MoonTextGen

MoonTextGen is a lightweight MoonBit text synthesis engine built around deterministic n-gram Markov statistics.

It is designed for:

• placeholder text generation
• gentle text obfuscation
• repeatable test data synthesis
• a clean, extensible MoonBit codebase for the OSC2026 competition

Repository Links

• GitLink: https://gitlink.org.cn/xkmxkm68/moontextgen
• GitHub: https://github.com/xkmxkm68/MoonTextGen

Project Name

MoonTextGen

Project Identifier

moontextgen

Project Intro

MoonTextGen turns a small set of curated corpus snippets into a deterministic text model. It learns local token transitions, then samples from those transitions to produce readable, reproducible output. The implementation stays intentionally small and dependency-free so it can be reviewed, extended, and adapted into more advanced text tooling later.

Creator

• Author and sole contributor: xkmxkm68

Features

• 2-gram and 3-gram model support
• deterministic generation from a seed
• preset profiles for mixed English, security-flavored text, and Chinese placeholder output
• compact MoonBit-only implementation
• testable pure core, with a thin CLI wrapper

Usage

moon test
moon run cmd/main
moon run cmd/main chinese
moon run cmd/main security

Repository Layout

• moontextgen.mbt core model and generation logic
• profiles.mbt built-in corpus profiles
• cmd/main/main.mbt CLI entry point
• moontextgen_wbtest.mbt regression tests
• docs/osc2026-self-check.md competition self-check notes
• LICENSE Apache-2.0

Source Notes

• The corpora in this repository are handcrafted for the project.
• No external copyrighted corpus is bundled.
• The implementation is intentionally small so it can grow into a broader text-processing toolkit later.
• The project history is intentionally incremental so reviewers can inspect the evolution of the implementation.

License

Apache-2.0