Track traits and overrides with FactoryProf

[palkan]

Context

Currently, we only track the factory name/id when collecting information about factories usage. However, for refactoring and context extraction purposes it would be useful to also see common variations (traits and attribute overrides).

TODO

Add variations information to the FactorProf data and print it in the FPROF=1 report. Example format:

[TEST PROF INFO] Factories usage

Total: 15285
...

    name           total   top-level   total time    time per call      top-level time           
    user            6091        2715    115.7671s          0.0426s            50.2517s 
     .admin          123         715     15.7671s          0.0466s             5.2517s 
     [name,role]      25          11       7.671s          0.0666s             1.2517s 
    post            2142        2098     93.3152s          0.0444s            92.1915s       
     .draft[tags]     12          12      9.3152s          0.164s             42.1915s        
  ...

In the example above, .xxx indicates a trait and [a,b] indicates the overrides keys, e.g., create(:user, :admin) is an .admin variation, while create(:post, :draft, tags: ["a"])—.draft[tags]. We do not distinguish values, only schemas.

Implementation notes

• FactoryProf collector should receive the variation name from the outside and should not deal with traits, overrides, etc. Thus, we need to extends the #track methods with the variation: keyword argument

• Overrides keys and traits must be sorted (since the order doesn't matter)

• It makes sense to add a specific variation id for variations with too many traits or overrides ("too many" must be configurable); e.g., create(:post, title: "RailsConf", publish_at: 2.days.since, tags: ["a", "b"], author: user) would be marked as [...].