This change ensure's the HAR file is saved at `recordHar.path` on the
client instead of the server.
NB: The goal was to make this change transparent to the user and NOT
introduce any new APIs. Namely, I want to leave the API open for
potential `context.har.start()` and `context.har.stop()`.
This does BREAK servers that expect the HAR to be at the `recordHar.path`
on the server, but I think that's OK since there haven't been reports
of missing HAR on client making me think not many users are getting
HAR with client and server on different hosts anyways.
Closes#8355
chore: migrate tracing to har
- `HarTracer` is used by both `HarRecorder` that implements
`recordHar` context option, and by tracing.
- We keep the `trace.network` format for now, so it is not
yet a valid har file, but it contains har entries.
Instead of filtering the whole trace file on export, we write
into separate trace file for each chunk. We also write a separate
trace.network file with all resources, because it is reused between
chunks.
This brings us towards `tracing.startFile()/stopFile()` api.
When browser receives multiple header values for the same header name,
we present them as LF-separated value. This is not considered valid in
Node, so we should split by LF when serving a snapshot.
There more invalid characters in headers, so just in case we try/catch it.
Without this, Playwright's CDP feature leaves unreachable
targets (namely OOPIFs).
This change allows for more advanced experimentation in user-land
without relying on out-of-band CDP connections and clients.
Now you can, for example, call `DOM.getDocument` on the
page OR main frame, observe there is an iframe node with
no `contentDocument` (i.e. OOPIF), make note of the referenced
`frameId`, and then iterate of page.frames() calling `Target.getInfo`
on each to link the Playwright Frame with the CDP `frameId` and
then recurse.
Relates #8113
This ensures that any tracing operations can access the recording state.
When stopping, we await for all operations to finish and then clear the state.
To avoid reentrancy, a new flag `isStopping` is introduced.
This patch adds support for the `vue` selector engine that allows
selecting DOM elements based on the component name.
> **NOTE**: `vue` engine supports Vue2 and Vue2.
References #7189
This patch adds support for the `react` selector engine that allows
selecting DOM elements based on the component name.
> **NOTE**: in case of multi-root components (React.Fragment), `react`
engine will select all root DOM elements.
> **NOTE**: `react` engine supports react v15+.
References #7189
Subsequent invocations reset the trace. This removes the `tracing._reset` method.
Also fixed a bug where BASE element was not reset properly.
Also fixed a bug where tracing would affect the result of protocol call
by setting `callMetadata.error` property.
`tracing._export({ path })` exports current tracing state into a file
and does not require tracing to be stopped.
`tracing._reset()` resets current tracing state, but keeps resources
around so they can be referenced in the future snapshots. Does not stop.
The usage pattern is:
```js
await tracing.start({ screenshots: true, snapshots: true });
// ...
await tracing._reset();
// Do stuff, it will all be in the export below.
await tracing._export({ path });
// ...
await tracing.stop();
```
It proved to be unmaintainable due to the following reasons:
- determining upstream branch for the technology preview was a manual
and tedious process
- once determined, actual Safari Technology Preview occasionally
diverted from the branch
This is a speculative fix that moves "background pages cleanup"
to `_didCloseInternal` so that it is only run once, but on both
context closure and browser closure.
Symptom from a flaky test:
```log
browserContext.close: page@18087c372d32819222707ca5e8fd1030 is sending "close" event after being disposed
at PageDispatcher._dispatchEvent (D:\a\playwright\playwright\src\dispatchers\dispatcher.ts:86:15)
at Page.<anonymous> (D:\a\playwright\playwright\src\dispatchers\pageDispatcher.ts:59:12)
at Page.emit (events.js:314:20)
at Page._didClose (D:\a\playwright\playwright\src\server\page.ts:220:10)
at CRPage.didClose (D:\a\playwright\playwright\src\server\chromium\crPage.ts:165:16)
at CRBrowserContext._onClosePersistent (D:\a\playwright\playwright\src\server\chromium\crBrowser.ts:476:24)
at CRBrowserContext.close (D:\a\playwright\playwright\src\server\browserContext.ts:288:20)
```
When new page turns into download, we can have `_markAsError` followed
by the `Page.ready` signal and report the page twice (as an error and
as a real page). This is flaky and depends on whether `Page.ready` is
fast enough before the page closes or not.
Exposed by the "should report new window downloads" test.
This is exposed by the flaky "should report new window downloads" test.
In this test a new page is created, initialized and closed before initialization
finishes. If `lifecycleEventsEnabled` fails with "Target closed error",
we correctly ignore the initialization failure, but a single usage of the
failed promise with `.then` fails anyway.
This patch:
- rolls Firefox to v1269 which is now a firefox-stable build.
- introduces a new channel, `firefox-beta`, for our internal needs.
Fixes#6817
This reverts commit a25b11659be8887b700311180fcd3653aa9e472b.
In a discussion with Dmitry Gozman we decided to revert this and instead
proceed with the following approach:
- rename `//browser_patches/firefox` to `//browser_patches/firefox-beta`
- rename `//browser_patches/firefox-stable` folder to
`//browser_patches/firefox`
In all of the folders, we will keep the `BUILD_NUMBER` original so that
it doesn't clash on the CDN.
This patch:
- starts downloading Firefox Stable equivalent by default
- starts running Firefox-Stable on our smoke tests (tests-1)
- starts running Firefox-Beta on our CQ1 tests (tests-2)
Note: there's a little confusion right now with browser names:
- `firefox-stable` - firefox-stable equivalent
- `firefox`- firefox-beta equivalent
I'll rename `firefox` to `firefox-beta` in a follow-up.
Fixes#6817
