Free · Snappy · Browser-local
Compress Parquet Files Online
Rewrite supported flat Parquet files with internal Snappy compression, validate the schema and values, then download a normal .parquet file.
Parquet → Snappy Parquet
Supported files are decoded, rewritten, and validated in this browser tab. File contents are not uploaded.
What the Parquet compressor preserves
The tool reads typed columns separately from the viewer’s display strings. It preserves the supported schema, column order, row count, values, and null positions, then reopens the result before enabling its download.
Inputs are limited to 25 MB because decoded columns can use much more browser memory than their encoded source. Unsupported schemas receive a controlled error rather than being flattened or converted.
Plain answers
Parquet viewer questions
Does this create a .parquet.gz file?+
No. It rewrites compression inside the Parquet column chunks, so the result remains a directly readable .parquet file rather than a gzip-wrapped archive.
Is my Parquet file uploaded?+
No. Decoding, Snappy rewriting, validation, and download creation run in this browser tab.
Which Parquet schemas are supported?+
This first version supports flat primitive Boolean, integer, floating-point, byte-array, and string columns. It rejects nested, repeated, list, map, INT96, fixed-length byte-array, and unsupported logical types instead of changing them.
Will Snappy always make the file smaller?+
No. Snappy favors broad compatibility and decoding speed. Its output can be larger than Zstandard, Brotli, or a source file that was already well optimized.