spellcure 1.0.3

SpellCure — correction engine for highly scrambled text

SpellCure

SpellCure is a mathematical correction engine for highly scrambled or distorted text, created by Saheban Khan (GitHub: Lsaheban) and maintained by Tohid Khan (GitHub: Tohid096).

Rather than using machine learning, SpellCure applies a position-weighted ratio algorithm to match noisy tokens with valid dictionary words — enabling high-accuracy recovery even from severely jumbled text.

---

✨ Features

• Corrects heavily scrambled or distorted words
• Pure mathematical algorithm (no ML required)
• Supports:
  • Small built-in vocabulary (~10k curated words)
  • Large NLTK vocabulary (~200k+ words)
• Works with single words, sentences, or mixed noisy text
• Fast, deterministic, and lightweight
• Extensible word bank (users may request custom additions)

---

🧠 How SpellCure Works

SpellCure analyzes each token using:

• Position-based character similarity
• Ratio scoring
• Multi-stage refinement
• Optional large NLTK dataset

Examples:

Input	Output
aplpe	apple
tihs wrod is msesdy	this word is messy
olve is evryetign	love is everything

---

🚀 Quick Start

Importing SpellCure

from spellcure import corrector
---
## 🧪 Example Usage

Here is a minimal working example using the **small** vocabulary mode:

```python
from spellcure import corrector

def test_small():
    model = corrector(mode="small")   # Use small curated word bank
    output = model.correct("olve is evryetign")
    print(output)

test_small()
#Outpu: love is everything


# small = ~10k curated words
# large = ~200k NLTK words
model = corrector(mode="large")

bash pip install spellcure