Android check pte exists

[kirklandsign]

Raise RuntimeException if not.

cc @cbilgin

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/10931

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

✅ You can merge normally! (1 Unrelated Failure)

As of commit 905ccd0 with merge base 24789c8 ():

BROKEN TRUNK - The following job failed but were present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

• pull / test-llava-runner-linux / linux-job (gh) (trunk failure)
  test_llava_export

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[facebook-github-bot]

@kirklandsign has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

[facebook-github-bot]

@kirklandsign has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

[Copilot]

Pull Request Overview

Adds explicit file existence and readability checks for the model and tokenizer paths in both LlmModule and Module.load, and centralizes constructor logic in LlmModule by delegating to a primary constructor.

• Introduces a primary LlmModule constructor that validates file paths before invoking initHybrid.
• Refactors all other LlmModule constructors to delegate to this primary constructor.
• Adds a file validation check in Module.load to throw a RuntimeException if the model file is missing or unreadable.

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
extension/android/executorch_android/src/main/java/org/pytorch/executorch/extension/llm/LlmModule.java	Added file existence/readability checks and refactored constructors to delegate to a single primary constructor.
extension/android/executorch_android/src/main/java/org/pytorch/executorch/Module.java	Added model file validation in load(...) before initializing the native peer.

Comments suppressed due to low confidence (4)

extension/android/executorch_android/src/main/java/org/pytorch/executorch/extension/llm/LlmModule.java:53

• There are no tests covering the scenario where the model or tokenizer file is invalid; adding unit tests for these exception paths would improve test coverage.

throw new RuntimeException("Cannot load model path " + modulePath);

extension/android/executorch_android/src/main/java/org/pytorch/executorch/Module.java:58

• No tests currently exercise the failure path when the model file is unreadable or missing; adding tests for this case would enhance reliability.

throw new RuntimeException("Cannot load model path " + modelPath);

extension/android/executorch_android/src/main/java/org/pytorch/executorch/extension/llm/LlmModule.java:53

• [nitpick] Throwing a generic RuntimeException makes it harder to handle specific error cases; consider using FileNotFoundException or IllegalArgumentException for clearer intent.

throw new RuntimeException("Cannot load model path " + modulePath);

extension/android/executorch_android/src/main/java/org/pytorch/executorch/Module.java:58

• [nitpick] Using a generic RuntimeException for file load failures may be too broad; consider a more specific exception type such as FileNotFoundException to signal the root cause.

throw new RuntimeException("Cannot load model path " + modelPath);

[Copilot]

The file existence and readability check is duplicated in Module.java; consider extracting this into a shared utility method to reduce duplication.

Suggested change

	File modelFile = new File(modulePath);
	if (!modelFile.canRead() || !modelFile.isFile()) {
	throw new RuntimeException("Cannot load model path " + modulePath);
	}
	File tokenizerFile = new File(tokenizerPath);
	if (!tokenizerFile.canRead() || !tokenizerFile.isFile()) {
	throw new RuntimeException("Cannot load tokenizer path " + tokenizerPath);
	}
	validateFile(modulePath, "model path");
	validateFile(tokenizerPath, "tokenizer path");

[Copilot]

Duplicate file validation logic appears here and in LlmModule.java; extracting a shared helper method could improve maintainability.

[facebook-github-bot]

@kirklandsign has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.