Counting 2,987 Big Data & Machine Learning Frameworks, Toolsets, and Examples...
Suggestion? Feedback? Tweet @stkim1


Build Status codecov

A SQLite virtual table extension to expose Parquet files as SQL tables. You may also find csv2parquet useful.

This blog post provides some context on why you might use this.



You can fetch a version built for Ubuntu 16.04 at



The first run will git clone a bunch of libraries, patch them to be statically linkable and build them.

Subsequent builds will only build the parquet virtual table extension.

Building (release)

Run ./make-linx-pgo to build an instrumented binary, run tests to collect real-life usage samples, then build an optimized binary. PGO seems to give a 5-10% reduction in query times.





$ sqlite/sqlite3
sqlite> .load build/linux/libparquet
sqlite> CREATE VIRTUAL TABLE demo USING parquet('parquet-generator/99-rows-1.parquet');
sqlite> SELECT * FROM demo;
...if all goes well, you'll see data here!...

Note: if you get an error like:

sqlite> .load build/linux/libparquet
Error: parquet/ wrong ELF class: ELFCLASS64

You have the 32-bit SQLite installed. To fix this, do:

sudo apt-get remove --purge sqlite3
sudo apt-get install sqlite3:amd64

Supported features

Row group filtering

Row group filtering is supported for strings and numerics so long as the SQLite type matches the Parquet type.

e.g. if you have a column foo that is an INT32, this query will skip row groups whose statistics prove that it does not contain relevant rows:

SELECT * FROM tbl WHERE foo = 123;

but this query will devolve to a table scan:

SELECT * FROM tbl WHERE foo = '123';

This is laziness on my part and could be fixed without too much effort.

Row filtering

For common constraints, the row is checked to see if it satisfies the query's constraints before returning control to SQLite's virtual machine. This minimizes the number of allocations performed when many rows are filtered out by the user's criteria.

Memoized slices

Individual clauses are mapped to the row groups they match.

eg going on row group statistics, which store minimum and maximum values, a clause like WHERE city = 'Dawson Creek' may match 80% of row groups.

In reality, it may only be present in one or two row groups.

This is recorded in a shadow table so future queries that contain that clause can read only the necessary row groups.


These Parquet types are supported:

  • INT96 timestamps (exposed as milliseconds since the epoch)
  • INT8/INT16/INT32/INT64
  • UTF8 strings
  • Variable- and fixed-length byte arrays

These are not currently supported: