icy2utf8.c File Reference

#include "config.h"
#include "intsym.h"
#include "compat.h"

Include dependency graph for icy2utf8.c:

Go to the source code of this file.

Macros
#define	uint8_t   unsigned char
#define	uint16_t   unsigned short

Functions
static int	is_utf8 (const char *src)
char *	icy2utf8 (const char *src, int force)

Variables
static const uint8_t	cp1252_utf8 []
static const uint16_t	tblofs [257]

Macro Definition Documentation

uint16_t

#define uint16_t   unsigned short

Definition at line 37 of file icy2utf8.c.

uint8_t

#define uint8_t   unsigned char

Definition at line 36 of file icy2utf8.c.

Function Documentation

icy2utf8()

char * icy2utf8	(	const char *	src,
		int	force
	)

Definition at line 378 of file icy2utf8.c.

{
    const uint8_t *s = (const uint8_t *)src;
    size_t srclen, dstlen, i, k;
    uint8_t ch, *d;
    char *dst;
 
    /* Some funny streams from Apple/iTunes give ICY info in UTF-8 already.
       So, be prepared and don't try to re-encode such. Unless forced. */
    if(!force && is_utf8(src)) return (compat_strdup(src));
 
    srclen = strlen(src) + 1;
    /* allocate conservatively */
    if ((d = malloc(srclen * 3)) == NULL)
        return (NULL);
 
    i = 0;
    dstlen = 0;
    while (i < srclen) {
        ch = s[i++];
        k = tblofs[ch];
        while (k < tblofs[ch + 1])
            d[dstlen++] = cp1252_utf8[k++];
    }
 
    /* dstlen includes trailing NUL since srclen also does */
    if ((dst = realloc(d, dstlen)) == NULL) {
        free(d);
        return (NULL);
    }
    return (dst);
}

is_utf8()

static int is_utf8 ( const char *  src )

static

Definition at line 336 of file icy2utf8.c.

{
    uint8_t ch;
    size_t i;
    const uint8_t* s = (const uint8_t*) src;
 
    /* We make a loop over every character, until we find a null one.
       Remember: The string is supposed to end with a NUL, so ahead checks are safe. */
    while ((ch = *s++)) {
        /* Ye olde 7bit ASCII chars 'rr fine for anything */
        if(ch < 0x80) continue;
 
        /* Now, we watch out for non-UTF conform sequences. */
        else if ((ch < 0xC2) || (ch > 0xFD))
            return 0;
        /* check for some misformed sequences */
        if (((ch == 0xC2) && (s[0] < 0xA0)) ||
            ((ch == 0xEF) && (s[0] == 0xBF) && (s[1] > 0xBD)))
            /* XXX add more for outside the BMP */
            return 0;
 
        /* Check the continuation bytes. */
        if      (ch < 0xE0) i = 1;
        else if (ch < 0xF0) i = 2;
        else if (ch < 0xF8) i = 3;
        else if (ch < 0xFC) i = 4;
        else
            i = 5;
 
        while (i--)
            if ((*s++ & 0xC0) != 0x80)
                return 0;
    }
 
    /* If no check failed, the string indeed looks like valid UTF-8. */
    return 1;
}

Referenced by icy2utf8().

Variable Documentation

cp1252_utf8

const uint8_t cp1252_utf8[]

static

Definition at line 39 of file icy2utf8.c.

Referenced by icy2utf8().

tblofs

const uint16_t tblofs[257]

static

Definition at line 298 of file icy2utf8.c.

Referenced by icy2utf8().