datayoga-core 1.1.0

DataYoga for Python

DataYoga Core

Introduction

datayoga-py is the transformation engine used in DataYoga, a framework for building and generating data pipelines.

Installation

pip install datayoga

Quick Start

This demonstrates how to transform data using a DataYoga job.

Create a Job

Use this example.yaml:

- uses: add_field
  with:
    fields:
      - field: full_name
        language: jmespath
        expression: concat([fname, ' ' , lname])
      - field: country
        language: sql
        expression: country_code || ' - ' || UPPER(country_name)
- uses: rename_field
  with:
    fields:
      - from_field: fname
        to_field: first_name
      - from_field: lname
        to_field: last_name
- uses: remove_field
  with:
    fields:
      - field: credit_card
      - field: country_name
      - field: country_code
- uses: map
  with:
    expression:
      {
        first_name: first_name,
        last_name: last_name,
        greeting: "'Hello ' || CASE WHEN gender = 'F' THEN 'Ms.' WHEN gender = 'M' THEN 'Mr.' ELSE 'N/A' END || ' ' || full_name",
        country: country,
        full_name: full_name
      }
    language: sql

Transform Data Using datayoga-py

Use this code snippet to transform a data record using the job defined above:

import datayoga_core as dy
from datayoga_core.job import Job
from datayoga_core.utils import read_yaml

job_settings = read_yaml("example.yaml")
job = dy.compile(job_settings)

assert job.transform({"fname": "jane", "lname": "smith", "country_code": 1, "country_name": "usa", "credit_card": "1234-5678-0000-9999", "gender": "F"}) == {"first_name": "jane", "last_name": "smith", "country": "1 - USA", "full_name": "jane smith", "greeting": "Hello Ms. jane smith"}

As can be seen, the record has been transformed based on the job:

• fname field renamed to first_name.
• lname field renamed to last_name.
• country field added based on an SQL expression.
• full_name field added based on a JMESPath expression.
• greeting field added based on an SQL expression.

Examples

• Add a new field country out of an SQL expression that concatenates country_code and country_name fields after upper case the later:

  uses: add_field
  with:
    field: country
    language: sql
    expression: country_code || ' - ' || UPPER(country_name)
  

• Rename fname field to first_name and lname field to last_name:

  uses: rename_field
  with:
    fields:
      - from_field: fname
        to_field: first_name
      - from_field: lname
        to_field: last_name
  

• Remove credit_card field:

  uses: remove_field
  with:
    field: credit_card
  

For a full list of supported block types see reference.

Expression Language

DataYoga supports both SQL and JMESPath expressions. JMESPath are especially useful to handle nested JSON data, while SQL is more suited to flat row-like structures.

Notes

• Dot notation in expression represents nesting fields in the object, for example name.first_name refers to { "name": { "first_name": "John" } }.
• In order to refer to a field that contains a dot in its name, escape it, for example name\.first_name refers to { "name.first_name": "John" }.

JMESPath Custom Functions

DataYoga adds the following custom functions to the standard JMESPath library:

Function	Description	Example	Comments
capitalize	Capitalizes all the words in the string	Input: {"name": "john doe"} Expression: capitalize(name) Output: John Doe
concat	Concatenates an array of variables or literals	Input: {"fname": "john", "lname": "doe"} Expression: concat([fname, ' ' ,lname]) Output: john doe	This is equivalent to the more verbose built-in expression: ' '.join([fname,lname])
hash	Calculates a hash using the hash_name hash function and returns its hexadecimal representation	Input: {"some_str": "some_value"} Expression: hash(some_str, `sha1`) Output: 8c818171573b03feeae08b0b4ffeb6999e3afc05	Supported algorithms: sha1 (default), sha256, md5, sha384, sha3_384, blake2b, sha512, sha3_224, sha224, sha3_256, sha3_512, blake2s
left	Returns a specified number of characters from the start of a given text string	Input: {"greeting": "hello world!"} Expression: left(greeting, `5`) Output: hello
lower	Converts all uppercase characters in a string into lowercase characters	Input: {"fname": "John"} Expression: lower(fname) Output: john
mid	Returns a specified number of characters from the middle of a given text string	Input: {"greeting": "hello world!"} Expression: mid(greeting, `4`, `3`) Output: o w
replace	Replaces all the occurrences of a substring with a new one	Input: {"sentence": "one four three four!"} Expression: replace(sentence, 'four', 'two') Output: one two three two!
right	Returns a specified number of characters from the end of a given text string	Input: {"greeting": "hello world!"} Expression: right(greeting, `6`) Output: world!
split	Splits a string into a list of strings after breaking the given string by the specified delimiter (comma by default)	Input: {"departments": "finance,hr,r&d"} Expression: split(departments) Output: ['finance', 'hr', 'r&d']	Default delimiter is comma - a different delimiter can be passed to the function as the second argument, for example: split(departments, ';')
time_delta_days	Returns the number of days between a given dt and now (positive) or the number of days that have passed from now (negative)	Input: {"dt": '2021-10-06T18:56:16.701670+00:00'} Expression: time_delta_days(dt) Output: 365	If dt is a string, ISO datetime (2011-11-04T00:05:23+04:00, for example) is assumed. If dt is a number, Unix timestamp (1320365123, for example) is assumed.
time_delta_seconds	Returns the number of seconds between a given dt and now (positive) or the number of seconds that have passed from now (negative)	Input: {"dt": '2021-10-06T18:56:16.701670+00:00'} Expression: time_delta_days(dt) Output: 31557600	If dt is a string, ISO datetime (2011-11-04T00:05:23+04:00, for example) is assumed. If dt is a number, Unix timestamp (1320365123, for example) is assumed.
upper	Converts all lowercase characters in a string into uppercase characters	Input: {"fname": "john"} Expression: upper(fname) Output: JOHN
uuid	Generates a random UUID4 and returns it as a string in standard format	Input: None Expression: uuid() Output: 3264b35c-ff5d-44a8-8bc7-9be409dac2b7