A minimalist Apache Parquet file reader implemented in Python that converts Parquet files to JSONL (JSON Lines) format.

• Reads Apache Parquet files
• Supports common data types:
  • INT32
  • INT64
  • FLOAT
  • DOUBLE
  • BYTE_ARRAY
• Handles dictionary encoding
• Supports compression:
  • Uncompressed
  • GZIP
  • Snappy
• Outputs to JSONL format

• Python 3.x
• Required packages:
  • thrift
  • snappy

1. Clone the repository
2. Install dependencies:

pip install thrift
pip install python-snappy

python main.py input.parquet output.jsonl

• input.parquet: Path to the input Parquet file
• output.jsonl: Path where the output JSONL file will be written

python main.py iris.parquet iris.jsonl

This will convert the iris.parquet file to JSONL format, where each line is a JSON object representing one row of data.

• Currently supports only DataPageV1
• Supports basic data types and encodings
• Assumes flat schema structure
• Limited to PLAIN and RLE_DICTIONARY encodings

pyparquet/
├── main.py          # Main implementation
├── parquet.thrift   # Thrift definition file
└── src/
    └── thrift/      # Generated Thrift code

This project is open source and available under the MIT License.