puralang-engine 1.0.0

An AI-driven domain-specific language engine for automated data cleaning pipelines.

PuraLang Engine 🚀

An AI-powered Domain-Specific Language for automated data cleaning pipelines.
Describe your data problem in plain English — PuraLang writes and runs the pipeline for you.

---

What is PuraLang?

PuraLang is a custom programming language built specifically for data cleaning. Instead of writing 30–50 lines of Python/Pandas code every time you need to clean a dataset, you write a clean, human-readable .pura script — or better yet, just describe what you want in English and let the AI generate the script for you.

The Problem It Solves

Every data engineer and ML practitioner spends hours writing repetitive boilerplate code like this:

import pandas as pd
df = pd.read_csv("users.csv")
df = df.drop_duplicates(subset=["user_id"])
df["age"] = df["age"].fillna(24)
df["email"] = df["email"].str.strip().str.lower()
df.to_csv("clean_users.csv", index=False)

With PuraLang, the same result is achieved in 5 readable lines:

LOAD "users.csv"
  |> DROP_DUPLICATES "user_id"
  |> FILL_NULLS "age" VALUE 24
  |> FORMAT_STRINGS "email" TO LOWERCASE
  |> EXPORT_CSV "clean_users.csv"

And with AI Mode, you don't even need to write that:

puralang ask "clean users.csv — remove duplicate user IDs, fill missing ages with 24, lowercase all emails"

---

Features

• Custom DSL Syntax — Clean, readable pipeline syntax with |> operators
• AI Mode — Describe your cleaning task in plain English; the engine generates and runs the script automatically
• Visual Execution Trace — Beautiful terminal output showing row counts before and after every operation
• Multiple Operations — DROP_DUPLICATES, FILL_NULLS, FORMAT_STRINGS, FILTER_ROWS, EXPORT_CSV
• Zero Boilerplate — No Pandas knowledge required to use it

---

Installation

pip install puralang-engine

Or clone and run locally:

git clone https://github.com/SaiDarsini/puralang_engine.git
cd puralang_engine
pip install -r requirements.txt

---

Quick Start

Manual Mode — Write a .pura script

Create a file called pipeline.pura:

LOAD "dirty_data.csv"
  |> DROP_DUPLICATES "user_id"
  |> FILL_NULLS "age" VALUE 24
  |> FORMAT_STRINGS "email" TO LOWERCASE
  |> EXPORT_CSV "cleaned_output.csv"

Run it:

puralang run pipeline.pura

AI Mode — Describe it in English

puralang ask "load sales.csv, remove duplicate order IDs, fill missing prices with 0, export to clean_sales.csv"

PuraLang will:

1. Send your description to an AI model
2. Show you the generated .pura script
3. Execute it automatically
4. Print the trace report

---

Execution Output

Every pipeline run produces a visual trace table in your terminal:

 ┌─────────────────────────── PuraLang Execution Trace ───────────────────────────┐
 │ Operation                      │ Rows Before │ Rows After │
 ├────────────────────────────────┼─────────────┼────────────┤
 │ LOAD SOURCE DATA               │ -           │ 4          │
 │ DROP DUPLICATES [user_id]      │ 4           │ 3          │
 │ FILL NULL FIELDS [age]         │ 3           │ 3          │
 │ STRING TRANSFORM [email]       │ 3           │ 3          │
 │ EXPORT COMPILED FILE           │ 3           │ 3          │
 └────────────────────────────────┴─────────────┴────────────┘

---

Supported Operations

Operation	Syntax	Description
Load CSV	LOAD "file.csv"	Load a CSV file into the pipeline
Drop Duplicates	DROP_DUPLICATES "column"	Remove duplicate rows based on a column
Fill Nulls	FILL_NULLS "column" VALUE 0	Fill missing values with a default
Format Strings	FORMAT_STRINGS "column" TO LOWERCASE	Normalize text casing
Filter Rows	FILTER_ROWS "column" GREATER_THAN 18	Filter rows by condition
Export CSV	EXPORT_CSV "output.csv"	Save cleaned data to a new file

---

Project Architecture

puralang_engine/
│
├── puralang/
│   ├── __init__.py       # Package version
│   ├── core.py           # Lark grammar, parser, and transformer engine
│   └── cli.py            # Typer CLI — run and ask commands
│
├── tests/
│   └── sample.pura       # Example PuraLang script
│
├── setup.py              # PyPI packaging config
└── README.md

The engine works in 3 stages:

.pura script → [Lark Parser] → Abstract Syntax Tree → [Transformer] → Pandas execution → Clean CSV

For AI Mode:

English prompt → [LLM API] → .pura script → [Engine] → Clean CSV

---

Tech Stack

• Lark — Grammar definition and parsing
• Pandas — Underlying data manipulation engine
• Rich — Beautiful terminal output and trace tables
• Typer — CLI command interface
• Google Gemini API — AI script generation (AI Mode)

---

Roadmap

• Core DSL parser and transformer
• CLI with run command
• AI Mode with ask command
• Visual execution trace table
• PyPI public release (pip install puralang-engine)
• Support for JSON and Excel input formats
• Web UI for non-technical users
• VS Code extension with .pura syntax highlighting

---

About the Author

Sai Darsini Sathuluru
B.Tech Student | Mohan Babu University | Generative AI Intern @ Prodigy InfoTech
Founder & Core Architect of PuraLang Engine

• GitHub: @SaiDarsini

---

License

This project is licensed under the MIT License — you are free to use, modify, and distribute it.
See the LICENSE file for details.

---

Built with ❤️ by Sai Darsini · If this helped you, please ⭐ the repo!