Quillpad Transliteration Server

Quillpad is an indic language input technology that revolutionized the Indian language typing scene. It is one of the most popular Indic input technologies with more than a billion words typed on the website alone.

Quillpad pioneered the successful use of machine learning for building a predictive language input technology. Quillpad has been rated as the best by many organisations that have embraced Quillpad.

Version

1.0.1

Preparation

Installing Dependencies

sudo apt install -y gcc python-dev python-pip python-mysqldb unzip bzip2 mysql-server default-libmysqlclient-dev
pip install cherrypy Pyrex-real MySQL-python mysql-connector-python cherrypy_cors Nevow tidy
mkdir logs

Setting up required files

There are several archive files in the repository which have to be extracted, these include trained transliteration models and additional text files necessary for the Quillpad Server.
Kindly extract all of these archives into the repository folder itself:

unzip additional_text_files.zip
unzip -d unique_word_files unique_word_files.zip

tar -xvf EnglishPronouncingTrees.tar.bz2
tar -xvf IndianPronouncingTrees.tar.bz2
tar -xvf bengali.tar.bz2
tar -xvf gujarati.tar.bz2
tar -xvf hindi.tar.bz2
tar -xvf kannada.tar.bz2
tar -xvf malayalam.tar.bz2
tar -xvf marathi.tar.bz2
tar -xvf nepali.tar.bz2
tar -xvf punjabi.tar.bz2
tar -xvf tamil.tar.bz2
tar -xvf telugu.tar.bz2

Installation

Quillpad Server requires Python 2.7 to run.

First, we need to compile the Quillpad Model loader that will be used to load the trained transliteration models

$ cd Python\ Cart/python
$ python setup.py build_ext --inplace
$ cp QuillCCart.so ../../
$ cd ../../

Now, the Quillpad Server is ready to run

$ python startquill_cherry.py

Additional Information

• Quillpad runs on port number 8090 (Additional configuration parameters are in quill_cherry8088.conf)

• processWordJSON and processWord are the API endpoints over which the transliteration server can be accessed.

Example:

  * localhost:8090/processWordJSON?inString=hello&lang=hindi
  * localhost:8090/processWordJSON?inString=hello&lang=kannada

Development

Additional Quillpad Documentation coming soon. Thanks for your patience.

FAQ

• How to run on port 80?
  • Set port in quill_cherry8088.conf and run sudo -E python startquill_cherry.py
• Issue with CherryPy? Use CherryPy-3.2.2 included within the repo
• Issue with Pyrex? Download the source code from here and use python setup.py install
• Issue with MySQL and MySQL-python?
  • If error during Quillpad server startup, then check this page
  • If above link also does not work then you may have to download & install via zip files
    • pip install MySQL-python-1.2.5.zip --> Download Link
    • pip install http://cdn.mysql.com/Downloads/Connector-Python/mysql-connector-python-2.0.4.zip

References

An invited talk paper for the Quillpad project: https://www.aclweb.org/anthology/W12-4810

Similar Google service integration: https://stackoverflow.com/questions/28707899/how-to-use-google-input-tools-in-website

More materials

• https://www.cse.iitb.ac.in/~anoopk/pages/softwares.html
• https://github.com/anoopkunchukuttan/indic_nlp_library
• https://github.com/anoopkunchukuttan/transliterator
• http://www.cfilt.iitb.ac.in/brahminet/static/publications/brahminet_naacl2015.pdf

To Do

• Clean the root dir and reorganize data files
• DB connectivity
• Python3 support?