[neptune]FEAT: new module that reconstructs diff function from mono and libra

[AidCheng]

This pull request refactors how diffs are generated and handled across the codebase, moving from an internal implementation in libra to a unified diff engine provided by the new neptune crate.

Diff engine refactor and dependency updates:

• Replaces the internal diff logic in libra and ceres with the Diff engine from the new neptune crate, simplifying the code and ensuring consistency. The diff computation is now done in-memory and returned as a string, removing the need for external process calls and file system operations (ceres/src/api_service/mono_api_service.rs, libra/src/command/diff.rs). [1] [2] [3] [4] [5]

• Adds the neptune crate as a dependency and workspace member in Cargo.toml files for the root project, ceres, and libra, enabling use of the new diff engine throughout the workspace (Cargo.toml, ceres/Cargo.toml, libra/Cargo.toml). [1] [2] [3] [4]

Dependency cleanup and configuration:

• Removes unused or redundant dependencies such as imara-diff, infer, tokio-test, tracing-test, color-backtrace, git2, and home from the workspace and project Cargo.toml files, streamlining the dependency tree (Cargo.toml, libra/Cargo.toml). [1] [2] [3] [4] [5]

• Updates the axum dependency to enable the "macros" and "json" features for improved API ergonomics (Cargo.toml).

Code simplification and documentation:

• Removes complex subprocess logic and file system operations from the diff generation path in ceres, making the process more robust and easier to maintain. The new diff function is documented and leverages pre-fetched blob data for efficiency (ceres/src/api_service/mono_api_service.rs).

• Minor typo fixes and code cleanup for clarity and maintainability (ceres/src/api_service/mono_api_service.rs).

These changes modernize the diff infrastructure, reduce maintenance overhead, and improve reliability and performance for diff-related operations.

Closes Issue #1298

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
mega	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Aug 6, 2025 11:01am

[Copilot]

Pull Request Overview

This PR introduces a new neptune crate that provides a unified diff engine, replacing the existing diff implementations in both libra and ceres with a centralized solution.

• Adds the neptune crate with a Diff engine that handles in-memory diff computation
• Refactors libra and ceres to use the new diff engine instead of external process calls
• Removes unused dependencies and streamlines the codebase

Reviewed Changes

Copilot reviewed 10 out of 10 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
neptune/src/neptune_engine.rs	Implements the core diff engine with unified diff generation
neptune/src/lib.rs	Library interface exposing the Diff struct
neptune/Cargo.toml	Package configuration for the neptune crate
libra/src/command/diff.rs	Refactored to use neptune's diff engine instead of imara-diff
ceres/src/api_service/mono_api_service.rs	Replaced subprocess-based diff with neptune's in-memory diff
Cargo.toml files	Added neptune as workspace member and dependency

Comments suppressed due to low confidence (1)

libra/src/command/diff.rs:129

• The variable file shadows the parameter file, which reduces code clarity. Consider using a different name like workdir_file or absolute_file.

                let file = to_workdir_path(file);

[Copilot]

The fallback hash "0000000" should be "0000000" (7 characters) to match the 8-character truncation pattern used for valid hashes. This inconsistency could cause formatting issues in git diff output.

Suggested change

	.unwrap_or_else(|| "0000000".into());
	let new_index = new_hash
	.map(|h| h.to_string()[0..8].to_string())
	.unwrap_or_else(|| "0000000".into());
	.unwrap_or_else(|| "00000000".into());
	let new_index = new_hash
	.map(|h| h.to_string()[0..8].to_string())
	.unwrap_or_else(|| "00000000".into());

[Copilot]

The fallback hash "0000000" should be "0000000" (7 characters) to match the 8-character truncation pattern used for valid hashes. This inconsistency could cause formatting issues in git diff output.

Suggested change

	.unwrap_or_else(|| "0000000".into());
	let new_index = new_hash
	.map(|h| h.to_string()[0..8].to_string())
	.unwrap_or_else(|| "0000000".into());
	.unwrap_or_else(|| "00000000".into());
	let new_index = new_hash
	.map(|h| h.to_string()[0..8].to_string())
	.unwrap_or_else(|| "00000000".into());

[Copilot]

The algorithm parameter is ignored (prefixed with underscore) but the function documentation mentions support for different algorithms. Either implement algorithm selection or remove the parameter and update the documentation.

Suggested change

_algorithm: &str,

[Copilot]

[nitpick] Unnecessary parentheses around PathBuf import. Should be path::PathBuf for consistency with Rust conventions.

Suggested change

	path::{PathBuf},
	path::PathBuf,

[Copilot]

The struct comment refers to DiffEngine but the actual struct is named Diff. The documentation should be updated to match the struct name.

Suggested change

	/// `DiffEngine` provides static methods to compare files between two states (old and new)
	/// `Diff` provides static methods to compare files between two states (old and new)

[Copilot]

[nitpick] The function documentation is incomplete. It should include error conditions and examples of the diff output format for better API usability.

Suggested change

	/// Generate diff content directly from two blob sets using the diff engine
	///
	/// # Arguments
	/// * `mr_link` - Merge request link identifier
	///
	/// # Returns
	/// String containing the unified diff output
	/// Generate diff content directly from two blob sets using the diff engine.
	///
	/// # Arguments
	/// * `mr_link` - Merge request link identifier.
	///
	/// # Returns
	/// Returns a `Result<String, GitError>`, where the `String` contains the unified diff output.
	///
	/// # Error Conditions
	/// Returns an `Err(GitError)` in the following cases:
	/// - The merge request with the given `mr_link` does not exist.
	/// - There is a storage backend error while retrieving the merge request or blobs.
	/// - The diff computation fails due to invalid or missing blob data.
	///
	/// # Diff Output Format Example
	/// The returned string is a unified diff, similar to the output of `git diff`. For example:
	/// ```
	/// --- a/file.txt
	/// +++ b/file.txt
	/// @@ -1,3 +1,4 @@
	/// line 1
	/// +added line
	/// line 2
	/// line 3
	/// ```
	/// Each diff hunk shows the changes between the two blob sets.