If the value from a boxing method (e.g. in a custom category) returns nil no mapping on the instance happens and the old value persists. (see: https://github.com/krzysztofzablocki/KZPropertyMapper/blob/master/KZPropertyMapper/KZPropertyMapper.m#L164)

Is there any chance to set a property to nil for custom boxed value?

It seems that the method +mapValuesFrom:toInstance:usingMapping:errors: appears only to be tested when passing nil to this parameter. There is no testing of the existence of a returned array of errors and their expected contents/quality.

Hi, this is more of a question.

From the call stack below, how can I figure out which property causes the problem? Usually it happens when the API returns something I did not account for, but I never know where to start.

Thanks!

Cheers Simon

Is it possible, to somehow disable warnings like: KZPropertyMapper: Ignoring mapping from matchTime to matchTime because type (null) does NOT match NSString

I'm avare that this might be null, NSString out of null is empty. I'm ready for that, but it just overflows my log and would not like to see it.

Hello!

To start: great library! I went to use this today on my Carthage-driven project and noticed the only supported methods of installation were Cocoapods and drag-and-drop, so I got to work. :)

This pull request adds an .xcodeproj with a shared iOS and Mac OS X Framework target. I've attempted to get this working with the most minimal changes to the original source/structure so as not to affect others using KZPropertyMapper via other installation methods.

I've tested this with the latest Xcode 7 build and Carthage version 0.8 (brew upgrade carthage to fetch, if needed).

All you need to do to get this working in an example project is have a Cartfile with the following contents:

github "cfdrake/KZPropertyMapper" "master"

From here, run carthage bootstrap. You'll see there is a .framework file in both Carthage/Build/iOS and Carthage/Build/Mac. You can now follow the instructions on the Carthage README to include the compiled frameworks in an app. After doing this, a simple @import KZPropertyMapper at the top of your source file should include the relevant header files.

Let me know if this seems to be working/you have any suggestions/you're open to adding support for this dependency manager. Thanks!

I was wondering, how would I support compound property, for example, if the values of latitude and longitude are in the root dictionary and I want them to create a CLLocation instance. I couldn't thought of an elegant way to do it though except falling back to hand crafting loading.

Also, I was wondering if there's support for explicitly ignoring fields.

To whom it may concern

I have been using a couple of data mapping libraries and this one seems to be very light weight and easy to use. However one of the thing I would love to have is relationship mapping. Like one-one, one-many relationship

I.e: With this dictionary @{ @"videoURL" : @"http://test.com/video.mp4", @"name" : @"Some Cool Video", @"videoType" : [NSNull null], @"sub_object" : @{ @"title" : @616, @"arbitraryData" : @"data" } }

Can I map the sub_object to a class that I already have

@interface MyClass: NSObject

@property NSString title; @property NSString arbitraryData

@end

Kind regards Leon

Adds tests for #52.

When mapping anonymous objects that adhere to some protocol, prefer using id<SomeProtocol> as the @property's type.

Avoid generic type definitions, such as NSObject<SomeProtocol>, because type-encoding is more difficult to reason about. If a concrete type declaration is publically available, such as a NSObject subclass adhering to SomeProtocol, it is safe to use the exact @class type.

Mapping can fail for valid properties that include only protocol information. For instance, id<MyProtocol> will fail, despite the fact that object conforms to SomeProtocol.

@encode(id<SomeProtocol>) has undocumented behavior which includes a property's Protocol information, as discussed in this StackOverflow thread.

In the event of an id<SomeProtocol>, I've used an NSScanner to enumerate each Protocol and test for protocol conformance on object.

Hi, I just updated to 2.5 and I'm really excited with the new features :)

Before the update I was using a modified version in which I surrounded some methods with a try/catch to avoid crashes in the testers devices because of the API changing constantly. Now I was thinking of use the new version to accomplish the same purpose, but I encountered a problem with an assertion.

JSON I expect:

{
    ...
    thing: 
        {
            a1 : "hello",
            ...
        },
    ...
}

Mapping:

    [KZPropertyMapper mapValuesFrom:properties 
                          toInstance:self
                       usingMapping:@{
                                  @"a1" : KZProperty(attribute1),
                                  }
     ];

JSON I'm reciving:

{
    ...
    thing: [],
    ...
}

This Asserts on line 252: AssertTrueOrReturn([key isKindOfClass:NSNumber.class]); inside the enumeration block.

The problem is natural. I'm giving an array to the mapper when it expects a dictionary. The crash is absolutely normal but with the new feature I was hoping that this won't happen anymore. If this is the desired behaviour there is no problem ;)

This addresses the issue raised by @andrewsardone in krzysztofzablocki/KZPropertyMapper#10, using his proposed solution.

Users (CocoaPods users in particular) should now use the class method +[KZPropertyMapper logIgnoredValues:] to configure whether to log ignored values.

The previous method, via #define KZPropertyMapperLogIgnoredValues, will still work but is no longer exposed in the public interface.

Credit to @andrewsardone for the idea and for suggesting that the class method be placed in a Debug category, to separate it from the class's main concern (that is, mapping stuff).

We could modify the Assertion code to handle better cases where the developer forgot to write the code to handle the custom Box.

    SEL mappingSelector = NSSelectorFromString([NSString stringWithFormat:@"boxValueAs%@:", mappingType]);
    AssertTrueOrReturnNoBlock([self respondsToSelector:mappingSelector], ^(NSError *error) {
    });

In the AssertTrueOrReturnNoBlock we could print the expected selector.

Lets say I have a JSON object like this:

  { 
       success: false,
       error: {code: 1, message: "Something was wrong!"},
       data: nil
  }

And I wan't to map object like this:

@interface Response : NSObject
@property (assign, nonatomic) BOOL    isSuccess;
@property (strong, nonatomic) id      data;
@property (strong, nonatomic) NSError *error;
@end

I can map error with KZCall or by custom boxing. But this methods unable to specify any errors during certain class instantiation or mapping. Is there any way to fix it with current library implementation?

Let's say I have an User object, and an user belongs to a group.

@interface User : NSObject
@property (strong, nonatomic) NSString *firstName;
@property (strong, nonatomic) NSString *lastName;
@property (strong, nonatomic) Group *group;
@end

@interface Group : NSObject
@property (strong, nonatomic) NSString *name;
@end

Now, what I'd like to be able to do is something like this:

@implementation User

- (instancetype)initWithDict:(NSDictionary *)dict {
    self = [super init];

    KZPropertyMapper mapValuesFrom:dict toInstance:self usingMapping:@{
        @"name": KZProperty(firstName),
        @"lastname": KZProperty(lastName),
        @"group": KZCall(Group, groupWithDict:, group),
    }];

    return self;
}

@end

So, call [Group groupWithDict:dict] and save that result to self.group. However, I now constantly have to create an intermediary helper method on the object I'm parsing, so the user object now gets something like this:

- (id)groupWithDict:(NSDictionary *)dict {
    return [Group groupWithDict:dict];
}

After writing a bunch of classes like this, it gets really tedious to duplicate these methods like this. Is there any solution? It looked like KZCallT would be handy, but the property then needs to be on the target object instead of self.