1.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
     1.2 +++ b/src/utf8.c	Thu Sep 15 10:47:38 2011 +0300
     1.3 @@ -0,0 +1,138 @@
     1.4 +#include "drawtext.h"
     1.5 +
     1.6 +#define	U8_IS_FIRST(x)		(((((x) >> 7) & 1) == 0) || ((((x) >> 6) & 3) == 3))
     1.7 +
     1.8 +static const char first_mask[] = {
     1.9 +	0,
    1.10 +	0x7f,	/* single byte, 7 bits valid */
    1.11 +	0x1f,	/* two-bytes, 5 bits valid */
    1.12 +	0xf,	/* three-bytes, 4 bits valid */
    1.13 +	0x7		/* four-bytes, 3 bits valid */
    1.14 +};
    1.15 +static const char first_shift[] = { 7, 5, 4, 3 };	/* see above */
    1.16 +
    1.17 +#define CONT_PREFIX	0x80
    1.18 +#define CONT_MASK	0x3f
    1.19 +#define CONT_SHIFT	6
    1.20 +
    1.21 +/* last charcodes for 1, 2, 3 or 4-byte utf8 chars */
    1.22 +static const int utf8_lastcode[] = { 0x7f, 0x7ff, 0xfff, 0x1fffff };
    1.23 +
    1.24 +#define prefix_mask(x)	(~first_mask[x])
    1.25 +#define prefix(x)		((prefix_mask(x) << 1) & 0xff)
    1.26 +
    1.27 +
    1.28 +char *dtx_utf8_next_char(char *str)
    1.29 +{
    1.30 +	return str + dtx_utf8_nbytes(str);
    1.31 +}
    1.32 +
    1.33 +int dtx_utf8_char_code(const char *str)
    1.34 +{
    1.35 +	int i, nbytes, shift, code = 0;
    1.36 +	char mask;
    1.37 +
    1.38 +	if(!U8_IS_FIRST(*str)) {
    1.39 +		return -1;
    1.40 +	}
    1.41 +
    1.42 +	nbytes = dtx_utf8_nbytes(str);
    1.43 +	mask = first_mask[nbytes];
    1.44 +	shift = 0;
    1.45 +
    1.46 +	for(i=0; i<nbytes; i++) {
    1.47 +		if(!*str) {
    1.48 +			break;
    1.49 +		}
    1.50 +
    1.51 +		code = (code << shift) | (*str++ & mask);
    1.52 +		mask = 0x3f;
    1.53 +		shift = i == 0 ? first_shift[nbytes] : 6;
    1.54 +	}
    1.55 +
    1.56 +	return code;
    1.57 +}
    1.58 +
    1.59 +int dtx_utf8_nbytes(const char *str)
    1.60 +{
    1.61 +	int i, numset = 0;
    1.62 +	int c = *str;
    1.63 +
    1.64 +	if(!U8_IS_FIRST(c)) {
    1.65 +		for(i=0; !U8_IS_FIRST(str[i]); i++);
    1.66 +		return i;
    1.67 +	}
    1.68 +
    1.69 +	/* count the leading 1s */
    1.70 +	for(i=0; i<4; i++) {
    1.71 +		if(((c >> (7 - i)) & 1) == 0) {
    1.72 +			break;
    1.73 +		}
    1.74 +		numset++;
    1.75 +	}
    1.76 +
    1.77 +	if(!numset) {
    1.78 +		return 1;
    1.79 +	}
    1.80 +	return numset;
    1.81 +}
    1.82 +
    1.83 +int dtx_utf8_char_count(const char *str)
    1.84 +{
    1.85 +	int n = 0;
    1.86 +
    1.87 +	while(*str) {
    1.88 +		n++;
    1.89 +		str = dtx_utf8_next_char((char*)str);
    1.90 +	}
    1.91 +	return n;
    1.92 +}
    1.93 +
    1.94 +size_t dtx_utf8_from_char_code(int code, char *buf)
    1.95 +{
    1.96 +	size_t nbytes = 0;
    1.97 +	int i;
    1.98 +
    1.99 +	for(i=0; i<4; i++) {
   1.100 +		if(code <= utf8_lastcode[i]) {
   1.101 +			nbytes = i + 1;
   1.102 +			break;
   1.103 +		}
   1.104 +	}
   1.105 +
   1.106 +	if(!nbytes && buf) {
   1.107 +		for(i=0; i<nbytes; i++) {
   1.108 +			int idx = nbytes - i - 1;
   1.109 +			int mask, shift, prefix;
   1.110 +
   1.111 +			if(idx > 0) {
   1.112 +				mask = CONT_MASK;
   1.113 +				shift = CONT_SHIFT;
   1.114 +				prefix = CONT_PREFIX;
   1.115 +			} else {
   1.116 +				mask = first_mask[nbytes];
   1.117 +				shift = first_shift[nbytes];
   1.118 +				prefix = prefix(nbytes);
   1.119 +			}
   1.120 +
   1.121 +			buf[idx] = (code & mask) | (prefix & ~mask);
   1.122 +			code >>= shift;
   1.123 +		}
   1.124 +	}
   1.125 +	return nbytes;
   1.126 +}
   1.127 +
   1.128 +size_t dtx_utf8_from_string(const wchar_t *str, char *buf)
   1.129 +{
   1.130 +	size_t nbytes = 0;
   1.131 +	char *ptr = buf;
   1.132 +
   1.133 +	while(*str) {
   1.134 +		int cbytes = dtx_utf8_from_char_code(*str++, ptr);
   1.135 +		if(ptr) {
   1.136 +			ptr += cbytes;
   1.137 +		}
   1.138 +		nbytes += cbytes;
   1.139 +	}
   1.140 +	return nbytes;
   1.141 +}