Evaluator#key? forwards key name argument #664
Conversation
When testing the presence of multiple, optional keys, it would be nice to be able to specify the key name to `key?` as you can with `values.key?`. Since this merely delegates to that method, I can't see any reason why this omission was intentional. `#key_name` is preserved as the default argument, so this will not change any existing behavior.
| The `key?` method supports passing an explicit key name for rules that have multiple keys. | ||
|
|
||
| ```ruby | ||
| class DistanceContract < Dry::Validation::Contract |
There was a problem hiding this comment.
Can we have a spec at contract level (/spec/integration/contract/) mimicking this same scenario where we are asking of an specific keys?
There was a problem hiding this comment.
Sure, I think the test coverage is pretty comprehensive already but that would probably better communicate the intent of this.
There was a problem hiding this comment.
I merged it as-is because having this example in the docs already covers the intended usage and the implementation is so simple that the unit test will suffice.
| # @return [Boolean] | ||
| # | ||
| # @api public | ||
| def key? | ||
| values.key?(key_name) | ||
| def key?(name = key_name) |
There was a problem hiding this comment.
I think we want to support the same type of args as Values#key?, WDYT? Values supports symbols, arrays, path objects etc. so this could be something like:
def key?(*args)
args.empty? ? values.key?(key_name) : values.key?(*args)
endThere was a problem hiding this comment.
I was intending this to work with Schema::Path, yes.
The Values implementation doensn't rely on splatting to work, is that necessary here?
def key?(key, hash = data)
return hash.key?(key) if key.is_a?(Symbol)
Schema::Path[key].reduce(hash) do |a, e|
if e.is_a?(Array)
result = e.all? { |k| key?(k, a) }
return result
elsif e.is_a?(Symbol) && a.is_a?(Array)
return false
else
return false unless a.is_a?(Array) ? (e >= 0 && e < a.size) : a.key?(e)
end
a[e]
end
true
endThere was a problem hiding this comment.
@alassek ah right, doesn't have to be *args, good point!
😄| class DistanceContract < Dry::Validation::Contract | ||
| schema do | ||
| optional(:kilometers).filled(:integer) | ||
| optional(:miles).filled(:integer) |
There was a problem hiding this comment.
| optional(:miles).filled(:integer) | |
| optional(:miles).value(:integer) |
| ```ruby | ||
| class DistanceContract < Dry::Validation::Contract | ||
| schema do | ||
| optional(:kilometers).filled(:integer) |
There was a problem hiding this comment.
| optional(:kilometers).filled(:integer) | |
| optional(:kilometers).value(:integer) |
|
Sorry, was not blowing this off just had a busy week. Thanks for fixing the doc example for me! |
|
@alassek no worries! |
When testing the presence of multiple, optional keys, it would be nice to be able to specify the key name to
key?as you can withvalues.key?.Since this merely delegates to that method, I can't see any reason why this omission was intentional.
#key_nameis preserved as the default argument, so this will not change any existing behavior.discussed in the forum here