WebAssembly bindings to read and write the Apache Parquet format to and from Apache Arrow.
This is designed to be used alongside a JavaScript Arrow implementation, such as the canonical JS Arrow library.
Including all compression codecs, the brotli-encoded WASM bundle is 907KB.
parquet-wasm
is published to NPM. Install with
yarn add parquet-wasm
# or
npm install parquet-wasm
These bindings expose two APIs to users because there are two separate implementations of Parquet and Arrow in Rust.
parquet
andarrow
: These are the "official" Rust implementations of Arrow and Parquet. These projects started earlier and may be more feature complete.parquet2
andarrow2
: These are safer (in terms of memory access) and claim to be faster, though I haven't written my own benchmarks yet.
Since these parallel projects exist, why not give the user the choice of which to use? In general the reading API is identical in both APIs, however the write options differ between the two projects.
Presumably no one wants to use both parquet
and parquet2
at once, so the default bundles separate parquet
and parquet2
into separate entry points to keep bundle size as small as possible. The following describe the six bundles available:
Entry point | Rust crates used | Description | Documentation |
---|---|---|---|
parquet-wasm/bundler/arrow1 |
parquet and arrow |
"Bundler" build, to be used in bundlers such as Webpack | Link |
parquet-wasm/node/arrow1 |
parquet and arrow |
Node build, to be used with require in NodeJS |
Link |
parquet-wasm/esm/arrow1 |
parquet and arrow |
ESM, to be used directly from the Web as an ES Module | Link |
parquet-wasm or parquet-wasm/bundler/arrow2 |
parquet2 and arrow2 |
"Bundler" build, to be used in bundlers such as Webpack | Link |
parquet-wasm/node/arrow2 |
parquet2 and arrow2 |
Node build, to be used with require in NodeJS |
Link |
parquet-wasm/esm/arrow2 |
parquet2 and arrow2 |
ESM, to be used directly from the Web as an ES Module | Link |
Note that when using the esm
bundles, the default export must be awaited. See here for an example.
This implementation uses the arrow2
and parquet2
Rust crates.
This is the default implementation and is more full-featured, including metadata handling and async reading. Refer to the API documentation for more details and examples.
This implementation uses the arrow
and parquet
Rust crates.
Refer to the API documentation for more details and examples.
These functions are not present in normal builds to cut down on bundle size. To create a custom build, see Custom Builds below.
setPanicHook(): void
Sets console_error_panic_hook
in Rust, which provides better debugging of panics by having more informative console.error
messages. Initialize this first if you're getting errors such as RuntimeError: Unreachable executed
.
The WASM bundle must be compiled with the console_error_panic_hook
for this function to exist.
import { tableFromArrays, tableFromIPC, tableToIPC } from "apache-arrow";
import {
readParquet,
writeParquet,
Compression,
WriterPropertiesBuilder,
} from "parquet-wasm";
// Create Arrow Table in JS
const LENGTH = 2000;
const rainAmounts = Float32Array.from({ length: LENGTH }, () =>
Number((Math.random() * 20).toFixed(1))
);
const rainDates = Array.from(
{ length: LENGTH },
(_, i) => new Date(Date.now() - 1000 * 60 * 60 * 24 * i)
);
const rainfall = tableFromArrays({
precipitation: rainAmounts,
date: rainDates,
});
// Write Arrow Table to Parquet
const writerProperties = new WriterPropertiesBuilder()
.setCompression(Compression.ZSTD)
.build();
const parquetBuffer = writeParquet(
tableToIPC(rainfall, "stream"),
writerProperties
);
// Read Parquet buffer back to Arrow Table
const table = tableFromIPC(readParquet(parquetBuffer));
console.log(table.schema.toString());
// Schema<{ 0: precipitation: Float32, 1: date: Date64<MILLISECOND> }>
Tl;dr: Try the new
readParquetFFI
API, new in 0.4.0. This API is less well tested than the "normal"readParquet
API, but should be faster and have much less memory overhead (by a factor of 2). If you hit any bugs, please create a reproducible issue.
Under the hood, parquet-wasm
first decodes a Parquet file into Arrow in WebAssembly memory. But
then that WebAssembly memory needs to be copied into JavaScript for use by Arrow JS. The "normal"
read APIs (e.g. readParquet
) use the Arrow IPC
format to get the data back to JavaScript. But this
requires another memory copy inside WebAssembly to assemble the various arrays into a single
buffer to be copied back to JS.
Instead, the new readParquetFFI
API uses Arrow's C Data
Interface to be able to copy or view
Arrow arrays from within WebAssembly memory without any serialization.
Note that this approach uses the arrow-js-ffi
library to parse the Arrow C Data Interface definitions. This library has not yet been tested in
production, so it may have bugs!
I wrote an interactive blog post on this approach and the Arrow C Data Interface if you want to read more!
import { Table } from "apache-arrow";
import { parseRecordBatch } from "arrow-js-ffi";
// Edit the `parquet-wasm` import as necessary
import { readParquetFFI, __wasm } from "parquet-wasm/node2";
// A reference to the WebAssembly memory object. The way to access this is different for each
// environment. In Node, use the __wasm export as shown below. In ESM the memory object will
// be found on the returned default export.
const WASM_MEMORY = __wasm.memory;
const resp = await fetch("https://example.com/file.parquet");
const parquetUint8Array = new Uint8Array(await resp.arrayBuffer());
const wasmArrowTable = readParquetFFI(parquetUint8Array);
const recordBatches = [];
for (let i = 0; i < wasmArrowTable.numBatches(); i++) {
// Note: Unless you know what you're doing, setting `true` below is recommended to _copy_
// table data from WebAssembly into JavaScript memory. This may become the default in the
// future.
const recordBatch = parseRecordBatch(
WASM_MEMORY.buffer,
wasmArrowTable.arrayAddr(i),
wasmArrowTable.schemaAddr(),
true
);
batches.push(recordBatch);
}
const table = new Table(batches);
The Parquet specification permits several compression codecs. This library currently supports:
- Uncompressed
- Snappy
- Gzip
- Brotli
- ZSTD
- LZ4 (deprecated)
- LZ4_RAW. Supported in
arrow2
only.
LZ4 support in Parquet is a bit messy. As described here, there are two LZ4 compression options in Parquet (as of version 2.9.0). The original version LZ4
is now deprecated; it used an undocumented framing scheme which made interoperability difficult. The specification now reads:
It is strongly suggested that implementors of Parquet writers deprecate this compression codec in their user-facing APIs, and advise users to switch to the newer, interoperable
LZ4_RAW
codec.
It's currently unknown how widespread the ecosystem support is for LZ4_RAW
. As of pyarrow
v7, it now writes LZ4_RAW
by default and presumably has read support for it as well.
In some cases, you may know ahead of time that your Parquet files will only include a single compression codec, say Snappy, or even no compression at all. In these cases, you may want to create a custom build of parquet-wasm
to keep bundle size at a minimum. If you install the Rust toolchain and wasm-pack
(see Development), you can create a custom build with only the compression codecs you require.
Note that this project uses Cargo syntax newly released in version 1.60. So you need version 1.60 or higher to compile this project. To upgrade your toolchain, use rustup update stable
.
Reader-only bundle with Snappy compression using the arrow
and parquet
crates:
wasm-pack build --no-default-features --features arrow1 --features snappy --features reader
Writer-only bundle with no compression support using the arrow2
and parquet2
crates, targeting Node:
wasm-pack build --target nodejs --no-default-features --features arrow2 --features writer
Debug bundle with reader and writer support, targeting Node, using arrow
and parquet
crates with all their supported compressions, with console_error_panic_hook
enabled:
wasm-pack build \
--dev \
--target nodejs \
--no-default-features \
--features arrow1 \
--features reader \
--features writer \
--features all_compressions \
--features debug
# Or, given the fact that the default feature includes several of these features, a shorter version:
wasm-pack build --dev --target nodejs --features debug
Refer to the wasm-pack
documentation for more info on flags such as --release
, --dev
, target
, and to the Cargo documentation for more info on how to use features.
By default, arrow
, all_compressions
, reader
, and writer
features are enabled. Use --no-default-features
to remove these defaults.
arrow1
: Use thearrow
andparquet
cratesarrow2
: Use thearrow2
andparquet2
cratesreader
: Activate read support.writer
: Activate write support.async
: Activate asynchronous read support (only applies to thearrow2
endpoints).all_compressions
: Activate all supported compressions for the crate(s) in use.brotli
: Activate Brotli compression.gzip
: Activate Gzip compression.snappy
: Activate Snappy compression.zstd
: Activate ZSTD compression.lz4
: Activate LZ4_RAW compression (only applies to thearrow2
endpoints).debug
: Expose thesetPanicHook
function for better error messages for Rust panics. Additionally compiles CLI debug functions.
- Example of pushdown predicate filtering, to download only chunks that match a specific condition
- Column filtering, to download only certain columns
- More tests
A starting point of my work came from @my-liminal-space's read-parquet-browser
(which is also dual licensed MIT and Apache 2).
@domoritz's arrow-wasm
was a very helpful reference for bootstrapping Rust-WASM bindings.