iithlp 0.2.5

A comprehensive transliteration scheme for Indian languages

IITHLP

A comprehensive transliteration scheme for Indian languages with a strict one-to-one character mapping.

Overview

IITHLP is a Python package that provides a comprehensive transliteration scheme for 9 major Indian scripts covering 17 scheduled Indian languages. It ensures a strict one-to-one mapping between each Indic character and a single Roman character (with diacritics as needed), ensuring perfect reversibility while maintaining phonetic similarity.

Supported Scripts and Languages

IITHLP v0.2.2 provides comprehensive transliteration support for 9 Indian scripts covering 17 of the 22 scheduled Indian languages:

Script	Languages
Devanagari	Hindi, Sanskrit, Marathi, Nepali, Konkani, Bodo, Maithili, Dogri
Bengali	Bengali, Assamese
Tamil	Tamil
Malayalam	Malayalam
Telugu	Telugu
Kannada	Kannada
Gurmukhi	Punjabi
Gujarati	Gujarati
Odia	Odia

Languages Not Covered

The following scheduled Indian languages are not currently supported:

• Urdu (uses Perso-Arabic script)
• Kashmiri (primarily uses Perso-Arabic script)
• Sindhi (uses Perso-Arabic script)
• Santli (uses Ol Chiki script)
• Manipuri (uses Meitei Mayek script)

All supported scripts feature a strict one-to-one mapping between each Indic character and a single Roman character (with diacritics as needed), ensuring perfect reversibility while maintaining phonetic similarity.

Features

• Strict one-to-one mapping: Each Indic character maps to exactly one Roman character (with diacritics as needed)
• Perfect reversibility: Guaranteed round-trip conversion without information loss
• Script detection: Automatically detects the script of input text
• Comprehensive coverage: Supports all characters, including vowels, consonants, matras, and special symbols
• Easy to use: Simple API for both script-to-Roman and Roman-to-script conversion

Installation

# You can install directly from the provided Python file
pip install iithlp

Usage

Basic Usage

from iithlp import to_roman, to_script

# Transliterate from an Indian script to Roman
roman_text = to_roman("नमस्ते दुनिया")  # Devanagari
print(roman_text)  # "nmsⁿté dúníyā́"

# Transliterate from Roman to an Indian script
devanagari_text = to_script("nmsⁿté dúníyā́", script="devanagari")
print(devanagari_text)  # "नमस्ते दुनिया"

Script Detection

from iithlp import IITHLP

# Create a transliterator instance
transliterator = IITHLP()

# Detect script of text
text = "ಕನ್ನಡ"
detected_script = transliterator.detect_script(text)
print(detected_script)  # 'kannada'

Getting Supported Scripts

from iithlp import get_supported_scripts

# Get all supported scripts
scripts = get_supported_scripts()
print(scripts)  # ['devanagari', 'bengali', 'tamil', ...]

Mapping Scheme

The mapping scheme follows these principles:

1. Phonetic similarity: Similar sounds have similar representations
2. Diacritics for disambiguation: Diacritics are used to distinguish between similar sounds
3. Consistency: Similar patterns across scripts are represented consistently
4. Reversibility: The scheme ensures perfect round-trip conversion

Example Mappings

Vowels (Independent Form)

Roman	Devanagari	Bengali	Tamil	Malayalam	Telugu	Kannada	Gurmukhi	Gujarati	Odia
a	अ	অ	அ	അ	అ	ಅ	ਅ	અ	ଅ
ā	आ	আ	ஆ	ആ	ఆ	ಆ	ਆ	આ	ଆ
i	इ	ই	இ	ഇ	ఇ	ಇ	ਇ	ઇ	ଇ
ī	ई	ঈ	ஈ	ഈ	ఈ	ಈ	ਈ	ઈ	ଈ
u	उ	উ	உ	ഉ	ఉ	ಉ	ਉ	ઉ	ଉ
ū	ऊ	ঊ	ஊ	ഊ	ఊ	ಊ	ਊ	ઊ	ଊ
e	-	এ	எ	എ	ఎ	ಎ	ਏ	એ	ଏ
ê	ए	ঐ	ஐ	ഐ	ఐ	ಐ	ਐ	ઐ	ଐ
o	-	ও	ஒ	ഒ	ఒ	ಒ	ਓ	ઓ	ଓ
ô	ओ	ঔ	ஔ	ഔ	ఔ	ಔ	ਔ	ઔ	ଔ

Consonants (Sample)

Roman	Devanagari	Bengali	Tamil	Malayalam	Telugu	Kannada	Gurmukhi	Gujarati	Odia
k	क	ক	க	ക	క	ಕ	ਕ	ક	କ
g	ग	গ	-	ഗ	గ	ಗ	ਗ	ગ	ଗ
c	च	চ	ச	ച	చ	ಚ	ਚ	ચ	ଚ
j	ज	জ	ஜ	ജ	జ	ಜ	ਜ	જ	ଜ
ṭ	ट	ট	ட	ട	ట	ಟ	ਟ	ટ	ଟ
t	त	ত	த	ത	త	ತ	ਤ	ત	ତ
p	प	প	ப	പ	ప	ಪ	ਪ	પ	ପ
m	म	ম	ம	മ	మ	ಮ	ਮ	મ	ମ
y	य	য	ய	യ	య	ಯ	ਯ	ય	ଯ
l	ल	ল	ல	ല	ల	ಲ	ਲ	લ	ଲ

License

MIT License

Acknowledgements and Ownership

This package is a product of work from the department of Heritage Science and Technology, IIT Hyderabad. It provides comprehensive support for 9 major Indian scripts and a strict one-to-one mapping scheme. For any concerns please raise an issue on the github repository. You may reach out to the author at ramanan93'at'gmail'dot'com.