dta-tools

dta-tools provides TypeScript and R libraries for working with Stata .dta files. Both readers cover Stata 5 through 19 and preserve labels, long strings, display formats, and Stata missing values. The R package also writes Stata 18/19 .dta files. Its Arrow-based .arrow format preserves supported R column classes alongside Stata storage types and metadata in the same data frame. The R package also supplies Stata-aware metadata, storage, recoding, tabulation, summary statistics, merge, and data-signature operations.

Choose a library

Language	Package	Start here
TypeScript	@jbearak/dta-parser, DTA and Arrow IPC readers	npm package README
R	dtatools	R package README

The TypeScript package has its own parser and works with either an ArrayBuffer or a Node filesystem-backed reader.

For Stata imports in R, dtatools::read_dta() follows haven's common read interface and returns dibbles, tibbles, or data tables with haven-compatible labels and tagged missing values. In the September 16, 2026 corpus run, its multicore reader completed the 641-file, 46.9 GB DHS corpus in 38.7 seconds, about 70 times as fast as the recorded haven comparison. In the ten-read India comparison, median read time for the 5.2 GB file was well under a second, versus several minutes with haven. See the R package README for benchmarks and methods.

The Rust crate is the internal read/write core used by the R package. It is not published to crates.io. Its interface is documented with Rustdoc:

cargo doc -p dta-tools --no-deps --open

Compatibility

Install the R package from the dtatools package repository:

install.packages("dtatools", repos = c(
  dtatools = "https://jbearak.github.io/dta-parser",
  CRAN = "https://cloud.r-project.org"
))

See installation and publishing details for binary support, source requirements, and automatic release updates.

Package-owned underscore identifiers use dta_ and _dta. See the naming migration for renamed classes, constants, and metadata fields.

The TypeScript and Rust readers follow the same compatibility contract and are checked against shared fixtures. See DTA compatibility for supported format releases, text encodings, missing values, and language-specific result shapes.

Project documentation

• Contributing covers repository layout, development, testing, conformance, and releases.
• Benchmarks covers methodology and links to dated TypeScript, Rust, R, haven, and Stata results.
• R guides: intentional differences from Stata, including numeric replacement and identifier precision, mutation by reference, containers, egen calculations, including equivalent gen(), egen(), and := forms, and the tab() parity matrix against Stata's tabulate.

License

GPL-3.0. See LICENSE.