NSString category for GSM7 encoding and GSM7/UCS-2BE string length evaluation

gistfile1.m

//
//  NSString+SMSCount.m
//  SmsProsto
//
//  Created by Farcaller on 28.11.10.
//  Copyright 2010 Codeneedle LLC. All rights reserved.
//

#import "NSString+SMSCount.h"
#include <iconv.h>
#include <errno.h>

static int gsm7_wctomb(unichar *r, unichar wc)
{
	unsigned char s1, s2;
	s1 = (wc & 0xff00) >> 8;
	s2 = wc & 0x00ff;
	if( s1 == 0x00 ) {
		if( s2 == 0xA || s2 == 0xD ||
		   (s2 >= 0x20 && s2 <= 0x23 ) || (s2 >= 0x25 && s2 <= 0x3f)
		   || (s2 >= 0x41 && s2 <= 0x5A) || (s2 >= 0x61 && s2 <= 0x7A) ) {
			*r = s2;
			return 1;
		} else {
			switch( s2 ) {
				case 0x24: *r = 0x02; return 1;
				case 0x40: *r = 0x00; return 1;
				case 0x5b: *r = 0x1b3c; return 2;
				case 0x5c: *r = 0x1b2f; return 2;
				case 0x5d: *r = 0x1b3e; return 2;
				case 0x5e: *r = 0x1b14; return 2;
				case 0x5f: *r = 0x11; return 1;
				case 0x7b: *r = 0x1b28; return 2;
				case 0x7c: *r = 0x1b40; return 2;
				case 0x7d: *r = 0x1b29; return 2;
				case 0x7e: *r = 0x1b3d; return 2;
				case 0xa3: *r = 0x01; return 1;
				case 0xa4: *r = 0x24; return 1;
				case 0xa5: *r = 0x03; return 1;
				case 0xa7: *r = 0x5f; return 1;
				case 0xb0: *r = 0x24; return 1;
				case 0xbf: *r = 0x60; return 1;
				case 0xc5: *r = 0x0e; return 1;
				case 0xc6: *r = 0x1c; return 1;
				case 0xc7: *r = 0x09; return 1;
				case 0xc9: *r = 0x1f; return 1;
				case 0xc4: *r = 0x5b; return 1;
				case 0xd1: *r = 0x5d; return 1;
				case 0xd6: *r = 0x5c; return 1;
				case 0xd8: *r = 0x0b; return 1;
				case 0xdc: *r = 0x5e; return 1;
				case 0xdf: *r = 0x1e; return 1;
				case 0xe0: *r = 0x7f; return 1;
				case 0xe4: *r = 0x7b; return 1;
				case 0xe5: *r = 0x0f; return 1;
				case 0xe6: *r = 0x1d; return 1;
				case 0xe7: *r = 0x09; return 1;
				case 0xe8: *r = 0x04; return 1;
				case 0xe9: *r = 0x05; return 1;
				case 0xec: *r = 0x07; return 1;
				case 0xf1: *r = 0x7d; return 1;
				case 0xf2: *r = 0x08; return 1;
				case 0xf6: *r = 0x7c; return 1;
				case 0xf8: *r = 0x0c; return 1;
				case 0xf9: *r = 0x06; return 1;
				case 0xfc: *r = 0x7e; return 1;
					/* one way mappings */
				case 0xc0: *r = 0x41; return 1;
				case 0xc1: *r = 0x41; return 1;
				case 0xc2: *r = 0x41; return 1;
				case 0xc3: *r = 0x41; return 1;
				case 0xc8: *r = 0x45; return 1;
				case 0xca: *r = 0x45; return 1;
				case 0xcb: *r = 0x45; return 1;
				case 0xcc: *r = 0x49; return 1;
				case 0xcd: *r = 0x49; return 1;
				case 0xce: *r = 0x49; return 1;
				case 0xcf: *r = 0x49; return 1;
				case 0xd2: *r = 0x4f; return 1;
				case 0xd3: *r = 0x4f; return 1;
				case 0xd4: *r = 0x4f; return 1;
				case 0xd5: *r = 0x4f; return 1;
				case 0xd9: *r = 0x55; return 1;
				case 0xda: *r = 0x55; return 1;
				case 0xdb: *r = 0x55; return 1;
				case 0xdd: *r = 0x59; return 1;
				case 0xe1: *r = 0x61; return 1;
				case 0xe2: *r = 0x61; return 1;
				case 0xe3: *r = 0x61; return 1;
				case 0xea: *r = 0x65; return 1;
				case 0xeb: *r = 0x65; return 1;
				case 0xed: *r = 0x69; return 1;
				case 0xee: *r = 0x69; return 1;
				case 0xef: *r = 0x69; return 1;
				case 0xf3: *r = 0x6f; return 1;
				case 0xf4: *r = 0x6f; return 1;
				case 0xf5: *r = 0x6f; return 1;
				case 0xfa: *r = 0x75; return 1;
				case 0xfb: *r = 0x75; return 1;
				case 0xfd: *r = 0x79; return 1;
				case 0xff: *r = 0x79; return 1;
			}
		}
		return -1;
    } else if( s1 == 0x20 ) {
		if( s2 == 0xac ) {
            *r = 0x1b65; return 2;
		}
    } else if( s1 == 0x03 ) {
		switch( s2 ) {
			case 0x94: *r = 0x10; return 1;
			case 0xa6: *r = 0x12; return 1;
			case 0x93: *r = 0x13; return 1;
			case 0x9b: *r = 0x14; return 1;
			case 0xa9: *r = 0x15; return 1;
			case 0xa0: *r = 0x16; return 1;
			case 0xa8: *r = 0x17; return 1;
			case 0xa3: *r = 0x18; return 1;
			case 0x98: *r = 0x19; return 1;
			case 0x9e: *r = 0x1a; return 1;
			case 0x91: *r = 0x41; return 1;
			case 0x92: *r = 0x42; return 1;
			case 0x95: *r = 0x45; return 1;
			case 0x97: *r = 0x48; return 1;
			case 0x99: *r = 0x49; return 1;
			case 0x9a: *r = 0x4b; return 1;
			case 0x9c: *r = 0x4d; return 1;
			case 0x9d: *r = 0x4e; return 1;
			case 0x9f: *r = 0x4f; return 1;
			case 0xa1: *r = 0x50; return 1;
			case 0xa4: *r = 0x54; return 1;
			case 0xa5: *r = 0x55; return 1;
			case 0xa7: *r = 0x58; return 1;
			case 0x96: *r = 0x5a; return 1;
		}
		return -1;
    }
	return 0;
};

@implementation NSString (SMSCount)

- (NSUInteger)smsLength
{
	char *b = (char*)[self GSM7String];
	if(b) {
		int l = strlen(b);
		free(b);
		return l;
	} else {
		// can't encode, fallback to UCS-2BE
		return [self length]*2;
	}
}

- (const char *)GSM7String
{
	char *ret = NULL;
	NSUInteger len = [self length];
	unichar *ibuf = malloc(len*sizeof(unichar));
	assert(ibuf);
	char *obuf = malloc(len*sizeof(unichar));
	assert(obuf);
	bzero(obuf, len*sizeof(unichar));
	NSRange r;
	r.length = len;
	r.location = 0;
	[self getCharacters:ibuf range:r];
	
	for(int i=0, j=0; i<len; ++i) {
		unichar ic = ibuf[i];
		unichar oc = 0;
		int l = gsm7_wctomb(&oc, ic);
		if(l < 1)
			goto final;
		unsigned char s1 = (oc & 0xff00) >> 8;
		unsigned char s2 = oc & 0x00ff;
		if(s1) {
			obuf[j++] = s1;
		}
		obuf[j++] = s2;
	}
	
	ret = strdup(obuf);

final:
	free(ibuf);
	free(obuf);
	return ret;
}

@end