polars-readstat 0.13.0

Read SAS (sas7bdat), Stata (dta), and SPSS (sav) files with polars

polars_readstat

Polars plugin for SAS (.sas7bdat), Stata (.dta), and SPSS (.sav/.zsav) files.

The Python package wraps the Rust core in polars_readstat_rs and exposes a Polars-first API. The project includes cross-library parity tests and roundtrip checks to reduce regressions.

The Rust engine is generally faster for many workloads, but performance varies by file shape and options. If you need the legacy C/C++ engine, use version 0.11.1 (see the prior version).

Why use this?

• In project benchmarks, the new Rust-backed engine is typically faster than pandas/pyreadstat on large SAS/Stata files, especially for subset/filter workloads.
• It avoids the older C/C++ toolchain complexity and ships as standard Python wheels.
• API is Polars-first (scan_readstat, read_readstat, write_readstat).

Install

pip install polars-readstat

Core API

1) Lazy scan

import polars as pl
from polars_readstat import scan_readstat

lf = scan_readstat("/path/file.sas7bdat", preserve_order=True)
df = lf.select(["SERIALNO", "AGEP"]).filter(pl.col("AGEP") >= 18).collect()

Key parameters:

Parameter	Default	Description
preserve_order	False	Return rows in original file order. Set True when order matters; may be slower with multiple threads.
missing_string_as_null	False	Convert empty strings to null.
value_labels_as_strings	False	For labeled numeric columns (Stata/SPSS), return the string label instead of the numeric code.
schema_overrides	None	Dict mapping column names to Polars types (e.g. {"id": pl.Int64}). Useful when the file header reports a narrower type than the data requires.
batch_size	100_000	Number of rows per internal chunk during collect.
informative_nulls	None	Capture user-defined missing value indicators. See Informative Nulls.

2) Informative Nulls

SAS, Stata, and SPSS files support user-defined missing value codes (SAS .A–.Z, Stata .a–.z, SPSS discrete/range missings). By default these are read as null. The informative_nulls option captures the missing-value indicator alongside the data value.

from polars_readstat import scan_readstat, read_readstat, InformativeNullOpts

# Track all eligible columns; add a "<col>_null" String indicator column after each
lf = scan_readstat("file.dta", informative_nulls=InformativeNullOpts(columns="all"))
df = read_readstat("file.sas7bdat", informative_nulls={"columns": "all"})

Three output modes (set via the mode parameter):

Mode	Description
"separate_column" (default)	Adds a parallel String column <col><suffix> after each tracked column
"struct"	Wraps each (value, indicator) pair into a Struct column
"merged_string"	Merges into a single String column (value as string, or the indicator code)

from polars_readstat import InformativeNullOpts

opts = InformativeNullOpts(
    columns=["income", "age"],      # or "all"
    mode="separate_column",         # "separate_column", "struct", or "merged_string"
    suffix="_missing",              # indicator column suffix (separate_column mode only)
    use_value_labels=True,          # use value label for indicator string when defined
)

informative_nulls accepts either an InformativeNullOpts dataclass or a plain dict. It is supported on scan_readstat, read_readstat, and ScanReadstat.

3) Eager read

from polars_readstat import read_readstat

df = read_readstat("/path/file.dta")

4) Metadata + schema

from polars_readstat import ScanReadstat

reader = ScanReadstat(path="/path/file.sav")
schema = reader.schema      # polars.Schema
metadata = reader.metadata  # dict with file info and per-column details
lf = reader.df              # LazyFrame — same as calling scan_readstat(path)

metadata is a dict with a columns list. Each column entry includes:

• "name" — column name
• "label" — variable label (description), if present
• "value_labels" — dict mapping coded values to label strings, if present

5) Write (Stata/SPSS) - EXPERIMENTAL

Writing support is experimental and compatibility varies across tools. Stata roundtrip tests are included; SPSS roundtrip coverage is limited. Please report issues.

from polars_readstat import write_readstat

write_readstat(df, "/path/out.dta")
write_readstat(df, "/path/out.sav")

# With value labels and variable labels (both formats)
write_readstat(
    df,
    "/path/out.dta",
    value_labels={"sex": {1: "Male", 2: "Female"}},
    variable_labels={"sex": "Sex of respondent", "age": "Age in years"},
)

# Stata-only options
write_readstat(df, "/path/out.dta", compress=True, threads=8)

write_readstat supports Stata (dta) and SPSS (sav). SAS writing is not supported.

Parameter	Formats	Description
value_labels	dta, sav	Dict mapping column names to {coded_value: label_string}.
variable_labels	dta, sav	Dict mapping column names to descriptive label strings.
compress	dta only	Write compressed Stata file.
threads	dta only	Number of threads for writing.

Tests run

Test coverage includes:

• Cross-library comparisons on the pyreadstat and pandas test data, checking results against polars-readstat==0.11.1, pyreadstat, and pandas.
• Stata/SPSS read/write roundtrip tests.
• Large-file read/write benchmark runs on real-world data (results below).

If you want to run the same checks locally, helper scripts and tests are in scripts/ and tests/.

Benchmark

Benchmarks compare four scenarios: 1) load the full file, 2) load a subset of columns (Subset:True), 3) filter to a subset of rows (Filter: True), 4) load a subset of columns and filter to a subset of rows (Subset:True, Filter: True).

Benchmark context:

• Machine: AMD Ryzen 7 8845HS (16 cores), 14 GiB RAM, Linux Mint 22
• Storage: external SSD
• polars-readstat (rust engine v0.12.4) last run: February 24, 2026; comparison library timings for SAS/Stata (v0.11.1) last run August 31, 2025
• Version tested: polars-readstat 0.12.4 (new Rust engine) against polars-readstat 0.11.1 (prior C++ and C engines) and pandas and pyreadstat
• Method: wall-clock timings via Python time.time()

Compared to Pandas and Pyreadstat (using read_file_multiprocessing for parallel processing in Pyreadstat)

SAS

all times in seconds (speedup relative to pandas in parenthesis below each)

Library	Full File	Subset: True	Filter: True	Subset: True, Filter: True
polars_readstat New rust engine	0.72 (2.9×)	0.04 (51.5×)	1.04 (2.9×)	0.04 (52.5×)
polars_readstat engine="cpp" (fastest for 0.11.1)	1.31 (1.6×)	0.09 (22.9×)	1.56 (1.9×)	0.09 (23.2×)
pandas	2.07	2.06	3.03	2.09
pyreadstat	10.75 (0.2×)	0.46 (4.5×)	11.93 (0.3×)	0.50 (4.2×)

Stata

all times in seconds (speedup relative to pandas in parenthesis below each)

Library	Full File	Subset: True	Filter: True	Subset: True, Filter: True
polars_readstat New rust engine	0.17 (6.7×)	0.12 (9.8×)	0.24 (4.1×)	0.11 (8.7×)
polars_readstat engine="readstat" (the only option for 0.11.1)	1.80 (0.6×)	0.27 (4.4×)	1.31 (0.8×)	0.29 (3.3×)
pandas	1.14	1.18	0.99	0.96
pyreadstat	7.46 (0.2×)	2.18 (0.5×)	7.66 (0.1×)	2.24 (0.4×)

SPSS

all times in seconds (speedup relative to pandas in parenthesis below each)

Library	Full File	Subset: True	Filter: True	Subset: True, Filter: True
polars_readstat New rust engine	0.22 (6.6×)	0.15 (9.1×)	0.25 (6.0×)	0.26 (4.5×)
pandas	1.46	1.36	1.49	1.16
pyreadstat	9.25 (0.2×)	4.85 (0.3×)	9.39 (0.2×)	4.75 (0.2×)

Detailed benchmark notes and dataset descriptions are in BENCHMARKS.md.