Indic-PersoArabic-Script-Converter

Indo-Pakistani Transliteration

A python library to convert from Indian scripts to Pakistani scripts and vice-versa.

Currently supported methods

1. Rule-based conversion

• Faster, but does not support short vowels
• Will not be accurate, especially for Arabic-to-Indic

2. Sangam Project's online transliteration API

• Uses an online endpoint for the conversion
• Produces much better results, but much slower

Usage

Installation

Pre-requisites:

• Use Python 3.7+
• pip install git+https://github.com/GokulNC/indic_nlp_library

pip install indo-arabic-transliteration

Using rule-based conversion

from indo_arabic_transliteration.mapper import script_convert
script_convert(text: str, from_script: str, to_script: str)

Using Sangam API

from indo_arabic_transliteration.sangam_api import online_transliterate
online_transliterate(text: str, from_script: str, to_script: str)

Languages

We use the standard BCP 47 language tags to refer to the language-script combinations.

Hindi-Urdu (Hindustani)

Language	Script	Code
Hindi	Devanagari	hi-IN
Urdu	Perso-Arabic	ur-PK

Example:

# Rule-based
script_convert("हैदराबाद‎", 'hi-IN', 'ur-PK') # حیدرآباد
script_convert("حيدرآباد‎", 'ur-PK', 'hi-IN') # हीदराबाद‎

# Online-API
online_transliterate("حيدرآباد‎", 'ur-PK', 'hi-IN') # हैदराबाद‎
online_transliterate("हैदराबाद‎", 'hi-IN', 'ur-PK') # حیدرآباد‎

Notes & Resources:

• Both the nations share a common national language (Hindustani) but written in different scripts and also registered as different languages.
• Official Tools
  • Software by Pakistani Center for Language Engineering
  • Online Tool by Indian Center for Development of Advanced Computing
• Devanagari to PersoArabic mapping
  • Note: This same rule-based function can be used for Saraiki and Shina languages also
    • TODO: Shina characters here seems to be bit different. So use with caution

Panjabi

Language	Script	Code
East Punjabi	Gur'Mukhi	pa-IN
West Punjabi	ShahMukhi	pa-PK

Example:

# Rule-based
script_convert("ਸਿੰਘ", 'pa-IN', 'pa-PK') # سںگھ
script_convert("سںگھ", 'pa-PK', 'pa-IN') # ਸਂਘ

# Online-API
online_transliterate("سنگھ", 'pa-PK', 'pa-IN') # ਸਿੰਘ
online_transliterate("ਸਿੰਘ", 'pa-IN', 'pa-PK') # سِنگھ

Notes & Resources:

• You can also use these JavaScript libraries:
  • Anvaad-JS by KhalisFoundation
  • Gurmukhi-Utils by ShabadOS (Demo)
• Gurmukhi to Shahmukhi mapping

Sindhi

Language	Script	Code
Indian Sindhi	Devanagari	sd-IN
Pakistani Sindhi	Perso-Arabic	sd-PK

Example:

# Rule-based
script_convert("हैदराबाद‎", 'sd-IN', 'sd-PK') # حیدرآباد
script_convert("حيدرآباد‎", 'sd-PK', 'sd-IN') # हीदराबाद‎

# Online-API
online_transliterate("حيدرآباد‎", 'sd-PK', 'sd-IN') # हैदराबाद‎
online_transliterate("हैदराबाद‎", 'sd-IN', 'sd-PK') # حیدرآباد‎

Notes & Resources:

• Before Devanagari standardization, Sindhi was written in Landa scripts like Khojki, Khudawadi, Multani, Gurmukhi, etc. depending upon the region.
  • To convert from Devanagari to the above legacy scripts, use AksharaMukha's python library.
• You can also use this JavaScript library or online converter.
• Sindhi-PersoArabic to Devanagari mapping

---

Other Methods

MachineLearning-based Transliteration

• Uses LibIndicTrans library for models
  • Install it by pip install git+https://github.com/libindic/indic-trans
• Currently supports only Hindi-Urdu languages

API:

from indo_arabic_transliteration.ml_based import ml_transliterate
# Same interface as script_convert()

Indic-to-Arabic with Diacritics

• Indic scripts are mostly phonetic. Use this to retain diacritics in PersoArabic
  • Currently only supports Hindustani (Hindi to Urdu) and Punjabi (Gurmukhi to Shahmukhi)
  • Uses AksharaMukhi library

API:

from indo_arabic_transliteration.lossless_converter import convert_with_diacritics
# Same interface as script_convert()

---

Support

• For help in using the library, please use the GitHub Issues section.
• For script conversion errors from the online API, please write directly to the Sangam team. We are not related to them in anyway and this is not an official library.