Andrew Datsenko

This commit **refactors** the **Yoga layout engine's JNI finalizer classes** within **React Native's Android implementation**, converting `YogaConfigJNIFinalizer` and `YogaNodeJNIFinalizer` from Java to Kotlin. This **code modernization** is part of the ongoing **Kotlinification effort**, improving consistency and maintainability in the `yoga` package. The conversion specifically migrates these two small classes, which handle native memory cleanup, and removes redundant `super.finalize()` calls that are not directly callable in Kotlin. This **maintenance** task ensures the codebase continues to evolve without altering existing functionality.

This commit **refactors** the **Fantom CI/CD workflow** by separating the build and test phases into distinct jobs to improve efficiency. A new **GitHub Action**, `build-fantom-runner`, is introduced to compile the Fantom runner binary and upload it as an artifact. The existing `run-fantom-tests` action is updated to download and utilize this pre-built binary, removing redundant build dependencies and ccache configurations from the test job. This **maintenance** effort, reflected in `test-all.yml` and `scripts/fantom.sh`, ensures the binary is built only once, leading to **faster CI execution** and more streamlined resource utilization for Fantom tests.

This commit introduces **inline snapshot testing** to **Fantom's custom `expect` implementation**, mirroring Jest's `toMatchInlineSnapshot()` functionality. On the first test run, snapshot values are automatically written directly into the test source file as template literals; subsequent runs compare values and fail on mismatch unless an update is forced. This **new capability** spans Fantom's two-process architecture, with the **runtime (Hermes VM)** handling snapshot comparison and result tracking, and the **runner (Node.js)** managing source map symbolication, direct source file rewriting, and result persistence. This significantly enhances the developer experience for testing by providing a convenient way to manage and update snapshot expectations directly within test files.

Add LLVM coverage conversion to LCOV format (#55860)

This commit introduces a **new capability** to collect **C++ code coverage** for the **Fantom project**, significantly enhancing its testing infrastructure. It involves **refactoring** the **Fantom build and test runner** (`hermesc.js`, `tester.js`, `runner.js`) to accept and propagate coverage-related flags and environment variables, such as `LLVM_PROFILE_FILE`, during test execution. A new module, `processLLVMCoverage.js`, is added to handle the merging of `profraw` files and exporting of LLVM C++ coverage reports. This comprehensive update allows developers to gain deeper insights into the test coverage of Fantom's C++ components, improving code quality and test effectiveness.

This commit **enhances test coverage** for the **`getBoundingClientRect` API** within the **`react-native` web APIs DOM implementation**. It introduces new integration tests for the **`Text` component**, specifically verifying its behavior with this layout API. The tests confirm that standalone `Text` elements return valid `DOMRect` dimensions, while nested `Text` elements (virtual text) correctly yield zero dimensions, reflecting their lack of independent layout. This **maintenance** work ensures the **reliability and expected behavior** of `getBoundingClientRect` for `Text` components, preventing regressions in layout calculations.

This commit **enables render debug strings** within the **React Native renderer** when operating in "opt mode". It introduces a **new capability** by modifying the preprocessor directive in `packages/react-native/ReactCommon/react/renderer/debug/flags.h` to activate `RN_DEBUG_STRING_CONVERTIBLE` when `RN_ENABLE_DEBUG_STRING_CONVERTIBLE` is defined. This change is crucial for internal testing, allowing the team to achieve better code coverage by utilizing debug string functionality in environments where "opt mode" is the only available option. Ultimately, this enhances the **debuggability and testability of the React Native rendering pipeline**.

This commit performs **maintenance cleanup** by **removing an outdated and unused script** from the project's **tooling infrastructure**. This action streamlines the codebase, eliminating unnecessary files that no longer serve a purpose, thereby improving overall project hygiene. The change has no functional impact on the application or its features, as the script was already defunct, but it reduces clutter and potential confusion for developers.

This commit delivers a **bug fix** addressing an `EXC_BAD_ACCESS` crash within the **Animated module** on **Apple platforms**. The crash occurred because C++ lambdas in `packages/react-native/ReactApple/RCTAnimatedModuleProvider/RCTAnimatedModuleProvider.mm`, specifically around the `getModule` symbol, were directly capturing Objective-C `self` without retaining it, leading to premature deallocation. Consequently, when these lambdas were invoked later, accessing instance variables like `_onRender` resulted in a crash. The fix **implements weak/strong self capturing** within these C++ lambdas, ensuring `self` is safely retained for the duration of the lambda's execution. This **prevents animation-related crashes** and significantly improves the **stability** of animated components in React Native on iOS and macOS.

This commit introduces a **performance improvement** for **Android** by implementing **caching for `PlatformColor` resolution**, significantly reducing the number of JNI calls. It adds an `EvictingCacheMap` to the `PlatformColorParser` in `ReactCommon` to store resolved colors, along with a mechanism to invalidate this cache. The `AppearanceModule` in `ReactAndroid` is updated to call a new `invalidatePlatformColorCache` method whenever the device's appearance settings change, ensuring cache consistency. This **optimization** directly benefits Android applications by making `PlatformColor` resolution more efficient and responsive.

Add Platform Color example (#53908)

This commit introduces a **new capability** to **conditionally disable code coverage collection** within the **Fantom test runner**. It adds a `fantom_disable_coverage` pragma and integrates logic in `coverageUtils.js` and `runner.js` to prevent coverage from being collected for specific tests. This change primarily impacts **benchmark tests**, where coverage is now automatically disabled, and **memory-sensitive tests** like `MemoryBaseline-itest.js`, which can explicitly opt out using the new pragma. The goal is to **improve test performance** and ensure more accurate results for tests sensitive to overhead.

This commit **fixes** an issue with **code coverage patterns** within the **Metro Babel transformer configuration**. It adjusts the `include` and `exclude` patterns for `babel-plugin-istanbul` in `private/react-native-fantom/config/metro-babel-transformer.flow.js` by adding a glob prefix. This **maintenance** change ensures that code coverage is correctly inserted and reported, particularly when running in diverse Metro setups where pattern matching might otherwise be inconsistent. The **testing infrastructure** will now provide more reliable and accurate coverage data across different environments.

This commit delivers a **bug fix** addressing an issue where command discovery failed due to a conflict with "dotslash" in `node_modules`, specifically preventing `buck2` invocation. It **updates the `PATH` environment variable** within the **React Native Fantom runner utility functions** `runCommand` and `runCommandSync` in `private/react-native-fantom/runner/utils.js`. By explicitly including `/usr/bin` in the `PATH`, this change ensures that necessary system commands are correctly located and executed. This resolves environment-related failures and improves the reliability of the React Native build and runtime processes.

This commit introduces **JavaScript code coverage reporting** by integrating `babel-plugin-istanbul` into the **Metro bundler's Babel transformer**. It enables conditional code instrumentation during bundling when a `collectCoverage` flag is active, allowing the **test runner** to collect and report coverage data. The `runner.js` now passes this flag to `createBundle` and the `runtime/setup.js` updates `TestSuiteResult` to include the `coverageMap`. This **new capability** provides developers with essential metrics for assessing test completeness and code quality.

This commit introduces a **fix** to the **ReactPerfettoLogger** by modifying the `logWebPerfEvent` function. It now **ignores web performance events** that have start or end timestamps occurring before the program's initialization. This change addresses an issue where artificial `console.timeStamp` marks, used for web track ordering, were incorrectly included in **Perfetto traces**, even though they were ignored by RNDT on Chrome. By filtering these unusually early timestamps, the commit significantly **improves the accuracy of performance data** collected for React Native applications.

This commit **integrates `babel-plugin-istanbul`** into the project's build system, a crucial **internal tooling improvement**. This addition enables the collection of **code coverage metrics**, providing essential insights into the thoroughness of existing tests and highlighting areas of the codebase that may lack sufficient test coverage. The change primarily affects the **development environment and testing infrastructure**, enhancing **quality assurance processes** by facilitating more comprehensive code analysis.

This commit **adds support for Perfetto tracing on Windows**, significantly enhancing performance analysis capabilities for React Native applications on that platform. It involves **refactoring** the `ReactCommon/cxxreact` tracing infrastructure by adding preprocessor directives to prevent macro clashes with `perfetto.h` during Windows compilation. Furthermore, Android-specific logging and file writing logic within `reactperflogger/FuseboxPerfettoDataSource.cpp` are now encapsulated using platform-specific preprocessor directives, ensuring proper behavior and enabling this **new capability** across different operating systems. This work improves the cross-platform compatibility of the tracing system.

This commit **refactors** and **generalizes accessibility testing** within React Native by introducing a new, reusable `accessibilityPropsSuite` utility. Existing accessibility tests for **`Pressable`**, **`TouchableWithoutFeedback`**, and **`Image`** components have been updated to leverage this shared suite, improving test consistency and maintainability. This **new capability** streamlines the process of verifying accessibility properties across the component library. Additionally, a `toJSONObject` method was added to `FantomRenderedOutput` to enhance the structured output of rendered components, likely aiding in test assertions. This change significantly improves the efficiency and robustness of the **accessibility testing infrastructure**.

This commit **introduces a new debug prop for `accessibilityState`** within the **React Native renderer's accessibility system**. It adds utility functions in `AccessibilityPrimitives.h` to convert `AccessibilityState` and its `CheckedState` enum into debug strings, and includes `accessibilityState` in the debug properties via `AccessibilityProps.cpp`. This **new capability** enhances **developer tooling** by allowing easier inspection of accessibility states, which is reflected in updated integration tests for `Pressable` and `Image` components. This change will aid in **debugging and developing more accessible React Native applications**.