Re: NSString bug with test and really dodgy patch.

[Ibadinov Marat]

On Oct 3, 2012, at 12:19 PM, Jens Ayton wrote:

In summary, "figure out what Cocoa does." :-)

Under OSX 10.6.8 it misbehaves badly with wide strings, but has no hard feelings about char arrays (not terminated by \0);

I've got following output:

2012-10-03 15:51:34.667 printf[3894:903] abcd

кириллица

2012-10-03 15:51:34.671 printf[3894:903] –∫–∏—Ä–∏–

2012-10-03 15:51:34.671 printf[3894:903] –

2012-10-03 15:51:34.672 printf[3894:903] –

кириллица

кири?

кириллица

For the program: 

#import <Foundation/Foundation.h>

#include "wchar.h"

static

void test_string(const char *cstr, int precision) 

{

    char *array = malloc(strlen(cstr));

    memcpy(array, cstr, strlen(cstr));

    

    NSLog(@"%@", [NSString stringWithFormat:@"%.*s", precision, array]);

}

static

void test_string_long(const wchar_t *wcstr, int precision) 

{

    wchar_t *array = malloc(wcslen(wcstr) * sizeof(wchar_t));

    memcpy(array, wcstr, wcslen(wcstr) * sizeof(wchar_t));

    

    NSLog(@"%@", [NSString stringWithFormat:@"%.*ls", precision, array]);

}

static

void test_printf_long(const wchar_t *wcstr, int precision) 

{

    wchar_t *array = malloc(wcslen(wcstr) * sizeof(wchar_t));

    memcpy(array, wcstr, wcslen(wcstr) * sizeof(wchar_t));

    

    printf("%.*ls\n", precision, array);

}

int main (int argc, const char * argv[])

{

    @autoreleasepool {

        

        test_string("abcd", 4);

        

        printf("%s\n", "кириллица");

        test_string("кириллица", 9);

        

        wchar_t *wcstr = L"кириллица";

        

        test_string_long(wcstr, 9);

        

        NSLog(@"%@", [NSString stringWithFormat:@"%ls", wcstr]);

        

        printf("%ls\n", wcstr);

        

        test_printf_long(wcstr, 9);

        

        test_printf_long(wcstr, 18);

        

    }

    return 0;

}

I presume there is no need to test Apple's Foundation behavior with surrogate UTF-16 pairs.