void *X(malloc_plain)(size_t n)
{
void *p;
if (n == 0)
n = 1;
p = real_malloc(n);
CK(p);
#ifdef MIN_ALIGMENT
A((((uintptr_t)p) % MIN_ALIGNMENT) == 0);
#endif
return p;
}
/*
* UTF-16 => wchar
*/
int mbfl_filt_conv_utf16_wchar(int c, mbfl_convert_filter *filter)
{
int n, endian;
endian = filter->status & 0xff00;
switch (filter->status & 0x0f) {
case 0:
if (endian) {
n = c & 0xff;
} else {
n = (c & 0xff) << 8;
}
filter->cache |= n;
filter->status++;
break;
default:
if (endian) {
n = (c & 0xff) << 8;
} else {
n = c & 0xff;
}
n |= filter->cache & 0xffff;
filter->status &= ~0x0f;
if (n >= 0xd800 && n < 0xdc00) {
filter->cache = ((n & 0x3ff) << 16) + 0x400000;
} else if (n >= 0xdc00 && n < 0xe000) {
n &= 0x3ff;
n |= (filter->cache & 0xfff0000) >> 6;
filter->cache = 0;
if (n >= MBFL_WCSPLANE_SUPMIN && n < MBFL_WCSPLANE_SUPMAX) {
CK((*filter->output_function)(n, filter->data));
} else { /* illegal character */
n &= MBFL_WCSGROUP_MASK;
n |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(n, filter->data));
}
} else {
/*
* wchar => HZ
*/
int
mbfl_filt_conv_wchar_hz(int c, mbfl_convert_filter *filter)
{
int s;
s = 0;
if (c >= ucs_a1_cp936_table_min && c < ucs_a1_cp936_table_max) {
s = ucs_a1_cp936_table[c - ucs_a1_cp936_table_min];
} else if (c >= ucs_a2_cp936_table_min && c < ucs_a2_cp936_table_max) {
s = ucs_a2_cp936_table[c - ucs_a2_cp936_table_min];
} else if (c >= ucs_a3_cp936_table_min && c < ucs_a3_cp936_table_max) {
s = ucs_a3_cp936_table[c - ucs_a3_cp936_table_min];
} else if (c >= ucs_i_cp936_table_min && c < ucs_i_cp936_table_max) {
s = ucs_i_cp936_table[c - ucs_i_cp936_table_min];
} else if (c >= ucs_hff_cp936_table_min && c < ucs_hff_cp936_table_max) {
s = ucs_hff_cp936_table[c - ucs_hff_cp936_table_min];
}
if (s & 0x8000) {
s -= 0x8080;
}
if (s <= 0) {
if (c == 0) {
s = 0;
} else if (s <= 0) {
s = -1;
}
} else if ((s >= 0x80 && s < 0x2121) || (s > 0x8080)) {
s = -1;
}
if (s >= 0) {
if (s < 0x80) { /* ASCII */
if ((filter->status & 0xff00) != 0) {
CK((*filter->output_function)(0x7e, filter->data)); /* '~' */
CK((*filter->output_function)(0x7d, filter->data)); /* '}' */
}
filter->status = 0;
if (s == 0x7e){
CK((*filter->output_function)(0x7e, filter->data));
}
CK((*filter->output_function)(s, filter->data));
} else { /* GB 2312-80 */
if ((filter->status & 0xff00) != 0x200) {
CK((*filter->output_function)(0x7e, filter->data)); /* '~' */
CK((*filter->output_function)(0x7b, filter->data)); /* '{' */
}
filter->status = 0x200;
CK((*filter->output_function)((s >> 8) & 0x7f, filter->data));
CK((*filter->output_function)(s & 0x7f, filter->data));
}
} else {
if (filter->illegal_mode != MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
int
main (void)
{
size_t b_size = 2;
char *b = malloc (b_size);
ASSERT (b);
/* Ideally we would rely on the snprintf-posix module, in which case
this guard would not be required, but due to limitations in gnulib's
implementation (see modules/snprintf-posix), we cannot. */
if (snprintf (b, b_size, "%ju", (uintmax_t) 3) == 1
&& b[0] == '3' && b[1] == '\0')
{
CK (int, inttostr);
CK (unsigned int, uinttostr);
CK (off_t, offtostr);
CK (uintmax_t, umaxtostr);
CK (intmax_t, imaxtostr);
return 0;
}
/* snprintf doesn't accept %ju; skip this test. */
return 77;
}
/*
* UCS-2BE => wchar
*/
int mbfl_filt_conv_ucs2be_wchar(int c, mbfl_convert_filter *filter)
{
int n;
if (filter->status == 0) {
filter->status = 1;
n = (c & 0xff) << 8;
filter->cache = n;
} else {
filter->status = 0;
n = (c & 0xff) | filter->cache;
CK((*filter->output_function)(n, filter->data));
}
return c;
}
void FlatCodeGen::COND_TRANSLATE()
{
out <<
" _widec = " << GET_KEY() << ";\n";
out <<
" _keys = " << ARR_OFF( CK(), "(" + vCS() + "<<1)" ) << ";\n"
" _conds = " << ARR_OFF( C(), CO() + "[" + vCS() + "]" ) << ";\n"
"\n"
" _slen = " << CSP() << "[" << vCS() << "];\n"
" _cond = _slen > 0 && _keys[0] <=" << GET_WIDE_KEY() << " &&\n"
" " << GET_WIDE_KEY() << " <= _keys[1] ?\n"
" _conds[" << GET_WIDE_KEY() << " - _keys[0]] : 0;\n"
"\n";
out <<
" switch ( _cond ) {\n";
for ( CondSpaceList::Iter csi = condSpaceList; csi.lte(); csi++ ) {
GenCondSpace *condSpace = csi;
out << " case " << condSpace->condSpaceId + 1 << ": {\n";
out << TABS(2) << "_widec = " << CAST(WIDE_ALPH_TYPE()) << "(" <<
KEY(condSpace->baseKey) << " + (" << GET_KEY() <<
" - " << KEY(keyOps->minKey) << "));\n";
for ( GenCondSet::Iter csi = condSpace->condSet; csi.lte(); csi++ ) {
out << TABS(2) << "if ( ";
CONDITION( out, *csi );
Size condValOffset = ((1 << csi.pos()) * keyOps->alphSize());
out << " ) _widec += " << condValOffset << ";\n";
}
out << " }\n";
out << " break;\n";
}
SWITCH_DEFAULT();
out <<
" }\n";
}
int mbfl_filt_conv_byte4be_wchar(int c, mbfl_convert_filter *filter)
{
int n;
if (filter->status == 0) {
filter->status = 1;
n = (c & 0xff) << 24;
filter->cache = n;
} else if (filter->status == 1) {
filter->status = 2;
n = (c & 0xff) << 16;
filter->cache |= n;
} else if (filter->status == 2) {
filter->status = 3;
n = (c & 0xff) << 8;
filter->cache |= n;
} else {
filter->status = 0;
n = (c & 0xff) | filter->cache;
CK((*filter->output_function)(n, filter->data));
}
return c;
}
/*
* armscii8 => wchar
*/
int mbfl_filt_conv_armscii8_wchar(int c, mbfl_convert_filter *filter)
{
int s;
if (c >= 0 && c < armscii8_ucs_table_min) {
s = c;
} else if (c >= armscii8_ucs_table_min && c < 0x100) {
s = armscii8_ucs_table[c - armscii8_ucs_table_min];
if (s <= 0) {
s = c;
s &= MBFL_WCSPLANE_MASK;
s |= MBFL_WCSPLANE_ARMSCII8;
}
} else {
s = c;
s &= MBFL_WCSGROUP_MASK;
s |= MBFL_WCSGROUP_THROUGH;
}
CK((*filter->output_function)(s, filter->data));
return c;
}
/*
* ISO-8859-13 => wchar
*/
int mbfl_filt_conv_8859_13_wchar(int c, mbfl_convert_filter *filter)
{
int s;
if (c >= 0 && c < 0xa0) {
s = c;
} else if (c >= 0xa0 && c < 0x100) {
s = iso8859_13_ucs_table[c - 0xa0];
if (s <= 0) {
s = c;
s &= MBFL_WCSPLANE_MASK;
s |= MBFL_WCSPLANE_8859_13;
}
} else {
s = c;
s &= MBFL_WCSGROUP_MASK;
s |= MBFL_WCSGROUP_THROUGH;
}
CK((*filter->output_function)(s, filter->data));
return c;
}
/*
* any => HTML
*/
int mbfl_filt_conv_html_enc(int c, mbfl_convert_filter *filter)
{
int tmp[64];
int i;
unsigned int uc;
const mbfl_html_entity_entry *e;
if (c < sizeof(htmlentitifieds) / sizeof(htmlentitifieds[0]) &&
htmlentitifieds[c] != 1) {
CK((*filter->output_function)(c, filter->data));
} else {
CK((*filter->output_function)('&', filter->data));
for (i = 0; (e = &mbfl_html_entity_list[i])->name != NULL; i++) {
if (c == e->code) {
char *p;
for (p = e->name; *p != '\0'; p++) {
CK((*filter->output_function)((int)*p, filter->data));
}
goto last;
}
}
{
int *p = tmp + sizeof(tmp) / sizeof(tmp[0]);
CK((*filter->output_function)('#', filter->data));
uc = (unsigned int)c;
*(--p) = '\0';
do {
*(--p) = "0123456789"[uc % 10];
uc /= 10;
} while (uc);
for (; *p != '\0'; p++) {
CK((*filter->output_function)(*p, filter->data));
}
}
last:
CK((*filter->output_function)(';', filter->data));
}
return c;
}
/*
* cp1254 => wchar
*/
int
mbfl_filt_conv_cp1254_wchar(int c, mbfl_convert_filter *filter)
{
int s;
if (c >= 0 && c < cp1254_ucs_table_min) {
s = c;
} else if (c >= cp1254_ucs_table_min && c < 0x100) {
s = cp1254_ucs_table[c - cp1254_ucs_table_min];
if (s <= 0) {
s = c;
s &= MBFL_WCSPLANE_MASK;
s |= MBFL_WCSPLANE_CP1254;
}
} else {
s = c;
s &= MBFL_WCSGROUP_MASK;
s |= MBFL_WCSGROUP_THROUGH;
}
CK((*filter->output_function)(s, filter->data));
return c;
}
int mbfl_filt_conv_html_dec(int c, mbfl_convert_filter *filter)
{
int pos, ent = 0;
mbfl_html_entity_entry *entity;
char *buffer = (char*)filter->opaque;
if (!filter->status) {
if (c == '&' ) {
filter->status = 1;
buffer[0] = '&';
} else {
CK((*filter->output_function)(c, filter->data));
}
} else {
if (c == ';') {
if (buffer[1]=='#') {
if (filter->status > 2 && (buffer[2] == 'x' || buffer[2] == 'X')) {
if (filter->status > 3) {
/* numeric entity */
for (pos=3; pos<filter->status; pos++) {
int v = buffer[pos];
if (v >= '0' && v <= '9') {
v = v - '0';
} else if (v >= 'A' && v <= 'F') {
v = v - 'A' + 10;
} else if (v >= 'a' && v <= 'f') {
v = v - 'a' + 10;
} else {
ent = -1;
break;
}
ent = ent * 16 + v;
}
} else {
ent = -1;
}
} else {
/* numeric entity */
if (filter->status > 2) {
for (pos=2; pos<filter->status; pos++) {
int v = buffer[pos];
if (v >= '0' && v <= '9') {
v = v - '0';
} else {
ent = -1;
break;
}
ent = ent*10 + v;
}
} else {
ent = -1;
}
}
if (ent >= 0 && ent < 0x110000) {
CK((*filter->output_function)(ent, filter->data));
} else {
for (pos = 0; pos < filter->status; pos++) {
CK((*filter->output_function)(buffer[pos], filter->data));
}
CK((*filter->output_function)(c, filter->data));
}
filter->status = 0;
/*php_error_docref("ref.mbstring" TSRMLS_CC, E_NOTICE, "mbstring decoded '%s'=%d", buffer, ent);*/
} else {
/* named entity */
buffer[filter->status] = 0;
entity = (mbfl_html_entity_entry *)mbfl_html_entity_list;
while (entity->name) {
if (!strcmp(buffer+1, entity->name)) {
ent = entity->code;
break;
}
entity++;
}
if (ent) {
/* decoded */
CK((*filter->output_function)(ent, filter->data));
filter->status = 0;
/*php_error_docref("ref.mbstring" TSRMLS_CC, E_NOTICE,"mbstring decoded '%s'=%d", buffer, ent);*/
} else {
/* failure */
buffer[filter->status++] = ';';
buffer[filter->status] = 0;
/* php_error_docref("ref.mbstring" TSRMLS_CC, E_WARNING, "mbstring cannot decode '%s'", buffer); */
mbfl_filt_conv_html_dec_flush(filter);
}
}
} else {
/* add character */
buffer[filter->status++] = c;
/* add character and check */
if (!strchr(html_entity_chars, c) || filter->status+1==html_enc_buffer_size || (c=='#' && filter->status>2))
{
/* illegal character or end of buffer */
if (c=='&')
filter->status--;
buffer[filter->status] = 0;
/* php_error_docref("ref.mbstring" TSRMLS_CC, E_WARNING, "mbstring cannot decode '%s'", buffer)l */
mbfl_filt_conv_html_dec_flush(filter);
if (c=='&')
{
buffer[filter->status++] = '&';
}
}
}
}
return c;
}
/*
* EUC-JP => wchar
*/
int
mbfl_filt_conv_eucjp_wchar(int c, mbfl_convert_filter *filter)
{
int c1, s, w;
switch (filter->status) {
case 0:
if (c >= 0 && c < 0x80) { /* latin */
CK((*filter->output_function)(c, filter->data));
} else if (c > 0xa0 && c < 0xff) { /* X 0208 first char */
filter->status = 1;
filter->cache = c;
} else if (c == 0x8e) { /* kana first char */
filter->status = 2;
} else if (c == 0x8f) { /* X 0212 first char */
filter->status = 3;
} else {
w = c & MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
case 1: /* got first half */
filter->status = 0;
c1 = filter->cache;
if (c > 0xa0 && c < 0xff) {
s = (c1 - 0xa1)*94 + c - 0xa1;
if (s >= 0 && s < jisx0208_ucs_table_size) {
w = jisx0208_ucs_table[s];
} else {
w = 0;
}
if (w <= 0) {
w = ((c1 & 0x7f) << 8) | (c & 0x7f);
w &= MBFL_WCSPLANE_MASK;
w |= MBFL_WCSPLANE_JIS0208;
}
CK((*filter->output_function)(w, filter->data));
} else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
CK((*filter->output_function)(c, filter->data));
} else {
w = (c1 << 8) | c;
w &= MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
case 2: /* got 0x8e */
filter->status = 0;
if (c > 0xa0 && c < 0xe0) {
w = 0xfec0 + c;
CK((*filter->output_function)(w, filter->data));
} else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
CK((*filter->output_function)(c, filter->data));
} else {
w = 0x8e00 | c;
w &= MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
case 3: /* got 0x8f, X 0212 first char */
if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
CK((*filter->output_function)(c, filter->data));
filter->status = 0;
} else {
filter->status++;
filter->cache = c;
}
break;
case 4: /* got 0x8f, X 0212 second char */
filter->status = 0;
c1 = filter->cache;
if (c1 > 0xa0 && c1 < 0xff && c > 0xa0 && c < 0xff) {
s = (c1 - 0xa1)*94 + c - 0xa1;
if (s >= 0 && s < jisx0212_ucs_table_size) {
w = jisx0212_ucs_table[s];
} else {
w = 0;
}
if (w <= 0) {
w = ((c1 & 0x7f) << 8) | (c & 0x7f);
w &= MBFL_WCSPLANE_MASK;
w |= MBFL_WCSPLANE_JIS0212;
}
CK((*filter->output_function)(w, filter->data));
} else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
CK((*filter->output_function)(c, filter->data));
} else {
w = (c1 << 8) | c | 0x8f0000;
w &= MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
default:
filter->status = 0;
break;
}
return c;
}
void TabCodeGen::writeData()
{
/* If there are any transtion functions then output the array. If there
* are none, don't bother emitting an empty array that won't be used. */
if ( redFsm->anyActions() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxActArrItem), A() );
ACTIONS_ARRAY();
CLOSE_ARRAY() <<
"\n";
}
if ( redFsm->anyConditions() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxCondOffset), CO() );
COND_OFFSETS();
CLOSE_ARRAY() <<
"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxCondLen), CL() );
COND_LENS();
CLOSE_ARRAY() <<
"\n";
OPEN_ARRAY( WIDE_ALPH_TYPE(), CK() );
COND_KEYS();
CLOSE_ARRAY() <<
"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxCondSpaceId), C() );
COND_SPACES();
CLOSE_ARRAY() <<
"\n";
}
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxKeyOffset), KO() );
KEY_OFFSETS();
CLOSE_ARRAY() <<
"\n";
OPEN_ARRAY( WIDE_ALPH_TYPE(), K() );
KEYS();
CLOSE_ARRAY() <<
"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxSingleLen), SL() );
SINGLE_LENS();
CLOSE_ARRAY() <<
"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxRangeLen), RL() );
RANGE_LENS();
CLOSE_ARRAY() <<
"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxIndexOffset), IO() );
INDEX_OFFSETS();
CLOSE_ARRAY() <<
"\n";
if ( useIndicies ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxIndex), I() );
INDICIES();
CLOSE_ARRAY() <<
"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxState), TT() );
TRANS_TARGS_WI();
CLOSE_ARRAY() <<
"\n";
if ( redFsm->anyActions() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxActionLoc), TA() );
TRANS_ACTIONS_WI();
CLOSE_ARRAY() <<
"\n";
}
}
else {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxState), TT() );
TRANS_TARGS();
CLOSE_ARRAY() <<
"\n";
if ( redFsm->anyActions() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxActionLoc), TA() );
TRANS_ACTIONS();
CLOSE_ARRAY() <<
"\n";
}
}
if ( redFsm->anyToStateActions() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxActionLoc), TSA() );
TO_STATE_ACTIONS();
CLOSE_ARRAY() <<
"\n";
}
if ( redFsm->anyFromStateActions() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxActionLoc), FSA() );
FROM_STATE_ACTIONS();
CLOSE_ARRAY() <<
"\n";
}
if ( redFsm->anyEofActions() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxActionLoc), EA() );
EOF_ACTIONS();
CLOSE_ARRAY() <<
"\n";
}
if ( redFsm->anyEofTrans() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxIndexOffset+1), ET() );
EOF_TRANS();
CLOSE_ARRAY() <<
"\n";
}
STATE_IDS();
}
/*
* UTF-8 => wchar
*/
int mbfl_filt_conv_utf8_wchar(int c, mbfl_convert_filter *filter)
{
int s, c1, w = 0, flag = 0;
retry:
switch (filter->status & 0xff) {
case 0x00:
if (c < 0x80) {
CK((*filter->output_function)(c, filter->data));
} else if (c >= 0xc2 && c <= 0xdf) { /* 2byte code first char: 0xc2-0xdf */
filter->status = 0x10;
filter->cache = c & 0x1f;
} else if (c >= 0xe0 && c <= 0xef) { /* 3byte code first char: 0xe0-0xef */
filter->status = 0x20;
filter->cache = c & 0xf;
} else if (c >= 0xf0 && c <= 0xf4) { /* 3byte code first char: 0xf0-0xf4 */
filter->status = 0x30;
filter->cache = c & 0x7;
} else {
mbfl_filt_put_invalid_char(c, filter);
}
break;
case 0x10: /* 2byte code 2nd char: 0x80-0xbf */
case 0x21: /* 3byte code 3rd char: 0x80-0xbf */
case 0x32: /* 4byte code 4th char: 0x80-0xbf */
filter->status = 0;
if (c >= 0x80 && c <= 0xbf) {
s = (filter->cache<<6) | (c & 0x3f);
filter->cache = 0;
CK((*filter->output_function)(s, filter->data));
} else {
mbfl_filt_put_invalid_char(filter->cache, filter);
goto retry;
}
break;
case 0x20: /* 3byte code 2nd char: 0:0xa0-0xbf,D:0x80-9F,1-C,E-F:0x80-0x9f */
s = (filter->cache<<6) | (c & 0x3f);
c1 = filter->cache & 0xf;
if ((c >= 0x80 && c <= 0xbf) &&
((c1 == 0x0 && c >= 0xa0) ||
(c1 == 0xd && c < 0xa0) ||
(c1 > 0x0 && c1 != 0xd))) {
filter->cache = s;
filter->status++;
} else {
mbfl_filt_put_invalid_char(filter->cache, filter);
goto retry;
}
break;
case 0x30: /* 4byte code 2nd char: 0:0x90-0xbf,1-3:0x80-0xbf,4:0x80-0x8f */
s = (filter->cache<<6) | (c & 0x3f);
c1 = filter->cache & 0x7;
if ((c >= 0x80 && c <= 0xbf) &&
((c1 == 0x0 && c >= 0x90) ||
(c1 == 0x4 && c < 0x90) ||
(c1 > 0x0 && c1 != 0x4))) {
filter->cache = s;
filter->status++;
} else {
mbfl_filt_put_invalid_char(filter->cache, filter);
goto retry;
}
break;
case 0x31: /* 4byte code 3rd char: 0x80-0xbf */
if (c >= 0x80 && c <= 0xbf) {
filter->cache = (filter->cache<<6) | (c & 0x3f);
filter->status++;
} else {
mbfl_filt_put_invalid_char(filter->cache, filter);
goto retry;
}
break;
default:
filter->status = 0;
break;
}
return c;
}
/*
* wchar => eucJP-win
*/
int
mbfl_filt_conv_wchar_eucjpwin(int c, mbfl_convert_filter *filter)
{
int c1, c2, s1;
s1 = 0;
if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
s1 = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
} else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
s1 = ucs_a2_jis_table[c - ucs_a2_jis_table_min];
} else if (c >= ucs_i_jis_table_min && c < ucs_i_jis_table_max) {
s1 = ucs_i_jis_table[c - ucs_i_jis_table_min];
} else if (c >= ucs_r_jis_table_min && c < ucs_r_jis_table_max) {
s1 = ucs_r_jis_table[c - ucs_r_jis_table_min];
} else if (c >= 0xe000 && c < (0xe000 + 10*94)) { /* user (X0208 85ku - 94ku) */
s1 = c - 0xe000;
c1 = s1/94 + 0x75;
c2 = s1%94 + 0x21;
s1 = (c1 << 8) | c2;
} else if (c >= (0xe000 + 10*94) && c < (0xe000 + 20*94)) { /* user (X0212 85ku - 94ku) */
s1 = c - (0xe000 + 10*94);
c1 = s1/94 + 0xf5;
c2 = s1%94 + 0xa1;
s1 = (c1 << 8) | c2;
}
if (s1 == 0xa2f1) {
s1 = 0x2d62; /* NUMERO SIGN */
}
if (s1 <= 0) {
c1 = c & ~MBFL_WCSPLANE_MASK;
if (c1 == MBFL_WCSPLANE_WINCP932) {
s1 = c & MBFL_WCSPLANE_MASK;
if (s1 >= ((85 + 0x20) << 8)) { /* 85ku - 120ku */
s1 = -1;
}
} else if (c1 == MBFL_WCSPLANE_JIS0208) {
s1 = c & MBFL_WCSPLANE_MASK;
if (s1 >= ((85 + 0x20) << 8)) { /* 85ku - 94ku */
s1 = -1;
}
} else if (c1 == MBFL_WCSPLANE_JIS0212) {
s1 = c & MBFL_WCSPLANE_MASK;
if (s1 >= ((83 + 0x20) << 8)) { /* 83ku - 94ku */
s1 = -1;
} else {
s1 |= 0x8080;
}
} else if (c == 0xa5) { /* YEN SIGN */
s1 = 0x216f; /* FULLWIDTH YEN SIGN */
} else if (c == 0x203e) { /* OVER LINE */
s1 = 0x2131; /* FULLWIDTH MACRON */
} else if (c == 0xff3c) { /* FULLWIDTH REVERSE SOLIDUS */
s1 = 0x2140;
} else if (c == 0xff5e) { /* FULLWIDTH TILDE */
s1 = 0x2141;
} else if (c == 0x2225) { /* PARALLEL TO */
s1 = 0x2142;
} else if (c == 0xff0d) { /* FULLWIDTH HYPHEN-MINUS */
s1 = 0x215d;
} else if (c == 0xffe0) { /* FULLWIDTH CENT SIGN */
s1 = 0x2171;
} else if (c == 0xffe1) { /* FULLWIDTH POUND SIGN */
s1 = 0x2172;
} else if (c == 0xffe2) { /* FULLWIDTH NOT SIGN */
s1 = 0x224c;
} else if (c == 0xff5e) { /* FULLWIDTH TILDE */
s1 = 0x2141;
} else {
s1 = -1;
c1 = 0;
c2 = cp932ext1_ucs_table_max - cp932ext1_ucs_table_min;
while (c1 < c2) { /* CP932 vendor ext1 (13ku) */
const int oh = cp932ext1_ucs_table_min / 94;
if (c == cp932ext1_ucs_table[c1]) {
s1 = ((c1 / 94 + oh + 0x21) << 8) + (c1 % 94 + 0x21);
break;
}
c1++;
}
if (s1 < 0) {
c1 = 0;
c2 = cp932ext3_ucs_table_max - cp932ext3_ucs_table_min;
while (c1 < c2) { /* CP932 vendor ext3 (115ku - 119ku) */
if (c == cp932ext3_ucs_table[c1]) {
if (c1 < cp932ext3_eucjp_table_size) {
s1 = cp932ext3_eucjp_table[c1];
}
break;
}
c1++;
}
}
}
if (c == 0) {
s1 = 0;
} else if (s1 <= 0) {
s1 = -1;
}
}
if (s1 >= 0) {
if (s1 < 0x80) { /* latin */
CK((*filter->output_function)(s1, filter->data));
} else if (s1 < 0x100) { /* kana */
CK((*filter->output_function)(0x8e, filter->data));
CK((*filter->output_function)(s1, filter->data));
} else if (s1 < 0x8080) { /* X 0208 */
CK((*filter->output_function)(((s1 >> 8) & 0xff) | 0x80, filter->data));
CK((*filter->output_function)((s1 & 0xff) | 0x80, filter->data));
} else { /* X 0212 */
/*
* HZ => wchar
*/
int
mbfl_filt_conv_hz_wchar(int c, mbfl_convert_filter *filter)
{
int c1, s, w;
switch (filter->status & 0xf) {
/* case 0x00: ASCII */
/* case 0x10: GB2312 */
case 0:
if (c == 0x7e) {
filter->status += 2;
} else if (filter->status == 0x10 && c > 0x20 && c < 0x7f) { /* DBCS first char */
filter->cache = c;
filter->status += 1;
} else if (c >= 0 && c < 0x80) { /* latin, CTLs */
CK((*filter->output_function)(c, filter->data));
} else {
w = c & MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
/* case 0x11: GB2312 second char */
case 1:
filter->status &= ~0xf;
c1 = filter->cache;
if (c1 > 0x20 && c1 < 0x7f && c > 0x20 && c < 0x7f) {
s = (c1 - 1)*192 + c + 0x40; /* GB2312 */
if (s >= 0 && s < cp936_ucs_table_size) {
w = cp936_ucs_table[s];
} else {
w = 0;
}
if (w <= 0) {
w = (c1 << 8) | c;
w &= MBFL_WCSPLANE_MASK;
w |= MBFL_WCSPLANE_GB2312;
}
CK((*filter->output_function)(w, filter->data));
} else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
CK((*filter->output_function)(c, filter->data));
} else {
w = (c1 << 8) | c;
w &= MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
/* '~' */
case 2:
if (c == 0x7d) { /* '}' */
filter->status = 0x0;
} else if (c == 0x7b) { /* '{' */
filter->status = 0x10;
} else if (c == 0x7e) { /* '~' */
filter->status = 0x0;
CK((*filter->output_function)(0x007e, filter->data));
}
break;
default:
filter->status = 0;
break;
}
return c;
}
/*
* wchar => cp51932
*/
int
mbfl_filt_conv_wchar_cp51932(int c, mbfl_convert_filter *filter)
{
int c1, c2, s1;
s1 = 0;
if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
s1 = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
} else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
s1 = ucs_a2_jis_table[c - ucs_a2_jis_table_min];
} else if (c >= ucs_i_jis_table_min && c < ucs_i_jis_table_max) {
s1 = ucs_i_jis_table[c - ucs_i_jis_table_min];
} else if (c >= ucs_r_jis_table_min && c < ucs_r_jis_table_max) {
s1 = ucs_r_jis_table[c - ucs_r_jis_table_min];
}
if (s1 >= 0x8080) s1 = -1; /* we don't support JIS X0213 */
if (s1 <= 0) {
c1 = c & ~MBFL_WCSPLANE_MASK;
if (c1 == MBFL_WCSPLANE_WINCP932) {
s1 = c & MBFL_WCSPLANE_MASK;
if (s1 >= ((85 + 0x20) << 8)) { /* 85ku - 120ku */
s1 = -1;
}
} else if (c1 == MBFL_WCSPLANE_JIS0208) {
s1 = c & MBFL_WCSPLANE_MASK;
if ((s1 >= ((85 + 0x20) << 8) && /* 85ku - 94ku */
s1 <= ((88 + 0x20) << 8)) ||/* IBM extension */
(s1 >= ((93 + 0x20) << 8) && /* 89ku - 92ku */
s1 <= ((94 + 0x20) << 8))) {
s1 = -1;
}
} else if (c == 0xa5) { /* YEN SIGN */
s1 = 0x005c; /* YEN SIGN */
} else if (c == 0x203e) { /* OVER LINE */
s1 = 0x007e; /* FULLWIDTH MACRON */
} else if (c == 0xff3c) { /* FULLWIDTH REVERSE SOLIDUS */
s1 = 0x2140;
} else if (c == 0xff5e) { /* FULLWIDTH TILDE */
s1 = 0x2141;
} else if (c == 0x2225) { /* PARALLEL TO */
s1 = 0x2142;
} else if (c == 0xff0d) { /* FULLWIDTH HYPHEN-MINUS */
s1 = 0x215d;
} else if (c == 0xffe0) { /* FULLWIDTH CENT SIGN */
s1 = 0x2171;
} else if (c == 0xffe1) { /* FULLWIDTH POUND SIGN */
s1 = 0x2172;
} else if (c == 0xffe2) { /* FULLWIDTH NOT SIGN */
s1 = 0x224c;
} else {
s1 = -1;
c1 = 0;
c2 = cp932ext1_ucs_table_max - cp932ext1_ucs_table_min;
while (c1 < c2) { /* CP932 vendor ext1 (13ku) */
if (c == cp932ext1_ucs_table[c1]) {
s1 = ((c1/94 + 0x2d) << 8) + (c1%94 + 0x21);
break;
}
c1++;
}
if (s1 < 0) {
c1 = 0;
c2 = cp932ext2_ucs_table_max - cp932ext2_ucs_table_min;
while (c1 < c2) { /* CP932 vendor ext3 (115ku - 119ku) */
if (c == cp932ext2_ucs_table[c1]) {
s1 = ((c1/94 + 0x79) << 8) +(c1%94 + 0x21);
break;
}
c1++;
}
}
}
if (c == 0) {
s1 = 0;
} else if (s1 <= 0) {
s1 = -1;
}
}
if (s1 >= 0) {
if (s1 < 0x80) { /* latin */
CK((*filter->output_function)(s1, filter->data));
} else if (s1 < 0x100) { /* kana */
CK((*filter->output_function)(0x8e, filter->data));
CK((*filter->output_function)(s1, filter->data));
} else if (s1 < 0x8080) { /* X 0208 */
CK((*filter->output_function)(((s1 >> 8) & 0xff) | 0x80, filter->data));
CK((*filter->output_function)((s1 & 0xff) | 0x80, filter->data));
} else {
if (filter->illegal_mode != MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
/*
* JIS-ms => wchar
*/
int
mbfl_filt_conv_jis_ms_wchar(int c, mbfl_convert_filter *filter)
{
int c1, s, w;
retry:
switch (filter->status & 0xf) {
/* case 0x00: ASCII */
/* case 0x10: X 0201 latin */
/* case 0x20: X 0201 kana */
/* case 0x80: X 0208 */
/* case 0x90: X 0212 */
case 0:
if (c == 0x1b) {
filter->status += 2;
} else if (c == 0x0e) { /* "kana in" */
filter->status = 0x20;
} else if (c == 0x0f) { /* "kana out" */
filter->status = 0;
} else if (filter->status == 0x10 && c == 0x5c) { /* YEN SIGN */
CK((*filter->output_function)(0xa5, filter->data));
} else if (filter->status == 0x10 && c == 0x7e) { /* OVER LINE */
CK((*filter->output_function)(0x203e, filter->data));
} else if (filter->status == 0x20 && c > 0x20 && c < 0x60) { /* kana */
CK((*filter->output_function)(0xff40 + c, filter->data));
} else if ((filter->status == 0x80 || filter->status == 0x90) && c > 0x20 && c < 0x93) { /* kanji first char */
filter->cache = c;
filter->status += 1;
} else if (c >= 0 && c < 0x80) { /* latin, CTLs */
CK((*filter->output_function)(c, filter->data));
} else if (c > 0xa0 && c < 0xe0) { /* GR kana */
CK((*filter->output_function)(0xfec0 + c, filter->data));
} else {
w = c & MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
/* case 0x81: X 0208 second char */
/* case 0x91: X 0212 second char */
case 1:
filter->status &= ~0xf;
c1 = filter->cache;
if (c > 0x20 && c < 0x7f) {
s = (c1 - 0x21)*94 + c - 0x21;
if (filter->status == 0x80) {
if (s >= 0 && s < jisx0208_ucs_table_size) {
w = jisx0208_ucs_table[s];
} else if (s >= cp932ext1_ucs_table_min && s < cp932ext1_ucs_table_max) {
w = cp932ext1_ucs_table[s - cp932ext1_ucs_table_min];
} else if (s >= cp932ext2_ucs_table_min && s < cp932ext2_ucs_table_max) {
w = cp932ext2_ucs_table[s - cp932ext2_ucs_table_min];
} else if (s >= cp932ext3_ucs_table_min && s < cp932ext3_ucs_table_max) {
w = cp932ext3_ucs_table[s - cp932ext3_ucs_table_min];
} else if (s >= 94 * 94 && s < 114 * 94) {
/* user-defined => PUA (Microsoft extended) */
w = s - 94*94 + 0xe000;
} else {
w = 0;
}
if (w <= 0) {
w = (c1 << 8) | c;
w &= MBFL_WCSPLANE_MASK;
w |= MBFL_WCSPLANE_JIS0208;
}
} else {
if (s >= 0 && s < jisx0212_ucs_table_size) {
w = jisx0212_ucs_table[s];
} else {
w = 0;
}
if (w <= 0) {
w = (c1 << 8) | c;
w &= MBFL_WCSPLANE_MASK;
w |= MBFL_WCSPLANE_JIS0212;
}
}
CK((*filter->output_function)(w, filter->data));
} else if (c == 0x1b) {
filter->status += 2;
} else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
CK((*filter->output_function)(c, filter->data));
} else {
w = (c1 << 8) | c;
w &= MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
/* ESC */
/* case 0x02: */
/* case 0x12: */
/* case 0x22: */
/* case 0x82: */
/* case 0x92: */
case 2:
if (c == 0x24) { /* '$' */
filter->status++;
} else if (c == 0x28) { /* '(' */
filter->status += 3;
} else {
filter->status &= ~0xf;
CK((*filter->output_function)(0x1b, filter->data));
goto retry;
}
break;
/* ESC $ */
/* case 0x03: */
/* case 0x13: */
/* case 0x23: */
/* case 0x83: */
/* case 0x93: */
case 3:
if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
filter->status = 0x80;
} else if (c == 0x28) { /* '(' */
filter->status++;
} else {
filter->status &= ~0xf;
CK((*filter->output_function)(0x1b, filter->data));
CK((*filter->output_function)(0x24, filter->data));
goto retry;
}
break;
/* ESC $ ( */
/* case 0x04: */
/* case 0x14: */
/* case 0x24: */
/* case 0x84: */
/* case 0x94: */
case 4:
if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
filter->status = 0x80;
} else if (c == 0x44) { /* 'D' */
filter->status = 0x90;
} else {
filter->status &= ~0xf;
CK((*filter->output_function)(0x1b, filter->data));
CK((*filter->output_function)(0x24, filter->data));
CK((*filter->output_function)(0x28, filter->data));
goto retry;
}
break;
/* ESC ( */
/* case 0x05: */
/* case 0x15: */
/* case 0x25: */
/* case 0x85: */
/* case 0x95: */
case 5:
if (c == 0x42 || c == 0x48) { /* 'B' or 'H' */
filter->status = 0;
} else if (c == 0x4a) { /* 'J' */
filter->status = 0x10;
} else if (c == 0x49) { /* 'I' */
filter->status = 0x20;
} else {
filter->status &= ~0xf;
CK((*filter->output_function)(0x1b, filter->data));
CK((*filter->output_function)(0x28, filter->data));
goto retry;
}
break;
default:
filter->status = 0;
break;
}
return c;
}
void FFlatCodeGen::writeData()
{
if ( redFsm->anyConditions() ) {
OPEN_ARRAY( WIDE_ALPH_TYPE(), CK() );
COND_KEYS();
CLOSE_ARRAY() <<
L"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxCondSpan), CSP() );
COND_KEY_SPANS();
CLOSE_ARRAY() <<
L"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxCond), C() );
CONDS();
CLOSE_ARRAY() <<
L"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxCondIndexOffset), CO() );
COND_INDEX_OFFSET();
CLOSE_ARRAY() <<
L"\n";
}
OPEN_ARRAY( WIDE_ALPH_TYPE(), K() );
KEYS();
CLOSE_ARRAY() <<
L"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxSpan), SP() );
KEY_SPANS();
CLOSE_ARRAY() <<
L"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxFlatIndexOffset), IO() );
FLAT_INDEX_OFFSET();
CLOSE_ARRAY() <<
L"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxIndex), I() );
INDICIES();
CLOSE_ARRAY() <<
L"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxState), TT() );
TRANS_TARGS();
CLOSE_ARRAY() <<
L"\n";
if ( redFsm->anyActions() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxActListId), TA() );
TRANS_ACTIONS();
CLOSE_ARRAY() <<
L"\n";
}
if ( redFsm->anyToStateActions() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxActionLoc), TSA() );
TO_STATE_ACTIONS();
CLOSE_ARRAY() <<
L"\n";
}
if ( redFsm->anyFromStateActions() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxActionLoc), FSA() );
FROM_STATE_ACTIONS();
CLOSE_ARRAY() <<
L"\n";
}
if ( redFsm->anyEofActions() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxActListId), EA() );
EOF_ACTIONS();
CLOSE_ARRAY() <<
L"\n";
}
if ( redFsm->anyEofTrans() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxIndexOffset+1), ET() );
EOF_TRANS();
CLOSE_ARRAY() <<
L"\n";
}
STATE_IDS();
}
/*
* UHC => wchar
*/
int
mbfl_filt_conv_uhc_wchar(int c, mbfl_convert_filter *filter)
{
int c1, w = 0, flag = 0;
switch (filter->status) {
case 0:
if (c >= 0 && c < 0x80) { /* latin */
CK((*filter->output_function)(c, filter->data));
} else if (c > 0x80 && c < 0xff && c != 0xc9) { /* dbcs lead byte */
filter->status = 1;
filter->cache = c;
} else {
w = c & MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
case 1: /* dbcs second byte */
filter->status = 0;
c1 = filter->cache;
if ( c1 >= 0x81 && c1 <= 0xa0){
w = (c1 - 0x81)*190 + (c - 0x41);
if (w >= 0 && w < uhc1_ucs_table_size) {
flag = 1;
w = uhc1_ucs_table[w];
} else {
w = 0;
}
} else if ( c1 >= 0xa1 && c1 <= 0xc6){
w = (c1 - 0xa1)*190 + (c - 0x41);
if (w >= 0 && w < uhc2_ucs_table_size) {
flag = 2;
w = uhc2_ucs_table[w];
} else {
w = 0;
}
} else if ( c1 >= 0xc7 && c1 <= 0xfe){
w = (c1 - 0xc7)*94 + (c - 0xa1);
if (w >= 0 && w < uhc3_ucs_table_size) {
flag = 3;
w = uhc3_ucs_table[w];
} else {
w = 0;
}
}
if (flag > 0){
if (w <= 0) {
w = (c1 << 8) | c;
w &= MBFL_WCSPLANE_MASK;
w |= MBFL_WCSPLANE_UHC;
}
CK((*filter->output_function)(w, filter->data));
} else {
if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
CK((*filter->output_function)(c, filter->data));
} else {
w = (c1 << 8) | c;
w &= MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
}
break;
default:
filter->status = 0;
break;
}
return c;
}
R X(float_property)(const float_property p)
{
const R base = FLT_RADIX;
const R eps = EPSILON;
const R t = MANT_DIG;
const R emin = MIN_EXP;
const R emax = MAX_EXP;
const R prec = eps * base;
static R rmin = K(1.0);
static R rmax = K(1.0);
const R rnd = FLTROUND;
static R sfmin = K(-1.0);
static short first = TRUE;
if (first)
{
/* Compute rmin */
{
const INT n = 1 - MIN_EXP;
INT i;
for (i = 0; i < n; i++)
rmin /= base;
}
/* Compute rmax */
{
INT i;
rmax -= eps;
for (i = 0; i < emax; i++)
rmax *= base;
}
/* Compute sfmin */
{
R small = K(1.0) / rmax;
sfmin = rmin;
if (small >= sfmin)
sfmin = small * (eps + K(1.0));
}
first = FALSE;
}
if (p == NFFT_EPSILON)
return eps;
else if (p == NFFT_SAFE_MIN)
return sfmin;
else if (p == NFFT_BASE)
return base;
else if (p == NFFT_PRECISION)
return prec;
else if (p == NFFT_MANT_DIG)
return t;
else if (p == NFFT_FLTROUND)
return rnd;
else if (p == NFFT_E_MIN)
return emin;
else if (p == NFFT_R_MIN)
return rmin;
else if (p == NFFT_E_MAX)
return emax;
else if (p == NFFT_R_MAX)
return rmax;
else
CK(0 /* cannot happen */);
return K(-1.0);
} /* dlamch_ */
void CSharpFlatCodeGen::writeData()
{
/* If there are any transtion functions then output the array. If there
* are none, don't bother emitting an empty array that won't be used. */
if ( redFsm->anyActions() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxActArrItem), A() );
ACTIONS_ARRAY();
CLOSE_ARRAY() <<
"\n";
}
if ( redFsm->anyConditions() ) {
OPEN_ARRAY( WIDE_ALPH_TYPE(), CK() );
COND_KEYS();
CLOSE_ARRAY() <<
"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxCondSpan), CSP() );
COND_KEY_SPANS();
CLOSE_ARRAY() <<
"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxCond), C() );
CONDS();
CLOSE_ARRAY() <<
"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxCondIndexOffset), CO() );
COND_INDEX_OFFSET();
CLOSE_ARRAY() <<
"\n";
}
OPEN_ARRAY( WIDE_ALPH_TYPE(), K() );
KEYS();
CLOSE_ARRAY() <<
"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxSpan), SP() );
KEY_SPANS();
CLOSE_ARRAY() <<
"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxFlatIndexOffset), IO() );
FLAT_INDEX_OFFSET();
CLOSE_ARRAY() <<
"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxIndex), I() );
INDICIES();
CLOSE_ARRAY() <<
"\n";
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxState), TT() );
TRANS_TARGS();
CLOSE_ARRAY() <<
"\n";
if ( redFsm->anyActions() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxActionLoc), TA() );
TRANS_ACTIONS();
CLOSE_ARRAY() <<
"\n";
}
if ( redFsm->anyToStateActions() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxActionLoc), TSA() );
TO_STATE_ACTIONS();
CLOSE_ARRAY() <<
"\n";
}
if ( redFsm->anyFromStateActions() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxActionLoc), FSA() );
FROM_STATE_ACTIONS();
CLOSE_ARRAY() <<
"\n";
}
if ( redFsm->anyEofActions() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxActionLoc), EA() );
EOF_ACTIONS();
CLOSE_ARRAY() <<
"\n";
}
if ( redFsm->anyEofTrans() ) {
OPEN_ARRAY( ARRAY_TYPE(redFsm->maxIndexOffset+1), ET() );
EOF_TRANS();
CLOSE_ARRAY() <<
"\n";
}
STATE_IDS();
}
/*
* EUC-TW => wchar
*/
int
mbfl_filt_conv_euctw_wchar(int c, mbfl_convert_filter *filter)
{
int c1, s, w, plane;
switch (filter->status) {
case 0:
if (c >= 0 && c < 0x80) { /* latin */
CK((*filter->output_function)(c, filter->data));
} else if (c > 0xa0 && c < 0xff) { /* dbcs first byte */
filter->status = 1;
filter->cache = c;
} else if (c == 0x8e) { /* mbcs first byte */
filter->status = 2;
filter->cache = c;
} else {
w = c & MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
case 1: /* mbcs second byte */
filter->status = 0;
c1 = filter->cache;
if (c > 0xa0 && c < 0xff) {
w = (c1 - 0xa1)*94 + (c - 0xa1);
if (w >= 0 && w < cns11643_1_ucs_table_size) {
w = cns11643_1_ucs_table[w];
} else {
w = 0;
}
if (w <= 0) {
w = (c1 << 8) | c;
w &= MBFL_WCSPLANE_MASK;
w |= MBFL_WCSPLANE_CNS11643;
}
CK((*filter->output_function)(w, filter->data));
} else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
CK((*filter->output_function)(c, filter->data));
} else {
w = (c1 << 8) | c;
w &= MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
case 2: /* got 0x8e, first char */
c1 = filter->cache;
if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
CK((*filter->output_function)(c, filter->data));
filter->status = 0;
} else if (c > 0xa0 && c < 0xaf) {
filter->status = 3;
filter->cache = c - 0xa1;
} else {
w = (c1 << 8) | c;
w &= MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
case 3: /* got 0x8e, third char */
filter->status = 0;
c1 = filter->cache;
if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
CK((*filter->output_function)(c, filter->data));
filter->status = 0;
} else if (c > 0xa0 && c < 0xff) {
filter->status = 4;
filter->cache = (c1 << 8) + c - 0xa1;
} else {
w = (c1 << 8) | c;
w &= MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
case 4: /* mbcs fourth char */
filter->status = 0;
c1 = filter->cache;
if (c1 >= 0x100 && c1 <= 0xdff && c > 0xa0 && c < 0xff) {
plane = (c1 & 0xf00) >> 8;
s = (c1 & 0xff)*94 + c - 0xa1;
w = 0;
if (s >= 0) {
if (plane == 1 && s < cns11643_2_ucs_table_size) {
w = cns11643_2_ucs_table[s];
}
if (plane == 13 && s < cns11643_14_ucs_table_size) {
w = cns11643_14_ucs_table[s];
}
}
if (w <= 0) {
w = ((c1 & 0x7f) << 8) | (c & 0x7f);
w &= MBFL_WCSPLANE_MASK;
w |= MBFL_WCSPLANE_CNS11643;
}
CK((*filter->output_function)(w, filter->data));
} else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
/*
* UTF-8 => wchar
*/
int mbfl_filt_conv_utf8_wchar(int c, mbfl_convert_filter *filter)
{
int s;
if (c < 0x80) {
if (c >= 0) {
CK((*filter->output_function)(c, filter->data));
}
filter->status = 0;
} else if (c < 0xc0) {
int status = filter->status & 0xff;
switch (status) {
case 0x10: /* 2byte code 2nd char */
case 0x21: /* 3byte code 3rd char */
case 0x32: /* 4byte code 4th char */
case 0x43: /* 5byte code 5th char */
case 0x54: /* 6byte code 6th char */
filter->status = 0;
s = filter->cache | (c & 0x3f);
if ((status == 0x10 && s >= 0x80) ||
(status == 0x21 && s >= 0x800 && (s < 0xd800 || s > 0xdfff)) ||
(status == 0x32 && s >= 0x10000) ||
(status == 0x43 && s >= 0x200000) ||
(status == 0x54 && s >= 0x4000000 && s < MBFL_WCSGROUP_UCS4MAX)) {
CK((*filter->output_function)(s, filter->data));
}
break;
case 0x20: /* 3byte code 2nd char */
case 0x31: /* 4byte code 3rd char */
case 0x42: /* 5byte code 4th char */
case 0x53: /* 6byte code 5th char */
filter->cache |= ((c & 0x3f) << 6);
filter->status++;
break;
case 0x30: /* 4byte code 2nd char */
case 0x41: /* 5byte code 3rd char */
case 0x52: /* 6byte code 4th char */
filter->cache |= ((c & 0x3f) << 12);
filter->status++;
break;
case 0x40: /* 5byte code 2nd char */
case 0x51: /* 6byte code 3rd char */
filter->cache |= ((c & 0x3f) << 18);
filter->status++;
break;
case 0x50: /* 6byte code 2nd char */
filter->cache |= ((c & 0x3f) << 24);
filter->status++;
break;
default:
filter->status = 0;
break;
}
} else if (c < 0xe0) { /* 2byte code first char */
filter->status = 0x10;
filter->cache = (c & 0x1f) << 6;
} else if (c < 0xf0) { /* 3byte code first char */
filter->status = 0x20;
filter->cache = (c & 0xf) << 12;
} else if (c < 0xf8) { /* 4byte code first char */
filter->status = 0x30;
filter->cache = (c & 0x7) << 18;
} else if (c < 0xfc) { /* 5byte code first char */
filter->status = 0x40;
filter->cache = (c & 0x3) << 24;
} else if (c < 0xfe) { /* 6 byte code first char */
filter->status = 0x50;
filter->cache = (c & 0x1) << 30;
} else {
filter->status = 0;
filter->cache = 0;
}
return c;
}
/*
* ISO-2022-JP-Mobile => wchar
*/
int
mbfl_filt_conv_2022jp_mobile_wchar(int c, mbfl_convert_filter *filter)
{
int c1, s, w, snd = 0;
retry:
switch (filter->status & 0xf) {
/* case 0x00: ASCII */
/* case 0x10: X 0201 latin */
/* case 0x20: X 0201 kana */
/* case 0x80: X 0208 */
case 0:
if (c == 0x1b) {
filter->status += 2;
} else if (filter->status == 0x20 && c > 0x20 && c < 0x60) { /* kana */
CK((*filter->output_function)(0xff40 + c, filter->data));
} else if (filter->status == 0x80 && c > 0x20 && c < 0x80) { /* kanji first char */
filter->cache = c;
filter->status += 1;
} else if (c >= 0 && c < 0x80) { /* latin, CTLs */
CK((*filter->output_function)(c, filter->data));
} else if (c > 0xa0 && c < 0xe0) { /* GR kana */
CK((*filter->output_function)(0xfec0 + c, filter->data));
} else {
w = c & MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
/* case 0x81: X 0208 second char */
case 1:
w = 0;
filter->status &= ~0xf;
c1 = filter->cache;
if (c > 0x20 && c < 0x7f) {
s = (c1 - 0x21)*94 + c - 0x21;
if (s <= 137) {
if (s == 31) {
w = 0xff3c; /* FULLWIDTH REVERSE SOLIDUS */
} else if (s == 32) {
w = 0xff5e; /* FULLWIDTH TILDE */
} else if (s == 33) {
w = 0x2225; /* PARALLEL TO */
} else if (s == 60) {
w = 0xff0d; /* FULLWIDTH HYPHEN-MINUS */
} else if (s == 80) {
w = 0xffe0; /* FULLWIDTH CENT SIGN */
} else if (s == 81) {
w = 0xffe1; /* FULLWIDTH POUND SIGN */
} else if (s == 137) {
w = 0xffe2; /* FULLWIDTH NOT SIGN */
}
}
if (w == 0) {
if (s >= cp932ext1_ucs_table_min && s < cp932ext1_ucs_table_max) { /* vendor ext1 (13ku) */
w = cp932ext1_ucs_table[s - cp932ext1_ucs_table_min];
} else if (s >= 0 && s < jisx0208_ucs_table_size) {
w = jisx0208_ucs_table[s];
} else {
w = 0;
}
}
if (s >= (84*94) && s < 91*94) {
s += 22*94;
if (filter->from->no_encoding == mbfl_no_encoding_2022jp_kddi) {
w = mbfilter_sjis_emoji_kddi2unicode(s, &snd);
}
if (w > 0 && snd > 0) {
CK((*filter->output_function)(snd, filter->data));
}
}
if (w <= 0) {
w = (c1 << 8) | c;
w &= MBFL_WCSPLANE_MASK;
w |= MBFL_WCSPLANE_JIS0208;
}
CK((*filter->output_function)(w, filter->data));
} else if (c == 0x1b) {
filter->status += 2;
} else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
CK((*filter->output_function)(c, filter->data));
} else {
w = (c1 << 8) | c;
w &= MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
/* ESC */
/* case 0x02: */
/* case 0x12: */
/* case 0x22: */
/* case 0x82: */
case 2:
if (c == 0x24) { /* '$' */
filter->status++;
} else if (c == 0x28) { /* '(' */
filter->status += 3;
} else {
filter->status &= ~0xf;
CK((*filter->output_function)(0x1b, filter->data));
goto retry;
}
break;
/* ESC $ */
/* case 0x03: */
/* case 0x13: */
/* case 0x23: */
/* case 0x83: */
case 3:
if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
filter->status = 0x80;
} else if (c == 0x28) { /* '(' */
filter->status++;
} else {
filter->status &= ~0xf;
CK((*filter->output_function)(0x1b, filter->data));
CK((*filter->output_function)(0x24, filter->data));
goto retry;
}
break;
/* ESC $ ( */
/* case 0x04: */
/* case 0x14: */
/* case 0x24: */
/* case 0x84: */
case 4:
if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
filter->status = 0x80;
} else {
filter->status &= ~0xf;
CK((*filter->output_function)(0x1b, filter->data));
CK((*filter->output_function)(0x24, filter->data));
CK((*filter->output_function)(0x28, filter->data));
goto retry;
}
break;
/* ESC ( */
/* case 0x05: */
/* case 0x15: */
/* case 0x25: */
/* case 0x85: */
case 5:
if (c == 0x42) { /* 'B' */
filter->status = 0;
} else if (c == 0x4a) { /* 'J' */
filter->status = 0;
} else if (c == 0x49) { /* 'I' */
filter->status = 0x20;
} else {
filter->status &= ~0xf;
CK((*filter->output_function)(0x1b, filter->data));
CK((*filter->output_function)(0x28, filter->data));
goto retry;
}
break;
default:
filter->status = 0;
break;
}
return c;
}
/*
* eucJP-win => wchar
*/
int
mbfl_filt_conv_eucjpwin_wchar(int c, mbfl_convert_filter *filter)
{
int c1, s, w, n;
switch (filter->status) {
case 0:
if (c >= 0 && c < 0x80) { /* latin */
CK((*filter->output_function)(c, filter->data));
} else if (c > 0xa0 && c < 0xff) { /* CP932 first char */
filter->status = 1;
filter->cache = c;
} else if (c == 0x8e) { /* kana first char */
filter->status = 2;
} else if (c == 0x8f) { /* X 0212 first char */
filter->status = 3;
} else {
w = c & MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
case 1: /* got first half */
filter->status = 0;
c1 = filter->cache;
if (c > 0xa0 && c < 0xff) {
w = 0;
s = (c1 - 0xa1)*94 + c - 0xa1;
if (s <= 137) {
if (s == 31) {
w = 0xff3c; /* FULLWIDTH REVERSE SOLIDUS */
} else if (s == 32) {
w = 0xff5e; /* FULLWIDTH TILDE */
} else if (s == 33) {
w = 0x2225; /* PARALLEL TO */
} else if (s == 60) {
w = 0xff0d; /* FULLWIDTH HYPHEN-MINUS */
} else if (s == 80) {
w = 0xffe0; /* FULLWIDTH CENT SIGN */
} else if (s == 81) {
w = 0xffe1; /* FULLWIDTH POUND SIGN */
} else if (s == 137) {
w = 0xffe2; /* FULLWIDTH NOT SIGN */
}
}
if (w == 0) {
if (s >= cp932ext1_ucs_table_min && s < cp932ext1_ucs_table_max) { /* vendor ext1 (13ku) */
w = cp932ext1_ucs_table[s - cp932ext1_ucs_table_min];
} else if (s >= 0 && s < jisx0208_ucs_table_size) { /* X 0208 */
w = jisx0208_ucs_table[s];
} else if (s >= (84 * 94)) { /* user (85ku - 94ku) */
w = s - (84 * 94) + 0xe000;
}
}
if (w <= 0) {
w = ((c1 & 0x7f) << 8) | (c & 0x7f);
w &= MBFL_WCSPLANE_MASK;
w |= MBFL_WCSPLANE_WINCP932;
}
CK((*filter->output_function)(w, filter->data));
} else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
CK((*filter->output_function)(c, filter->data));
} else {
w = (c1 << 8) | c;
w &= MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
case 2: /* got 0x8e, X0201 kana */
filter->status = 0;
if (c > 0xa0 && c < 0xe0) {
w = 0xfec0 + c;
CK((*filter->output_function)(w, filter->data));
} else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
CK((*filter->output_function)(c, filter->data));
} else {
w = 0x8e00 | c;
w &= MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
case 3: /* got 0x8f, X 0212 first char */
if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
CK((*filter->output_function)(c, filter->data));
filter->status = 0;
} else {
filter->status++;
filter->cache = c;
}
break;
case 4: /* got 0x8f, X 0212 second char */
filter->status = 0;
c1 = filter->cache;
if (c1 > 0xa0 && c1 < 0xff && c > 0xa0 && c < 0xff) {
s = (c1 - 0xa1)*94 + c - 0xa1;
if (s >= 0 && s < jisx0212_ucs_table_size) {
w = jisx0212_ucs_table[s];
if (w == 0x007e) {
w = 0xff5e; /* FULLWIDTH TILDE */
}
} else if (s >= (82*94) && s < (84*94)) { /* vender ext3 (83ku - 84ku) <-> CP932 (115ku -120ku) */
s = (c1<< 8) | c;
w = 0;
n = 0;
while (n < cp932ext3_eucjp_table_size) {
if (s == cp932ext3_eucjp_table[n]) {
if (n < (cp932ext3_ucs_table_max - cp932ext3_ucs_table_min)) {
w = cp932ext3_ucs_table[n];
}
break;
}
n++;
}
} else if (s >= (84*94)) { /* user (85ku - 94ku) */
w = s - (84*94) + (0xe000 + (94*10));
} else {
w = 0;
}
if (w == 0x00A6) {
w = 0xFFE4; /* FULLWIDTH BROKEN BAR */
}
if (w <= 0) {
w = ((c1 & 0x7f) << 8) | (c & 0x7f);
w &= MBFL_WCSPLANE_MASK;
w |= MBFL_WCSPLANE_JIS0212;
}
CK((*filter->output_function)(w, filter->data));
} else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
CK((*filter->output_function)(c, filter->data));
} else {
w = (c1 << 8) | c | 0x8f0000;
w &= MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
default:
filter->status = 0;
break;
}
return c;
}
/*
* wchar => ISO-2022-JP-Mobile
*/
int
mbfl_filt_conv_wchar_2022jp_mobile(int c, mbfl_convert_filter *filter)
{
int c1, c2, s1, s2;
s1 = 0;
s2 = 0;
if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
s1 = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
} else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
s1 = ucs_a2_jis_table[c - ucs_a2_jis_table_min];
} else if (c >= ucs_i_jis_table_min && c < ucs_i_jis_table_max) {
s1 = ucs_i_jis_table[c - ucs_i_jis_table_min];
} else if (c >= ucs_r_jis_table_min && c < ucs_r_jis_table_max) {
s1 = ucs_r_jis_table[c - ucs_r_jis_table_min];
} else if (c >= 0xe000 && c < (0xe000 + 20*94)) { /* user (95ku - 114ku) */
s1 = c - 0xe000;
c1 = s1/94 + 0x7f;
c2 = s1%94 + 0x21;
s1 = (c1 << 8) | c2;
}
if (s1 <= 0) {
c1 = c & ~MBFL_WCSPLANE_MASK;
if (c1 == MBFL_WCSPLANE_WINCP932) {
s1 = c & MBFL_WCSPLANE_MASK;
s2 = 1;
} else if (c1 == MBFL_WCSPLANE_JIS0208) {
s1 = c & MBFL_WCSPLANE_MASK;
} else if (c1 == MBFL_WCSPLANE_JIS0212) {
s1 = c & MBFL_WCSPLANE_MASK;
s1 |= 0x8080;
} else if (c == 0xa5) { /* YEN SIGN */
s1 = 0x216f; /* FULLWIDTH YEN SIGN */
} else if (c == 0x203e) { /* OVER LINE */
s1 = 0x2131; /* FULLWIDTH MACRON */
} else if (c == 0xff3c) { /* FULLWIDTH REVERSE SOLIDUS */
s1 = 0x2140;
} else if (c == 0xff5e) { /* FULLWIDTH TILDE */
s1 = 0x2141;
} else if (c == 0x2225) { /* PARALLEL TO */
s1 = 0x2142;
} else if (c == 0xff0d) { /* FULLWIDTH HYPHEN-MINUS */
s1 = 0x215d;
} else if (c == 0xffe0) { /* FULLWIDTH CENT SIGN */
s1 = 0x2171;
} else if (c == 0xffe1) { /* FULLWIDTH POUND SIGN */
s1 = 0x2172;
} else if (c == 0xffe2) { /* FULLWIDTH NOT SIGN */
s1 = 0x224c;
}
}
if ((s1 <= 0) || (s1 >= 0xa1a1 && s2 == 0)) { /* not found or X 0212 */
s1 = -1;
c1 = 0;
c2 = cp932ext1_ucs_table_max - cp932ext1_ucs_table_min;
while (c1 < c2) { /* CP932 vendor ext1 (13ku) */
if (c == cp932ext1_ucs_table[c1]) {
s1 = ((c1/94 + 0x2d) << 8) + (c1%94 + 0x21);
break;
}
c1++;
}
if (c == 0) {
s1 = 0;
} else if (s1 <= 0) {
s1 = -1;
}
}
if (filter->to->no_encoding == mbfl_no_encoding_2022jp_kddi &&
mbfilter_unicode2sjis_emoji_kddi(c, &s1, filter) > 0) {
CODE2JIS(c1,c2,s1,s2);
s1 -= 0x1600;
}
if (filter->status == 1 && filter->cache > 0) {
return c;
}
if (s1 >= 0) {
if (s1 < 0x80) { /* latin */
if ((filter->status & 0xff00) != 0) {
CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
CK((*filter->output_function)(0x28, filter->data)); /* '(' */
CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
}
CK((*filter->output_function)(s1, filter->data));
filter->status = 0;
} else if (s1 > 0xa0 && s1 < 0xe0) { /* kana */
if ((filter->status & 0xff00) != 0x100) {
CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
CK((*filter->output_function)(0x28, filter->data)); /* '(' */
CK((*filter->output_function)(0x49, filter->data)); /* 'I' */
}
filter->status = 0x100;
CK((*filter->output_function)(s1 & 0x7f, filter->data));
} else if (s1 < 0x7e7f) { /* X 0208 */
if ((filter->status & 0xff00) != 0x200) {
CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
CK((*filter->output_function)(0x24, filter->data)); /* '$' */
CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
}
filter->status = 0x200;
CK((*filter->output_function)((s1 >> 8) & 0xff, filter->data));
CK((*filter->output_function)(s1 & 0x7f, filter->data));
}
} else {
if (filter->illegal_mode != MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
/*
* cp51932 => wchar
*/
int
mbfl_filt_conv_cp51932_wchar(int c, mbfl_convert_filter *filter)
{
int c1, s, w;
switch (filter->status) {
case 0:
if (c >= 0 && c < 0x80) { /* latin */
CK((*filter->output_function)(c, filter->data));
} else if (c > 0xa0 && c < 0xff) { /* CP932 first char */
filter->status = 1;
filter->cache = c;
} else if (c == 0x8e) { /* kana first char */
filter->status = 2;
} else {
w = c & MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
case 1: /* got first half */
filter->status = 0;
c1 = filter->cache;
if (c > 0xa0 && c < 0xff) {
w = 0;
s = (c1 - 0xa1)*94 + c - 0xa1;
if (s <= 137) {
if (s == 31) {
w = 0xff3c; /* FULLWIDTH REVERSE SOLIDUS */
} else if (s == 32) {
w = 0xff5e; /* FULLWIDTH TILDE */
} else if (s == 33) {
w = 0x2225; /* PARALLEL TO */
} else if (s == 60) {
w = 0xff0d; /* FULLWIDTH HYPHEN-MINUS */
} else if (s == 80) {
w = 0xffe0; /* FULLWIDTH CENT SIGN */
} else if (s == 81) {
w = 0xffe1; /* FULLWIDTH POUND SIGN */
} else if (s == 137) {
w = 0xffe2; /* FULLWIDTH NOT SIGN */
}
}
if (w == 0) {
if (s >= cp932ext1_ucs_table_min && s < cp932ext1_ucs_table_max) { /* vendor ext1 (13ku) */
w = cp932ext1_ucs_table[s - cp932ext1_ucs_table_min];
} else if (s >= 0 && s < jisx0208_ucs_table_size) { /* X 0208 */
w = jisx0208_ucs_table[s];
} else if (s >= cp932ext2_ucs_table_min && s < cp932ext2_ucs_table_max) { /* vendor ext2 (89ku - 92ku) */
w = cp932ext2_ucs_table[s - cp932ext2_ucs_table_min];
}
}
if (w <= 0) {
w = ((c1 & 0x7f) << 8) | (c & 0x7f);
w &= MBFL_WCSPLANE_MASK;
w |= MBFL_WCSPLANE_WINCP932;
}
CK((*filter->output_function)(w, filter->data));
} else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
CK((*filter->output_function)(c, filter->data));
} else {
w = (c1 << 8) | c;
w &= MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
case 2: /* got 0x8e, X0201 kana */
filter->status = 0;
if (c > 0xa0 && c < 0xe0) {
w = 0xfec0 + c;
CK((*filter->output_function)(w, filter->data));
} else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
CK((*filter->output_function)(c, filter->data));
} else {
w = 0x8e00 | c;
w &= MBFL_WCSGROUP_MASK;
w |= MBFL_WCSGROUP_THROUGH;
CK((*filter->output_function)(w, filter->data));
}
break;
default:
filter->status = 0;
break;
}
return c;
}
/*
* wchar => JIS
*/
int
mbfl_filt_conv_wchar_jis_ms(int c, mbfl_convert_filter *filter)
{
int c1, s;
s = 0;
if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
s = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
} else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
s = ucs_a2_jis_table[c - ucs_a2_jis_table_min];
} else if (c >= ucs_i_jis_table_min && c < ucs_i_jis_table_max) {
s = ucs_i_jis_table[c - ucs_i_jis_table_min];
} else if (c >= ucs_r_jis_table_min && c < ucs_r_jis_table_max) {
s = ucs_r_jis_table[c - ucs_r_jis_table_min];
} else if (c >= 0xe000 && c < (0xe000 + 10 * 94)) {
/* PUE => Microsoft extended (pseudo 95ku - 114ku) */
/* See http://www.opengroup.or.jp/jvc/cde/ucs-conv.html#ch4_2 */
s = c - 0xe000;
s = (s / 94 + 0x75) << 8 | (s % 94 + 0x21);
} else if (c >= (0xe000 + 10 * 94) && c <= (0xe000 + 20 * 94)) {
/* PUE => JISX0212 user-defined (G3 85ku - 94ku) */
/* See http://www.opengroup.or.jp/jvc/cde/ucs-conv.html#ch4_2 */
s = c - (0xe000 + 10 * 94);
s = (s / 94 + 0xf5) << 8 | (s % 94 + 0xa1);
}
/* do some transliteration */
if (s <= 0) {
c1 = c & ~MBFL_WCSPLANE_MASK;
if (c1 == MBFL_WCSPLANE_JIS0208) {
s = c & MBFL_WCSPLANE_MASK;
} else if (c1 == MBFL_WCSPLANE_JIS0212) {
s = c & MBFL_WCSPLANE_MASK;
s |= 0x8080;
} else if (c == 0xa5) { /* YEN SIGN */
s = 0x1005c;
} else if (c == 0x203e) { /* OVER LINE */
s = 0x1007e;
} else if (c == 0xff3c) { /* FULLWIDTH REVERSE SOLIDUS */
s = 0x2140;
} else if (c == 0xff5e) { /* FULLWIDTH TILDE */
s = 0x2141;
} else if (c == 0x2225) { /* PARALLEL TO */
s = 0x2142;
} else if (c == 0xff0d) { /* FULLWIDTH HYPHEN-MINUS */
s = 0x215d;
} else if (c == 0xffe0) { /* FULLWIDTH CENT SIGN */
s = 0x2171;
} else if (c == 0xffe1) { /* FULLWIDTH POUND SIGN */
s = 0x2172;
} else if (c == 0xffe2) { /* FULLWIDTH NOT SIGN */
s = 0x224c;
}
}
if (s <= 0 || (s >= 0x8080 && s < 0x10000)) {
int i;
s = -1;
for (i = 0;
i < cp932ext1_ucs_table_max - cp932ext1_ucs_table_min; i++) {
const int oh = cp932ext1_ucs_table_min / 94;
if (c == cp932ext1_ucs_table[i]) {
s = ((i / 94 + oh + 0x21) << 8) + (i % 94 + 0x21);
break;
}
}
if (s < 0) {
const int oh = cp932ext2_ucs_table_min / 94;
const int cp932ext2_ucs_table_size =
cp932ext2_ucs_table_max - cp932ext2_ucs_table_min;
for (i = 0; i < cp932ext2_ucs_table_size; i++) {
if (c == cp932ext2_ucs_table[i]) {
s = ((i / 94 + oh + 0x21) << 8) + (i % 94 + 0x21);
break;
}
}
}
if (s < 0) {
const int cp932ext3_ucs_table_size =
cp932ext3_ucs_table_max - cp932ext3_ucs_table_min;
const int limit = cp932ext3_ucs_table_size >
cp932ext3_eucjp_table_size ?
cp932ext3_eucjp_table_size:
cp932ext3_ucs_table_size;
for (i = 0; i < limit; i++) {
if (c == cp932ext3_ucs_table[i]) {
s = cp932ext3_eucjp_table[i];
break;
}
}
}
if (c == 0) {
s = 0;
} else if (s <= 0) {
s = -1;
}
}
if (s >= 0) {
if (s < 0x80) { /* ASCII */
if ((filter->status & 0xff00) != 0) {
CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
CK((*filter->output_function)(0x28, filter->data)); /* '(' */
CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
}
filter->status = 0;
CK((*filter->output_function)(s, filter->data));
} else if (s < 0x100) { /* kana */
if ((filter->status & 0xff00) != 0x100) {
CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
CK((*filter->output_function)(0x28, filter->data)); /* '(' */
CK((*filter->output_function)(0x49, filter->data)); /* 'I' */
}
filter->status = 0x100;
CK((*filter->output_function)(s & 0x7f, filter->data));
} else if (s < 0x8080) { /* X 0208 */
if ((filter->status & 0xff00) != 0x200) {
CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
CK((*filter->output_function)(0x24, filter->data)); /* '$' */
CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
}
filter->status = 0x200;
CK((*filter->output_function)((s >> 8) & 0x7f, filter->data));
CK((*filter->output_function)(s & 0x7f, filter->data));
} else if (s < 0x10000) { /* X 0212 */
if ((filter->status & 0xff00) != 0x300) {
CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
CK((*filter->output_function)(0x24, filter->data)); /* '$' */
CK((*filter->output_function)(0x28, filter->data)); /* '(' */
CK((*filter->output_function)(0x44, filter->data)); /* 'D' */
}
filter->status = 0x300;
CK((*filter->output_function)((s >> 8) & 0x7f, filter->data));
CK((*filter->output_function)(s & 0x7f, filter->data));
} else { /* X 0201 latin */
if ((filter->status & 0xff00) != 0x400) {
