solcore-rs

solcore-rs is a Rust implementation of Solcore.

Warning

This project is a work in progress and is not ready for production use.

Try it in the online playground.

Language and compatibility targets

The source-language and standard-library target is the Haskell implementation at argotorg/solcore@a7b7afb0. Both implementations now use the canonical .sol source syntax. The eight library sources in std/ come from that revision, with local whitespace-only formatting applied by solcore-fmt.

See SEMANTIC_DIFFERENCES.md for intentional corrections to upstream defects, ABI limitations, and validation scope. Rust retains structured diagnostics and safety checks; compatibility does not imply identical compiler output or production readiness. The 499-source parser corpus remains a historical 2f372bde regression set, as documented in its README.

Build and test

Rust is required. The checked-in rust-toolchain.toml selects Rust 1.97.0 automatically when using rustup.

cargo build --workspace --locked
cargo test --workspace --locked

Formatting Solcore

The standalone solcore-fmt command formats UTF-8 Solcore source files:

# Print formatted source without changing the file.
cargo run -p solcore-fmt -- main.sol

# Read stdin and print the result.
cargo run -p solcore-fmt -- < main.sol

# Format one or more files in place.
cargo run -p solcore-fmt -- --write main.sol library.sol

# Check formatting in CI without changing files.
cargo run -p solcore-fmt -- --check main.sol

# Install the standalone command locally.
cargo install --path crates/fmt

Use solcore-fmt --help for all options. Indentation defaults to four spaces; --indent-width N changes it (1–16 spaces). A - input reads stdin. Multiple inputs are printed in argument order unless --write or --check is selected. Inputs are individual files; directories are not searched recursively. --check exits with status 1 when formatting differs; invalid syntax, arguments, or I/O errors exit with status 2.

Formatting reuses the existing lexer and parser through solcore_parser::format::format_source. It changes whitespace while retaining the original token, literal, and comment text, and verifies that the token and comment sequence is unchanged. Syntax errors prevent formatting. The initial style preserves explicit continuation line breaks and existing CRLF line endings; automatic line wrapping and a configurable maximum line width are not implemented yet.

The native and browser language servers use the same formatter for document formatting, respecting the editor's indentation and final-newline preferences. Documents with syntax errors are left unchanged in the editor.

The existing CI tests also require Solcore test fixtures, E2E sources, fuzz seeds, playground examples, and the standard library to be formatted. Only unparseable fixtures in the designated syntax-error test directories are exempt. Run this check locally with cargo test -p solcore-parser --test formatting --locked.

Before --write replaces any file, every input is read and formatted, and all changed results are staged in the destination directories. Individual files are replaced atomically with their permissions preserved. Symlinks and read-only files are rejected. A later filesystem failure during replacement can still leave earlier files formatted; this is not a multi-file transaction.

License

Licensed under the Apache License 2.0.