft/playwright/issues/31133
Historically, this plugin was important to translate dynamic imports
into require calls so that we can intercept them and transpile.
This is not needed anymore with ESM loader enabled by default, so we can
avoid this transformation and support dynamic imports of ESM-only
packages/files.
Fixes#17075, fixes#23255, fixes#31140, references #23662.
This allows snapshots of file:/// pages with external stylesheets,
images, etc to be rendered correctly in the trace viewer. (Otherwise, it
tries to request the file:/// URIs directly and the requests get blocked
by the browser.)
Fixes#31112.
When merging blob reports test ids are patched to make sure there is no
collision when merging reports that might have overlapping test ids.
However, even if you were merging reports that had no overlapping ids,
all test ids will be modified, which is an undesirable side effect.
This PR only modify test ids when the same test id has already been used
in a previous blob report.
----
This change is also part of
https://github.com/microsoft/playwright/pull/30962
The spirit of this change is reverting #23153. Since that time, we have
moved tracing and `artifactsDir` lifetime into the test runner, so the
reason for revert is mitigated.
Fixes#30287, fixes#30718, fixes#30959.
This includes two major changes:
- reuse `SerializedFS` for live test runner tracing;
- merge scheduled `appendFile` operations into a single `fs` call.
In some cases, this improves performance of UI mode by 61% and
performance of `trace: on` mode by 38%. Note that performance
improvement on the average test will not be as noticeable.
References #30875, #30635.
- Documents `PLAYWRIGHT_FORCE_TTY` and `FORCE_COLOR` across terminal
reporters.
- New `PLAYWRIGHT_LIST_PRINT_STEPS`. Removes undocumented test-only
`PW_TEST_DEBUG_REPORTERS_PRINT_STEPS`.
- Replaces `PLAYWRIGHT_HTML_REPORT` with `PLAYWRIGHT_HTML_OUTPUT_DIR`
and `PW_TEST_HTML_REPORT_OPEN` with `PLAYWRIGHT_HTML_OPEN` for
consistency, supports older versions for backwards compatibility.
- New `PLAYWRIGHT_HTML_HOST`, `PLAYWRIGHT_HTML_PORT` and
`PLAYWRIGHT_HTML_ATTACHMENTS_BASE_URL`.
- New `PLAYWRIGHT_JUNIT_STRIP_ANSI` and
`PLAYWRIGHT_JUNIT_INCLUDE_PROJECT_IN_TEST_NAME`.
- Removes `PW_HTML_REPORT` that was set for unknown reason.
Previously, terminal reporters consulted `process.stdout.isTTY`. Now it
is possible to control the tty behavior:
- `PLAYWRIGHT_FORCE_TTY=0` or `PLAYWRIGHT_FORCE_TTY=false` to disable
TTY;
- `PLAYWRIGHT_FORCE_TTY=1` or `PLAYWRIGHT_FORCE_TTY=true` to enable TTY,
defaults to 100 columns when real columns are unavailable;
- `PLAYWRIGHT_FORCE_TTY=<number>` to force enable TTY and set the
columns.
Fixes#29422.
Previously, terminating worker always had a 30 seconds force exit.
Now, regular worker termination assumes that process will eventually
finish tearing down all the fixtures and exits. However, the
self-destruction routine keeps the 30 seconds timeout to avoid zombies.
Fixes#30504.
Implements feature requested in
https://github.com/microsoft/playwright/issues/30457
The test runner treats flaky tests as failures when the flag is enabled,
but still reports flaky tests as flaky in the reporting interface. It
feels like something worth discussing as this behaviour makes sense to
me, but looked a bit odd to @BJSS-russell-pollock when I ran this past
him.
Closes#30457.
Matching bu `apiName@wallTime` fails when two actions start at the same
time, e.g. two parallel api requests. Moreover, it results in trace
actions that have parent set to themselves, which in turn causes
infinite loop in the trace viewer. To avoid this problems we write
stepId explicitly to the library trace and use those step ids to find
corresponding test runner steps.
The stepId is passed via zone in case of expect, because the protocol
step is quite deep in the call chain after or explicitly in case of API
call steps created by the test runner instrumentation.
Pre Node.js `18.20.0`:
- `assert` is supported
Post Node.js `18.20.0`
- `assert` and `with` is supported.
Before https://github.com/microsoft/playwright/pull/30482 we kept
`asserts` in the JS code, Node.js was interpreting them. The `with`
keyword was not supported, this was what the PR was fixing.
After https://github.com/microsoft/playwright/pull/30482 Babel is
converting `assert` (deprecated) into `with` (successor) since we use
the `deprecatedAssertSyntax` option.
This means, that the minimum Node.js version we support in order to use
import attributes is now `18.20.0` where they added the `with` support.
This follows our principle of supporting only the latest minor release
for Node.js versions.
See here for the 18.20 changelog:
> #### Added support for import attributes
>
> Support has been added for import attributes, to replace the old
import
> assertions syntax. This will aid migration by making the new syntax
available
> across all currently supported Node.js release lines.
>
> This adds the `with` keyword which should be used in place of the
previous
> `assert` keyword, which will be removed in a future semver-major
Node.js
> release.
>
> For example,
>
> ```console
> import "foo" assert { ... }
> ```
>
> should be replaced with
>
> ```console
> import "foo" with { ... }
> ```
Fixes https://github.com/microsoft/playwright/pull/30482 - the tests
were a noop before, since they were tree-shaked by Babel.
Similarly to page.close, we pass test-runner specific reason to
facilitate better error messages.
```
1) a.test.ts:10:11 › test
Error: apiRequestContext.fetch: Fixture { request } from beforeAll cannot be reused in a test.
- Recommended fix: use a separate { request } in the test.
- Alternatively, manually create APIRequestContext in beforeAll and dispose it in afterAll.
See https://playwright.dev/docs/api-testing#sending-api-requests-from-ui-tests for more details.
9 |
10 | test('test', async () => {
> 11 | await context.fetch('http://example.com');
| ^
12 | });
13 |
```
Closes#29260.
* Use only monotonicTime for sorting, do not use wallTime for that
* Since test runner and the library can be running on different
machines, those machines may have clocks which are not synchronized. To
avoid problems in such cases we compute delta between test runner and
and library contexts based on a start time of action that exists in both
contexts.
When comparing `outputDir` and html-reporter `outputFolder`, we now make
sure that both paths end with a forward-slash.
Fixes#28677
---------
Co-authored-by: Georg Unterholzner <georg.unterholzner@dynatrace.com>
There are plenty of edge cases in this area:
- interrupted test run;
- did not run because of serial mode failure;
- failed before `test.skip()` call (e.g. in `beforeEach`) in one of the
retries;
- and more...
Related issues: #28322, #28321, #27455, #17652.
Prior changes: #27762, #26385, #28360, probably more.
There is still some duplication between `outcome()` and similar logic in
`base.ts`, which might be deduped in a follow-up.
Fixes#28322.
- Automatically waiting for the overlay locator to be hidden, with
`allowStayingVisible` opt-out.
- `times: 1` option.
- `removeLocatorHandler(locator, handler)` method.
- Passing `locator` as first argument to `handler`.
Fixes#30471. Fixes#30424. Fixes#29779.
Generate 'global timeout' and 'max failures' errors in the runner. Avoid
reading `config.globalTimeout` and `config.maxFailures` in the base
reporters.
Reference https://github.com/microsoft/playwright/issues/29768
There are new "non-manual" WPT accname tests that we now mostly pass,
which required a few tweeks in calculating role and name.
Also implemented accessible description computation, which is just a
small addition on top of accessible name, and passed respective wpt
tests.
References #18332.
Instead of filtering tests assuming there are no two projects with same
name we always rebuild test tree from scratch and restore previos test
results in the list mode.
Fixes https://github.com/microsoft/playwright/issues/30396
According to the spec, such controls should use the native value as long
as they have "aria-label". The relevant spec section is 2D.
However, there is an open issue that claims this should always apply,
and all browsers and wpt test actually do that:
https://github.com/w3c/accname/issues/64.
Fixes#28848.
The accessible name computation spec has changed to explicitly mention
this case:
Step 2A. Hidden Not Referenced. If the current node is hidden and is:
- Not part of an aria-labelledby or aria-describedby traversal, where
the node directly referenced by that relation was hidden.
- Nor part of a native host language text alternative element (e.g.
label in HTML) or attribute traversal, where the root of that traversal
was hidden.
See https://w3c.github.io/accname/#computation-steps. Chromium, Firefox
and Safari all agree with the spec.
Fixes#29796.
Repro: `cd tests/components/ct-svelte && rm -rf node_modules
package-lock.json && npm i && npx playwright test --project=chromium`
Follow-up based on
https://github.com/microsoft/playwright/pull/28624#issuecomment-1858608101.
Svelte has no router by default, only SvelteKit - so lets remove the
package which is not maintained anymore and not recommended.
* FullConfig split into FullConfig and ConfigInWorker
* FullProject split into FullProject and ProjectInWorker
* Tests can only access {Config, Project}InWorker, while reporters can
only access Full{Config,Project}. This will allow in the future to
change the reporter API independently from the tests.
* Added documentation for the 4 classes.
When websocket disconnects during `browserType.connect()` call, the
error looks like this now:
```
browserType.connect: Custom error message received over WebSocket
```
Previously, it was a generic error:
```
browserType.connect: Target page, context or browser has been closed
```
Fixes#30040
# Context
In the `playwright.config` file, we can custom the default `timeout`
property for `expect.toPass` by the property:
```ts title='playwright.config.ts'
export default defineConfig({
// ...
expect: {
toPass: {
timeout: 60_000,
},
},
});
```
However, we can't customize the `intervals` option. So the default value
is always `[100, 250, 500, 1000]`.
# Goal
Add the possibility to customize the intervals option from the
`playwright.config` file.
```ts title='playwright.config.ts'
export default defineConfig({
// ...
expect: {
toPass: {
timeout: 60_000,
intervals: [1000, 2000, 3000]
},
},
});
```
* Add `Suite.entries` that returns tests and suites in their declaration
order
* Exposed `Suite.type` and `TestCase.type` for discriminating between
different entry types.
* Blob report format is updated to store entries instead of separate
lists for suites and tests.
* Bumped blob format version to 2, added modernizer.
Fixes https://github.com/microsoft/playwright/issues/29984
This brings up the question that we would show the tag name twice if its
a tag in a title. This would be aligned to how HTML report is doing it.
Fixes https://github.com/microsoft/playwright/issues/29927
---------
Signed-off-by: Max Schmitt <max@schmitt.mx>
Co-authored-by: Dmitry Gozman <dgozman@gmail.com>
With the recent change that gave after hooks a separate timeout, fixture
teardown does not imply that "test finished successfully, but fixture
teardown was slow".
This was historically done to make `console.log()` have colors. However,
this makes any other code that checks `process.stdout.isTTY` incorrectly
assume real TTY support.
Node18 and Node20 now respect `FORCE_COLOR=1` in console, so our default
behavior of forcing colors in the worker process just works out of the
box. See https://github.com/nodejs/node/pull/48034.
Previously, there was at most one "hard error", as opposite to multiple
"soft errors". This was done to preserve the historic behavior at the
time of introducing multiple `TestInfo.errors`.
With this change, every user callback that is executed `withRunnable()`
can throw an error and/or timeout, and both of these will end up in
`TestInfo.errors`.
Additionally, there is at most one "unhandled exception" error, to avoid
flooding the report with mass failures.
Drive-by: remove boolean arguments from `_failWithError()`.
Fixes#29876.
When collecting dependencies both from CJS loader and from ESM loader,
the latter would overwrite the dependencies set instead of appending.
Also make sure cts/cjs/mts/mjs are all supported equally.
References #29747.
Test-specific output dir and snapshot names are trimmed to 60 chars
instead of just 100. The snapshot names are still trimmed at 100 chars
for backwards compatibility.
Reference https://github.com/microsoft/playwright/issues/29719
This helps in a case where navigating to an origin fails for some
reason, for example because a registered service worker loads some
content into the supposedly blank page.
Fixes#29402.
This covers blocked requests, e.g. mixed-content, that receive
`loadingFailed` with empty `errorText`.
Also, forcefully resolve `allHeaders()` in this case, since we know
there will be no actual network headers.
Fixes#29833.
This fixes a case where we incorrectly used the final config's base path
when resolving relative path mappings in the absence of the baseUrl.
Fixes#29816.
This PR is a fix proposal for a bug when trying to record a omnibox
navigation after a recorded action (e.g., `fill`).
The following test, included in this PR, reproduces the problem:
```ts
test('should record omnibox navigations after recordAction', async ({ page, openRecorder, server }) => {
const recorder = await openRecorder();
await recorder.setContentAndWait(`<textarea></textarea>`);
await Promise.all([
recorder.waitForOutput('JavaScript', 'fill'),
page.locator('textarea').fill('Hello world'),
]);
// for performed actions, 5 seconds is the time needed to ensure they are committed
await page.waitForTimeout(5000);
await page.goto(server.PREFIX + `/empty.html`);
await recorder.waitForOutput('JavaScript', `await page.goto('${server.PREFIX}/empty.html');`);
});
```
After performed actions (e.g., `click`), it successfully records the
navigation as long as there's at least a 5 sec. gap between both
actions. That happens because after that 5 sec. interval the performed
action is automatically commited and therefore the navigation is not
stored as a signal of that action.
The proposed fix for recorded actions also forces that action to be
automatically commited after 5 sec (for testing, I'm using 500ms to
speed up the test execution).
It was already handling worker sessions, but not OOPIFs. As a result,
some functionality was properly implemented only for OOPIFs and not for
workers.
This change removes OOPIFs fanout for network-related calls from CRPage
and moves that to the CRNetworkManager, now also covering workers.
This will be useful to run `beforeAll`/`afterAll` hooks with a separate
`TestInfo` instance, as well as run use helpers like
`_runAndFailOnError()` during scope teardown.
This fixes some edge cases where fixtures and steps inside them were
attached to the wrong parent (see the new test).
This will also allow to replace some `runAsStep` calls with a flat list
of tasks to do that do not rely on lexical scope.
API changes:
- `test(title, details, body)` where details contain `tag` and
`annotation`.
- similar `details` property added to `test.skip`, `test.fail`,
`test.fixme`, `test.only`, `test.describe` and other `test.describe.*`
variations.
- `TestProject.tagFilter`/`TestConfig.tagFilter` that supports logical
tag expressions with `(`, `)`, `and`, `or` and `not`.
- `--tag` CLI option to filter by tags.
- New annotations are available in `TestInfo.annotations` and
`TestCase.annotations`.
- New tags are available in `TestCase.tags`.
Reporter changes:
- `json` reporter includes new tags in addition to old `@smoke`-style
tags. **Breaking**: tags are now listed with the leading `@` symbol.
- `html` reporter filters by old and new tags with the same `@smoke`
token.
Fixes#29229, fixes#23180.
The metadata.error change was brought back in
https://github.com/microsoft/playwright/pull/29271and it broke java port
as we could have error and result set simulteniously. This PR moves the
logic to the trace recorder instead and keeps the protocol contract
clear that either error or result is present, but not both.
- Modifiers that only depend on the worker fixtures are implemented as
`beforeAll` hooks.
- Modifiers that depend on test fixtures are implemented as `beforeEach`
hooks.
- Pushed `_runAndFailOnError` down the stack, wrapping individual hooks
instead of the whole "before hooks" section.
- Reused the same code to run `beforeAll` and `afterAll` hooks and
modifiers.
**Behavior change**: `test.skip()` inside a `beforeAll` now skips the
hook and all tests in the suite.
When `updateSnapshots === 'missing'` we generate new expectations on the
first attempt and don't retry the test afterwards instead of trying it
retries-1 times and only writing new expectation on the last attempt.
This logic infects all serial mode suites that contain the test with
missing expectations, so they also will not be retried.
Reference https://github.com/microsoft/playwright/issues/29073
Motivation: On Windows we call around 50 times `PrintDeps.exe` which
takes on a very fast machine 500+ms. On Linux we do it around 120 times
(`ldd`) which takes around 150ms.
This change validates the dependencies once on browser install (`npx
playwright install`). In case its failing, it will emit a warning, in
case of a success, it will create a marker file that the binary has been
validated. For future `launch()` calls, we'll read this file and if
exists, we'll not validate again. Otherwise we'll validate again.
Note: If the marker file is older than 30 days, the browser will be
validated again.
Previously, new `Recorder` instance was given an existing
`InjectedScript`. However, we built a separate source for
`InjectedScript` vs `Recorder`, and both bundles contain their own copy
of all helper modules, e.g. `roleUtils`.
This resulted in two copies of helper modules, which is troublesome for
any module-level globals like a top-level cache. Depending on whether
`Recorder` or `InjectedScript` called into the helper, they would access
the different value of a module global, which lead to bugs.
To prevent this, we force any external dependencies to be imported
through the `InjectedScript.utils`.
This supports mixed quotes locators in JavaScript where we are not sure
what quote is the correct one, so we normalize to unescaped single quote
when comparing with the original.
Drive-by: we were allowing single quotes in Python, Java and .NET, but
these are actually not allowed.
Regressed in #27718.
Fixes#28630.
Otherwise, we forever block SIGTERM and SIGHUP by registering a handler
that does not do anything (due to no browsers to close) and prevents
default handler that exits from running.
Fixes#28091.
- Keep main frames in a separate bucket, so that page methods that
redirect to the main frame continue to work.
- Increase default dispatchers limit to `10_000`.
- Increase dispatchers limit for `JSHandle`/`ElementHandle` to
`100_000`.
Fixes#28320, #28503.
If request gets cancelled by the page before we fulfill it, we receive
`loadingFailed` event. In that case we'll ignore interception error if
any, otherwise the error will be propagated to the caller.
Fixes https://github.com/microsoft/playwright/issues/28490
