Backend Python API for 'What' Chrome extension

Concordia University Hacks 2020

Team: Gavin Dove, Yaniv Silberman, Olga Sanchis, Kees Vandenberg

Authored by: Gavin Dove, January 2020

'What' enables frictionless practice of french reading, right in your browser. By letting you select nouns, adjectives, verbs, or entire sentences, 'What' uses NLP to scan the main body of text in any web page, selects appropriate words for translation from english to french, and re-inserts them in the mainbody of text, for a seemless user experience. Additional features include a translation frequency selector, a difficulty selector (based on a DB of 30,000 words and the percentage of people that know them), and translation of any word on the web page that the user highlights.

See devpost page for further project details: https://devpost.com/software/what-wp4cxu
See repo for 'What' front end code at: https://github.com/YanivSilberman/what-client

Set-up:

Note: Requires ngrok to be running on same port as gunicorn https://ngrok.com/

1. Activate virtual environment inside outer what_API folder

2. Add environment variable for path to google translate API key:
   $ export GOOGLE_APPLICATION_CREDENTIALS="path/to/key"

3. Add environment variable for path to nltk dataset
   $ export NLTK_DATA="path/to/nltk_data"

4. Start gunicorn:
   $ gunicorn --reload what_API.app

5. In a new terminal, activate ngrok on same port as gunicorn
   
   

API routes:

/sentence: logic for selecting and translating random sentences on web page

/word_lookup: logic for translating highlighted english or french words on web page

/word: logic for translating individual words according to user selected word-type, frequency & difficulty