ambers

Pure Rust SPSS .sav/.zsav reader — Arrow-native, zero C dependencies.

Features

• Read .sav (bytecode) and .zsav (zlib) files
• Arrow RecordBatch output — zero-copy to Polars, DataFusion, DuckDB
• Rich metadata: variable labels, value labels, missing values, MR sets, measure levels
• Lazy reader via scan_sav() — returns Polars LazyFrame with projection and row limit pushdown
• No PyArrow dependency — uses Arrow PyCapsule Interface for zero-copy transfer
• 4–8x faster than pyreadstat, on par with polars_readstat
• Python + Rust dual API from a single crate

Installation

Python:

pip install ambers

Rust:

cargo add ambers

Quick Start

Python

import ambers as am

# Eager read — data + metadata
df, meta = am.read_sav("survey.sav")

# Lazy read — returns Polars LazyFrame
lf, meta = am.scan_sav("survey.sav")
df = lf.select(["Q1", "Q2", "age"]).head(1000).collect()

# Explore metadata
meta.summary()
meta.describe("Q1")
meta.value("Q1")

# Read metadata only (fast, skips data)
meta = am.read_sav_metadata("survey.sav")

Rust

use ambers::{read_sav, read_sav_metadata};

// Read data + metadata
let (batch, meta) = read_sav("survey.sav")?;
println!("{} rows, {} cols", batch.num_rows(), meta.number_columns);

// Read metadata only
let meta = read_sav_metadata("survey.sav")?;
println!("{}", meta.label("Q1").unwrap_or("(no label)"));

Metadata API (Python)

All variable-name methods raise KeyError for unknown variables.

Streaming Reader (Rust)

let mut scanner = ambers::scan_sav("survey.sav")?;
scanner.select(&["age", "gender"])?;
scanner.limit(1000);

while let Some(batch) = scanner.next_batch()? {
    println!("Batch: {} rows", batch.num_rows());
}

Performance

Eager Read

All results return a Polars DataFrame. Average of 5 runs on Windows 11, Python 3.13, 24-core machine.

• vs pyreadstat: 4–6x faster across all file sizes
• vs pyreadstat multiprocess (4 workers): ambers single-threaded still faster on every file
• vs polars_readstat: tied on large files, 2–6x faster on small/medium files (lower startup overhead)
• No PyArrow dependency — uses Arrow PyCapsule Interface for zero-copy transfer

pyreadstat multiprocess returns pandas; timing includes pl.from_pandas() conversion.

Lazy Read with Pushdown

scan_sav() returns a Polars LazyFrame. Unlike eager reads, it only reads the data you ask for:

On the 1.1 GB file, selecting 5 columns and 1000 rows completes in 22ms — 76x faster than reading the full dataset.

License

MIT