Better way to observe in ReactiveCocoa - objective-c

I'm still new to ReactiveCocoa. I wanted to simply add an observer to a field, so did it like this:
[_countryPicker rac_observeKeyPath:#"value" options:nil observer:self block:^(VBCountry* value, NSDictionary* change)
{
if ([_mobileField.textField.text length] == 0)
{
[_mobileField.textField setText:[NSString stringWithFormat:#"+%i", value.dialCode]];
}
}];
With block callback, and no need to explicitly detach the observer, this is already better than old-style KVO.
However, is this a low-level method to which there's a higher level of abstraction? If so, is it OK to call this method directly? And what's the better/higher way to do it?

I'd advise against depending on the direct KVO methods. They're really an implementation detail.
Let's progressively re-write that with idiomatic RAC operators.
First we'll just replace the direct KVO code with RACObserve:
[RACObserve(_countryPicker, value) subscribeNext:^(VBCountry *value) {
if ([_mobileField.textField.text length] == 0)
{
[_mobileField.textField setText:[NSString stringWithFormat:#"+%i", value.dialCode]];
}
}];
Then we'll replace the if and string formatting with -filter: and -map::
[[[RACObserve(_countryPicker, value) filter:^(id _) {
return [_mobileField.textField.text length] > 0;
}] map:^(VBCountry *value) {
return [NSString stringWithFormat:#"+%i", value.dialCode];
}] subscribeNext:^(NSString *text) {
[_mobileField.textField setText:text];
}];
Finally we'll use the RAC macro to make the assignment over time explicit:
RAC(_mobileField.textField, text) = [[RACObserve(_countryPicker, value) filter:^(id _) {
return [_mobileField.textField.text length] > 0;
}] map:^(VBCountry *value) {
return [NSString stringWithFormat:#"+%i", value.dialCode];
}];

Related

Reactive Cocoa - Convert certain signal values into error or completed

Perhaps I'm still struggling on the reactive learning curve but I am having a hard time figuring out how to bridge a non reactive class with the rest of my reactive code. I am using a category to extend the non-reactive class.
The property is just an Enum representing the current state of a network action, states like New, Submitted, Processing and Completed. Right now I have written the following method in my category:
#implementation JRequestBase (RACExtensions)
- (RACSignal*) rac_RequestStateSignal
{
return RACAble(self, state);
}
#end
However, when state transitions from Processing -> Completed or from any state to Errored I want this signal to send Completed or Error instead of Next Value. How can I accomplish this in a category? I want to do something like:
#implementation JRequestBase (RACExtensions)
- (RACSignal*) rac_RequestStateSignal
{
return [RACAble(self, state) map:^(NSNumber *state){
if ([state intValue] == iRequestStateComplete)
{
# SEND COMPLETE
}
else if ([state intValue] == iRequestStateErrored)
{
# SEND ERROR
}
else
{
return state;
}
}];
}
#end
edit: I took a look at the GHAPIDemo and have come up with the following:
- (RACSignal*) rac_RequestSignal
{
RACSubject *subject = [[RACReplaySubject alloc] init];
[[RACAble(self, state) subscribeNext:^(NSNumber* s){
if ( [s intValue] == JRequestStateCompleted)
{
[subject sendNext:self];
[subject sendCompleted];
}
else if ([s intValue] == JRequestStateErrored)
{
NSMutableDictionary *dict = [NSMutableDictionary dictionary];
// .. Set up dict with necessary values.
NSError *error = [NSError errorWithDomain:#"blah" code:1 userInfo:dict];
[subject sendError:error];
}
}];
return subject;
}
I'm not 100% sure this is the right way but it seems to be working.
Whenever you want to map values → signal events, instead of values → values, you should use -flattenMap: to return a signal corresponding to each input value. Then, as the "flatten" in the name implies, they'll be combined into one resulting signal.
However, this case is a little different, because you want to terminate the signal as soon as you get the Complete value. We'll use -takeUntilBlock: to represent that part.
The resulting code looks something like this:
- (RACSignal*) rac_RequestStateSignal
{
return [[RACObserve(self, state)
takeUntilBlock:^ BOOL (NSNumber *state){
return [state intValue] == iRequestStateComplete;
}]
flattenMap:^(NSNumber *state){
if ([state intValue] == iRequestStateErrored)
{
// Create a meaningful NSError here if you can.
return [RACSignal error:nil];
}
else
{
return [RACSignal return:state];
}
}];
}
(I used RACObserve because ReactiveCocoa 2.0 is now the only supported version, but you can use RACAble until you're ready to upgrade.)
As a general rule, you should avoid using subjects when possible, since they make code more stateful and reduce laziness.

Refactoring similar methods for objective C

How do I refactor similar methods for the following (Objective C)?
- (void)insertNewSong:(Song *)newSong forArtist:(Artist *)artist {
NSMutableArray *newSongList = [[artist songs] mutableCopy];
BOOL hasInserted = NO;
for (int i = 0; i < [[artist songs] count]; i++) {
Song *existingSong = [[artist songs] objectAtIndex:i];
if ([[newSong title] caseInsensitiveCompare:[existingSong title]] == NSOrderedAscending) {
[newSongList insertObject:newSong atIndex:i];
hasInserted = YES;
break;
}
}
if (hasInserted == NO) {
[newSongList addObject:newSong];
}
artist.songs = newSongList;
}
- (void)insertNewArtistToSongList:(Artist *)newArtist {
BOOL hasInserted = NO;
for (int i = 0; i < [_artists count]; i++) {
Artist *existingArtist = [_artists objectAtIndex:i];
if ([[newArtist name] caseInsensitiveCompare:[existingArtist name]] == NSOrderedAscending) {
[_artists insertObject:newArtist atIndex:i];
hasInserted = YES;
break;
}
}
if (hasInserted == NO) {
[_artists addObject:newArtist];
}
}
For the insertNewSong method, a NSMutableArray [artist songs] containing each Song object is used.
For the insertNewArtist method, a NSMutableArray instance variable _artists containing each Artist Object is used.
Both methods insert an object into an NSMutableArray by comparing the text property of the input object against the text property found within the arrays.
Currently the above methods contain some duplication but is easy to understand (in my case). I was thinking whether there might be a way of simplifying it into a more general method, and does not hurt readability?
There is no general rule, but here are some general rules:
Sometimes it makes sense to combine code like this, sometimes not. Lots of pluses/minuses.
Sometimes it's best to abstract PART of the operation, and leave the other part custom.
Generally, if you have a lot of "if thingA then do this, else that" logic, you've done it wrong (or should not do it at all).
It's best when you can write a single routine and just pass in different parameters (that aren't simply Boolean switches) to differentiate the multiple cases.
It's hard.
And, as a general rule, I don't try too hard to abstract until I have the third instance of nearly the same logic.
(Generally speaking.)

Can't validate null string in Objective-C

I have the following code:
NSString *content = [[NSUserDefaults standardUserDefaults] stringForKey:#"mykey"];
NSLog(#"string is %#",content);
if ([content stringIsEmpty]){
NSLog(#"empty string");
}else{
NSLog(#"string is not empty");
}
stringIsEmpty is class category on NSString:
- (BOOL ) stringIsEmpty {
if ((NSNull *) self == [NSNull null]) {
return YES;
}
if (self == nil) {
return YES;
} else if ([self length] == 0) {
return YES;
}
return NO;
}
The output is:
string is (null)
string is not empty
How could it be null and not empty at the same time?
What happens is that:
[content stringIsEmpty:YES]
will return false (NO), when content is nil. So your code will take the
NSLog(#"string is not empty");
branch. This would be better:
if (content && [content stringIsEmpty:YES]){
...
A better way of doing this would be reversing the semantics of the method:
if ([content stringIsNotEmpty]) {
this would work finely because when content is nil it would return NO, when it is not nil, it would execute your method.
EDIT:
In Objective-C, sending a message to nil is legal and by definition will evaluate to nil. Google for "objective c sending message to nil".
In another language (C++), your code would crash (actually undefined behaviour, but to make things simple).
I use a small function to test for emptiness. It works on more than just strings:
static inline BOOL isEmpty(id thing) {
return thing == nil
|| ([thing respondsToSelector:#selector(length)]
&& [(NSData *)thing length] == 0)
|| ([thing respondsToSelector:#selector(count)]
&& [(NSArray *)thing count] == 0);
}
I usually import it in my pch file - you can see it along with attribution: https://gist.github.com/325926
As #sergio has already pointed out - when your string is nil you can't send it messages that test it for nil-ness - as sending messages to nil will do nothing for void methods, and return nil where the method returns something.
also
you are calling your method with a parameter
if ([content stringIsEmpty:YES])
but your method declaration doesn't take one:
- (BOOL ) stringIsEmpty {
What's that all about?
You have to check for the 'content == nil' case outside of the method.
If you want to be able to call just one method, change the method to something that tests for a positive, such as "stringHasContent", returning YES if self.length > 0.

NSString "nil or empty" check -- is this complete?

I was writing a small Category on NSString, and I wanted to know if this method is accurately handles all potential use cases:
Update: to clarify -- I wanted to make sure I'm not missing some oddball case involving character encodings, etc..
#implementation NSString (Helpers)
+(BOOL)stringIsNilOrEmpty:(NSString*)aString {
if (!aString)
return YES;
return [aString isEqualToString:#""];
}
#end
Sample usage:
-(void) sampleUsage {
NSString *emptyString = #"";
NSString *nilString = nil;
NSAssert([NSString stringIsNilOrEmpty:nilString] == YES, #"String is nil/empty");
NSAssert([NSString stringIsNilOrEmpty:emptyString] == YES, #"String is nil/empty");
}
#end
I only use the next conditional and do not even need a category:
if (!aString.length)
{
...
}
Using Objective-C theory, a message to NIL will return nil or zero, so basically you do not have to test for nil.
You can simplify the code by removing conditional:
+(BOOL)stringIsNilOrEmpty:(NSString*)aString {
return !(aString && aString.length);
}
#dasblinkenlight's answer is fine, but a much more readable conditional check I would use is:
NSString *string = ...; // define the string
if ([string length] == 0) {
// Do stuff with the string
} else {
// The string is empty or nil here
}
Very concise and does not require a separate convenience function definition. It's easy enough to remember.
EDIT: #Michael G. Emmons posted this as the last comment to that answer... credit to him but I'm listing this as an answer in its own right.
Some examples of this sort of "is not empty or blank" tests as a category on NSString.
// Please note that in general I advocate using a prefix on category methods
// to avoid category collisions. I've not done this here for clarity.
// The #interface is also excluded from this example for brevity.
#implementation NSString (MyAdditions)
- (BOOL)isNotEmpty
{
return [self length] != 0;
}
- (BOOL)isNotBlank
{
if ([self isNotEmpty])
{
NSCharacterSet *nonWhitespaceSet = [[NSCharacterSet whitespaceAndNewlineCharacterSet] invertedSet];
NSRange range = [self rangeOfCharactersFromSet:nonWhitespaceSet];
return range.location != NSNotFound;
}
return NO;
}
#end
Simply Check your string length
> if (!yourString.length){
> //your code } a
message to NIL will return nil or 0, so no need to test for nil :).
Happy coding ...
Make sure to check for spaces, trim white spaces before calculating length.
+(BOOL)stringIsNilOrEmpty:(NSString*)aString {
return !aString || [[aString stringByTrimmingCharactersInSet:
[NSCharacterSet whitespaceCharacterSet]] length] == 0;
}

Can Objective-C switch on NSString?

Is there a more intelligent way to rewrite this?
if ([cardName isEqualToString:#"Six"]) {
[self setValue:6];
} else if ([cardName isEqualToString:#"Seven"]) {
[self setValue:7];
} else if ([cardName isEqualToString:#"Eight"]) {
[self setValue:8];
} else if ([cardName isEqualToString:#"Nine"]) {
[self setValue:9];
}
Unfortunately they cannot. This is one of the best and most sought after utilizations of switch statements, so hopefully they hop on the (now) Java (and others) bandwagon!
If you are doing card names, perhaps assign each card object an integer value and switch on that. Or perhaps an enum, which is considered as a number and can therefore be switched upon.
e.g.
typedef enum{
Ace, Two, Three, Four, Five ... Jack, Queen, King
} CardType;
Done this way, Ace would be be equal to case 0, Two as case 1, etc.
You could set up a dictionary of blocks, like this:
NSString *lookup = #"Hearts"; // The value you want to switch on
typedef void (^CaseBlock)();
// Squint and this looks like a proper switch!
NSDictionary *d = #{
#"Diamonds":
^{
NSLog(#"Riches!");
},
#"Hearts":
^{
self.hearts++;
NSLog(#"Hearts!");
},
#"Clubs":
^{
NSLog(#"Late night coding > late night dancing");
},
#"Spades":
^{
NSLog(#"I'm digging it");
}
};
((CaseBlock)d[lookup])(); // invoke the correct block of code
To have a 'default' section, replace the last line with:
CaseBlock c = d[lookup];
if (c) c(); else { NSLog(#"Joker"); }
Hopefully Apple will teach 'switch' a few new tricks.
For me, a nice easy way:
NSString *theString = #"item3"; // The one we want to switch on
NSArray *items = #[#"item1", #"item2", #"item3"];
int item = [items indexOfObject:theString];
switch (item) {
case 0:
// Item 1
break;
case 1:
// Item 2
break;
case 2:
// Item 3
break;
default:
break;
}
Unfortunately, switch statements can only be used on primitive types. You do have a few options using collections, though.
Probably the best option would be to store each value as an entry in an NSDictionary.
NSDictionary *stringToNumber = [NSDictionary dictionaryWithObjectsAndKeys:
[NSNumber numberWithInt:6],#"Six",
[NSNumber numberWithInt:7],#"Seven",
[NSNumber numberWithInt:8],#"Eight",
[NSNumber numberWithInt:9],#"Nine",
nil];
NSNumber *number = [stringToNumber objectForKey:cardName];
if(number) [self setValue:[number intValue]];
A bit late but for anyone in the future I was able to get this to work for me
#define CASE(str) if ([__s__ isEqualToString:(str)])
#define SWITCH(s) for (NSString *__s__ = (s); ; )
#define DEFAULT
Here is the more intelligent way to write that. It's to use an NSNumberFormatter in the "spell-out style":
NSString *cardName = ...;
NSNumberFormatter *nf = [[NSNumberFormatter alloc] init];
[nf setNumberStyle:NSNumberFormatterSpellOutStyle];
NSNumber *n = [nf numberFromString:[cardName lowercaseString]];
[self setValue:[n intValue]];
[nf release];
Note that the number formatter wants the string to be lowercased, so we have to do that ourselves before passing it in to the formatter.
There are other ways to do that, but switch isn't one of them.
If you only have a few strings, as in your example, the code you have is fine. If you have many cases, you could store the strings as keys in a dictionary and look up the corresponding value:
NSDictionary *cases = #{#"Six" : #6,
#"Seven" : #7,
//...
};
NSNumber *value = [cases objectForKey:cardName];
if (value != nil) {
[self setValue:[value intValue]];
}
BY FAR.. my FAVORITE "ObjC Add-On" is ObjectMatcher
objswitch(someObject)
objcase(#"one") { // Nesting works.
objswitch(#"b")
objcase(#"a") printf("one/a");
objcase(#"b") printf("one/b");
endswitch // Any code can go here, including break/continue/return.
}
objcase(#"two") printf("It's TWO."); // Can omit braces.
objcase(#"three", // Can have multiple values in one case.
nil, // nil can be a "case" value.
[self self], // "Case" values don't have to be constants.
#"tres", #"trois") { printf("It's a THREE."); }
defaultcase printf("None of the above."); // Optional default must be at end.
endswitch
AND it works with non-strings, TOO... in loops, even!
for (id ifNumericWhatIsIt in #[#99, #0, #"shnitzel"])
objswitch(ifNumericWhatIsIt)
objkind(NSNumber) printf("It's a NUMBER.... ");
objswitch([ifNumericWhatIsIt stringValue])
objcase(#"3") printf("It's THREE.\n");
objcase(#"99") printf("It's NINETY-NINE.\n");
defaultcase printf("some other Number.\n");
endswitch
defaultcase printf("It's something else entirely.\n");
endswitch
It's a NUMBER.... It's NINETY-NINE.
It's a NUMBER.... some other Number.
It's something else entirely.
Best of all, there are SO few {...}'s, :'s, and ()'s
Objective-c is no different from c in this aspect, it can only switch on what c can (and the preproc def's like NSInteger, NSUInteger, since they ultimately are just typedef'd to an integral type).
Wikipedia:
c syntax:
The switch statement causes control to be transferred to one of several statements depending on the value of an expression, which must have integral type.
Integral Types:
In computer science, an integer is a datum of integral data type, a
data type which represents some finite subset of the mathematical
integers. Integral data types may be of different sizes and may or may
not be allowed to contain negative values.
I'm kind of late to the party, but to answer the question as stated, there's a more intelligent way:
NSInteger index = [#[#"Six", #"Seven", #"Eight", #"Nine"] indexOfObject:cardName];
if (index != NSNotFound) [self setValue: index + 6];
Note that indexOfObject will look for the match using isEqual:, exactly as in the question.
Building on #Graham Perks idea posted earlier, designed a simple class to make switching on strings fairly simple and clean.
#interface Switcher : NSObject
+ (void)switchOnString:(NSString *)tString
using:(NSDictionary<NSString *, CaseBlock> *)tCases
withDefault:(CaseBlock)tDefaultBlock;
#end
#implementation Switcher
+ (void)switchOnString:(NSString *)tString
using:(NSDictionary<NSString *, CaseBlock> *)tCases
withDefault:(CaseBlock)tDefaultBlock
{
CaseBlock blockToExecute = tCases[tString];
if (blockToExecute) {
blockToExecute();
} else {
tDefaultBlock();
}
}
#end
You would use it like this:
[Switcher switchOnString:someString
using:#{
#"Spades":
^{
NSLog(#"Spades block");
},
#"Hearts":
^{
NSLog(#"Hearts block");
},
#"Clubs":
^{
NSLog(#"Clubs block");
},
#"Diamonds":
^{
NSLog(#"Diamonds block");
}
} withDefault:
^{
NSLog(#"Default block");
}
];
The correct block will execute according to the string.
Gist for this solution
You can use macros approach to achieve it:
#define CASE(str) if ([__s__ isEqualToString:(str)])
#define SWITCH(s) for (NSString *__s__ = (s); ; )
#define DEFAULT
SWITCH (string) {
CASE (#"TestString") {
break;
}
CASE (#"YetAnotherString") {
break;
}
CASE (#"Test") {
break;
}
DEFAULT {
break;
}
}
I can't Comment on cris's answer on #Cris answer but i would like to say that:
There is an LIMITATION for #cris's method:
typedef enum will not take alphanumeric values
typedef enum
{
12Ace, 23Two, 23Three, 23Four, F22ive ... Jack, Queen, King
} CardType;
So here is another One:
Link Stack over flow Go to this user answer "user1717750"
typedef enum
{
Six,
Seven,
Eight
} cardName;
- (void) switchcardName:(NSString *) param {
switch([[cases objectForKey:param] intValue]) {
case Six:
NSLog(#"Six");
break;
case Seven:
NSLog(#"Seven");
break;
case Eight:
NSLog(#"Eight");
break;
default:
NSLog(#"Default");
break;
}
}
Enjoy Coding.....