A VS Code extension for inspecting Parquet files below the "just show me a data grid" level — schema, row-group and column-chunk layout, encoding, compression, statistics, dictionary/bloom-filter presence, and common anti-patterns (uncompacted row groups, missing dictionary/bloom coverage, encodings that vary across row groups).
Parquet reading is powered by Rugo,
opteryx-core's Parquet/file engine. The core is not reimplemented here: it's
a pinned, vendored copy of Rugo's Python-free C++ metadata reader
(rugo/src/parquet), compiled to WASM with no Cython/nanobind/libpython
involved. See native/README.md.
Real metadata reads (schema, row groups, column-chunk stats, bloom-filter presence) working end to end against real files.
npm install
npm run build:native # requires emcc (brew install emscripten)
npm run watch # esbuild watch mode
Then launch the "Run Extension" debug target in VS Code (F5).
Apache-2.0. Includes vendored Apache-2.0 code from opteryx-core/Rugo — see
native/README.md for provenance.