Styjun

Does anyone know how to write a unit test for NSLog functionality?

I am writing unit tests for all my framework components and it's very import to me to test if logging occurs in various scenarios.

Ex: A configuration that allows a user of your framework to enable or disable logging. It is important that logging occurs if they enable it but this is difficult to verify in unit tests.

Can someone please help me?

Simple way to test is, do NSLog and then check the console log. And code to check that

 NSMutableArray *consoleLog = [NSMutableArray array];

aslclient client = asl_open(NULL, NULL, ASL_OPT_STDERR);

aslmsg query = asl_new(ASL_TYPE_QUERY);
asl_set_query(query, ASL_KEY_MSG, NULL, ASL_QUERY_OP_NOT_EQUAL);
aslresponse response = asl_search(client, query);

asl_free(query);

aslmsg message;
while((message = aslresponse_next(response)))

 const char *msg = asl_get(message, ASL_KEY_MSG);
 [consoleLog addObject:[NSString stringWithFormat:@"%s" , msg]];


NSLog(@"Console: %@", consoleLog);

I would recommend writing a wrapper for NSLog and testing that wrapper thoroughly. Example: you have a class URLStringValidator that exists to validate url strings.

You want to log a message if an empty String is passed in. You would make a protocol to represent your logger.

protocol Logging 
 func log(_ message: String, args: CVarArg...)


struct Logger: Logging 
 func log(_ message: String, args: CVarArg...) 
 NSLog(message, args)
 


class URLValidator 
 let logger: Logging

 init(logger: Logging = Logger()) 
 self.logger = logger
 

 func isEmpty(urlString: String) -> Bool 
 guard !urlString.isEmpty else 
 logger.log("Don't pass an empty string")
 return false
 
 return true
 

Now you can use a fake instead of Logger. I would also argue that you need an additional test to assert that logger has a concrete implementation of the correct type. ex:

class FakeLogger: Logging 
 var capturedMessage: String?

 func log(_ message: String, args: CVarArg...) 
 capturedMessage = message
 


class URLValidatorTests 
 func testLoggerDefaults() 
 XCTAssertTrue(URLValidator().logger is Logger,
 "A url validator should use the correct concrete class for its logger")
 

 func testValidatingEmptyString() 
 let fakeLogger = FakeLogger()
 let validator = URLValidator(logger: fakeLogger)

 validator.isEmpty(urlString: "")
 XCTAssertNil(fakeLogger.capturedMessage)

 validator.isEmpty(urlString: "Not Empty")
 XCTAssertEqual(fakeLogger.capturedMessage, "Don't pass an empty string")
 

Now you can test your logic (the part you care about and should be testing) and leave NSLog as a configuration detail. You could just as easy use os_log or write to a custom file.