⚡ Key Features

• 🚀 Lazy-First Architecture: Built on Polars LazyFrames. Nothing executes until you say so.
• 🛡️ Strict Type Safety: Every transform, every parameter, and every I/O operation is validated with Pydantic models. No more stringly-typed chaos.
• 🔌 Universal I/O:
  • Readers: CSV, Parquet, Excel, JSON, IPC.
  • Healers: Auto-detects encoding issues and "heals" broken CSVs on the fly.
• 🧪 Analytics Module:
  • Built-in scikit-learn integration for Clustering and Regression.
  • Automatic "What-If" simulation engines.
• 🔧 Transform Registry: A modular plugin system for registering data transformation steps.

📦 Installation

pip install pyquery-core

💻 Usage (The SDK)

This is a library for builders. Use it to construct your own data pipelines.

1. The Engine

The PyQueryEngine is the orchestrator.

from pyquery_core.core import PyQueryEngine
from pyquery_core.io.files import FileLoader

# Initialize
engine = PyQueryEngine()

# Load Data (Lazy)
df = FileLoader.read_csv("massive_data.csv")

# Register a Pipeline
pipeline = [
    {"type": "filter", "params": {"column": "revenue", "operator": ">", "value": 1000}},
    {"type": "group_by", "params": {"by": "region", "agg": {"revenue": "sum"}}}
]

# Execute
result = engine.run(df, pipeline)
print(result.collect())

2. Analytics & ML

Run complex statistical analysis without the boilerplate.

from pyquery_core.analytics.ml import ClusterEngine

# Auto-Clustering
model = ClusterEngine(data=df, n_clusters=3)
segments = model.fit_predict()
print(segments)

📂 Architecture

The library is structured for modularity:

🤝 Contributing

This is the Core. Code quality here is paramount.

1. Fork it.
2. Branch it (git checkout -b feature/fancy-algo).
3. Test it. (If it breaks the engine, we break your PR).
4. Push it.

📜 License

GPL-3.0. Open source forever. 💖