Import Table (Parquet)

Overview

The Import Table (Parquet) node reads Apache Parquet data from an upstream file and converts it into a DataFrame.

Purpose

Use this node when a workflow receives columnar data from data pipelines, warehouses, Spark jobs, Hadoop exports, or other systems that produce Parquet files.

Configuration

This node requires a file input and has no additional configuration fields. When you drag it onto the canvas:

1. Click the + icon on the node to add an input source.
2. Select a source node from the list (for example, a Read File or File Listener node that provides a .parquet file).
3. Configure the source node to point to the Parquet file you want to read.
4. Open the Parquet node settings to preview the output.

Inputs

Input	Data Type	Input Type	Description
InputFile	NodeFile	Single	Input Parquet file that been read

Outputs

Output	Data Type	Collection	Description
ParquetDataSource	DataFrame	False	DataFrame created from the Parquet schema and rows.

Processing Logic

The node reads the upstream file with DuckDB. If the input is an archive file, it extracts the archive into the working directory and reads all extracted *.parquet files from the extracted folder. It builds the output schema from a sample result set and writes the full Parquet query result into the output DataFrame. If no rows are found, the node adds a warning.

Examples

• Each column corresponds to a field defined in the Parquet schema
• Data types are preserved as defined in the Parquet file (integers, doubles, strings, timestamps, etc.)
• All rows from the file are included by default

Notes / Limitations

• No additional configuration is required on this node.
• When reading an archive, the node reads Parquet files from the first extracted folder.
• For large Parquet files, use Query In-Memory downstream to filter or aggregate the data before passing it further in the workflow.