Linguist: Objective-C is being confused with Mathematica

[tbodt]

This file is actually Objective-C, but the Linguist thinks it's Mathematica. There are no other files that I know of that are like this.

[arfon]

We have a heuristic for this common mis-detection but it looks like the Objective-C regex isn't matching here. There _is_ a match though for Mathematica:

http://rubular.com/r/14CnutbFQq

I wonder if we should add things like NSString NSArray to this regex. Any suggestions on how to improve this @tbodt would be very welcome!

[pchaigno]

In the context of M, MUF, Mathematica, Matlab and Mercury couldn't we use a simpler regex? For instance #(include|import|define).

[tbodt]

How about ^[ \t]*(@(interface|class|protocol|property|end|synchronised|selector|implementation))|(#(include|import|define))\b? A combination of the existing regex and @pchaigno's idea. Or we could just look for message-passing syntax:

/\[ [^\[]+? \s+ \w+(:.+?)? \]/x

This looks for [, some stuff that is not [ (the message target), some space, some word characters (the message name), an optional : followed by some stuff (message arguments), and a ]. I think it would match most Objective-C code. I tried it on a random app delegate from an app I'm working on, and the results can be found at http://rubular.com/r/NJhCNXkjOV.

[pchaigno]

Wouldn't the three keywords include, import and define be sufficient? I'm worried about the complexity (and performance implications) of these regex...

[tbodt]

@pchaigno Hmm. Technically, you can write Objective-C code without any of those:

int main() {
}

But if you want to call any library functions, you need either include or import.

Plus, I now realize that the standard main.m doesn't match the Objective-C regex:

#import <Cocoa/Cocoa.h>

int main(int argc, const char * argv[]) {
    return NSApplicationMain(argc, argv);
}

No special Objective-C keywords. I don't know how that's disambiguated correctly.

[phase]

I think \#(include|import|define) should work fine. If someone is going to create an Objective-C file with only int main(){}, they deserve to have the wrong language. That's probably legal in 10+ languages.

[revolter]

I have this problem too, but I can't share the whole file. It starts like this:

//
//  FileName.m
//  AppName
//
//  Created by Creator on 12/13/13.
//  Copyright (c) 2013 Creator. All rights reserved.
//

#import "FileName.h"

#import "ControllerManager.h"
#import "UserManager.h"
#import "Util.h"

@implementation FileName

[pchaigno]

@tbodt Is this still an issue?

[tbodt]

@pchaigno not for me...

[elmindreda]

This is still a problem for this file.

[Luflosi]

The Objective-C file in the comment above from @elmindreda is still being mis-classified as Mathematica. Could this be because of the comment at the top of the file?

[tbodt]

Right now it's being syntax highlighted as objective-c. My guess is that the classifier is fixed and you just need to wait for the cache to clear.