Implement fitness function that considers response object keys+values

[lisv]

Following the meeting on the 15th of February:

• Parse entire JSON response and look at all properties

• Create ResponseObjectFitness: an individual (i.e. request) that results in a rare JSON object (looking first at unseen keys and then at unseen value categories) will have a high fitness

• For integers categories like negative, positive and zero can be used. Or look at range, values outside the range seen so far should get higher fitness value.

• For strings look at special cases: empty string, non-empty string or length

• Clustering can be used for both integer values and strings. Only store cluster instead of all values or keys. See if new values or response objects fall outside of cluster -> high fitness.

[lisv]

Update

• ResponseObjectFitness is created. It differentiates between methods, their params, and the value category of the param values.
• Currently a dictionary is kept that stores the 'path' of a value category, i.e. method/param/value-category. This means that the distance (based on an empty dictionary) between method/param1/value-category and method/param2/value-category is the same as the distance between method1/param1/value-category and method2/param1/value-category, which is not ideal.