PrivacySpy API

The repository contains the source code powering the API service of PrivacySpy. It is used for text highlighting based on an individual sentence's importance (assuming the sentence is a part of a privacy policy). Dockerfile is included.

Installation

Before proceeding, make sure you have Python 3.6 installed.

pip install -r requirements.txt
pip install -U spacy==2.0.3
python3 -m spacy download en

Docker Build

docker build -t privacyspy-api .
docker run -p 5000:5000 -t privacyspy-api

API endpoints

/analyze GET

Parameters:

• url - URL of a privacy policy in application/x-www-form-urlencoded format
• token - PrivacySpy token that corresponds to the environment variable token of the service

OR

/analyze POST

Parameters:

• plain_text - plain text of a privacy policy
• raw_html - raw html of a web page in case plain_text is not submitted
• token - PrivacySpy token that corresponds to the environment variable token of the service

Returns:

• In case of success, the API endpoint returns a JSON object of the following format:

{
  "response": [
    {
      "score": 0.554556940669308, 
      "sentence": "What do we mean by \"personal information?\""
    }, 
    {
      "score": 0.8253833221800004, 
      "sentence": "For us, \"personal information\" means information which identifies you, like your name or email address."
    }, ...
  ], 
  "status": "success", 
  "version": "1.0"
}

• If there was an error, status will contain error, and the response will contain the error message instead. Additionally, there will be an error_code field containing integer values corresponding to the error as follows:
  • 1 -- no URL provided
  • 2 -- no token provided
  • 3 -- invalid token
  • 4 -- failed to extract content from the URL
  • 5 -- privacy policy is not in English