Skip to content

A simplistic, general purpose pipeline framework.

License

Notifications You must be signed in to change notification settings

tamasgal/thepipe

Repository files navigation

thepipe

Documentation Status Codacy Badge Travis-CI Build Status Test-coverage PyPI Package latest release

A simplistic, general purpose pipeline framework, which can easily be integrated into existing (analysis) chains and workflows.

Installation

thepipe can be installed via pip:

pip install thepipe

Features

  • Easy to use interface and integration into existing workflows
  • Automatic provenance tracking (set Provenance().outfile to dump it upon program termination)
  • Modules can be either subclasses of Module or bare python functions
  • Data is passed via a simple Python dictionary from module to module (wrapped in a class called Blob which adds some visual candy and error reporting)
  • Integrated hierarchical logging system
  • Colour coded log and print messages (self.log() and self.cprint() in Modules)
  • Performance statistics for the whole pipeline and each module individually
  • Clean exit when interrupting the pipeline with CTRL+C

The Pipeline

Here is a basic example how to create a pipeline, add some modules to it, pass some parameters and drain the pipeline.

Note that pipeline modules can either be vanilla (univariate) Python functions or Classes which derive from thepipe.Module.

import thepipe as tp


class AModule(tp.Module):
    def configure(self):
        self.cprint("Configuring AModule")
        self.max_count = self.get("max_count", default=23)
        self.index = 0

    def process(self, blob):
        self.cprint("This is cycle #%d" % self.index)
        blob['index'] = self.index
        self.index += 1

        if self.index > self.max_count:
            self.log.critical("That's enough...")
            raise StopIteration
        return blob

    def finish(self):
        self.cprint("I'm done!")


def a_function_based_module(blob):
    print("Here is the blob:")
    print(blob)
    return blob


pipe = tp.Pipeline()
pipe.attach(AModule, max_count=5)  # pass any parameters to the module
pipe.attach(a_function_based_module)
pipe.drain()  # without arguments it will drain until a StopIteration is raised

This will produce the following output:

2020-05-26 12:43:12 ++ AModule: Configuring AModule
Pipeline and module initialisation took 0.001s (CPU 0.001s).
2020-05-26 12:43:12 ++ AModule: This is cycle #0
Here is the blob:
Blob (1 entries):
'index' => 0
2020-05-26 12:43:12 ++ AModule: This is cycle #1
Here is the blob:
Blob (1 entries):
'index' => 1
2020-05-26 12:43:12 ++ AModule: This is cycle #2
Here is the blob:
Blob (1 entries):
'index' => 2
2020-05-26 12:43:12 ++ AModule: This is cycle #3
Here is the blob:
Blob (1 entries):
'index' => 3
2020-05-26 12:43:12 ++ AModule: This is cycle #4
Here is the blob:
Blob (1 entries):
'index' => 4
2020-05-26 12:43:12 ++ AModule: This is cycle #5
2020-05-26 12:43:12 CRITICAL ++ AModule: That's enough...
2020-05-26 12:43:12 ++ AModule: I'm done!
============================================================
5 cycles drained in 0.001284s (CPU 0.001475s). Memory peak: 27.01 MB
wall  mean: 0.000070s  medi: 0.000052s  min: 0.000042s  max: 0.000122s  std: 0.000031s
CPU   mean: 0.000070s  medi: 0.000052s  min: 0.000042s  max: 0.000124s  std: 0.000032s