awk_plus_plus

A language designed for data orchestration.

Features

• Fuzzy regex engine and Semantic search to retrieve information in an in-process DB.
• End-user programming.
• Orthogonal Persistence based on DuckDB
• Transparent reference with Jsonnet. We plan to execute this feature with Dask.
• URL interpreter to manage data sources.

Installation from pip

Install the package with:

pip install awk_plus_plus

CLI Usage

You output your data to JSON with the cti command.

Web service

The command runs a web service with Gradio, allowing you to execute your expressions through a user-friendly user interface or by making HTTP requests.

cti run-webservice

Jsonnet support

Hello world

cti i "Hello world" -p -v 4

Jsonnet support

cti i '{"keys":: ["AWK", "SED", "SHELL"], "languages": [std.asciiLower(x) for x in self.keys]}'

URL interpreter

Our step further is the URL interpreter which allows you to manage different data sources with an unique syntax across a set of plugins.

STDIN, STDOUT, STDERR

cti i '{"lines": interpret("stream://stdin?strip=true")}'

Imap

cti i '{"emails": interpret("imap://USER:PASSWORD@HOST:993/INBOX")}'

Keyring

cti i '{"email":: interpret("keyring://backend/awk_plus_plus/email"), "emails": interpret($.email)}'

Files

cti i 'interpret("**/*.csv")'

SQL

cti i 'interpret("sql:SELECT * FROM email")'

Leverage the Power of Reference with Jsonnet

Unlike other programming languages that require multiple steps to reference data, Jsonnet requires only one step, thanks to its reference mechanism. This is particularly useful for data engineers who want to connect different services in a topological order. The code below represents this scenario in Python:

import requests

def fetch_character(id):
    url = f"https://rickandmortyapi.com/api/character/{id}"
    response = requests.get(url)
    return response.json()

def process_character(character):
    # Add new 'image' field with processed URL
    character['image'] += f"?awk_download=data/{character['name'].replace(' ', '_').lower()}.jpeg"
    
    # Process 'episode' field, fetching additional data if necessary
    character['episode'] = [requests.get(episode).json() for episode in character['episode']]
    
    return character


print([process_character(fetch_character(id)) for id in [1, 2, 3, 4, 5, 6]])

Contrary to the previous Python code, Jsonnet allows you to leverage the power of referential transparency. The previous code is equivalent in Jsonnet to:

[
   i("https://rickandmortyapi.com/api/character/%s" % id) + 
    {image: i(super.image+"?awk_download=data/"+std.strReplace(std.asciiLower(super.name), " ", "_")+".jpeg")} + 
    {episode: [i(episode) for episode in super.episode]}
   for id in [1,2,3,4,5,6]
]

Connect and call different data sources in one expression

{
   "emails": i("sql:SELECT subject FROM `%s`" %  self.email),
   // This expression saves the unseen emails from your inbox, as defined in your keyring, using IMAP query criteria. It then returns the netloc hash, which refers to the table.
   "email": i(i("keyring://backend/awk_plus_plus/primary_email")+"?q=UNSEEN")
}

Protocols and Plugins

• pop3://
• imap://
• keyring://backend/{service}/{username}
• sql:{expression}
• https://
• file:/

Note

This project has been set up using PyScaffold 4.5 and the dsproject extension 0.0.post167+g4386552.