unipipeline

simple way to build the declarative and distributed data pipelines.

Why you should use it

Declarative strict config
Scaffolding
Fully typed
Python support 3.6+
Brokers support
- kafka
- rabbitmq
- inmemory simple pubsub
Interruption handling = safe user code transactions
CLI

How to Install

$ pip3 install unipipeline

Example

# dag.yml
---

service:
  name: "example"
  echo_colors: true
  echo_level: error


external:
  service_name: {}


brokers:
  default_broker:
    import_template: "unipipeline.brokers.uni_memory_broker:UniMemoryBroker"

  ender_broker:
    import_template: "example.brokers.uni_log_broker:LogBroker"


messages:
  __default__:
    import_template: "example.messages.{{name}}:{{name|camel}}"

  input_message: {}

  inetermediate_message: {}

  ender_message: {}


cron:
  my_super_task:
    worker: my_super_cron_worker
    when: 0/1 * * * *

  my_mega_task:
    worker: my_super_cron_worker
    when: 0/2 * * * *

  my_puper_task:
    worker: my_super_cron_worker
    when: 0/3 * * * *


waitings:
  __default__:
    import_template: example.waitings.{{name}}_wating:{{name|camel}}Waiting

  common_db: {}


workers:
  __default__:
    import_template: "example.workers.{{name}}:{{name|camel}}"

  my_super_cron_worker:
    input_message: uni_cron_message

  input_worker:
    input_message: input_message
    waiting_for:
      - common_db

  intermediate_first_worker:
    input_message: inetermediate_message
    output_workers:
      - ender_second_worker
    waiting_for:
      - common_db

  intermediate_second_worker:
    input_message: inetermediate_message
    external: service_name
    output_workers:
      - ender_frist_worker

  ender_frist_worker:
    input_message: ender_message

  ender_second_worker:
    input_message: ender_message
    broker: ender_broker
    waiting_for:
      - common_db

Get Started

create ./unipipeline.yml such as example above
run cli command

unipipeline -f ./unipipeline.yml scaffold

It should create all structure of your workers, brokers and so on

remove error raising from workers
correct message structure for make more usefull
correct broker connection (if need)
run cli command to run your consumer

unipipeline -f ./unipipeline.yml consume input_worker

or with python

from unipipeline import Uni
u = Uni(f'./unipipeline.yml')
u.init_consumer_worker(f'input_worker')
u.initialize()
u.start_consuming()

produce some message to the message broker by your self or with tools

unipipeline -f ./unipipeline.yml produce --worker input_worker --data='{"some": "prop"}'

or with python

# main.py
from unipipeline import Uni

u = Uni(f'./unipipeline.yml')
u.init_producer_worker(f'input_worker')
u.initialize()
u.send_to(f'input_worker', dict(some='prop'))

Definition

Service

service:
  name: some_name       # need for health-check file name
  echo_level: warning   # level of uni console logs (debug, info, warning, error)
  echo_colors: true     # show colors in console

External

external:
  some_name_of_external_service: {}

no props
it needs for declarative grouping the external workers with service

Worker

workers:
  __default__:                                        # each worker get this default props if defined
    retry_max_count: 10
    
  some_worker_name:
    retry_max_count: 3                                # just counter. message move to /dev/null if limit has reached 
    retry_delay_s: 1                                  # delay before retry
    topic: "{{name}}"                                 # template string
    error_payload_topic: "{{topic}}__error__payload"  # template string
    error_topic: "{{topic}}__error"                   # template string
    broker: "default_broker"                          # broker name. reference to message transport 
    external: null                                    # name of external service. reference in this config file 
    ack_after_success: true                           # automatic ack after process message
    waiting_for:                                      # list of references
      - some_waiting_name                             # name of block. this worker must wait for connection of this external service if need
    output_workers:                                   # list of references
      - some_other_worker_name                        # allow worker sending messages to this worker
    
    inport_template: "some.module.hierarchy.to.worker.{{name}}:{{name|camel}}OfClass"   # required module and classname for import

    input_message: "name_of_message"                  # required reference of input message type

Waiting

waitings:
  some_blocked_service_name:
    retry_max_count: 3                         # the same semantic as worker.retry_max_count
    retry_delay_s: 10                          # the same semantic as worker.retry_delay_s
    import_template: "some.module:SomeClass"   # required. the same semantic as worker.import_template

Broker

brokers:
  some_name_of_broker:
    retry_max_count: 3                         # the same semantic as worker.retry_max_count
    retry_delay_s: 10                          # the same semantic as worker.retry_delay_s
    content_type: application/json             # content type
    compression: null                          # compression (null, application/x-gzip, application/x-bz2, application/x-lzma)
    import_template: "some.module:SomeClass"   # required. the same semantic as worker.import_template

Message

messages:
  name_of_message:
    import_template: "some.module:SomeClass"   # required. the same semantic as worker.import_template

build in messages:

messages:
  uni_cron_message:
    import_template: unipipeline.messages.uni_cron_message:UniCronMessage

CLI

unipipeline

usage: unipipeline --help

UNIPIPELINE: simple way to build the declarative and distributed data pipelines. this is cli tool for unipipeline

positional arguments:
  {check,scaffold,init,consume,cron,produce}
                        sub-commands
    check               check loading of all modules
    scaffold            create all modules and classes if it is absent. no args
    init                initialize broker topics for workers
    consume             start consuming workers. connect to brokers and waiting for messages
    cron                start cron jobs, That defined in config file
    produce             publish message to broker. send it to worker

optional arguments:
  -h, --help            show this help message and exit
  --config-file CONFIG_FILE, -f CONFIG_FILE
                        path to unipipeline config file (default: ./unipipeline.yml)
  --verbose [VERBOSE]   verbose output (default: false)

unipipeline check

usage: 
    unipipeline -f ./unipipeline.yml check
    unipipeline -f ./unipipeline.yml --verbose=yes check

check loading of all modules

optional arguments:
  -h, --help  show this help message and exit

unipipeline init

usage: 
    unipipeline -f ./unipipeline.yml init
    unipipeline -f ./unipipeline.yml --verbose=yes init
    unipipeline -f ./unipipeline.yml --verbose=yes init --workers some_worker_name_01 some_worker_name_02

initialize broker topics for workers

optional arguments:
  -h, --help            show this help message and exit
  --workers INIT_WORKERS [INIT_WORKERS ...], -w INIT_WORKERS [INIT_WORKERS ...]
                        workers list for initialization (default: [])

unipipeline scaffold

usage: 
    unipipeline -f ./unipipeline.yml scaffold
    unipipeline -f ./unipipeline.yml --verbose=yes scaffold

create all modules and classes if it is absent. no args

optional arguments:
  -h, --help  show this help message and exit

unipipeline consume

usage: 
    unipipeline -f ./unipipeline.yml consume
    unipipeline -f ./unipipeline.yml --verbose=yes consume
    unipipeline -f ./unipipeline.yml consume --workers some_worker_name_01 some_worker_name_02
    unipipeline -f ./unipipeline.yml --verbose=yes consume --workers some_worker_name_01 some_worker_name_02

start consuming workers. connect to brokers and waiting for messages

optional arguments:
  -h, --help            show this help message and exit
  --workers CONSUME_WORKERS [CONSUME_WORKERS ...], -w CONSUME_WORKERS [CONSUME_WORKERS ...]
                        worker list for consuming

unipipeline produce

usage: 
    unipipeline -f ./unipipeline.yml produce --worker some_worker_name_01 --data {"some": "json", "value": "for worker"}
    unipipeline -f ./unipipeline.yml --verbose=yes produce --worker some_worker_name_01 --data {"some": "json", "value": "for worker"}
    unipipeline -f ./unipipeline.yml produce --alone --worker some_worker_name_01 --data {"some": "json", "value": "for worker"}
    unipipeline -f ./unipipeline.yml --verbose=yes produce --alone --worker some_worker_name_01 --data {"some": "json", "value": "for worker"}

publish message to broker. send it to worker

optional arguments:
  -h, --help            show this help message and exit
  --alone [PRODUCE_ALONE], -a [PRODUCE_ALONE]
                        message will be sent only if topic is empty
  --worker PRODUCE_WORKER, -w PRODUCE_WORKER
                        worker recipient
  --data PRODUCE_DATA, -d PRODUCE_DATA
                        data for sending

unipipeline cron

usage: 
    unipipeline -f ./unipipeline.yml cron
    unipipeline -f ./unipipeline.yml --verbose=yes cron

start cron jobs, That defined in config file

optional arguments:
  -h, --help  show this help message and exit

simple way to build the declarative and destributed data pipelines with python

Related tags

Overview

unipipeline

Why you should use it

How to Install

Example

Get Started

Definition

Service

External

Worker

Waiting

Broker

Message

CLI

unipipeline

unipipeline check

unipipeline init

unipipeline scaffold

unipipeline consume

unipipeline produce

unipipeline cron

Contributing

TODO LIST

Owner

aliaksandr-master

Data Analytics on Genomes and Genetics

In this project, ETL pipeline is build on data warehouse hosted on AWS Redshift.

Extract Thailand COVID-19 Cluster data from daily briefing pdf.

Stochastic Gradient Trees implementation in Python

BigDL - Evaluate the performance of BigDL (Distributed Deep Learning on Apache Spark) in big data analysis problems

Karate Club: An API Oriented Open-source Python Framework for Unsupervised Learning on Graphs (CIKM 2020)

PyPDC is a Python package for calculating asymptotic Partial Directed Coherence estimations for brain connectivity analysis.

Reading streams of Twitter data, save them to Kafka, then process with Kafka Stream API and Spark Streaming

Powerful, efficient particle trajectory analysis in scientific Python.

Single machine, multiple cards training; mix-precision training; DALI data loader.

Tkinter Izhikevich Neuron Model With Python

Spectacular AI SDK fuses data from cameras and IMU sensors and outputs an accurate 6-degree-of-freedom pose of a device.

Candlestick Pattern Recognition with Python and TA-Lib

Find exposed data in Azure with this public blob scanner

Gaussian processes in TensorFlow

Package for decomposing EMG signals into motor unit firings, as used in Formento et al 2021.

Tools for the analysis, simulation, and presentation of Lorentz TEM data.

This python script allows you to manipulate the audience data from Sl.ido surveys

pyETT: Python library for Eleven VR Table Tennis data

PostQF is a user-friendly Postfix queue data filter which operates on data produced by postqueue -j.