Crate parquet_lite 

parquet-lite

A lightweight, pure-Rust alternative to the official Apache Parquet crate.

Designed for projects where the full parquet crate is overkill. parquet-lite provides read-path essentials in a fraction of the size, with zero unsafe code and full WASM compatibility.

Key Differences from Official Crate

Feature	parquet (official)	parquet-lite
Binary size	Large	Small
Dependencies	~80	~15
Thrift dependency	Yes	No (hand-rolled)
Read support	Full	Flat schemas
Write support	Yes	Not yet
Arrow integration	Yes	Yes
WASM compatible	Partial	Full

Quick Start

use parquet_lite::*;
use std::fs;

let data = fs::read("data.parquet").unwrap();

// Read metadata
let metadata = read_metadata(&data).unwrap();
println!("Rows: {}, Columns: {}", metadata.num_rows, metadata.num_columns);

// Read as Arrow batches
let batches = read_to_arrow_batches(&data, 1024).unwrap();
for batch in batches {
    let batch = batch.unwrap();
    println!("Batch: {} rows", batch.num_rows());
}

Feature Flags

Feature	Default	Description
snappy	✅	Snappy compression/decompression
serde	❌	Serde serialization for metadata
wasm	❌	WASM bindings via wasm-bindgen
full	❌	All features enabled

Re-exports

pub use types::Compression;

pub use types::Encoding;

pub use types::ParquetError;

pub use types::ParquetMetadata;

pub use types::ParquetType;

pub use types::Result;

pub use types::ColumnMetadata;

pub use types::RowGroupMetadata;

pub use schema::ColumnSchema;

pub use schema::LogicalType;

pub use schema::SchemaBuilder;

pub use schema::TimestampUnit;

pub use reader::ColumnData;

pub use reader::ParquetReader;

pub use arrow_convert::ArrowConverter;

pub use batch_iter_advanced::SelectiveBatchIterator;

pub use streaming_reader::StreamingParquetReader;

pub use statistics::ColumnStatistics;

pub use statistics::StatisticsCollector;

Modules

arrow_convert

batch_iter_advanced

codecs

metadata

reader

schema

statistics

streaming_reader

types

Functions

print_stats

Print a formatted statistics summary to stdout.

read_columns_to_arrow_batches

Create a batch iterator reading only the specified columns.

read_metadata

Parse metadata from raw Parquet file bytes.

read_to_arrow_batches

Create a batch iterator over Arrow RecordBatches.