The REST server sample code is based on the Flask web framework.
You've installed Flask in lab5, but in this lab, you're going to extend the provided Flask server. If you want a "I just want to make it work" introduction, this tutorial on Flask routes is good. There are official flask tutorials that provide more details.
You're being provided with a starting flask servers in rest-server.py that provides the basics of the specifics of extracting an image file over REST was devised from an example here to implement the rawimage API. We also provide the add method. You'll need to add the jsonimage and dotproduct endpoints.
Both of the endpoints you implement can retrieve the JSON arguments using the request.get_json() method from the Flask request.
We've provided starter code in rest-client.py that construct a REST request for the rawimage and add query. The image query loads a 1.6MB JPG image of the flatirons for processing. The image is from Wikipedia and taken by Jesse Varner. Modified by AzaToth. - Self-made photo.Originally uploaded on 2006-04-19 by Molas. Uploaded edit 2007-12-23 by AzaToth., CC BY-SA 2.5, https://commons.wikimedia.org/w/index.php?curid=3267545. You should use the same image for the jsonimage endpoint.
You will need to install the base Python3 system as well as the libraries jsonpickle and Pillow libraries in addition to the Flask libraries. The easiest way to do this is using the pip command. You'll need to install some Python libraries in addition to the flask libraries installed in lab5. This command install the stuff I needed:
sudo pip3 install pillow jsonpickle
You should modify the client and server to implement the add endpoint and insure that the sample client code receives the correct answer(s). It may be useful to specify a debug flag that dumps out debug output.
You should then use the client to accept a command-line argument indicating the endpoint to be tested and the number of iterations to test. For example:
python rest-client.py localhost add 1000
would run the add endpoint 1000 times against the server on the localhost and then report the time taken divided by the number of queries (1000). This gives you a time-per-query, which should be expressed in milliseconds. The python perf_counter routine from the time module makes it easy to conduct such timing tests. We measure multiple queries because each query is fairly short and we want to average over many such queries.