Many `ts.LanguageService` APIs accept a filename, for example
```ts
getQuickInfoAtPosition(fileName: string, position: number)
```
The requirement is that `fileName` is agnostic to the platform (Linux, Mac,
Windows, etc), and is always normalized to TypeScript's internal
`NormalizedPath`.
This is evident from the way these APIs are called from the language server:
```ts
private onHover(params: lsp.TextDocumentPositionParams) {
const lsInfo = this.getLSAndScriptInfo(params.textDocument);
if (lsInfo === undefined) {
return;
}
const {languageService, scriptInfo} = lsInfo;
const offset = lspPositionToTsPosition(scriptInfo, params.position);
const info = languageService.getQuickInfoAtPosition(scriptInfo.fileName, offset);
// ...
}
```
9fca9c6651/server/src/session.ts (L594)
Here `scriptInfo.fileName` is always a `ts.server.NormalizedPath`.
However, https://github.com/angular/angular/pull/39917 accidentally leaked
the platform-specific paths, and caused a mismatch between the incoming paths
and the paths stored in the internal data structure `fileToComponent`.
This PR fixes the bug by always normalizing the paths, and updating the
type to reflect the format of the underlying data.
Fixes https://github.com/angular/vscode-ng-language-service/issues/1063
PR Close#40492
The initial implementation assumed that the consuming editors would
de-duplicate rename locations. In fact, vscode treats overlapping rename
locations as distinct and errors when trying to preview the renames.
This commit updates the language service to de-duplicate exact file+span
matches before returning rename and reference locations.
While vscode _does_ de-duplicate reference results, it still makes sense
to de-duplicate them on our side when possible to make tests more
understandable. If a template has 3 instances of a variable, it makes
sense to get get 3 reference results rather than 4+ with some duplicates.
PR Close#40454
The current "go to definition" is broken for template variables and
references when a template is overridden. This is because we get the
file url from the source span, which uses the overridden name
'override.html'. Instead, we can retrieve the template file from the
compiler in the same manner that is done for references.
Another way to fix this would have been to use the real template file path when
overriding a template, but this was the more straightforward fix since
the strategy was already used in find references and rename locations.
fixes https://github.com/angular/vscode-ng-language-service/issues/1054
PR Close#40455
When a form is reset, it goes through `_forEachChild` to call `reset` on each of its children.
The problem is that if a control is removed while the loop is running (e.g. by a subscription),
the form will throw an error, because it built up the list of available control before the loop
started.
These changes fix the issue by adding a null check before invoing the callback.
Fixes#33401.
PR Close#40462
Close#40215
`fesm2015/zone.js` is built to `esm` bundle with rollup, so the 'use strict';
statement is not generated in the bundle, even we put the 'use strict' in the src code,
rollup removes the code in the final bundle.
So if we load the `fesm2015/zone.js` as a module, such as `ng serve`, in the index.html
```
<script src="polyfills.js" type="module"></script>
```
Everything works fine, since polyfills.js is loaded as `module`, so it is always `strict`.
But in `ng test`, webpack concat the `zone.js` and loaded into the karma html. For other app and
test code, they are still `strict` since they are `module` because they have `export/import`
statement, but `zone.js` is a bundle without `export`, it is a `side effect` bundle, so after
loaded by webpack, it becomes non-strict. Which causes some issues, such as #40215,
the root cause is the `this` context should be `undefined` but treated as `Window` in `non-strict` mode.
```
Object.prototype.toString.apply(undefined);
// should be [object undefined], but it is [object Window] in non-strict mode.
// zone.js patched version of toString
Object.prototype.toString = function() {
...
// in non-strict mode, this is Window
return originalObjectPrototypeToString.call(this);
}
```
So in this commit, `'use strict';` is always added to the `esm` bundles.
PR Close#40456
The `getRenameInfo` action is used by consumers to
1. Determine if a location is a candidate for renames
2. Determine what text to use as the starting point for the rename
PR Close#40439
We should provide the completion when the cursor is in the attribute
name after the `@` and `animate-`, but now the `KeySpan` starts from the
`@` or `animate-`. For example, the animation event `(@name.done)="v"`,
we can know where the cursor is by the `KeySpan` of `name.done` exactly,
it's in the event name or in the phase name.
PR Close#40347
Reasons for change:
- css_parser, css_ast, and css_lexer are not used anywhere and there are
no entry points from compiler.ts
- tested by building Angular and building/running aio with build-local
PR Close#37463
When loading a module that doesn't provide `RouterModule.forChild()` preloader will get stuck
in an infinite loop and throw `ERROR Error: Maximum call stack size exceeded.`
The issue is that child module's `Injector` will look to its parent `Injector` when it doesn't
find any `ROUTES` so it will return routes for it's parent module instead. This will load the
child again that returns its parent's routes and so on.
Closes#29164
PR Close#36605
This commit lays the groundwork for potentially providing rename
locations from the Ivy native LS. The approach is very similar to what
was done with the feature to find references. One difference, however,
is that we did not require the references to be fully "correct". That
is, the exact text spans did not matter so much, as long as we provide a
location that logically includes the referenced item.
An example of a necessary difference between rename locations and references is
directives. The entire element in the template is a "reference" of the
directive's class. However, it's not a valid location to be renamed. The
same goes for aliased inputs/outputs. The locations in the template
directly map to the class property, which is correct for references, but
would not be correct for rename locations, which should instead map to
the string node fo the alias.
As an initial approach to address the aforementioned issues with rename
locations, we check that all the rename location nodes have the same text. If
_any_ node has text that differs from the request, we do not return any
rename locations. This works as a way to prevent renames that could
break the the program by missing some required nodes in the rename action, but
allowing other nodes to be renamed.
PR Close#40140
Because the query now has `flags` which specify the mode, the static query
instruction can now be remove. It is simply normal query with `static` flag.
PR Close#40091
Previous implementation would fire changes `QueryList.changes.subscribe`
whenever the `QueryList` was recomputed. This resulted in artificially
high number of change notifications, as it is possible that recomputing
`QueryList` results in the same list. When the `QueryList` gets recomputed
is an implementation detail and it should not be the thing which determines
how often change event should fire.
This change introduces a new `emitDistinctChangesOnly` option for
`ContentChildren` and `ViewChildren`.
```
export class QueryCompWithStrictChangeEmitParent {
@ContentChildren('foo', {
// This option will become the default in the future
emitDistinctChangesOnly: true,
})
foos!: QueryList<any>;
}
```
PR Close#40091
When using the `NewEntryPointWriter`, we must copy over all files from the
entry-point bundle to the new entry-point. But since we are going to
write out the modified files directly, there is no need to copy those.
This commit skips copying the files that have been modified.
PR Close#40429
When using the `NewEntryPointWriter` we copy unmodified files over to the new
entry-point in addition to writing out the source files that are processed by ngcc.
But we were not copying over associated source-map files for these unmodified
source files, leading to warnings in downstream tooling.
Now we will also copy over source-maps that reside as siblings of unmodified
source files. We have to make sure that the sources of the source-map point
to the correct files, so we also update the `sourceRoot` property of the copied
source-map.
Fixes#40358
PR Close#40429
This commit updates the logic that calculates `useFactory` function arguments to avoid relying on `instanceof`
checks (thus always retaining symbols) and relies on flags that DI decorators contain (as a monkey-patched property).
Another perf benefit is having less megamorphic reads while calculating args for the `useFactory` call: we used to
check whether a token has `ngMetadataName` property 4 times (in worst case), now we have just 1 megamorphic read in
all cases.
Closes#40143.
PR Close#40145
Report non-template diagnotics when calling `getDiagnotics` function of
the language service we only returned template diagnotics. This change
causes it to return all diagnotics, not just diagnostics from the
template type checker.
PR Close#40331
This commit fixes a bug in the **View Engine** implementation of
`getSemanticDiagnostics` and `getDefinitionAndBoundSpan` for node in the
decorator metadata that represents an external URL
(`templateUrl` or `styleUrls`).
The URL could be either relative or absolute, but the latter was not taken
into account.
Fix https://github.com/angular/vscode-ng-language-service/issues/1055
PR Close#40406
The `template` and `isInline` fields were previously stored in a nested
object, which was initially done to accommodate for additional template
information to support accurate source maps for external templates. In
the meantime the source mapping has been accomplished in a different
way, and I feel this flattened structure is simpler and smaller so is
preferable over the nested object. This change also makes the `isInline`
property optional with a default value of `false`.
PR Close#40383
The parser has a list of tag definitions that it uses when parsing the template. Each tag has a
`contentType` which tells the parser what kind of content the tag should contain. The problem is
that the browser has two separate `title` tags (`HTMLTitleElement` and `SVGTitleElement`) and each
of them has to have a different `contentType`, otherwise the parser will throw an error further down
the pipeline.
These changes update the tag definitions so that each tag name can have multiple content types
associated with it and the correct one can be returned based on the element's prefix.
Fixes#31503.
PR Close#40259
Do not warn that navigation was triggered outside Angular zone if the
Router was created outside Angular zone in the first place.
Closes#25837
PR Close#25839
Previously, the SW would wait to become idle before executing scheduled
tasks (including checks for newer app versions). It was considered idle
when it hadn't received any request for at least 5 seconds. As a result,
if the app performed polling (i.e. sent requests to the server) in a
shorter than 5 seconds interval, the SW would never detect and update to
a newer app version.
Related issue: #40207
This commit fixes this by adding a max delay to `IdleScheduler` to
ensure that no scheduled task will remain pending for longer than the
specified max delay.
PR Close#40234
This commit refactors `Driver#deleteAllCaches()` to use `Array#map()`
instead of `Array#reduce()` for running async operations in parallel.
This allows avoiding having to recursively wrap Promises with
`Promise.all()`.
PR Close#40234
Previously, clients were notified about updates sequentially. This
wasn't necessary.
This commit changes the `Driver#notifyClientsAboutUpdate()` method to
notify the clients in parallel (by switching from `Array#reduce()` to
`Array#map()` and `Promise.all()`).
This also aligns the `notifyClientsAboutUpdate()` method with the
`notifyClientsAboutUnrecoverableState()` method.
PR Close#40234
Previously, the `Driver#notifyClientsAboutUnrecoverableState()` method
would not wait for the completion of the promises created to notify the
clients. Theoretically, this could result in the SW instance's getting
destroyed by the browser before all clients have been notified. This is
extremely unlikely to happen in practice, since the async operations are
very quick, but it _is_ theoretically possible.
This commit ensures that the SW instance will remain alive while
notifying the clients by making `notifyClientsAboutUnrecoverableState()`
await the notification promises.
PR Close#40234
The decorator downleveling transform patches `ts.EmitResolver.isReferencedAliasDeclaration`
to prevent elision of value imports that occur only in a type-position, which would
inadvertently install the patch repeatedly for each source file in the program.
This could potentially result in a stack overflow when a very large number of files is
present in the program.
This commit fixes the issue by ensuring that the patch is only applied once.
This is also a slight performance improvement, as `isReferencedAliasDeclaration`
is no longer repeatedly calling into all prior installed patch functions.
Fixes#40276
PR Close#40374
Previously, if there were path-mapped entry-points, where one contaied the
string of another - for example `worker-client` and `worker` - then the
base paths were incorrectly computed resulting in the wrong package path
for the longer entry-point. This was because, when searching for a matching
base path, the strings were tested using `startsWith()`, whereas we should
only match if the path was contained in a directory from a file-system
point of view.
Now we not only check whether the target path "starts with" the base path
but then also whether the target path is actually contained in the base path
using `fs.relative()`.
Fixes#40352Fixes#40357
PR Close#40376
`Object.values` is not supported in IE11 without a polyfill. The quickest,
most straightfoward fix for this is to simply use `Object.keys` instead.
We may want to consider including the polyfill in the CLI in the future
or just wait until IE11 support is dropped before using
`Object.values`.
PR Close#40370
Currently the language service has to force `compileNonExportedClasses` to
`true` to handle inline NgModules in tests, regardless of the value in user's
tsconfig.json.
However, the override is not reinstated after the compiler option changes
(triggered by a change in tsconfig.json).
This commit fixes the bug.
PR Close#40364
This commit documents how to add a helper function which combines all the params
in the router state tree into a single object. It provides a starting point for
developers to reference if they require a more fine-tuned approach.
Fixes#11023
PR Close#40306
In #37182 the in-memory-web-api module was moved into this repo.
Copy the reamde into this repo with the following changes:
* Removed Travis badges
* Updated github links to point to Angular repo
* Removed 'running tests' as it is no longer relevant
Fixes#40190
PR Close#40203
When we attach a `ViewRef` to a `ViewContainerRef`, we save a reference to the container
onto the `ViewRef` so that we can remove it when the ref is destroyed. The problem is
that if the container's `hostView` is destroyed first, the `ViewRef` has no way of knowing
that it should stop referencing the container.
These changes remove the leak by not saving a reference at all. Instead, when a `ViewRef`
is destroyed, we clean it up through the `LContainer` directly. We don't need to worry
about the case where the container is destroyed before the view, because containers
automatically clean up all of their views upon destruction.
Fixes#38648.
PR Close#40219
The `NgControlStatusGroup` directive is shared between template-driven and reactive form modules. In cases when
only reactive forms module is present, the `NgControlStatusGroup` directive is still activated on all `<form>`
elements, but if there is no other reactive directive applied (such as `formGroup`), corresponding `ControlContainer`
token is missing, thus causing exceptions (since `NgControlStatusGroup` directive relies on it to determine the
status). This commit updates the logic to handle the case when no `ControlContainer` is present (effectively making
directive logic a noop in this case).
Alternative approach (more risky) worth considering in the future is to split the `NgControlStatusGroup` into
2 directives with different set of selectors and include them into template-driven and reactive modules separately.
The downside is that these directives might be activated simultaneously on the same element (e.g. `<form>`),
effectively doing the work twice.
Resolves#38391.
PR Close#40344
This class is refactored to extend the new `NodeJSReadonlyFileSystem`
which itself extends `NodeJSPathManipulation`. These new classes allow
consumers to create file-systems that provide a subset of the full file-system.
PR Close#40281
Now that `ReadonlyFileSystem` and `PathManipulation` interfaces are
available, this commit updates the localize package to use these more
focussed interfaces.
PR Close#40281
Now that `ReadonlyFileSystem` and `PathManipulation` interfaces are
available, this commit updates the compiler-cli to use these more
focussed interfaces.
PR Close#40281
Keen Yee Liau
Andrew Scott
Kristiyan Kostadinov
twerske
JiaLiPassion
ivanwonder
Scott Wang
Martin Sikora
Zach Arend
Misko Hevery
Alexey Elin
Pete Bacon Darwin
Max Schorradt
Andrew Kushnir
JoostK
markostanimirovic
Trotyl
Joey Perrott
George Kalpakas
Alan Agius
Che Liu
sijmenhuizenga