A command-line tool for streaming Parquet as line-delimited JSON.
It reads only required ranges from file, HTTP or S3 locations, and supports offset/limit and column selection.
It uses the Apache Parquet Official Native Rust Implementation which has excellent support for compression formats and complex types.
Install from crates.io and execute from the command line, e.g.:
$ cargo install parquet2json
$ parquet2json --help
USAGE:
parquet2json [OPTIONS] <FILE> <SUBCOMMAND>
ARGS:
<FILE> Location of Parquet input file (file path, HTTP or S3 URL)
OPTIONS:
-t, --timeout <TIMEOUT> Request timeout in seconds [default: 60]
-h, --help Print help information
-V, --version Print version information
SUBCOMMANDS:
cat Outputs data as JSON lines
schema Outputs the Thrift schema
rowcount Outputs only the total row count
help Print this message or the help of the given subcommand(s)
$ parquet2json cat --help
USAGE:
parquet2json <FILE> cat [OPTIONS]
OPTIONS:
-o, --offset <OFFSET> Starts outputting from this row (first row: 0, last row: -1)
[default: 0]
-l, --limit <LIMIT> Maximum number of rows to output [default: -1]
-c, --columns <COLUMNS> Select columns by name (comma,separated,?prefixed_optional)
-h, --help Print help information
Credentials are provided as per standard AWS toolchain, i.e. per environment variables (AWS_ACCESS_KEY_ID
and AWS_SECRET_ACCESS_KEY
), AWS credentials file or IAM ECS container/instance profile.
The default AWS region must be set per environment variable (AWS_DEFAULT_REGION
) in AWS credentials file and must match region of the object's bucket.
Use it to stream output to files and other tools such as grep
and jq.
$ parquet2json ./myfile.pq cat > output.jsonl
$ parquet2json s3:https://amazon-reviews-pds/parquet/product_category=Gift_Card/part-00000-495c48e6-96d6-4650-aa65-3c36a3516ddd.c000.snappy.parquet cat
$ parquet2json https://amazon-reviews-pds.s3.us-east-1.amazonaws.com/parquet/product_category%3DGift_Card/part-00000-495c48e6-96d6-4650-aa65-3c36a3516ddd.c000.snappy.parquet cat
$ parquet2json ./myfile.pq cat --columns=url,level | jq 'select(.level==3) | .url'