ScriptMark

Automated grading CLI for student programming assignments. Rust core, TOML test specifications, no pytest dependency.

Features

• Custom test engine -- runs student code in subprocess, no pytest/unittest required
• TOML-based specs -- declarative test cases with expected values, checkers, and parametrization
• 8 built-in checkers -- exact, approx, sorted, set_eq, contains, regex, text, plus Rhai expressions and external Python checkers
• Parallel execution -- tokio-based orchestrator grades all students concurrently
• Sandboxed execution -- env isolation, import allowlist, resource limits (setrlimit), timeout with process kill
• Parametrize + oracle -- auto-generate test cases with random inputs and teacher reference implementations
• Canvas LMS integration -- pull rosters, push grades
• Similarity detection -- style + structural code similarity scoring
• Interactive TUI -- browse students, sessions, and similarity in the terminal
• HTML reports -- standalone dashboard with per-student breakdowns
• SQLite storage -- persist grading history across sessions

Installation

From source (recommended for now)

git clone https://github.com/Acture/scriptmark.git
cd scriptmark
cargo install --path crates/scriptmark-cli

From crates.io (once published)

cargo install scriptmark-cli

Requires python3 on PATH for running student Python code.

Quick Start

1. Write a TOML test spec:

[meta]
name = "find_max"
file = "lab5.py"
function = "find_max"
language = "python"

[[cases]]
name = "basic"
args = [[3, 1, 5, 2]]
expect = 5

[[cases]]
name = "negative"
args = [[-3, -1, -5]]
expect = -1

[[cases]]
name = "random inputs"
[cases.parametrize]
count = 20
seed = 42
[cases.parametrize.args]
nums = "list(int(-100, 100), 5, 20)"
[cases.parametrize.oracle]
rhai = "nums.sort(); nums[nums.len() - 1]"

2. Grade submissions:

scriptmark grade submissions/ -t tests/

TOML Spec Format

Each .toml file defines tests for one function/file:

Checkers

CLI Commands

scriptmark grade      Run tests + summarize + display grades
scriptmark run        Run tests only, output JSON
scriptmark summarize  Re-analyze existing results
scriptmark similarity Detect code similarity between submissions
scriptmark report     Generate HTML report
scriptmark db         Database management (init, import-roster, sessions, history)
scriptmark tui        Interactive terminal UI
scriptmark roster-pull  Pull roster from Canvas LMS
scriptmark grades-push  Push grades to Canvas LMS

Architecture

8-crate Rust workspace:

scriptmark-core       Models, TOML parsing, discovery, grading, similarity
scriptmark-runner     PythonExecutor, Checker trait, orchestrator, parametrize, oracle
scriptmark-cli        clap CLI with 9 commands + HTML report template
scriptmark-db         SQLite persistence (rusqlite, bundled)
scriptmark-canvas     Canvas LMS API client (reqwest)
scriptmark-tui        ratatui terminal UI
scriptmark-py         PyO3 Python bindings (in development)

License

GPL-3.0-or-later