Summary: In this diff I handle the case where the key of an object may be nullable. To do that, I added a nullable param to the schema.
Reviewed By: rickhanlonii
Differential Revision: D16282081
fbshipit-source-id: cb7668d754e2ff30aef22df582351a512c988016
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/25673
This diff add e2e code generator validation. I added proper buck rule which compiles cpp code and test if it really compiles.
While testing I faced few minor issues:
- `getBool` and `getNumber` shouldn't have `rt` param.
- Generators now start considering whole module name instead of sliced part of their name.
- Fixed import structure in generated cpp.
- renamed `jsireact` to `ReactCommon` following D16231697
Reviewed By: rickhanlonii
Differential Revision: D16221277
fbshipit-source-id: aff4011ad52dd5e16546ffdb709d6a751ebfaced
Summary:
This diff allows for parsing more flexible AST format.
Reusing logic used for methods, I add similar support for props and commands in components. Now it's possible to define separated type for props and commands in another part of the file.
Also, extracted this method to another file for reusing purposes.
Added tests.
Reviewed By: rickhanlonii
Differential Revision: D16221445
fbshipit-source-id: 21553bf5ade66588dd7dc0320d25333260b0ada9
Summary:
Following out internal communication we decided to change logic of current parser to consider types which extend 'TurboModule' instead of looking at default exports.
It also cassed for minor changes in testing logic and updating snapshots.
Reviewed By: rickhanlonii
Differential Revision: D16205707
fbshipit-source-id: c75cd4febf773ba5101e6b60ed1a921773246009
Summary: Code generators for modules and components don't have a lot of in common so there's no point to keepo them in the same directory and mix their files.
Reviewed By: rickhanlonii
Differential Revision: D16183434
fbshipit-source-id: b6ee32c6b223e8d6e4bc843b2e111598bee94dc5
Summary:
Add generating cpp files following NativeModuleSpec.
We still need to figure out how to deal with module names, so I currently made in workable by extracting `Sample` from `SampleNativeModule` and filing into proper places, but it's a temporary solution.
Reviewed By: rickhanlonii
Differential Revision: D16181292
fbshipit-source-id: 0f50352d7610f99f0228045ae78309383d3e8bfa
Summary:
Add basic Native Module header generator
We still need to figure out how to deal with module names, so I currently made in workable by extracting `Sample` from `SampleNativeModule` and filing into proper places, but it's a temporary solution.
Reviewed By: rickhanlonii
Differential Revision: D16180561
fbshipit-source-id: bb27592d565ae439707d781fb2650f2bdb140146
Summary: I this diff I include basic codegen skeleton. It does not work yet, but I'd like to consider it as kind of playgroud for further changes. I have splited tests for component generation from test for module generation.
Reviewed By: rickhanlonii
Differential Revision: D16161909
fbshipit-source-id: a0ecb81898810fe6aa09f76238c433894c731b73
Summary:
Following out internal communication we decided to change logic of current parser to consider types which extend 'TurboModule' instead of looking at default exports.
It also cassed for minor changes i n testing logic and updating snapshots.
Reviewed By: rickhanlonii
Differential Revision: D16200939
fbshipit-source-id: a21cbfc461647df2b9210c0eca4c2c95c285dfe1
Summary: Int32 and Float can be represented on native site in a different after cpp code generation. Inspired by component codegen.
Reviewed By: rickhanlonii
Differential Revision: D16201358
fbshipit-source-id: a5c6c449d69f162ad9cd2a998d57e9c65bc17706
Summary:
This diff includes minor codestyle changes.
All these fixes was discussed with Rick.
Reviewed By: rickhanlonii
Differential Revision: D16169332
fbshipit-source-id: e561e2f2b116b6fdf8434c3dfc20c3e610d7ecad
Summary: It's pointless to handle non optional key if value has withDefaul so I'm adding a rule into parser preventing from such cases
Reviewed By: lunaleaps
Differential Revision: D16166709
fbshipit-source-id: 38cef522b217917a3a4886d857720932f2ebb475
Summary: This part of writing codegen was straightforward. I've added check for 'FunctionTypeAnnotation' and then reuse exisiting methods' parser for generating schema for callback.
Reviewed By: TheSavior
Differential Revision: D16131213
fbshipit-source-id: 2ec0e241f2174dee3a93857563db0626013d004e
Summary:
I'm not very confident with this part, but actually in existing codegen we wrap param into another object marked as `NullableTypeAnnotion`. It makes logic a little more complicated. Also it allows for multiple `?` before type which is useless in code generation as well as nullable types inside arrays which also does not have impact on a final code generation.
I suggest adding `nullable` field into param which covers all existing cases (and probably all cases needed).
Reviewed By: TheSavior
Differential Revision: D16121609
fbshipit-source-id: 6e086d4d26bbd0aab3015ec7ecae106ebbaa5a2c
Summary: Add support for Object reusing mostly the code for arrays
Reviewed By: TheSavior
Differential Revision: D16122314
fbshipit-source-id: c1b201c659e6fdc98bed553fb16280ed5ce9e647
Summary: This implementation is a bit different than current one since previously object annotation was always wrapped into `GenericTypeAnnotation` object which is not needed in every of current use cases.
Reviewed By: TheSavior
Differential Revision: D16121727
fbshipit-source-id: e0de1852e13c66f459efd7a3bcde449e412d8b95
Summary: Now `Array<T>` is supported as a returning values or a param of function's definition. Also, Array of Array is allowed.
Reviewed By: TheSavior
Differential Revision: D16121246
fbshipit-source-id: 59c484120c4025a152e3ba8044eecf11dbbea1f7
Summary: While working on D16090949 I discovered that the code being generated for an array of enums was not correctly converted to mask. The type that holds the mask can't be enum class but other container type, I used `uint32_t`.
Reviewed By: shergin
Differential Revision: D16109338
fbshipit-source-id: 237077adaafe631eda973bc76cefa49035bbcd66
Summary: This diff a adds generic mechansm for extracting types from types' difinitions defined in other part of the file.
Reviewed By: TheSavior
Differential Revision: D16131742
fbshipit-source-id: 006b6980fa9f6a064d5bb8734ba2740b802b6989
Summary:
In this I add support for primitive types as return
- String
- Number
- Boolean
- Any
- Void
Reviewed By: TheSavior
Differential Revision: D16108273
fbshipit-source-id: 2dbd1d5a852a8cf5baf378a73d860492fe2f87cb
Summary: It's a skeleton for preparing schema json from native modules. Right not it does nothing, but it should considered as a kind of workplace for further work
Reviewed By: TheSavior, cpojer, RSNara
Differential Revision: D16107216
fbshipit-source-id: 12717d015e7409d980ef16cccd81330fa978b0b1
Summary:
`WithDefault` appears not to be required to be prefixed with `?` because it's option value per se.
Fixed tests, removed `?` where needed, updated snapshots and review them. Added mechanism fro throwing error when `?WithDefault` found. Add tests for it.
Reviewed By: rubennorte
Differential Revision: D16048463
fbshipit-source-id: f55ed7454aacf0b8c42944a9b5c1037ad1b360fe
Summary:
It's not needed to keep required providing default values even if they are not actually relevant.
Here I add a support for `null`, `0` of `false` instead by default and remove throwing errors if no other default value provided.
Reviewed By: rubennorte
Differential Revision: D16049047
fbshipit-source-id: bc4961af3873190568f2753fc4ec975354896df5
Summary:
It appears that `(e: BubblingEvent<T>) = mixed` exists only in given context and it's pointless to keep in this way. It could be simplified to `BubblingEventHandler<T>` without any negative consequences and that's the motivation of this diff.
The only tradeoff of this decision is leaving an opportunity to declare Bubbling/Direct event in the top of the file bc then analysing the code becomes much more difficult. However, it's not used anywhere so it's not a problem now and probably any time.
Also, changes the names to `DirectEventHandler` and `BubblingEventHandler` which are more related to current state. The names were updated in many places in code.
Reviewed By: rubennorte
Differential Revision: D16054571
fbshipit-source-id: 741d075eb46b80bac8eb73a6b30fc0b448cb3902
Summary:
This diff adds a way for the codegen to handle view configs with non-standard top event names by adding a `paperTopLevelNameDeprecated` field to events in the schema.
## The problem
The problem this is solving is that Android host components build their own options for event settings in the view config. So instead of enforcing `onChange` to use the top level event name `topChange` like iOS does, Android can use `change` or `forbarChange` or anything the person adding the component wanted at the time:
```
// Expected
topChange: {
registrationName: 'onChange',
},
// Android
bringBackStargateYouCowards: {
registrationName: 'onChange',
},
```
This is possible because of the way Android builds these settings
```
// On iOS, notice that there's no option to change the top level name:
RCT_EXPORT_VIEW_PROPERTY(onChange, RCTDirectEventBlock);
// On Android, you provide the top level event name
Override
public Map getExportedCustomDirectEventTypeConstants() {
return MapBuilder.of(
"bringBackStargateYouCowards",
MapBuilder.of("registrationName", "onChange"));
}
```
Since the codegen does not allow us to specify the top level event name (similar to iOS), we don't have a way to customize the names to support android
## The solution
To fix this, we're adding an extra option the event flow types:
```
onBubblingChange: (event: BubblingEvent<Event, 'customBubblingName'>) => void,
onDirectChange: (event: DirectEvent<Event, 'customDirectName'>) => void,
```
These will register **both** top level names in the view config:
```
{
directEventTypes: {
// Notice the same registration name is configured for different top level events
topChange: {
registrationName: 'onChange',
},
bringBackStargateYouCowards: {
registrationName: 'onChange',
},
}
}
```
This means when either `topChange` or `bringBackStargateYouCowards` fires it will call the onChange listener. **This gives us the flexibility to rename the native event name without breaking backwards compatibility.** Old apps will work when `bringBackStargateYouCowards` is fired, and new apps with an update will work when `topChange` fires.
Note: only the correct name will be generated for Fabric so technically we don't even really need to migrate the paper names and this prop can be deleted when paper is gone.
Reviewed By: cpojer
Differential Revision: D16042065
fbshipit-source-id: 40d076b43ffbbfc6c65c3c19de481d922a2add62
Summary:
This diff switches the native codegen over to the flow parser
It does this by:
- Creating a new e2e directory
- Migrating the schema.js fixtures to flow types in e2e/
- Updating the buck tests to use the flow type fixtures
- Finally, updating the rest of rn_codegen to use *NativeComponent instead of *Schema.js
Removing all of the schemas in the next diff to keep this one clean
Reviewed By: cpojer
Differential Revision: D15960603
fbshipit-source-id: 3df28b31e618491301578ab7f6e28a80f55404b2
Summary: The schema for these view commands is lifted wholesale from the schema for TurboModules: P67239314
Reviewed By: rickhanlonii
Differential Revision: D15943109
fbshipit-source-id: a0ccd4e47067b62970218df6a32527c15868c4a5
Summary: Fixes an issue with the native output of components without props and events (a trailing comma)
Reviewed By: cpojer
Differential Revision: D15960962
fbshipit-source-id: 315276ef01484546da43954e6fd879350e214006
Summary:
This diff removes an option from the codegen and replaces it with two new options
Removes:
- `isDeprecatedPaperComponentNameRCT`
Adds:
- `paperComponentName`: a better version of the removed option that allows more than just adding RCT
- `paperComponentNameDeprecated`: a new option that allows migrating native code to a new name
```
// Use for components with no current paper rename in progress
// Does not check for new name
paperComponentName?: string,
// Use for components currently being renamed in paper
// Will use new name if it is available and fallback to this name
paperComponentNameDeprecated?: string,
```
For example, Slider uses `paperComponentName: 'RCTSlider'` because it has a different name in fabric but is not currently being migrated to a new name. Because of other work in progress, we don't want to use UIManager check if we don't need to
Reviewed By: shergin
Differential Revision: D15857629
fbshipit-source-id: ca0d3b7dc4a75e00d136ae1f5c84f7423960399d
Summary:
## Context
At the moment, the codegen process does not expect any types defined in a native component (*NativeComponent.js) to be exported. For example, the `NativeProps` type may be defined as:
```
// RCTSegmentedControlNativeComponent.js
type NativeProps = $ReadOnly<{|
...
onChange?: ?(event: BubblingEvent<Event>) => mixed,
|}>;
```
However, it would be helpful to be able to reuse `NativeProps` in the user facing component:
```
// SegmentedControlIOS.js
type Props = $ReadOnly<{|
...NativeProps
onValueChange?: ?(value: number) => mixed,
|}>;
```
## Changes
- updates the `getTypes` function to unwrap the type declaration inside exported declarations
- add test to verify that exported types are parsed as expected
Reviewed By: rickhanlonii
Differential Revision: D15851693
fbshipit-source-id: a9cd375c69cbb8fe9a38be3d2a681227444fb33d
Summary: `Stringish` is commonly used in components to accept either strings or values coming from Fbt tags. The codegen should support them as they can be used transparently as strings.
Reviewed By: rickhanlonii
Differential Revision: D15821882
fbshipit-source-id: fb94f702d82820677c5a737325b7b46ff4c88151
Summary:
This diff reimplements the prop parsing infrastructure in a part where it interacts with RawProps value.
Local synthetic tests show that the new way is 3x faster but the actual production result is quite unpredictable. MobileLab tests show some improvements about 10-20 ms on iPhone 6.
In short, the new way is faster because it inverts the lookup order and heavily relies on actual data types (and their properties) that we use. The old approach required about 130 hash-map lookups (where the key is `std::string`) to parse a single *Props object.
The new approach prepares concrete-props-specific tables with indexes of coming values ahead of time, iterates over raw data and puts it into those tables, and then performs a lookup in a very efficient manner.
Reviewed By: JoshuaGross
Differential Revision: D15752968
fbshipit-source-id: 847106e652eb7fc7ef7b99884a6f819ea3b9fd06
Summary:
This diff adds codegen support for flow types such as:
```
type ModuleProps = $ReadOnly<{|
size: $ReadOnlyArray<('small', 'large')>
|}>
```
These array enums are codegen'd as bitmaps in c++
Reviewed By: sammy-SC
Differential Revision: D15766763
fbshipit-source-id: 8c55303fb3a0ab151eae2b441119ab078e2c5d3d
Summary:
This diff adds support for:
```
propName: WithDefault<string, null>,
```
It will throw if null is used for any other type like boolean
Reviewed By: TheSavior, cpojer
Differential Revision: D15748556
fbshipit-source-id: 925457ca1739bfad08e4776ecb47c0beb3acacf5
Summary:
This diff adds a testing screen dev route to the facebook app for testing generated view configs
It's not pretty (i have 0 tetra experiance) but it gets the job done
There are three cases handled:
- No generated config �
- Invalid generated config (useful for dev) �
- Valid generated config �
On the description page we:
- Redbox it it's invalid (this could be used to redbox test all host components)
- Show diffs of the view config properties
- List all of the generated config properties
- List all of the native config properties
Using this tool, it's easy to see what the current config on native is, add correct flow types for the generated config, and validate the generated config
Coming later: adding all of the native configs to the list (will probably need filtering)
Reviewed By: cpojer
Differential Revision: D15683033
fbshipit-source-id: 5a566a56bef4f3f0bac3ea581c2e6acb2b9984e3
Summary:
Now that we have the babel plugin, we can remove the checked in view configs
Note that this requires switching the old NativeComponent.js files back to `requireNativeComponent` for open source support (until the babel plugin and codegen are published to a package)
The babel plugin replaces this export with the inline view config
Reviewed By: cpojer
Differential Revision: D15524779
fbshipit-source-id: ab819ce6f24cb2f15a1897ed6d510a0db6aff3a1
Summary:
This diff updated the codegen flow types syntax replacing:
```
type Options = {
isDeprecatedPaperComponentNameRCT: true,
};
type ActivityIndicatorNativeType = CodegenNativeComponent<
'ActivityIndicatorView',
NativeProps,
Options,
>;
module.exports = ((requireNativeComponent(
'RCTActivityIndicatorView',
): any): ActivityIndicatorNativeType);
```
with:
```
export default codegenNativeComponent<NativeProps>('ActivityIndicatorView', {
isDeprecatedPaperComponentNameRCT: true,
});
```
This is from Tim's comment in the [View Config Codegen Quip](https://fb.quip.com/jR2aASHad4Se):
> What it CodegenNativeComponent were instead `NativeComponent.fromFlow<T>('…')` that returned `'...'`?
>And the Babel plugin swapped it for NativeComponent.fromSchema('...', {…}) which would both register and return '...'?
I went with `codegenNativeComponent` because it has nice parity with `requireNativeComponent`
I also didn't update the babel output here (we can update that whenever) because I think `registerGeneratedViewConfig` is more clear for what it's doing
Reviewed By: cpojer
Differential Revision: D15602077
fbshipit-source-id: 2d24dc32136ba6d31724f8c929b51417ba625a58
Summary:
This diff adds a babel plugin for the generated view configs which will inline them in the file instead of needing to check the view configs in (fb only)
The way it works is:
- babel reads the code
- looks for type alias `CodegenNativeComponent` in `*NativeComponet.js` files
- run the flow parser on the file source to create a schema
- run the schema into codegen to get the view config source code
- inject the generated source code back into the NativeComponent.js file
- remove the original export
- profit
After this diff we will remove the `js1 build viewconfigs` command and the checked-in NativeViewConfig.js files
Note: since this plugin is not published to open source, for now OSS will continue using the `requireNativeComponent` function
Reviewed By: cpojer
Differential Revision: D15516062
fbshipit-source-id: a8efb077773e04fd9753a7036682eeaae9175e09
Summary:
This diff updated the format of generated view configs so that they don't need to spread View props into every config, by adding a new registerGeneratedConfig function which will spread them instead
This is a bit of a cleanup of the generated output but is primarily so that the view config babel plugin will not need to rely on object spreading or object.assigns
Reviewed By: TheSavior, cpojer
Differential Revision: D15517199
fbshipit-source-id: 08e575578177bad12d40ee3dcad9381974b6466d
Summary: This diff adds support for ColorArrayValue in the flow parser
Reviewed By: cpojer
Differential Revision: D15502923
fbshipit-source-id: 6a906b6d609168378fabeb49d0080de011a34d78
Michał Osadnik
Ben Cheng
Samuel Susla
Rick Hanlon
Eli White
Cameron Bourke
Rubén Norte
Valentin Shergin
Sam Goldman