Adding in support to get attributes by attribute

[necrolyte2]

Maybe I am doing it wrong, but there are a lot of cases that I want to get say a tracker or custom_field by name. Since the Redmine API does not support this it makes it difficult and you have to essentially fetch all records and then filter them down manually.

This should essentially do that process instead of having the api user write their own function

Does it make sense to support a non-lazy operation though?

[rconradharris]

@necrolyte2 Couldn't we have duplicates, for example two records with name as foo?

If thats the case, it feels like this should return a list, and maybe be called 'get_all_by_attribute`.

[necrolyte2]

Good point. I'll change it and fix the build error too.
I was so focused on that I'm always trying to get a custom field id for filtering and our custom fields never have duplicate names, that I never really thought about all the cases.

[coveralls]

Changes Unknown when pulling f82743c on necrolyte2:master into * on maxtepkeev:master*.

[rconradharris]

Should probably just return an empty list rather than raise an exception.

[necrolyte2]

So just let the caller decide if it is an exception if the attribute does not exist?

[necrolyte2]

N/m I see what you mean

[rconradharris]

Yeah I think that makes sense; if no items are found that match the criteria, just return an empty record set.

[coveralls]

Changes Unknown when pulling 584b524 on necrolyte2:master into * on maxtepkeev:master*.

[maxtepkeev]

I don't think that it should live in a ResourceManager. I believe that ResourceSet is the correct place for this method, otherwise you're hardcoding it to only work with all() method, but it can also be convenient to use it with any other method that returns a ResourceSet, e.g. filter().

[necrolyte2]

Makes sense. I won't be able to get to changing that until next week, so if you get time and want to do it go for it.

[maxtepkeev]

I'm closing this pull request, because I've implemented a little bit more powerful Django-like lookup system for ResourceSet's get() and filter() methods. It allows you to do things like:

issues = redmine.issue.filter(project_id='foobar').filter(status__id=(2, 7), tracker__id=5, done_ratio=50)

i.e. the first call to filter(project_id='foobar') is the ResourceManager's filter method which just works as usual and makes the request to Redmine and returns the ResourceSet object with all issues of the foobar project. The second call to filter(status__id=(2, 7), tracker__id=5, done_ratio=50) is a new thing. It calls the filter() method on the ResourceSet object which will return a new ResourceSet with resource objects that have status_id either 2 or 7, tracker_id 5 and done_ration 50. The __ notation allows you to go as deep as you need to make filtering on the nested resource object's attributes.

I'll push the commits with this functionality a little bit later, because I want to rewrite a documentation a little bit and write some more tests for this.