hinglishswd

A Python library for Hinglish (Hindi+English code-mixed) NLP.

Features

• Language detection — English / Hindi (Devanagari) / Hinglish (Latin-script Hindi)
• Tokenization — Punctuation-aware splitting for Hinglish, spaCy-based for Devanagari Hindi
• Transliteration — Hinglish → Hindi Devanagari (via indic-transliteration)
• Translation — Hinglish/Hindi → English (via deep-translator Google Translate)
• Stop word removal — Built-in Hindi + Hinglish stop word lists
• Pipeline API — Single-call for full processing

Installation

pip install hinglishswd

Quick Start

from hinglishswd import HinglishNLP

nlp = HinglishNLP()

# Language detection
nlp.detect("kal mai khaana khane gaya")    # "hinglish"
nlp.detect("Hello, how are you?")           # "english"
nlp.detect("आज मौसम बहुत अच्छा है")         # "hindi"

# Tokenization
nlp.tokenize("kal mai khaana khane gaya")   # ['kal', 'mai', 'khaana', 'khane', 'gaya']

# Transliteration (Hinglish -> Devanagari)
nlp.transliterate("kal mai khaana khane gaya")  # कल मै खान खने गय

# Translation (Hinglish -> English)
nlp.translate("kal mai khaana khane gaya")  # "I went to eat yesterday"

# Full pipeline
result = nlp.pipeline("aaj mausam bahut acha hai")
# {
#   "text": "aaj mausam bahut acha hai",
#   "language": "hinglish",
#   "tokens": ["aaj", "mausam", "bahut", "acha", "hai"],
#   "tokens_no_stopwords": ["aaj", "mausam", "acha"],
#   "devanagari": "आज मौसम बहुत अच्छा है",
#   "english": "the weather is very good today"
# }

Module-level API

from hinglishswd import (
    detect_language,
    tokenize, tokenize_sentences,
    transliterate, hinglish_to_devanagari,
    to_english, hinglish_to_english, translate_pipeline,
    remove_stopwords,
)

lang = detect_language("Aap kahan ho?")
tokens = tokenize("Mujhe paani chahiye")
dev = hinglish_to_devanagari("mera naam rahul hai")
en = hinglish_to_english("aaj kya kar rahe ho?")

Package Structure

hinglishswd/
├── __init__.py
├── core.py              # HinglishNLP class (main API)
├── detect.py            # Language detection
├── tokenize.py          # Tokenization
├── transliterate.py     # Script conversion (Indic-transliteration)
├── translate.py         # Translation (deep-translator)
└── stopwords.py         # Hindi + Hinglish stop words

Dependencies

• indic-transliteration — Hinglish ↔ Devanagari transliteration
• deep-translator — Google Translate-based translation (optional, for .translate())

License

Modified MIT License — see LICENSE. Large-scale commercial users (>100M MAU or >$20M/month revenue) must display "hinglishswd" in their UI.