new text conversion and string code (big changes, take care please)

This commit is contained in:
Vítor Santos Costa
2013-12-02 14:49:41 +00:00
parent 8b7fa9be36
commit d7397b43af
65 changed files with 1498 additions and 2098 deletions

View File

@@ -65,6 +65,54 @@ _PL__utf8_get_char(const char *in, int *chr)
return (char *)in+1;
}
unicode_type_t
_PL__utf8_type(const char *in0, size_t len)
{ /* 2-byte, 0x80-0x7ff */
int chr;
char *in = (char *) in0;
int type = S_ASCII;
while (in[0] != '\0' && in-in0 < len) {
if ( (in[0]&0xe0) == 0xc0 && CONT(1) )
{ chr = ((in[0]&0x1f) << 6)|VAL(1,0);
if (chr > 255) return S_WIDE;
if (chr > 127) type = S_LATIN;
in += 2;
break;
}
/* 3-byte, 0x800-0xffff */
if ( (in[0]&0xf0) == 0xe0 && CONT(1) && CONT(2) )
{ chr = ((in[0]&0xf) << 12)|VAL(1,6)|VAL(2,0);
if (chr > 255) return S_WIDE;
if (chr > 127) type = S_LATIN;
in += 3;
}
/* 4-byte, 0x10000-0x1FFFFF */
if ( (in[0]&0xf8) == 0xf0 && CONT(1) && CONT(2) && CONT(3) )
{ chr = ((in[0]&0x7) << 18)|VAL(1,12)|VAL(2,6)|VAL(3,0);
if (chr > 255) return S_WIDE;
if (chr > 127) type = S_LATIN;
in += 4;
}
/* 5-byte, 0x200000-0x3FFFFFF */
if ( (in[0]&0xfc) == 0xf8 && CONT(1) && CONT(2) && CONT(3) && CONT(4) )
{ chr = ((in[0]&0x3) << 24)|VAL(1,18)|VAL(2,12)|VAL(3,6)|VAL(4,0);
if (chr > 255) return S_WIDE;
if (chr > 127) type = S_LATIN;
in += 5;
}
/* 6-byte, 0x400000-0x7FFFFFF */
if ( (in[0]&0xfe) == 0xfc && CONT(1) && CONT(2) && CONT(3) && CONT(4) && CONT(5) )
{ chr = ((in[0]&0x1) << 30)|VAL(1,24)|VAL(2,18)|VAL(3,12)|VAL(4,6)|VAL(5,0);
if (chr > 255) return S_WIDE;
if (chr > 127) type = S_LATIN;
in += 6;
}
in ++;
}
return type;
}
char *
_PL__utf8_put_char(char *out, int chr)