Previously if only a component template changed then we would know to
rebuild its component source file. But the compilation was incorrect if the
component was part of an NgModule, since we were not capturing the
compilation scope information that had a been acquired from the NgModule
and was not being regenerated since we were not needing to recompile
the NgModule.
Now we register compilation scope information for each component, via the
`ComponentScopeRegistry` interface, so that it is available for incremental
compilation.
The `ComponentDecoratorHandler` now reads the compilation scope from a
`ComponentScopeReader` interface which is implemented as a compound
reader composed of the original `LocalModuleScopeRegistry` and the
`IncrementalState`.
Fixes#31654
PR Close#31932
Publishing of NGCC packages should not be allowed. It is easy for a user to publish an NGCC'd version of a library they have workspace libraries which are being used in a workspace application.
If a users builds a library and afterwards the application, the library will be transformed with NGCC and since NGCC taints the distributed files that should be published.
With this change we use the npm/yarn `prepublishOnly` hook to display and error and abort the process with a non zero error code when a user tries to publish an NGCC version of the package.
More info: https://docs.npmjs.com/misc/scripts
PR Close#32031
Previously, when run with `createNewEntryPointFormats: true`, `ngcc`
would only update `package.json` with the new entry-point for the first
format property that mapped to a format-path. Subsequent properties
mapping to the same format-path would be detected as processed and not
have their new entry-point format recorded in `package.json`.
This commit fixes this by ensuring `package.json` is updated for all
matching format properties, when writing an `EntryPointBundle`.
PR Close#32052
Remove the `formatProperty` property from the `EntryPointBundle`
interface, because the property is not directly related to that type.
It was only used in one place, when calling `fileWriter.writeBundle()`,
but we can pass `formatProperty` directrly to `writeBundle()`.
PR Close#32052
This refactoring more clearly separates the different phases of the work
performed by `ngcc`, setting the ground for being able to run each phase
independently in the future and improve performance via parallelization.
Inspired by/Based on @alxhub's prototype: alxhub/angular@cb631bdb1
PR Close#32052
This change basically moves some checks to happen up front and ensures
we don't try to process any more properties than we absolutely need.
(The properties would not be processed before either, but we would
consider them, before finding out that they have already been processed
or that they do not exist in the entry-point's `package.json`.)
This change should make no difference in the work done by `ngcc`, but it
transforms the code in a way that makes the actual work known earlier,
thus making it easier to parallelize the processing of each property in
the future.
PR Close#32052
In commit 7b55ba58b (part of PR #29092), the implementation of
`makeEntryPointBundle()` was changed such that it now always return
`EntryPointBundle` (and not `null`).
However, the return type was not updated and as result we continued to
unnecessarily handle `null` as a potential return value in some places.
This commit fixes the return type to reflect the implementation and
removes the redundant code that was dealing with `null`.
PR Close#32052
ngcc analyzes the dependency structure of the entrypoints it needs to
process, as the compilation of entrypoints is ordering sensitive: any
dependent upon entrypoint must be compiled before its dependees. As part
of the analysis of the dependency graph, it is detected when a
dependency of entrypoint is not installed, in which case that entrypoint
will be marked as ignored.
For libraries that work with Angular Universal to run in NodeJS, imports
into builtin NodeJS modules can be present. ngcc's dependency analyzer
can only resolve imports within the TypeScript compilation, which
builtin modules are not part of. Therefore, such imports would
erroneously cause the entrypoint to become ignored.
This commit fixes the problem by taking the NodeJS builtins into account
when dealing with missing imports.
Fixes#31522
PR Close#31872
ngcc analyzes the dependency structure of the entrypoints it needs to
process, as the compilation of entrypoints is ordering sensitive: any
dependent upon entrypoint must be compiled before its dependees. As part
of the analysis of the dependency graph, it is detected when a
dependency of entrypoint is not installed, in which case that entrypoint
will be marked as ignored.
When a target entrypoint to compile is provided, it could occur that
given target is considered ignored because one of its dependencies might
be missing. This situation was not dealt with currently, instead
resulting in a crash of ngcc.
This commit prevents the crash by taking the above scenario into account.
PR Close#31872
Previously, `ngcc` would avoid processing a `formatPath` that a property
in `package.json` mapped to, if either the _property_ was marked as
processed or the `formatPath` (i.e. the file(s)) was processed in the
same `ngcc` run (since the `compiledFormats` set was not persisted
across runs).
This could lead in a situation where a `formatPath` would be compiled
twice (if for example properties `a` and `b` both mapped to the same
`formatPath` and one would run `ngcc` for property `a` and then `b`).
This commit fixes it by ensuring that as soon as a `formatPath` has been
processed all corresponding properties are marked as processed (which
persists across `ngcc` runs).
PR Close#32003
Previously, when `ngcc` was called with `compileAllFormats === false`
(i.e. how `@angular/cli` calls it), it would not attempt to process
more properties, once the first was successfully processed. However, it
_would_ continue looping over them and perform some unnecessary
operations, such as:
- Determining the format each property maps to (which can be an
expensive operation for some properties mapping to either UMD or
CommonJS).
- Checking whether each property has been processed (which involves
checking whether any property has been processed with a different
version of `ngcc` each time).
- Potentially marking properties as processed (which involves a
file-write operation).
This commit avoids the unnecessary operations by entirely skipping
subsequent properties, once the first one has been successfully
processed. While this theoretically improves performance, it is not
expected to have any noticeable impact in practice, since the list of
`propertiesToConsider` is typically small and the most expensive
operation (marking a property as processed) has low likelihood of
happening (plus these operations are a tiny fraction of `ngcc`'s work).
PR Close#32003
Previously, when `ngcc` needed to mark multiple properties as processed
(e.g. a processed format property and `typings` or all supported
properties for a non-Angular entry-point), it would update each one
separately and write the file to disk multiple times.
This commit changes this, so that multiple properties can be updated at
once with one file-write operation. While this theoretically improves
performance (reducing the I/O operations), it is not expected to have
any noticeable impact in practice, since these operations are a tiny
fraction of `ngcc`'s work.
This change will be useful for a subsequent change to mark all
properties that map to the same `formatPath` as processed, once it is
processed the first time.
PR Close#32003
This commit changes the emit order of ngcc when a class has multiple static
fields being assigned. Previously, ngcc would emit each static field
followed immediately by any extra statements specified for that field. This
causes issues with downstream tooling such as build optimizer, which expects
all of the static fields for a class to be grouped together. ngtsc already
groups static fields and additional statements. This commit changes ngcc's
ordering to match.
PR Close#31933
In #31426 a fix was implemented to render namespaced decorator imports
correctly, however it turns out that the fix only worked when decorator
information was extracted from static properties, not when using
`__decorate` calls.
This commit fixes the issue by creating the decorator metadata with the
full decorator expression, instead of only its name.
Closes#31394
PR Close#31614
An identifier may become repeated when bundling multiple source files
into a single bundle, so bundlers have a strategy of suffixing non-unique
identifiers with a suffix like $2. Since ngcc operates on such bundles,
it needs to process potentially suffixed identifiers in their canonical
form without the suffix. The "ngx-pagination" package was previously not
compiled fully, as most decorators were not recognized.
This commit ensures that identifiers are first canonicalized by removing
the suffix, such that they are properly recognized and processed by ngcc.
Fixes#31540
PR Close#31614
Any decorator information present in TypeScript is emitted into the
generated JavaScript sources by means of `__decorate` call. This call
contains both the decorators as they existed in the original source
code, together with calls to `tslib` helpers that convey additional
information on e.g. type information and parameter decorators. These
different kinds of decorator calls were not previously distinguished on
their own, but instead all treated as `Decorator` by themselves. The
"decorators" that were actually `tslib` helper calls were conveniently
filtered out because they were not imported from `@angular/core`, a
characteristic that ngcc uses to drop certain decorators.
Note that this posed an inconsistency in ngcc when it processes
`@angular/core`'s UMD bundle, as the `tslib` helper functions have been
inlined in said bundle. Because of the inlining, the `tslib` helpers
appear to be from `@angular/core`, so ngcc would fail to drop those
apparent "decorators". This inconsistency does not currently cause any
issues, as ngtsc is specifically looking for decorators based on their
name and any remaining decorators are simply ignored.
This commit rewrites the decorator analysis of a class to occur all in a
single phase, instead of all throughout the `ReflectionHost`. This
allows to categorize the various decorate calls in a single sweep,
instead of constantly needing to filter out undesired decorate calls on
the go. As an added benefit, the computed decorator information is now
cached per class, such that subsequent reflection queries that need
decorator information can reuse the cached info.
PR Close#31614
Currently we always generate the `read` parameter for the view and content query instructions, however since most of the time the `read` parameter won't be set, we'll end up generating `null` which adds 5 bytes for each query when minified. These changes make it so that the `read` parameter only gets generated if it has a value.
PR Close#31667
The support for decorators that were imported via a namespace,
e.g. `import * as core from `@angular/core` was implemented
piecemeal. This meant that it was easy to miss situations where
a decorator identifier needed to be handled as a namepsaced
import rather than a direct import.
One such issue was that UMD processing of decorators was not
correct: the namespace was being omitted from references to
decorators.
Now the types have been modified to make it clear that a
`Decorator.identifier` could hold a namespaced identifier,
and the corresponding code that uses these types has been
fixed.
Fixes#31394
PR Close#31426
There are two places in the ngcc processing where it needs to load the
content of a file given by a general path:
* when determining the format of an entry-point.
To do this ngcc uses the value of the relevant property in package.json.
But in the case of `main` it must parse the contents of the entry-point
file to decide whether the format is UMD or CommonJS.
* when parsing the source files for dependencies to determine the order in
which compilation must occur. The relative imports in each file are parsed
and followed recursively, looking for external imports.
Previously, we naively assumed that the path would match the file name exactly.
But actually we must consider the standard module resolution conventions.
E.g. the extension (.js) may be missing, or the path may refer to a directory
containing an index.js file.
This commit fixes both places.
This commit now requires the `DependencyHost` instances to check
the existence of more files than before (at worst all the different possible
post-fixes). This should not create a significant performance reduction for
ngcc. Since the results of the checks will be cached, and similar work is
done inside the TS compiler, so what we lose in doing it here, is saved later
in the processing. The main performance loss would be where there are lots
of files that need to be parsed for dependencies that do not end up being
processed by TS. But compared to the main ngcc processing this dependency
parsing is a small proportion of the work done and so should not impact
much on the overall performance of ngcc.
// FW-1444
PR Close#31509
When determining if a `main` path points to a UMD or CommonJS
format, the contents of the file need to be loaded and parsed.
Previously, it was assumed that the path referred to the exact filename,
but did not account for normal module resolution semantics, where the
path may be missing an extension or refer to a directory containing an
`index.js` file.
// FW-1444
PR Close#31509
Paths can be mapped directly to files, which was not being taken
into account when computing `basePaths` for the `EntryPointFinder`s.
Now if a `pathMapping` pattern does not exist or is a file, then we try
the containing folder instead.
Fixes#31424
PR Close#30525
Previously, ngcc had to walk the entire `node_modules` tree looking for
entry-points, even if it only needed to process a single target entry-point
and its dependencies.
This added up to a few seconds to each execution of ngcc, which is noticeable
when being run via the CLI integration.
Now, if an entry-point target is provided, only that target and its entry-points
are considered rather than the whole folder tree.
PR Close#30525
When profiling ngcc it is notable that a large amount of time
is spent dealing with an exception that is thrown (and handled
internally by fs) when checking the existence of a file.
We check file existence a lot in both finding entry-points
and when TS is compiling code. This commit adds a simple
cached `FileSystem`, which wraps a real `FileSystem` delegate.
This will reduce the number of calls through to `fs.exists()` and
`fs.readFile()` on the delegate.
Initial benchmarks indicate that the cache is miss to hit ratio
for `exists()` is about 2:1, which means that we save about 1/3
of the calls to `fs.existsSync()`.
Note that this implements a "non-expiring" cache, so it is not suitable
for a long lived `FileSystem`, where files may be modified externally.
The cache will be updated if a file is changed or moved via
calls to `FileSystem` methods but it will not be aware of changes
to the files system from outside the `FileSystem` service.
For ngcc we must create a new `FileSystem` service
for each run of `mainNgcc` and ensure that all file operations
(including TS compilation) use the `FileSystem` service.
This ensures that it is very unlikely that a file will change
externally during `mainNgcc` processing.
PR Close#30525
This message gets called if a format has already been
compiled and we only want the first. So the message itself
is wrong but it is also not very useful anyway.
PR Close#30525
The ngcc tool adds namespaced imports to files when compiling. The ngtsc
tooling was not processing types correctly when they were imported via
such namespaces. For example:
```
export declare class SomeModule {
static withOptions(...): ModuleWithProviders<ɵngcc1.BaseModule>;
```
In this case the `BaseModule` was being incorrectly attributed to coming
from the current module rather than the imported module, represented by
`ɵngcc1`.
Fixes#31342
PR Close#31367
If a package delcares a class internally on an NgModule, ngcc
needs to be able to add a public export to this class's type.
Previously, if the typing file for the declared is not imported
from the typings entry-point file, then ngcc cannot find it.
Now we try to guess the .d.ts files from the equivalent .js
files.
PR Close#31411
Non-wild-card path-mappings were not being matched correctly.
Further path-mapped secondary entry-points that
were imported from the associated primary entry-point were not
being martched correctly.
Fixes#31274
PR Close#31450
Some formats of CommonJS put the decorator helper calls
outside the class IIFE as statements on the top level of the
source file.
This commit adds support to the `CommonJSReflectionHost`
for this format.
PR Close#31335
If an entry-point has missing dependencies then it cannot be
processed and is marked as invalid. Similarly, if an entry-point
has dependencies that have been marked as invalid then that
entry-point too is invalid. In all these cases, ngcc should quietly
ignore these entry-points and continue processing what it can.
Previously, if an entry-point had more than one entry-point that
was transitively invalid then ngcc was crashing rather than
ignoring the entry-point.
PR Close#31276
Our module resolution prefers `.js` files over `.d.ts` files because
occasionally libraries publish their typings in the same directory
structure as the compiled JS files, i.e. adjacent to each other.
The standard TS module resolution would pick up the typings
file and add that to the `ts.Program` and so they would be
ignored by our analyzers. But we need those JS files, if they
are part of the current package.
But this meant that we also bring in JS files from external
imports from outside the package, which is not desired.
This was happening for the `@fire/storage` enty-point
that was importing the `firebase/storage` path.
In this commit we solve this problem, for the case of imports
coming from a completely different package, by saying that any
file that is outside the package root directory must be an external
import and so we do not analyze those files.
This does not solve the potential problem of imports between
secondary entry-points within a package but so far that does
not appear to be a problem.
PR Close#30591
Rather than passing a number of individual arguments, we can
just pass an `EntryPointBundle`, which already contains them.
This is also a precursor to using more of the properties in the bundle.
PR Close#30591
This will allow users of the `EntryPointBundle` to use some of the `EntryPoint`
properties without us having to pass them around one by one.
PR Close#30591
Previously we expected the constructor parameter `decorators`
property to be an array wrapped in a function. Now we also support
an array not wrapped in a function.
PR Close#30591
Some packages do not actually provide a `typings` field in their
package.json. But TypeScript naturally infers the typings file from
the location of the JavaScript source file.
This commit modifies ngcc to do a similar inference when finding
entry-points to process.
Fixes#28603 (FW-1299)
PR Close#30591
There are scenarios where it is not possible for ngcc to guess the format
or configuration of an entry-point just from the files on disk.
Such scenarios include:
1) Unwanted entry-points: A spurious package.json makes ngcc think
there is an entry-point when there should not be one.
2) Deep-import entry-points: some packages allow deep-imports but do not
provide package.json files to indicate to ngcc that the imported path is
actually an entry-point to be processed.
3) Invalid/missing package.json properties: For example, an entry-point
that does not provide a valid property to a required format.
The configuration is provided by one or more `ngcc.config.js` files:
* If placed at the root of the project, this file can provide configuration
for named packages (and their entry-points) that have been npm installed
into the project.
* If published as part of a package, the file can provide configuration
for entry-points of the package.
The configured of a package at the project level will override any
configuration provided by the package itself.
PR Close#30591
Previously each test relied on large shared mock file-systems, which
makes it difficult to reason about what is actually being tested.
This commit breaks up these big mock file-systems into smaller more
focused chunks.
PR Close#30591
Kristiyan Kostadinov
Pete Bacon Darwin
Alan Agius
George Kalpakas
JoostK
Alex Rickabaugh
Matias Niemelä