Metadata-Version: 2.4
Name: danda
Version: 0.1.2a1
Summary: Automatically prepare pandas DataFrames for analysis.
Author: Mohammad Shalabi
License: MIT
Keywords: pandas,csv,dataframe,preprocessing,cleaning
Requires-Python: >=3.10
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: pandas>=2.2
Requires-Dist: numpy>=1.26
Requires-Dist: build
Requires-Dist: twine
Provides-Extra: excel
Requires-Dist: openpyxl; extra == "excel"
Provides-Extra: parquet
Requires-Dist: pyarrow; extra == "parquet"
Provides-Extra: dev
Requires-Dist: pytest; extra == "dev"
Requires-Dist: pytest-cov; extra == "dev"
Requires-Dist: ruff; extra == "dev"
Requires-Dist: mypy; extra == "dev"
Dynamic: license-file

# danda

> **_Stop writing the same pandas preprocessing code over and over._**
> 
> **Automatically clean and optimize pandas DataFrames with one line of code.**

`danda` is a lightweight extension for pandas that prepares messy data for analysis by automatically cleaning, converting data types, and reducing memory usage.

Instead of writing dozens of lines of preprocessing code, simply do:

```python
clean_df = df.dg.clean().dg.optimize()
```

---

## Why danda?

Every data scientist has written code like this:

```python
df = pd.read_csv("sales.csv")

df = df.dropna(how="all")
df = df.dropna(axis=1, how="all")
df = df.drop_duplicates()

df["created_at"] = pd.to_datetime(df["created_at"])
df["active"] = df["active"].astype(bool)
df["country"] = df["country"].astype("category")
df["price"] = pd.to_numeric(df["price"])
```

With **danda**, all of that becomes:

```python
df = pd.read_csv("sales.csv")

df = df.dg.clean().dg.optimize()
```

---

# Features

## 🧹 Data cleaning

Automatically:

- Remove empty rows
- Remove empty columns
- Remove duplicate rows
- Strip leading/trailing whitespace

Example:

| Before | After |
|---------|-------|
| `" John "` | `"John"` |
| `" Alice"` | `"Alice"` |
| `"Bob "` | `"Bob"` |

---

## 📅 Automatic type detection

Automatically converts string columns to their appropriate pandas dtypes.

### Boolean

```
"True"
"False"
"0"
"1"
0
1
```

↓

```
boolean
```

---

### Datetime

```
2024-01-01
2024/01/01
2024-01-01T12:30:00
Jan 1, 2024
```

↓

```
datetime64[ns]
```

---

### Numeric

```
"10"
"25"
"3.14"
```

↓

```
int64 / float64
```

---

### Category

Low-cardinality columns are automatically converted to pandas `category` dtype to reduce memory usage.

Example:

```
Country

Germany
France
Germany
Germany
France
```

↓

```
category
```

---

## 📉 Memory optimization

Optimizing dtypes can dramatically reduce memory usage.

```python
before.dg.compare_memory(after)
```

Example:

```
Before : 95 MB
After  : 21 MB

Saved : 74 MB (77.9%)
```

---

## 📄 Cleaning reports

Every operation produces a report describing what happened.

```python
optimized.dg.report
```

Example:

```python
{
    "clean": {
        "EmptyRowsPlugin":
            "Number of deleted rows: 15",

        "EmptyColumnsPlugin":
            "Number of deleted columns: 2",

        "DropDuplicates":
            "Number of deleted rows: 31"
    },

    "optimize": {
        "BooleanTypePlugin":
            ["active"],

        "DateTimeTypePlugin":
            ["created_at"],

        "NumericTypePlugin":
            ["price"],

        "CategoryTypePlugin":
            ["country"]
    }
}
```

---

# Installation

```bash
pip install danda
```

---

# Quick Start

```python
import pandas as pd
import danda

df = pd.read_csv("employees.csv")

clean = df.dg.clean()

optimized = clean.dg.optimize()

print(optimized.dg.report)

print(clean.dg.compare_memory(optimized))
```

---

# Pandas Accessor

`danda` integrates directly into pandas using a custom accessor.

```python
df.dg.clean()

df.dg.optimize()

df.dg.report

df.dg.compare_memory(other_df)
```

No new DataFrame class.
No wrapper objects.

Just pandas.

---

# Philosophy

`danda` follows a simple principle:

> **Prepare data for analysis automatically while preserving the familiar pandas workflow.**

The library is designed to feel like a natural extension of pandas rather than a replacement.

---

# Roadmap

Planned features include:

- Read functions (`danda.read_csv()`, `read_excel()`, `read_parquet()`)
- Missing value normalization
- Integer downcasting
- Float optimization
- Automatic ID detection
- Data validation
- Dataset profiling
- Configurable plugin pipeline
- Custom plugins

---

# Plugin Architecture

`danda` is built around a plugin system.

Current plugins include:

- EmptyRowsPlugin
- EmptyColumnsPlugin
- DropDuplicatesPlugin
- EmptySpacesPlugin
- BooleanTypePlugin
- DateTimeTypePlugin
- NumericTypePlugin
- CategoryTypePlugin

Creating your own plugins is straightforward, allowing you to customize the cleaning pipeline for your own datasets.

---

# Contributing

Contributions, feature requests, and bug reports are welcome!

If you have an idea that makes data cleaning easier, feel free to open an issue or submit a pull request.

#### Normalize missing values

