void
wtf_push_ucs(Str os, wc_uint32 ucs, wc_status *st)
{
wc_ccs ccs;
if (ucs >= WC_C_LANGUAGE_TAG0 && ucs <= WC_C_CANCEL_TAG) {
if (! WcOption.use_language_tag)
return;
if (ucs == WC_C_LANGUAGE_TAG)
st->tag = Strnew_size(4);
else if (ucs == WC_C_CANCEL_TAG) {
st->tag = NULL;
st->ntag = 0;
} else if (st->tag && ucs >= WC_C_TAG_SPACE)
Strcat_char(st->tag, (char)(ucs & 0x7f));
return;
}
if (st->tag) {
st->ntag = wc_ucs_put_tag(st->tag->ptr);
st->tag = NULL;
}
if (ucs < 0x80) {
if (st->ntag)
wtf_push(os, WC_CCS_UCS_TAG, wc_ucs_to_ucs_tag(ucs, st->ntag));
else
Strcat_char(os, (char)ucs);
} else {
ccs = wc_ucs_to_ccs(ucs);
if (st->ntag && ucs <= WC_C_UNICODE_END) {
ccs = wc_ccs_ucs_to_ccs_ucs_tag(ccs);
ucs = wc_ucs_to_ucs_tag(ucs, st->ntag);
}
wtf_push(os, ccs, ucs);
}
}
Str
wc_conv_from_viet(Str is, wc_ces ces)
{
Str os;
wc_uchar *sp = (wc_uchar *)is->ptr;
wc_uchar *ep = sp + is->length;
wc_uchar *p;
wc_ccs ccs1 = WcCesInfo[WC_CCS_INDEX(ces)].gset[1].ccs;
wc_ccs ccs2 = WcCesInfo[WC_CCS_INDEX(ces)].gset[2].ccs;
wc_uint8 *map = NULL;
switch (ces) {
case WC_CES_TCVN_5712:
map = wc_c0_tcvn57122_map;
break;
case WC_CES_VISCII_11:
map = wc_c0_viscii112_map;
break;
case WC_CES_VPS:
map = wc_c0_vps2_map;
break;
}
wc_create_detect_map(ces, WC_FALSE);
for (p = sp; p < ep && ! WC_DETECT_MAP[*p]; p++)
;
if (p == ep)
return is;
os = Strnew_size(is->length);
if (p > sp)
Strcat_charp_n(os, is->ptr, (int)(p - sp));
for (; p < ep; p++) {
if (*p & 0x80)
wtf_push(os, ccs1, (wc_uint32)*p);
else if (*p < 0x20 && map[*p])
wtf_push(os, ccs2, (wc_uint32)*p);
else
Strcat_char(os, (char)*p);
}
return os;
}
Str
wc_char_conv_from_gb18030(wc_uchar c, wc_status *st)
{
static Str os;
static wc_uchar gb[4];
wc_uint32 gbk;
wc_wchar_t cc;
#ifdef USE_UNICODE
wc_uint32 ucs;
#endif
if (st->state == -1) {
st->state = WC_GB18030_NOSTATE;
os = Strnew_size(8);
}
switch (st->state) {
case WC_GB18030_NOSTATE:
switch (WC_GB18030_MAP[c]) {
case UB:
gb[0] = c;
st->state = WC_GB18030_MBYTE1;
return NULL;
case C1:
break;
default:
Strcat_char(os, (char)c);
break;
}
break;
case WC_GB18030_MBYTE1:
if (WC_GB18030_MAP[c] & LB) {
gbk = ((wc_uint32)gb[0] << 8) | c;
if (wc_gbk_or_gbk_ext(gbk) == WC_CCS_GBK_EXT)
wtf_push(os, WC_CCS_GBK_EXT, gbk);
else if (gb[0] >= 0xA1 && c >= 0xA1)
wtf_push(os, wc_gb2312_or_gbk(gbk), gbk);
else
wtf_push(os, WC_CCS_GBK, gbk);
} else if (WC_GB18030_MAP[c] == L4) {
gb[1] = c;
st->state = WC_GB18030_MBYTE2;
return NULL;
}
break;
case WC_GB18030_MBYTE2:
if (WC_GB18030_MAP[c] == UB) {
gb[2] = c;
st->state = WC_GB18030_MBYTE3;
return NULL;
}
break;
case WC_GB18030_MBYTE3:
if (WC_GB18030_MAP[c] == L4) {
cc.ccs = WC_CCS_GB18030_W;
cc.code = ((wc_uint32)gb[0] << 24)
| ((wc_uint32)gb[1] << 16)
| ((wc_uint32)gb[2] << 8)
| c;
#ifdef USE_UNICODE
if (WcOption.gb18030_as_ucs &&
(ucs = wc_gb18030_to_ucs(cc)) != WC_C_UCS4_ERROR)
wtf_push(os, WC_CCS_GB18030 | (wc_ucs_to_ccs(ucs) & ~WC_CCS_A_SET), cc.code);
else
#endif
wtf_push(os, cc.ccs, cc.code);
}
break;
}
st->state = -1;
return os;
}
Str
wc_conv_from_gb18030(Str is, wc_ces ces)
{
Str os;
wc_uchar *sp = (wc_uchar *)is->ptr;
wc_uchar *ep = sp + is->length;
wc_uchar *p;
int state = WC_GB18030_NOSTATE;
wc_uint32 gbk;
wc_wchar_t cc;
#ifdef USE_UNICODE
wc_uint32 ucs;
#endif
for (p = sp; p < ep && *p < 0x80; p++)
;
if (p == ep)
return is;
os = Strnew_size(is->length);
if (p > sp)
Strcat_charp_n(os, (char *)is->ptr, (int)(p - sp));
for (; p < ep; p++) {
switch (state) {
case WC_GB18030_NOSTATE:
switch (WC_GB18030_MAP[*p]) {
case UB:
state = WC_GB18030_MBYTE1;
break;
case C1:
wtf_push_unknown(os, p, 1);
break;
default:
Strcat_char(os, (char)*p);
break;
}
break;
case WC_GB18030_MBYTE1:
if (WC_GB18030_MAP[*p] & LB) {
gbk = ((wc_uint32)*(p-1) << 8) | *p;
if (wc_gbk_or_gbk_ext(gbk) == WC_CCS_GBK_EXT)
wtf_push(os, WC_CCS_GBK_EXT, gbk);
else if (*(p-1) >= 0xA1 && *p >= 0xA1)
wtf_push(os, wc_gb2312_or_gbk(gbk), gbk);
else
wtf_push(os, WC_CCS_GBK, gbk);
} else if (WC_GB18030_MAP[*p] == L4) {
state = WC_GB18030_MBYTE2;
break;
} else
wtf_push_unknown(os, p-1, 2);
state = WC_GB18030_NOSTATE;
break;
case WC_GB18030_MBYTE2:
if (WC_GB18030_MAP[*p] == UB) {
state = WC_GB18030_MBYTE3;
break;
} else
wtf_push_unknown(os, p-2, 3);
state = WC_GB18030_NOSTATE;
break;
case WC_GB18030_MBYTE3:
if (WC_GB18030_MAP[*p] == L4) {
cc.ccs = WC_CCS_GB18030_W;
cc.code = ((wc_uint32)*(p-3) << 24)
| ((wc_uint32)*(p-2) << 16)
| ((wc_uint32)*(p-1) << 8)
| *p;
#ifdef USE_UNICODE
if (WcOption.gb18030_as_ucs &&
(ucs = wc_gb18030_to_ucs(cc)) != WC_C_UCS4_ERROR)
wtf_push(os, WC_CCS_GB18030 | (wc_ucs_to_ccs(ucs) & ~WC_CCS_A_SET), cc.code);
else
#endif
wtf_push(os, cc.ccs, cc.code);
} else
wtf_push_unknown(os, p-3, 4);
state = WC_GB18030_NOSTATE;
break;
}
}
switch (state) {
case WC_GB18030_MBYTE1:
wtf_push_unknown(os, p-1, 1);
break;
case WC_GB18030_MBYTE2:
wtf_push_unknown(os, p-2, 2);
break;
case WC_GB18030_MBYTE3:
wtf_push_unknown(os, p-3, 3);
break;
}
return os;
}
Str wc_conv_from_hz( Str is, wc_ces ces )
{
int eax;
Str os;
wc_uchar *sp = (wc_uchar*)is->ptr;
wc_uchar *ep = &sp[ is->length ];
wc_uchar *p;
int state = 0;
p = sp;
if ( p < ep )
{
if ( p[0] >= 0 )
{
if ( p[0] != '~' )
p++;
else
{
if ( p == ep )
{
is->ptr = (char*)is;
return os;
}
else
{
os = Strnew_size( is->length );
if ( sp < p )
Strcat_charp_n( os, is->ptr, p - sp );
p++;
if ( p < ep )
{
switch ( state )
{
default:
break;
case 0:
if ( p[0] == '~' )
state = 1;
else
{
if ( WC_ISO_MAP[ p[0] ] == '@' )
state = 5;
else
{
if ( p[0] < 0 )
{
wtf_push_unknown( os, p, 1 );
}
if ( os->area_size <= os->length + 1 )
Strgrow( os );
os->ptr[ os->length ] = p[0];
os->length++;
os->ptr[ os->length ] = 0;
else
{
os->ptr[ os->length ] = p[0];
os->length++;
os->ptr[ os->length ] = 0;
}
}
}
break;
case 1:
if ( p[0] == '{' )
state = 3;
else
{
if ( p[0] == '~' )
{
if ( os->area_size <= os->length + 1 )
Strgrow( os );
os->ptr[ os->length ] = p[0];
os->length++;
os->ptr[ os->length ] = 0;
state = 0;
else
{
os->ptr[ os->length ] = p[0];
os->length++;
os->ptr[ os->length ] = 0;
state = 0;
}
}
else
{
if ( p[0] == 10 )
{
}
else
wtf_push_unknown( os, &p[ -1 ], 2 );
state = 0;
}
}
break;
case 2:
if ( p[0] != '}' )
{
if ( p[0] == 10 )
state = 0;
else
{
if ( ( WC_ISO_MAP[ p[0] & 127 ] & 255 ) == 0 )
{
wtf_push( os, 33089, p[0] | ( p[ -1 ] << 8 ) );
}
wtf_push_unknown( os, &p[ -1 ], 2 );
state = 3;
}
}
else
state = 0;
break;
case 3:
if ( p[0] == '~' )
state = 2;
else
{
if ( ( WC_ISO_MAP[ p[0] & 127 ] & 255 ) == 0 )
state = 4;
else
wtf_push_unknown( os, p, 1 );
}
break;
case 4:
if ( ( WC_ISO_MAP[ p[0] & 127 ] & 255 ) == 0 )
{
wtf_push( os, 33089, p[0] | ( p[ -1 ] << 8 ) );
}
wtf_push_unknown( os, &p[ -1 ], 2 );
state = 3;
break;
case 5:
if ( WC_ISO_MAP[ p[0] ] == '@' )
{
wtf_push( os, 33089, p[0] | ( p[ -1 ] << 8 ) );
}
wtf_push_unknown( os, &p[ -1 ], 2 );
state = 0;
break;
}
}
else
{
if ( state <= 5 )
{
if ( ( ( 1 << state ) & 54 ) != 0 )
wtf_push_unknown( os, &p[ -1 ], 1 );
os->ptr = (char*)os;
return os;
}
}
else
{
}
}
}
}
}
}
