Make string comparison generically case-insensitive?

[jswelling]

Hi folks;
I have an application where I'd like to modify the behavior of string comparison so that it is always case insensitive. I know that I can just add .as_lower to every string in every comparison operation, but I'm hoping to avoid the additional verbiage. Is there a way to modify the resolvers or some other component to accomplish this?

[zeroSteiner]

Yeah a resolver would probably work in your case. There will likely be edge cases though like to_upper would still work and any functions that return strings would be unaffected. Going the resolver route would basically let you normalize the input to upper or lower case.

[jswelling]

Would this be as simple as initializing the Context with a resolver function that would lowercase any value that was a string? I'm really unclear on how resolvers work and have been unable to find good examples.

[zeroSteiner]

Alright, I looked into this a bit more closely and on second thought I don't think a resolver is going to do it. The issue is with compound types, the resolver isn't used for member values. In the following example you'll see name works just fine but members of the affiliations array do not.

In [1]: import_rule_engine
   ...: def my_resolver(thing, name):
   ...:     value = rule_engine.resolve_item(thing, name)
   ...:     if isinstance(value, str):
   ...:         value = value.lower()
   ...:     return value
   ...: context = rule_engine.Context(resolver=my_resolver)

In [2]: thing = { 'name': 'Luke Skywalker', 'affiliations': ['Rebel Alliance', 'Jedi Order']}

In [3]: rule_engine.Rule('name == "luke skywalker"', context=context).evaluate(thing)
Out[3]: True

In [4]: rule_engine.Rule('affiliations[0] == "rebel alliance"', context=context).evaluate(thing)
Out[4]: False

In [5]: rule_engine.Rule('affiliations[0] == "Rebel Alliance"', context=context).evaluate(thing)
Out[5]: True

An alternative solution might be to set the regex_flags of the context to re.IGNORECASE, then use the regex match operator =~. The main catch here is that the right side of the operator is a regex and can't contain invalid regex characters so you wouldn't want to use a variable. If you're searching simple strings though it might do the trick.

In [1]: import re, rule_engine

In [2]: context = rule_engine.Context(regex_flags=re.IGNORECASE)

In [3]: thing = { 'name': 'Luke Skywalker', 'affiliations': ['Rebel Alliance', 'Jedi Order']}

In [4]: rule_engine.Rule('affiliations[0] =~ "rebel alliance"', context=context).evaluate(thing)
Out[4]: True

In [5]: rule_engine.Rule('name =~ "luke skywalker"', context=context).evaluate(thing)
Out[5]: True

[jswelling]

Thank you; the regex suggestion may solve my problem!