How to hide #property(retain, nonatomic)NSString*str in some class?
If you want to hide it from being visible in the .h file, consider using a private class extension in the .m file:
#interface YourClass ()
#property(retain, nonatomic)NSString*str;
#end
You can use a feature called “categories,” which allows you to extend an existing class by adding properties and methods to it.
Typically, you use an empty category inside your class's .m file for private methods and properties, and define a separate named category in an external .h file for protected members. Categories are actually quite interesting in that they allow you to override existing properties. So, for example, you can define a property as read-only in your .h file:
#interface Whatever {
NSObject *aValue;
}
#property (nonatomic,retain,readonly) NSObject *aValue;
#end
And then you can make it read/write for your own private use in an empty category inside your .m file:
#interface Whatever()
#property (nonatomic,retain) NSObject *aValue;
#end
You can find more about categories here.
Related
Following this blog post, I saw a way to solve the problem I was facing.
My problem, like his, was that I have a class that has a property which must be inherited and accessed in its subclass:
#interface A : NSObject
#property (nonatomic, readonly) NSUInteger prop;
#end
#implementation A
// Don't need to synthesize nowadays
#end
#interface B : A
// No new properties
#end
#implementation B
- (void)establishValueForProp
{
_prop = 1; // PROBLEM !!!
}
#end
The solution was this:
#interface A : NSObject {
#protected
NSUInteger _prop;
}
#property (nonatomic, readonly) NSUInteger prop;
#end
What I'm wondering is if there is another way to declare properties as protected?
The way I typically do this is to create a second header, called e.g. "ClassName_ForSubclasses.h". Inside that header file, add a class extension with the properties and methods you want subclasses to be able to call and override. Then, subclasses can import that header, while other users of the class(es) don't see it.
Another common way of doing it, is by creating a separate .h file (ASubclass.h, for example) and adding something like this:
#interface A (Protected)
#property (nonatomic, readonly) NSUInteger prop;
#end
Subclassers can then import this .h and will have access to this method. Note that since categories can't add properties, you'll have to redefine this property in A's private interface (class extension). The category will then provide public access to this property. Apple takes this exact approach with UIGestureRecognizer.h and UIGestureRecognizerSubclass.h (where UIGestureRecognizer.h exposes more methods, intended for subclassers to override).
The concept of protected (or private, for that matter) methods doesn't really exist, and this is only a way to somewhat achieve similar functionality.
Personally, I just create a category in the (only) .h file of the class that exposes an otherwise public property. This way it's separated from the main public interface, and since it's possible to get to private properties anyway, I think this approach is good enough.
According to some official talk, a class in Objective-C should only expose public methods and properties in its header:
#interface MyClass : NSObject
#property (nonatomic, strong) MyPublicObject *publicObject;
- (void)publicMethod;
#end
and private methods/properties should be kept in class extension in .m file:
#interface MyClass()
#property (nonatomic, strong) MyPrivateObject *privateObject;
- (void) privateMethod;
#end
and I don't think there is a protected type for things that are private but accessible from subclasses. I wonder, is there anyway to achieve this, apart from declaring private properties/methods publicly?
One way to solve this is to re-declare the property in your subclass's class extension, and then add an #dynamic statement so that the compiler won't create an overriding implementation of that property. So something like:
#interface SuperClass ()
#property (nonatomic, strong) id someProperty;
#end
....
#interface SubClass ()
#property (nonatomic, strong) id someProperty;
#end
#implementation SubClass
#dynamic someProperty;
#end
This obviously isn't ideal because it duplicates a privately visible declaration. But it is quite convenient and helpful in some situations so I'd say evaluate on a case-by-case basis the dangers involved in this duplication vs. exposing the property in the public interface.
An alternative - that is used by Apple in UIGestureRecognizer - is to declare the property in a separate category header file explicitly named as "private" or "protected" e.g. "SomeClass+Protected.h". That way, other programmers will know they ought not import the file. But, if you don't control the code you're inheriting from, that's not an option.
This is possible by using a class extension (not category) that you include in the implementation files of both the base class and subclasses.
A class extension is defined similar to a category, but without the category name:
#interface MyClass ()
In a class extension, you can declare properties, which will be able to synthesize the backing ivars (XCode > 4.4 automatic synthesis of the ivars also works here).
In the extension class, you can override/refine properties (change readonly to readwrite etc.), and add properties and methods that will be "visible" to the implementation files (but note that the properties and methods aren't really private and can still be called by selector).
Others have proposed using a seperate header file MyClass_protected.h for this, but this can also be done in the main header file using #ifdef like this:
Example:
BaseClass.h
#interface BaseClass : NSObject
// foo is readonly for consumers of the class
#property (nonatomic, readonly) NSString *foo;
#end
#ifdef BaseClass_protected
// this is the class extension, where you define
// the "protected" properties and methods of the class
#interface BaseClass ()
// foo is now readwrite
#property (nonatomic, readwrite) NSString *foo;
// bar is visible to implementation of subclasses
#property (nonatomic, readwrite) int bar;
-(void)baz;
#end
#endif
BaseClass.m
// this will import BaseClass.h
// with BaseClass_protected defined,
// so it will also get the protected class extension
#define BaseClass_protected
#import "BaseClass.h"
#implementation BaseClass
-(void)baz {
self.foo = #"test";
self.bar = 123;
}
#end
ChildClass.h
// this will import BaseClass.h without the class extension
#import "BaseClass.h"
#interface ChildClass : BaseClass
-(void)test;
#end
ChildClass.m
// this will implicitly import BaseClass.h from ChildClass.h,
// with BaseClass_protected defined,
// so it will also get the protected class extension
#define BaseClass_protected
#import "ChildClass.h"
#implementation ChildClass
-(void)test {
self.foo = #"test";
self.bar = 123;
[self baz];
}
#end
When you call #import, it basically copy-pastes the .h file to where you are importing it.
If you have an #ifdef, it will only include the code inside if the #define with that name is set.
In your .h file, you don't set the define so any classes importing this .h wont see the protected class extention.
In the base class and subclass .m file, you use #define before using #import so that the compiler will include the protected class extension.
While the other answers are correct, I'd like to add...
Private, protected and public are available for instance variables as such:
#interface MyClass : NSObject {
#private
int varA;
#protected
int varB;
#public
int varC;
}
#end
Your only choice is to declare it as public in the header file. If you want to at least keep some method separation, you can create a category and have all your protected methods and attributes there, but in the end everything will still be public.
#import "MyClass.h"
#interface MyClass (Protected)
- (void) protectedMethods;
#end
Simply create a .h file with your class extension. Import this into your .m files. Incidentally, this is a great way to test private members without breaking encapsulation (I'm not saying you should test private methods :) ).
// MyClassProtectedMembers.h
#interface MyClass()
#property (nonatomic, strong) MyPrivateObject *privateObject;
- (void) privateMethod;
#end
/////////////////
#import "MyClassProtectedMembers.h"
#implementation MyClass
// implement privateMethod here and any setters or getters with computed values
#end
Here's a gist of the idea: https://gist.github.com/philosopherdog/6461536b99ef73a5c32a
I see good answers for making properties visible, but I don't see exposing the methods addressed very clearly in any of these answers. Here is how I have successfully exposed private methods to the subclass using a Category:
SomeSuperClass.m:
#implementation SomeSuperClass
-(void)somePrivateMethod:(NSString*)someArgument {
...
}
SomeChildClass.h
#interface SomeChildClass : SomeSuperClass
SomeChildClass.m
#interface SomeSuperClass (exposePrivateMethod)
-(void)somePrivateMethod:(NSString*)someArgument;
#end
#implementation SomeChildClass
-(void)doSomething {
[super somePrivateMethod:#"argument"];
}
#end
That's because there's not even a real distinction between private and public. While the compiler may warn you about an interface missing a certain method or instance variable, your program will still work.
What's preferable and a clean solution for attributes in a class. Use categories (anonymous) or #private annotation in the interface definition (.h file).
The attributes should not be inherited by other classes.
BR,
mybecks
Categories and #private are two different things. Categories extend the classes where you can add methods to a class and #private is where other classes can't access your attributes directly and will have to use your accessors. If you want ivars that subclasses can't access, use #private in your interface.
#interface MyClass : NSObject
{
#private
int num; //private attribute where subclass can't access
}
First up read this to understand the #private modifier What does "#private" mean in Objective-C?
And then to answer your questions.
An anonymous category is called a class extension and this is what I use to keep my API's clean and to not publicly announce my ivars.
e.g
// .h
#interface MyClass : NSObject
#end
// .m
#interface MyClass ()
#property (nonatomic, strong) NSDate *someDateImWorkingWith;
#end
#implementation MyClass
#synthesize someDateImWorkingWith = _someDateImWorkingWith;
#end
As a result I have the NSDate *someDateImWorkingWith to use within my class but no one importing my header will even know it exists without doing a little bit of digging. This works well for me your milage may vary.
I've noticed a variety of #interface declarations for Objective-c classes. I'd like to understand why developers declare #interface in the following ways:
// in the .h file
#interface MyClass : NSObject
// ...
#end
// in the .m file (what's the purpose of the parens?)
#interface MyClass ()
// more property declarations which seem like they can go in the .h file
#end
// again in the .m file (what's the purpose of private?)
#interface MyClass (Private)
// some method declarations
#end
This is just a normal class interface, inheriting from NSObject, where you declare ivars, properties and methods
// in the .h file
#interface MyClass : NSObject
// ...
#end
The following two are categories, which allow you to add methods to a class. It is not a subclass however (do not declare a method with the same name, as you won't be able to access the original one). If you have a named category of the interface (like #interface MyClass (Private)), then the implementation should be provided in #implementation MyClass (Private), in the case of unnamed categories (also called extensions), the implementation can be provided as usual. Note that extensions also allow you to add ivars to the class while (named) categories do not.
// in the .m file (what's the purpose of the parens?)
#interface MyClass ()
// more property declarations which seem like they can go in the .h file
#end
// again in the .m file (what's the purpose of private?)
#interface MyClass (Private)
// some method declarations
#end
It is used to declared private methods.
This response explain this in details: What are best practices that you use when writing Objective-C and Cocoa?
What ever goes in the .m file is private. the parens are for categories so you can segment your code into categories to make it more readable. because the code is in .m and private, they called the category Private.
Normally we use
#interface interface_name : parent_class <delegates>
{
......
}
#end
method in .h file and in .m file we synthesis the properties of variables declared in .h file.
But in some code, this #interface.....#end method is kept in the .m file also. What does it mean? What is the difference between them?
Also give some words about getters and setters for the interface file that is defined in .m file...
It's common to put an additional #interface that defines a category containing private methods:
Person.h:
#interface Person
{
NSString *_name;
}
#property(readwrite, copy) NSString *name;
-(NSString*)makeSmallTalkWith:(Person*)person;
#end
Person.m:
#interface Person () //Not specifying a name for the category makes compiler checks that these methods are implemented.
-(void)startThinkOfWhatToHaveForDinner;
#end
#implementation Person
#synthesize name = _name;
-(NSString*)makeSmallTalkWith:(Person*)person
{
[self startThinkOfWhatToHaveForDinner];
return #"How's your day?";
}
-(void)startThinkOfWhatToHaveForDinner
{
}
#end
The 'private category' (the proper name for a nameless category is not 'private category', it's 'class extension') .m prevents the compiler from warning that the methods are defined. However, because the #interface in the .m file is a category you can't define ivars in it.
Update 6th Aug '12: Objective-C has evolved since this answer was written:
ivars can be declared in a class extension (and always could be - the answer was incorrect)
#synthesize is not required
ivars can now be declared in braces at the top of #implementation:
that is,
#implementation {
id _ivarInImplmentation;
}
//methods
#end
The concept is that you can make your project much cleaner if you
limit the .h to the public interfaces of your class, and then put
private implementation details in this class extension.
when you declare variable methods or properties in ABC.h file , It
means these variables properties and methods can be access outside the
class
#interface Jain:NSObject
{
NSString *_name;
}
#property(readwrite, copy) NSString *name;
-(NSString*)makeSmallTalkWith:(Person*)jain;
#end
#Interface allows you to declare private ivars, properties and
methods. So anything you declare here cannot be accessed from outside
this class. In general, you want to declare all ivars, properties and
methods by default as private
Simply say when you declare variable methods or properties in ABC.m
file , It means these variables properties and methods can not be
access outside the class
#interface Jain()
{
NSString *_name;
}
#property(readwrite, copy) NSString *name;
-(NSString*)makeSmallTalkWith:(Person*)jain;
#end
you can even create other classes in .m file,
for instance other small classes which inherit from the class declared in .h file but having some slight different behaviour.
You could use this in a factory pattern