Run 50 times test_rizza
5 times failed - test_positive_nested_reach_goal
2 times failed - test_positive_organism

______________________________ test_positive_nested_reach_goal ______________________________

    def test_positive_nested_reach_goal():
        """Create a population and breed them until one matches """
        test_pop = genetics.Population(
            gene_base=NESTED_GENOME, population_count=10)
        for i in range(10):
            test_pop.population[i].points = grade_nested_list(
                test_pop.population[i].genes
            )
        test_pop.sort_population()
        initial_points = test_pop.population[0].points
        generations = 1
        while test_pop.population[0].points > 0 and generations < 100:
            test_pop.breed_population()
            assert len(test_pop.population) == test_pop.population_count
            generations += 1
            for i in range(10):
                test_pop.population[i].points = grade_nested_list(
                    test_pop.population[i].genes
                )
        test_pop.sort_population()
>       assert test_pop.population[0].points < initial_points
E       assert 12749855292 < 9660187854
E        +  where 12749855292 = Organism(genes=[[11, 11, 11, 76, 99], [4, 92, 66, 66, 92], [16, 16, 25, 3, 3]], points=12749855292).points

tests/helpers/test_genetics.py:92: AssertionError

__________________________________ test_positive_organism ___________________________________

    def test_positive_organism():
        test_org = genetics.Organism(genes=BASE_GENOME)
        assert test_org.genes == BASE_GENOME
        test_org.generate_genes()
        # This assertion has a 1:3628800 chance of failing
        assert test_org.genes != BASE_GENOME
        previous = test_org.genes[:]
        # We will make the mutation chance 100% to force a mutation
        test_org.mutate(gene_base=BASE_GENOME, mutation_chance=1)
>       assert test_org.genes != previous
E       assert [1, 7, 6, 14, 2, 22, ...] != [1, 7, 6, 14, 2, 22, ...]
E        +  where [1, 7, 6, 14, 2, 22, ...] = Organism(genes=[1, 7, 6, 14, 2, 22, 13, 12, 16, 17, 0, 24, 19, 10, 21, 18, 3, 5, 4, 9, 20, 23, 15, 8, 11], points=0).genes

tests/helpers/test_genetics.py:104: AssertionError

The file should be the fallback for all configuration options not explicitly set at run-time or by the existing dependencies'.
For example, nailgun has a configuration file it goes to if nothing it set at run-time. Rizza should try to load the config from that file, if no information is passed in. If that file doesn't exist, or is blank, Rizza should fall back to this file proposed in this issue.
Yaml is probably the best way to go for this.

In the beginning, we only care about finding the correct solution. It is always easier to improve it later on.

yake

• keyword extraction with probabilities to all words,
• the lower the probability the more important word
  example.

hammer activation-key create --name 'abc'
>> Error: Could not find organization, please set one of options --organization-id, --organization, --organization-title, --organization-label.

usage:

import yake
error_message = "output string"
kw_extractor = yake.KeywordExtractor()
keywords = kw_extractor.extract_keywords(doc)
>>> 
[('find organization', 0.025580763661104764),
 ('Error', 0.0528063806271324),
 ('organization-id', 0.0528063806271324),
 ('organization-title', 0.0528063806271324),
 ('organization-label', 0.0528063806271324),
 ('organization', 0.06839635777587466),
 ('options', 0.11145728654016383),
 ('find', 0.1823394614884183),
 ('set', 0.1823394614884183)]

The most important words sorted from the top, starting with find organization , Error and organization-id and so on.

• change the probability to seek maximum not a minimum (1 - np.array(data)
• can be easily installed via pip

for API/nailgun:
it can match some whether a word in the error message is one of the fields or arguments and if it is give it high probability (>0.8), so the organism is likely to change.

for CLI:
By using roullete during the selection process, or rather stochastic roulete, it is probable that we will find the correct keyword, that hammer requires. The another important question would how to match a type (str, list, int, ...) to those keywords, randomly, naah that takes a lot of time. Another option is to create a matching scheme between API and CLI, this option is also unsatisfying.

• the judge process can use additional information hammer -d
• the keyword extractor can be fine-tuned with the satellite error messages.

The fitness function:
Nailgun is using error states, which is a good start.
We can burst the learning with heuristics like, if the error changed from the previous one/ones, get obtain a better score.

hammer can use error states as well, if debug option is used.

Before this is done is worth to think about general genetic algorithm and map the steps to those components. It was more complicated than I thought as these terms have different naming in rizza.

It's better to install and upgrade nailgun from nailgun master,
pip install git+ssh://[email protected]/SatelliteQE/nailgun.git

This option, in the form of a --prune flag, would go through the saved positive tests for an entity and remove any that no longer work.
This function would mainly be used for removing tests were saved due to reaching their max generations, instead of actually passing.
The function should check all positive tests saved in an entity's file.

Usage

$ rizza genetic -e Organization --prune
Organization create positive has been removed.
Organization delete positive has been removed.

SSIA

The current argparsr-based cli is a bit busy. Refactoring it to use click would allow for a more simple and dynamic flow.

When rizza is run from any directory, outside of it's own main directory, the config class is unable to load the config.

FileNotFoundError: [Errno 2] No such file or directory: 'config/rizza.yaml'

Appending rizza's project directory to the config file path should fix this issue.

What would help for debugging is actual request send to API, so it's easy to compare it with request that is used inside the robottelo/foreman/api tests.

Any ideas on how to achieve the desired state ?

Same applies for a hammer, the constructed command can make life easier.

Often entities are successfully created with few fields, and IMO it's appropriate to consider such a factor.

Readme need to be updated to include all the changes since it was last pushed.

New additions

• config
• genetic
• list

Also, a description for the purpose of each would be nice.

Logging implementation should be able to allow a user to write logs to a specified or default file, stdout, etc. All modules should be able to easily use the logger to handle output and feedback. Once implemented, the project will need to be gone through to identify areas where logging information would be valuable.

All of this should be configurable by the standard config file and at run-time via the cli.