/kingsearch

Primary LanguagePython

README

Details

  • Python version

    • 3.6.1

  • Flask version

    • 0.12.2

  • System dependencies

    • virtualenv -v 15.1.0

  • Configuration

    • FLASK_APP=kingsearch
    • FLASK_DEBUG=true

  • Database creation

    • No DB required

  • Database initialization

    • N/A

  • How to run the test suite

    • N/A

  • Development instructions

    • Set environment variables in Configuration
    • install dependencies & run with:
    cd kingsearch
    pip install --editable .
    flask run
  • Usage
    • search is performed as a post request on root path '/'
    • pass query parameter with search string.
    post '/', query: 'search string'
  • Good to know
    • Containing sentence is not returned.

Considerations

  • Best Practices
    • There does not seem to be a agreed upon organization to Flask apps. I may be missing something being new to the language/framework.
    • I tried to follow the Python linter in SublimeText as much as possible for style.
    • Would be interested in reading well written projects & documentation to better organize both.
  • Currently no tests are implemented.
    • Corner cases:
      • Results spanning multi lines will not be returned.
      • May be missing other corner cases.
  • Route documentation:
    • While there isn't much to the application, I found there does not seem to be great examples of how to document routes for Flask. I would love to see examples of how to best model documentation for python/flask.
  • Data Structures:
    • I am not using any complex data structures at this point. Just an array of strings.
      • Scaling is limited by memory at this point.
      • To scale to larger texts I would index the text within a database, indexing on paragraph & line numbers, as well as tokenize each word. I assume that a natural language parser might perform a good deal of these tasks.
  • Rest Status codes:
    • I am not explicitly using any status codes. I possibly would return a 204 if the request is parsed and no data matching query is found.
  • Deployment:
    • I would likely opt to deploy to AWS Lambda as it can be run serverless as a simple endpoint, enabling the service to remain lightweight. This would permit storing source texts in S3 or a database.
  • NLTK (WIP):
    • Included some of the code experimenting with the toolkit as comments.