MoonFrame

A lightweight DataFrame and tabular-data library for the MoonBit ecosystem.

MoonFrame provides column-oriented DataFrame / Series types together with CSV reading, filtering, sorting, null handling, summary statistics, and Markdown / JSON export. The goal is not to clone every pandas feature, but to give MoonBit a small, well-tested, extensible foundation for data analysis.

Status

Under active development toward v0.1. The current focus is the types / column / frame / ops / io layering described in docs/api.md. See that document for the authoritative public-API surface.

v0.1 progress:

• P0 — project skeleton
• P1 — types/ core (DataError, DataType, Scalar)
• P2 — column/ BuiltinColumn (constructors, accessors, slice /

  take, casts, typed zero-boxing accessors)

• P3 — types/ Field + Schema (duplicate-name detection,

  select, rename, index_of)

• P4 — frame/ Series + stats
• P5 — frame/ DataFrame + RowView
• P6 – P10 — ops/
• P11 – P12 — io/ CSV / Markdown / JSON
• P13 — facade re-exports, integration, examples

GroupBy and Join land in v0.2; NDJSON also v0.2; HTML and chart-data export in v0.3; an expression / lazy query layer in v0.4.

Layout

moonframe/      facade package — re-exports the public API
types/          value types, errors, schemas
column/         column storage backends
frame/          Series, DataFrame, RowView
ops/            select / filter / sort / null / stats / describe
io/             CSV (NyaCSV-backed), Markdown, JSON
docs/api.md     public API reference (source of truth)

Each subpackage carries its sources, its *_test.mbt blackbox tests, and its pkg.generated.mbti interface snapshot.

Building

moon check     # type-check the workspace
moon test      # run all blackbox tests
moon info      # regenerate per-package .mbti interfaces
moon fmt       # format sources

Dependencies

• xunyoyo/NyaCSV — CSV parser
• moonbitlang/x — extra standard-library utilities (@fs for filesystem I/O)

License

Apache-2.0 — see LICENSE.