[path_provider_android] Changes internal implementation to use JNI

[bparrishMines]

https://developer.android.com/guide/components/processes-and-threads

Pre-Review Checklist

• I read the Contributor Guide and followed the process outlined there for submitting PRs.
• I read the Tree Hygiene page, which explains my responsibilities.
• I read and followed the relevant style guides and ran the auto-formatter.
• I signed the CLA.
• The title of the PR starts with the name of the package surrounded by square brackets, e.g. [shared_preferences]
• I linked to at least one issue that this PR fixes in the description above.
• I updated pubspec.yaml with an appropriate new version according to the pub versioning philosophy, or I have commented below to indicate which version change exemption this PR falls under¹.
• I updated CHANGELOG.md to add a description of the change, following repository CHANGELOG style, or I have commented below to indicate which CHANGELOG exemption this PR falls under¹.
• I updated/added any relevant documentation (doc comments with ///).
• I added new tests to check the change I am making, or I have commented below to indicate which test exemption this PR falls under¹.
• All existing and new tests are passing.

If you need help, consider asking for advice on the #hackers-new channel on Discord.

Note: The Flutter team is currently trialing the use of Gemini Code Assist for GitHub. Comments from the gemini-code-assist bot should not be taken as authoritative feedback from the Flutter team. If you find its comments useful you can update your code accordingly, but if you are unsure or disagree with the feedback, please feel free to wait for a Flutter team member's review for guidance on which automated comments should be addressed.

Footnotes

1. Regular contributors who have demonstrated familiarity with the repository guidelines only need to comment if the PR is not auto-exempted by repo tooling. ↩ ↩² ↩³

[gemini-code-assist]

Code Review

This pull request refactors the Android implementation of path_provider to use jnigen and JNI instead of Pigeon for platform communication. This is a significant architectural change. While the implementation looks mostly correct, there are several potential memory leaks due to unreleased JNI objects. I've added specific comments on how to fix them.

A major concern is the removal of all existing tests without adding new ones to cover the new JNI-based implementation. The repository's style guide states that "Code should be tested"[^1]. Given the low-level nature of JNI interop, having a solid test suite is critical to ensure correctness and prevent regressions. Please add tests for the new implementation.

[gemini-code-assist]

The JArray<File?> object returned by getExternalCacheDirs() is a JNI object that needs to be released to avoid a memory leak. The current implementation does not release it.

    final JArray<File?>? files = _applicationContext.getExternalCacheDirs();
    if (files == null) {
      return null;
    }
    try {
      return _toStringList(files);
    } finally {
      files.release();
    }

[gemini-code-assist]

The JArray<File?> object returned by getExternalFilesDirs() is a JNI object that needs to be released to avoid a memory leak. The current implementation does not release it.

    final JArray<File?>? files = _applicationContext.getExternalFilesDirs(
      type != null ? _toNativeStorageDirectory(type) : null,
    );
    if (files == null) {
      return null;
    }
    try {
      return _toStringList(files);
    } finally {
      files.release();
    }

[gemini-code-assist]

There are two potential memory leaks here:

1. The iterator returned by files.iterator is a JIterator, which is a JNI object that must be released.
2. If file.getAbsolutePath() or toDartString() throws an exception, file.release() will not be called.

Both can be fixed by using try...finally blocks to ensure release() is always called.

  final Iterator<File?> filesIterator = files.iterator;
  try {
    while (filesIterator.moveNext()) {
      final File? file = filesIterator.current;
      if (file != null) {
        try {
          final String? path =
              file.getAbsolutePath()?.toDartString(releaseOriginal: true);
          if (path != null) {
            paths.add(path);
          }
        } finally {
          file.release();
        }
      }
    }
  } finally {
    (filesIterator as JObject).release();
  }

[gemini-code-assist]

To ensure that the File object is always released, even if getPath() or toDartString() throws an exception, it's safer to use a try...finally block.

    final File? file = _applicationContext.getCacheDir();
    if (file == null) {
      return null;
    }
    try {
      return file.getPath()?.toDartString(releaseOriginal: true);
    } finally {
      file.release();
    }

[gemini-code-assist]

To ensure that the File object (dir) is always released, even if an exception occurs, it's best practice to use a try...finally block.

    final File? dir = _applicationContext.getExternalFilesDir(null);
    if (dir == null) {
      return null;
    }
    try {
      return dir.getAbsolutePath()?.toDartString(releaseOriginal: true);
    } finally {
      dir.release();
    }

[bparrishMines]

cc @stuartmorgan-g for early review

[bparrishMines]

@HosseinYousefi Have you seen this error on Windows for JNI:

Changed 65 dependencies!
10 packages have newer versions incompatible with dependency constraints.
Try `flutter pub outdated` for more information.
Building Windows application...                                 
Nuget.exe not found, trying to download or use cached version.
CMake Error at C:/Program Files (x86)/Microsoft Visual Studio/2019/Community/Common7/IDE/CommonExtensions/Microsoft/CMake/CMake/share/cmake-3.19/Modules/FindPackageHandleStandardArgs.cmake:218 (message):
  Could NOT find JNI (missing: JAVA_AWT_LIBRARY JAVA_JVM_LIBRARY
  JAVA_INCLUDE_PATH JAVA_INCLUDE_PATH2 JAVA_AWT_INCLUDE_PATH)
Call Stack (most recent call first):
  C:/Program Files (x86)/Microsoft Visual Studio/2019/Community/Common7/IDE/CommonExtensions/Microsoft/CMake/CMake/share/cmake-3.19/Modules/FindPackageHandleStandardArgs.cmake:582 (_FPHSA_FAILURE_MESSAGE)
  C:/Program Files (x86)/Microsoft Visual Studio/2019/Community/Common7/IDE/CommonExtensions/Microsoft/CMake/CMake/share/cmake-3.19/Modules/FindJNI.cmake:382 (FIND_PACKAGE_HANDLE_STANDARD_ARGS)
  flutter/ephemeral/.plugin_symlinks/jni/src/CMakeLists.txt:32 (find_package)


Building Windows application...                                    11.5s
Unable to generate build files

[HosseinYousefi]

This is similar to getsentry/sentry-dart#3033

So I write what I wrote there here as well:

@HosseinYousefi maybe you can confirm

issue: we're using JNI on Android but also means Flutter Desktop targets are affected due to flutter's plugin system. we don't use any JNI api during runtime on Desktop -> will end users need java installed in that case on Desktop using the app?

End users don't need to have Java installed as long as we're not using JNI on Windows (which I assume we're not here).

This is an issue with Flutter's plugin system. You could technically spawn JVM and use package:jni on Windows. This will be solved once we use the newly introduced build hooks for building package:jni's native library instead. Sentry could pass a parameter to specify that it's only using JNI on android and not on desktop.

Originally posted by @HosseinYousefi in #3033

I can fix this by making the windows build show a warning instead of an error if it couldn't find JDK.

[stuartmorgan-g]

Does jnigen support include-based filtering yet? It was added to jnigen, and instead of a generated file that was huge, like the 12k lines in this PR, it's only about 600.

[stuartmorgan-g]

This seems to be blocked on dart-lang/native#2962.

[stuartmorgan-g]

Oops, I lost a comment because GitHub won't let me save a comment on the deleted file for reason. I had this on the deleted Dart unit tests:

I would expect these tests to be moved to the integration test runner rather than removed, so that we are still unit testing how we handle specific responses to from the system layer.

[Piinks]

Greetings from stale PR triage! 👋
Is this change still on your radar?

[stuartmorgan-g]

The PR was blocked by jni breaking Windows, but I believe jni 0.15.2 resolves the blocker.

[stuartmorgan-g]

I pushed a bunch of changes:

• Merged in latest master
• Updated to latest jni/jnigen
• Switched to the new Dart API generator approach instead of the config yaml
• Removed all of the native plugin code (the entire android/ directory)
• Added a minimal Dart unit test that registration works, so that we still have the Dart unit test structure set up.
• Added a facade with a stub export for web, to avoid compilation errors from transitive includes, as with path_provider_foundation.
• Minor style changes.

Currently we're gated on jni moving to the main publisher (although we could decide to change that policy; nothing would actually prevent us from shipping with the labs dependency), and ideally on dart-lang/native#2962 allowing us to do filtering, but that shouldn't matter much in practice since the analyzer handles the current file reasonably, and tree shaking should prevent most of the code from ever ending up in a build.