Pipeline SDK
The FlaskTrack Pipeline SDK lets Python and R processing blocks read their declared inputs and publish structured outputs. It is already available inside the FlaskTrack pipeline runtime; you do not need to configure credentials or connect to FlaskTrack from your script.
Use the SDK whenever a processing block needs to:
- read an Arrow table produced by a source or earlier block;
- read JSON configuration or metadata;
- open a declared file input;
- publish a table for another block or for reporting;
- publish JSON or a generated file; or
- add useful progress details to the execution log.
The basic pattern
Every script follows the same three-part pattern:
- Read each declared input by its port name.
- Process the data with Python or R libraries.
- Publish each declared output by its port name.
Python
import flasktrack as ft
rows = ft.input_table("rows")
result = rows.slice(0, 100)
ft.output_table("result", result)
R
source("/opt/flasktrack/flasktrack.R")
rows <- ft_input_table("rows")
result <- head(rows, 100)
ft_output_table("result", result)
The names rows and result must match the ports declared on the block. The
input and output helper must also match the port's data type.
Important: AI-generated scripts are unreviewed drafts. Review their port names, library syntax, calculations, and output types before saving or publishing a pipeline version.
Where to go next
- Follow the quick start to create your first processing block.
- Learn how ports and data types work.
- Browse the Python or R SDK reference.
- See complete examples.
- Diagnose a failed run with troubleshooting.