diogo/yap-6.3
Archived
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

ok, even if the locale is C, unicode is most often supported.

Browse Source
This commit is contained in:
Vítor Santos Costa 2016-03-03 23:19:19 +00:00
parent 5f9752baff
commit d3796aba5a
1 changed files with 283 additions and 281 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

564
os/chartypes.c
View File

@ -21,10 +21,10 @@ static char SccsId[] = "%W% %G%";
* @file chartypes.c * @file chartypes.c
* @author VITOR SANTOS COSTA <vsc@VITORs-MBP.lan> * @author VITOR SANTOS COSTA <vsc@VITORs-MBP.lan>
* @date Thu Nov 19 12:05:14 2015 * @date Thu Nov 19 12:05:14 2015
* *
* @brief Character Properties * @brief Character Properties
* *
* *
*/ */
///{@ ///{@
@ -79,7 +79,7 @@ static Int p_change_type_of_char(USES_REGS1);
Term Yap_StringToNumberTerm(char *s, encoding_t *encp) { Term Yap_StringToNumberTerm(char *s, encoding_t *encp) {
CACHE_REGS CACHE_REGS
int sno; int sno;
Term t; Term t;
sno = Yap_open_buf_read_stream(s, strlen(s), encp, MEM_BUF_USER); sno = Yap_open_buf_read_stream(s, strlen(s), encp, MEM_BUF_USER);
@ -101,76 +101,78 @@ Term Yap_StringToNumberTerm(char *s, encoding_t *encp) {
const char *encvs[] = {"LANG", "LC_ALL", "LC_CTYPE", NULL}; const char *encvs[] = {"LANG", "LC_ALL", "LC_CTYPE", NULL};
// wher we can fins an encoding // where we can fins an encoding
typedef struct enc_map { typedef struct enc_map {
const char *s; const char *s;
encoding_t e; encoding_t e;
} enc_map_t; } enc_map_t;
static enc_map_t ematches[] = { static enc_map_t ematches[] = {
{"UTF-8", ENC_ISO_UTF8}, {"UTF-8", ENC_ISO_UTF8},
{"utf8", ENC_ISO_UTF8}, {"utf8", ENC_ISO_UTF8},
{"UTF-16", ENC_UTF16_LE}, // ok, this is a very bad name {"UTF-16", ENC_UTF16_LE}, // ok, this is a very bad name
{"UCS-2", ENC_UTF16_LE}, // ok, this is probably gone by now {"UCS-2", ENC_UTF16_LE}, // ok, this is probably gone by now
{"ISO-LATIN1", ENC_ISO_LATIN1}, {"ISO-LATIN1", ENC_ISO_LATIN1},
{"ISO-8859-1", ENC_ISO_LATIN1}, {"ISO-8859-1", ENC_ISO_LATIN1},
{"Windows-1252", ENC_ISO_LATIN1}, // almost, but not quite {"Windows-1252", ENC_ISO_LATIN1}, // almost, but not quite
{"CP-1252", ENC_ISO_LATIN1}, {"CP-1252", ENC_ISO_LATIN1},
{"C", ENC_ISO_ASCII}, {"C", ENC_ISO_ASCII},
#ifdef _WIN32 #ifdef _WIN32
{NULL, ENC_UTF16_LE} {NULL, ENC_UTF16_LE}
#else #else
{NULL, ENC_ISO_UTF8} {NULL, ENC_ISO_UTF8}
#endif #endif
}; };
static encoding_t DefaultEncoding(void) { static encoding_t enc_os_default( encoding_t rc)\
encoding_t rc; {
int i = 0, j; // by default, return UTF-8
char *enc; // except in _WIN32
while (encvs[i]) { // note that we match the C locale to UTF8/16, as all Unix maachines will work on UNICODE.
char *v = getenv(encvs[i]);
if (v) { if (rc == ENC_ISO_ASCII) {
enc = strrchr(v, '.');
/* that's how it is supposed to be, except in OSX */
if (!enc)
enc = v;
else
enc++;
// now that we have one name, try to match it
j = 0;
while (ematches[j].s) {
if (!strcmp(ematches[j].s, enc))
return ematches[j].e;
j++;
}
}
i++;
}
// by default, return UTF-8
// except in _WIN32
#ifdef _WIN32 #ifdef _WIN32
rc = ENC_UTF16_BE; return = ENC_UTF16_BE;
#else #else
rc = ENC_ISO_UTF8; return ENC_ISO_UTF8;
#endif #endif
{ }
int j = 0;
while (rc != ematches[j].e)
j++;
// Yap_Warning("YAP will use default encoding %s", ematches[j].s);
}
return rc; return rc;
} }
static encoding_t DefaultEncoding(void) {
int i = 0;
while (encvs[i]) {
char *v = getenv(encvs[i]);
if (v) {
int j = 0;
size_t sz = strlen(v);
const char *coding;
while ((coding = ematches[j].s) != NULL) {
size_t sz2 = strlen(coding);
if (sz2 > sz) {
j++;
continue;
}
if (!strcmp(coding+(sz-sz2), v) ) {
return enc_os_default(ematches[j].e);
}
j++;
}
}
i++;
}
return enc_os_default(ENC_ISO_ASCII);
}
encoding_t Yap_DefaultEncoding(void) { encoding_t Yap_DefaultEncoding(void) {
CACHE_REGS CACHE_REGS
return LOCAL_encoding; return LOCAL_encoding;
} }
void Yap_SetDefaultEncoding(encoding_t new_encoding) { void Yap_SetDefaultEncoding(encoding_t new_encoding) {
CACHE_REGS CACHE_REGS
LOCAL_encoding = new_encoding; LOCAL_encoding = new_encoding;
} }
static Int get_default_encoding(USES_REGS1) { static Int get_default_encoding(USES_REGS1) {
@ -185,9 +187,9 @@ static Int p_encoding(USES_REGS1) { /* '$encoding'(Stream,N) */
if (sno < 0) if (sno < 0)
return FALSE; return FALSE;
if (IsVarTerm(t)) { if (IsVarTerm(t)) {
UNLOCK(GLOBAL_Stream[sno].streamlock); UNLOCK(GLOBAL_Stream[sno].streamlock);
return Yap_unify(ARG2, MkIntegerTerm(GLOBAL_Stream[sno].encoding)); return Yap_unify(ARG2, MkIntegerTerm(GLOBAL_Stream[sno].encoding));
} }
GLOBAL_Stream[sno].encoding = IntegerOfTerm(Deref(ARG2)); GLOBAL_Stream[sno].encoding = IntegerOfTerm(Deref(ARG2));
UNLOCK(GLOBAL_Stream[sno].streamlock); UNLOCK(GLOBAL_Stream[sno].streamlock);
return TRUE; return TRUE;
@ -195,81 +197,81 @@ static Int p_encoding(USES_REGS1) { /* '$encoding'(Stream,N) */
static int get_char(Term t) { static int get_char(Term t) {
if (IsVarTerm(t = Deref(t))) { if (IsVarTerm(t = Deref(t))) {
Yap_Error(INSTANTIATION_ERROR, t, NULL); Yap_Error(INSTANTIATION_ERROR, t, NULL);
return 0; return 0;
} }
if (!IsAtomTerm(t)) { if (!IsAtomTerm(t)) {
Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, NULL); Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, NULL);
return 0; return 0;
} }
Atom at = AtomOfTerm(t); Atom at = AtomOfTerm(t);
if (IsWideAtom(at)) { if (IsWideAtom(at)) {
wchar_t *s0 = RepAtom(AtomOfTerm(t))->WStrOfAE; wchar_t *s0 = RepAtom(AtomOfTerm(t))->WStrOfAE;
if (s0[1] != '\0') { if (s0[1] != '\0') {
Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, NULL); Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, NULL);
return 0; return 0;
}
return s0[0];
} else {
char *s0 = RepAtom(AtomOfTerm(t))->StrOfAE;
if (s0[1] != '\0') {
Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, NULL);
return 0;
}
return s0[0];
} }
return s0[0];
} else {
char *s0 = RepAtom(AtomOfTerm(t))->StrOfAE;
if (s0[1] != '\0') {
Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, NULL);
return 0;
}
return s0[0];
}
return 0; return 0;
} }
static int get_code(Term t) { static int get_code(Term t) {
if (IsVarTerm(t = Deref(t))) { if (IsVarTerm(t = Deref(t))) {
Yap_Error(INSTANTIATION_ERROR, t, NULL); Yap_Error(INSTANTIATION_ERROR, t, NULL);
return 0; return 0;
} }
if (!IsIntegerTerm(t)) { if (!IsIntegerTerm(t)) {
Yap_Error(TYPE_ERROR_CHARACTER_CODE, t, NULL); Yap_Error(TYPE_ERROR_CHARACTER_CODE, t, NULL);
return 0; return 0;
} }
Int ch = IntegerOfTerm(t); Int ch = IntegerOfTerm(t);
if (ch < -1) { if (ch < -1) {
Yap_Error(REPRESENTATION_ERROR_CHARACTER_CODE, t, NULL); Yap_Error(REPRESENTATION_ERROR_CHARACTER_CODE, t, NULL);
return 0; return 0;
} }
return ch; return ch;
} }
static int get_char_or_code(Term t, bool *is_char) { static int get_char_or_code(Term t, bool *is_char) {
if (!IsAtomTerm(t)) { if (!IsAtomTerm(t)) {
if (!IsIntegerTerm(t)) { if (!IsIntegerTerm(t)) {
Yap_Error(TYPE_ERROR_CHARACTER, t, NULL); Yap_Error(TYPE_ERROR_CHARACTER, t, NULL);
return 0; return 0;
}
Int ch = IntegerOfTerm(t);
if (ch < -1) {
Yap_Error(REPRESENTATION_ERROR_CHARACTER_CODE, t, NULL);
return 0;
}
*is_char = false;
return ch;
} }
Int ch = IntegerOfTerm(t);
if (ch < -1) {
Yap_Error(REPRESENTATION_ERROR_CHARACTER_CODE, t, NULL);
return 0;
}
*is_char = false;
return ch;
}
Atom at = AtomOfTerm(t); Atom at = AtomOfTerm(t);
if (IsWideAtom(at)) { if (IsWideAtom(at)) {
wchar_t *s0 = RepAtom(AtomOfTerm(t))->WStrOfAE; wchar_t *s0 = RepAtom(AtomOfTerm(t))->WStrOfAE;
if (s0[1] != '\0') { if (s0[1] != '\0') {
Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, NULL); Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, NULL);
return 0; return 0;
}
*is_char = true;
return s0[0];
} else {
char *s0 = RepAtom(AtomOfTerm(t))->StrOfAE;
if (s0[1] != '\0') {
Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, NULL);
return 0;
}
*is_char = true;
return s0[0];
} }
*is_char = true;
return s0[0];
} else {
char *s0 = RepAtom(AtomOfTerm(t))->StrOfAE;
if (s0[1] != '\0') {
Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, NULL);
return 0;
}
*is_char = true;
return s0[0];
}
return 0; return 0;
} }
@ -277,35 +279,35 @@ static Int toupper2(USES_REGS1) {
bool is_char = false; bool is_char = false;
Term t; Term t;
if (!IsVarTerm(t = Deref(ARG1))) { if (!IsVarTerm(t = Deref(ARG1))) {
Int out = get_char_or_code(t, &is_char), uout; Int out = get_char_or_code(t, &is_char), uout;
if (out < 128) if (out < 128)
uout = toupper(out); uout = toupper(out);
else
uout = towupper(out);
if (is_char)
return Yap_unify(ARG2, MkCharTerm(uout));
else
return Yap_unify(ARG2, MkIntegerTerm(uout));
} else if (!IsVarTerm(t = Deref(ARG2))) {
Int uout = get_char_or_code(t, &is_char), out;
char_kind_t charp = Yap_wide_chtype(uout);
if (charp == UC) {
if (uout < 128)
out = tolower(uout);
else else
out = towlower(uout); uout = towupper(out);
} else if (charp == LC) { if (is_char)
return false; return Yap_unify(ARG2, MkCharTerm(uout));
else
return Yap_unify(ARG2, MkIntegerTerm(uout));
} else if (!IsVarTerm(t = Deref(ARG2))) {
Int uout = get_char_or_code(t, &is_char), out;
char_kind_t charp = Yap_wide_chtype(uout);
if (charp == UC) {
if (uout < 128)
out = tolower(uout);
else
out = towlower(uout);
} else if (charp == LC) {
return false;
} else {
out = uout;
}
if (is_char)
return Yap_unify(ARG2, MkCharTerm(out));
else
return Yap_unify(ARG2, MkIntegerTerm(out));
} else { } else {
out = uout; Yap_Error(INSTANTIATION_ERROR, ARG1, NULL);
} }
if (is_char)
return Yap_unify(ARG2, MkCharTerm(out));
else
return Yap_unify(ARG2, MkIntegerTerm(out));
} else {
Yap_Error(INSTANTIATION_ERROR, ARG1, NULL);
}
return false; return false;
} }
@ -313,41 +315,41 @@ static Int tolower2(USES_REGS1) {
bool is_char = false; bool is_char = false;
Term t; Term t;
if (!IsVarTerm(t = Deref(ARG1))) { if (!IsVarTerm(t = Deref(ARG1))) {
bool is_char = false; bool is_char = false;
Int out = get_char_or_code(ARG1, &is_char), uout; Int out = get_char_or_code(ARG1, &is_char), uout;
if (out < 128) if (out < 128)
uout = tolower(out); uout = tolower(out);
else
uout = towlower(out);
if (is_char)
return Yap_unify(ARG2, MkCharTerm(uout));
else
return Yap_unify(ARG2, MkIntegerTerm(uout));
} else if (IsVarTerm(t = Deref(ARG2))) {
Int uout = get_char_or_code(t, &is_char), out;
char_kind_t charp = Yap_wide_chtype(uout);
if (charp == LC) {
if (uout < 128)
out = toupper(uout);
else else
out = towupper(uout); uout = towlower(out);
} else if (charp == UC) { if (is_char)
return false; return Yap_unify(ARG2, MkCharTerm(uout));
else
return Yap_unify(ARG2, MkIntegerTerm(uout));
} else if (IsVarTerm(t = Deref(ARG2))) {
Int uout = get_char_or_code(t, &is_char), out;
char_kind_t charp = Yap_wide_chtype(uout);
if (charp == LC) {
if (uout < 128)
out = toupper(uout);
else
out = towupper(uout);
} else if (charp == UC) {
return false;
} else {
out = uout;
}
if (is_char)
return Yap_unify(ARG2, MkCharTerm(out));
else
return Yap_unify(ARG2, MkIntegerTerm(out));
} else { } else {
out = uout; Yap_Error(INSTANTIATION_ERROR, ARG1, NULL);
} }
if (is_char)
return Yap_unify(ARG2, MkCharTerm(out));
else
return Yap_unify(ARG2, MkIntegerTerm(out));
} else {
Yap_Error(INSTANTIATION_ERROR, ARG1, NULL);
}
return false; return false;
} }
static Int static Int
p_change_type_of_char(USES_REGS1) { /* change_type_of_char(+char,+type) */ p_change_type_of_char(USES_REGS1) { /* change_type_of_char(+char,+type) */
Term t1 = Deref(ARG1); Term t1 = Deref(ARG1);
Term t2 = Deref(ARG2); Term t2 = Deref(ARG2);
if (!IsVarTerm(t1) && !IsIntegerTerm(t1)) if (!IsVarTerm(t1) && !IsIntegerTerm(t1))
@ -390,9 +392,9 @@ static Int char_type_ascii(USES_REGS1) {
static Int char_type_white(USES_REGS1) { static Int char_type_white(USES_REGS1) {
int ch = get_char(ARG1); int ch = get_char(ARG1);
if (ch < 256) { if (ch < 256) {
char_kind_t k = Yap_chtype[ch]; char_kind_t k = Yap_chtype[ch];
return k == BS; return k == BS;
} }
utf8proc_category_t ct = utf8proc_category(ch); utf8proc_category_t ct = utf8proc_category(ch);
return ct == UTF8PROC_CATEGORY_ZS; return ct == UTF8PROC_CATEGORY_ZS;
} }
@ -440,18 +442,18 @@ static Int char_type_upper(USES_REGS1) {
static Int char_type_punct(USES_REGS1) { static Int char_type_punct(USES_REGS1) {
int ch = get_char(ARG1); int ch = get_char(ARG1);
if (ch < 256) { if (ch < 256) {
char_kind_t k = Yap_chtype[ch]; char_kind_t k = Yap_chtype[ch];
return k >= QT && k <= BK; return k >= QT && k <= BK;
} }
return false; return false;
} }
static Int char_type_space(USES_REGS1) { static Int char_type_space(USES_REGS1) {
int ch = get_char(ARG1); int ch = get_char(ARG1);
if (ch < 256) { if (ch < 256) {
char_kind_t k = Yap_chtype[ch]; char_kind_t k = Yap_chtype[ch];
return k == BS; return k == BS;
} }
utf8proc_category_t ct = utf8proc_category(ch); utf8proc_category_t ct = utf8proc_category(ch);
return (ct >= UTF8PROC_CATEGORY_ZS && ct <= UTF8PROC_CATEGORY_PO); return (ct >= UTF8PROC_CATEGORY_ZS && ct <= UTF8PROC_CATEGORY_PO);
} }
@ -464,8 +466,8 @@ static Int char_type_end_of_file(USES_REGS1) {
static Int char_type_end_of_line(USES_REGS1) { static Int char_type_end_of_line(USES_REGS1) {
Int ch = get_char(ARG1); Int ch = get_char(ARG1);
if (ch < 256) { if (ch < 256) {
return ch >= 10 && ch <= 13; return ch >= 10 && ch <= 13;
} }
utf8proc_category_t ct = utf8proc_category(ch); utf8proc_category_t ct = utf8proc_category(ch);
return (ct >= UTF8PROC_CATEGORY_ZL && ct <= UTF8PROC_CATEGORY_ZP); return (ct >= UTF8PROC_CATEGORY_ZL && ct <= UTF8PROC_CATEGORY_ZP);
} }
@ -473,8 +475,8 @@ static Int char_type_end_of_line(USES_REGS1) {
static Int char_type_newline(USES_REGS1) { static Int char_type_newline(USES_REGS1) {
Int ch = get_char(ARG1); Int ch = get_char(ARG1);
if (ch < 256) { if (ch < 256) {
return ch == 10; return ch == 10;
} }
return false; return false;
} }
@ -551,9 +553,9 @@ static Int code_type_ascii(USES_REGS1) {
static Int code_type_white(USES_REGS1) { static Int code_type_white(USES_REGS1) {
int ch = get_code(ARG1); int ch = get_code(ARG1);
if (ch < 256) { if (ch < 256) {
char_kind_t k = Yap_chtype[ch]; char_kind_t k = Yap_chtype[ch];
return k == BS; return k == BS;
} }
utf8proc_category_t ct = utf8proc_category(ch); utf8proc_category_t ct = utf8proc_category(ch);
return ct == UTF8PROC_CATEGORY_ZS; return ct == UTF8PROC_CATEGORY_ZS;
} }
@ -601,18 +603,18 @@ static Int code_type_upper(USES_REGS1) {
static Int code_type_punct(USES_REGS1) { static Int code_type_punct(USES_REGS1) {
int ch = get_char(ARG1); int ch = get_char(ARG1);
if (ch < 256) { if (ch < 256) {
char_kind_t k = Yap_chtype[ch]; char_kind_t k = Yap_chtype[ch];
return k >= QT && k <= BK; return k >= QT && k <= BK;
} }
return false; return false;
} }
static Int code_type_space(USES_REGS1) { static Int code_type_space(USES_REGS1) {
int ch = get_code(ARG1); int ch = get_code(ARG1);
if (ch < 256) { if (ch < 256) {
char_kind_t k = Yap_chtype[ch]; char_kind_t k = Yap_chtype[ch];
return k == BS; return k == BS;
} }
utf8proc_category_t ct = utf8proc_category(ch); utf8proc_category_t ct = utf8proc_category(ch);
return (ct >= UTF8PROC_CATEGORY_ZS && ct <= UTF8PROC_CATEGORY_PO); return (ct >= UTF8PROC_CATEGORY_ZS && ct <= UTF8PROC_CATEGORY_PO);
} }
@ -625,8 +627,8 @@ static Int code_type_end_of_file(USES_REGS1) {
static Int code_type_end_of_line(USES_REGS1) { static Int code_type_end_of_line(USES_REGS1) {
Int ch = get_code(ARG1); Int ch = get_code(ARG1);
if (ch < 256) { if (ch < 256) {
return ch >= 10 && ch <= 13; return ch >= 10 && ch <= 13;
} }
utf8proc_category_t ct = utf8proc_category(ch); utf8proc_category_t ct = utf8proc_category(ch);
return (ct >= UTF8PROC_CATEGORY_ZL && ct <= UTF8PROC_CATEGORY_ZP); return (ct >= UTF8PROC_CATEGORY_ZL && ct <= UTF8PROC_CATEGORY_ZP);
} }
@ -634,8 +636,8 @@ static Int code_type_end_of_line(USES_REGS1) {
static Int code_type_newline(USES_REGS1) { static Int code_type_newline(USES_REGS1) {
Int ch = get_code(ARG1); Int ch = get_code(ARG1);
if (ch < 256) { if (ch < 256) {
return ch == 10; return ch == 10;
} }
return false; return false;
} }
@ -684,11 +686,11 @@ int ISOWGetc(int sno) {
int ch = GLOBAL_Stream[sno].stream_wgetc(sno); int ch = GLOBAL_Stream[sno].stream_wgetc(sno);
if (ch != EOF && GLOBAL_CharConversionTable != NULL) { if (ch != EOF && GLOBAL_CharConversionTable != NULL) {
if (ch < NUMBER_OF_CHARS) { if (ch < NUMBER_OF_CHARS) {
/* only do this in ASCII */ /* only do this in ASCII */
return GLOBAL_CharConversionTable[ch]; return GLOBAL_CharConversionTable[ch];
}
} }
}
return ch; return ch;
} }
@ -699,9 +701,9 @@ static Int p_force_char_conversion(USES_REGS1) {
if (GLOBAL_CharConversionTable2 == NULL) if (GLOBAL_CharConversionTable2 == NULL)
return (TRUE); return (TRUE);
for (i = 0; i < MaxStreams; i++) { for (i = 0; i < MaxStreams; i++) {
if (!(GLOBAL_Stream[i].status & Free_Stream_f)) if (!(GLOBAL_Stream[i].status & Free_Stream_f))
GLOBAL_Stream[i].stream_wgetc_for_read = ISOWGetc; GLOBAL_Stream[i].stream_wgetc_for_read = ISOWGetc;
} }
GLOBAL_CharConversionTable = GLOBAL_CharConversionTable2; GLOBAL_CharConversionTable = GLOBAL_CharConversionTable2;
return (TRUE); return (TRUE);
} }
@ -710,9 +712,9 @@ static Int p_disable_char_conversion(USES_REGS1) {
int i; int i;
for (i = 0; i < MaxStreams; i++) { for (i = 0; i < MaxStreams; i++) {
if (!(GLOBAL_Stream[i].status & Free_Stream_f)) if (!(GLOBAL_Stream[i].status & Free_Stream_f))
GLOBAL_Stream[i].stream_wgetc_for_read = GLOBAL_Stream[i].stream_wgetc; GLOBAL_Stream[i].stream_wgetc_for_read = GLOBAL_Stream[i].stream_wgetc;
} }
GLOBAL_CharConversionTable = NULL; GLOBAL_CharConversionTable = NULL;
return (TRUE); return (TRUE);
} }
@ -722,54 +724,54 @@ static Int char_conversion(USES_REGS1) {
unsigned char *s0, *s1; unsigned char *s0, *s1;
if (IsVarTerm(t)) { if (IsVarTerm(t)) {
Yap_Error(INSTANTIATION_ERROR, t, "char_conversion/2"); Yap_Error(INSTANTIATION_ERROR, t, "char_conversion/2");
return (FALSE); return (FALSE);
} }
if (!IsAtomTerm(t)) { if (!IsAtomTerm(t)) {
Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, "char_conversion/2"); Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, "char_conversion/2");
return (FALSE); return (FALSE);
} }
s0 = RepAtom(AtomOfTerm(t))->UStrOfAE; s0 = RepAtom(AtomOfTerm(t))->UStrOfAE;
if (s0[1] != '\0') { if (s0[1] != '\0') {
Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, "char_conversion/2"); Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, "char_conversion/2");
return (FALSE); return (FALSE);
} }
if (IsVarTerm(t1)) { if (IsVarTerm(t1)) {
Yap_Error(INSTANTIATION_ERROR, t1, "char_conversion/2"); Yap_Error(INSTANTIATION_ERROR, t1, "char_conversion/2");
return (FALSE); return (FALSE);
} }
if (!IsAtomTerm(t1)) { if (!IsAtomTerm(t1)) {
Yap_Error(REPRESENTATION_ERROR_CHARACTER, t1, "char_conversion/2"); Yap_Error(REPRESENTATION_ERROR_CHARACTER, t1, "char_conversion/2");
return (FALSE); return (FALSE);
} }
s1 = RepAtom(AtomOfTerm(t1))->UStrOfAE; s1 = RepAtom(AtomOfTerm(t1))->UStrOfAE;
if (s1[1] != '\0') { if (s1[1] != '\0') {
Yap_Error(REPRESENTATION_ERROR_CHARACTER, t1, "char_conversion/2"); Yap_Error(REPRESENTATION_ERROR_CHARACTER, t1, "char_conversion/2");
return (FALSE); return (FALSE);
} }
/* check if we do have a table for converting characters */ /* check if we do have a table for converting characters */
if (GLOBAL_CharConversionTable2 == NULL) { if (GLOBAL_CharConversionTable2 == NULL) {
int i; int i;
/* don't create a table if we don't need to */ /* don't create a table if we don't need to */
if (s0[0] == s1[0]) if (s0[0] == s1[0])
return (TRUE); return (TRUE);
GLOBAL_CharConversionTable2 = GLOBAL_CharConversionTable2 =
Yap_AllocCodeSpace(NUMBER_OF_CHARS * sizeof(char)); Yap_AllocCodeSpace(NUMBER_OF_CHARS * sizeof(char));
while (GLOBAL_CharConversionTable2 == NULL) { while (GLOBAL_CharConversionTable2 == NULL) {
if (!Yap_growheap(FALSE, NUMBER_OF_CHARS * sizeof(char), NULL)) { if (!Yap_growheap(FALSE, NUMBER_OF_CHARS * sizeof(char), NULL)) {
Yap_Error(RESOURCE_ERROR_HEAP, TermNil, LOCAL_ErrorMessage); Yap_Error(RESOURCE_ERROR_HEAP, TermNil, LOCAL_ErrorMessage);
return (FALSE); return (FALSE);
} }
}
if (trueGlobalPrologFlag(CHAR_CONVERSION_FLAG)) {
CACHE_REGS
if (p_force_char_conversion(PASS_REGS1) == FALSE)
return (FALSE);
}
for (i = 0; i < NUMBER_OF_CHARS; i++)
GLOBAL_CharConversionTable2[i] = i;
} }
if (trueGlobalPrologFlag(CHAR_CONVERSION_FLAG)) {
CACHE_REGS
if (p_force_char_conversion(PASS_REGS1) == FALSE)
return (FALSE);
}
for (i = 0; i < NUMBER_OF_CHARS; i++)
GLOBAL_CharConversionTable2[i] = i;
}
/* just add the new entry */ /* just add the new entry */
GLOBAL_CharConversionTable2[(int)s0[0]] = s1[0]; GLOBAL_CharConversionTable2[(int)s0[0]] = s1[0];
/* done */ /* done */
@ -781,43 +783,43 @@ static Int p_current_char_conversion(USES_REGS1) {
unsigned char *s0, *s1; unsigned char *s0, *s1;
if (GLOBAL_CharConversionTable == NULL) { if (GLOBAL_CharConversionTable == NULL) {
return (FALSE); return (FALSE);
} }
t = Deref(ARG1); t = Deref(ARG1);
if (IsVarTerm(t)) { if (IsVarTerm(t)) {
Yap_Error(INSTANTIATION_ERROR, t, "current_char_conversion/2"); Yap_Error(INSTANTIATION_ERROR, t, "current_char_conversion/2");
return (FALSE); return (FALSE);
} }
if (!IsAtomTerm(t)) { if (!IsAtomTerm(t)) {
Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, "current_char_conversion/2"); Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, "current_char_conversion/2");
return (FALSE); return (FALSE);
} }
s0 = RepAtom(AtomOfTerm(t))->UStrOfAE; s0 = RepAtom(AtomOfTerm(t))->UStrOfAE;
if (s0[1] != '\0') { if (s0[1] != '\0') {
Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, "current_char_conversion/2"); Yap_Error(REPRESENTATION_ERROR_CHARACTER, t, "current_char_conversion/2");
return (FALSE); return (FALSE);
} }
t1 = Deref(ARG2); t1 = Deref(ARG2);
if (IsVarTerm(t1)) { if (IsVarTerm(t1)) {
char out[2]; char out[2];
if (GLOBAL_CharConversionTable[(int)s0[0]] == '\0') if (GLOBAL_CharConversionTable[(int)s0[0]] == '\0')
return (FALSE); return (FALSE);
out[0] = GLOBAL_CharConversionTable[(int)s0[0]]; out[0] = GLOBAL_CharConversionTable[(int)s0[0]];
out[1] = '\0'; out[1] = '\0';
return (Yap_unify(ARG2, MkAtomTerm(Yap_LookupAtom(out)))); return (Yap_unify(ARG2, MkAtomTerm(Yap_LookupAtom(out))));
} }
if (!IsAtomTerm(t1)) { if (!IsAtomTerm(t1)) {
Yap_Error(REPRESENTATION_ERROR_CHARACTER, t1, "current_char_conversion/2"); Yap_Error(REPRESENTATION_ERROR_CHARACTER, t1, "current_char_conversion/2");
return (FALSE); return (FALSE);
} }
s1 = RepAtom(AtomOfTerm(t1))->UStrOfAE; s1 = RepAtom(AtomOfTerm(t1))->UStrOfAE;
if (s1[1] != '\0') { if (s1[1] != '\0') {
Yap_Error(REPRESENTATION_ERROR_CHARACTER, t1, "current_char_conversion/2"); Yap_Error(REPRESENTATION_ERROR_CHARACTER, t1, "current_char_conversion/2");
return (FALSE); return (FALSE);
} else { } else {
return (GLOBAL_CharConversionTable[(int)s0[0]] == '\0' && return (GLOBAL_CharConversionTable[(int)s0[0]] == '\0' &&
GLOBAL_CharConversionTable[(int)s0[0]] == s1[0]); GLOBAL_CharConversionTable[(int)s0[0]] == s1[0]);
} }
} }
static Int p_all_char_conversions(USES_REGS1) { static Int p_all_char_conversions(USES_REGS1) {
@ -825,28 +827,28 @@ static Int p_all_char_conversions(USES_REGS1) {
int i; int i;
if (GLOBAL_CharConversionTable == NULL) { if (GLOBAL_CharConversionTable == NULL) {
return (FALSE); return (FALSE);
} }
for (i = NUMBER_OF_CHARS; i > 0;) { for (i = NUMBER_OF_CHARS; i > 0;) {
i--; i--;
if (GLOBAL_CharConversionTable[i] != '\0') { if (GLOBAL_CharConversionTable[i] != '\0') {
Term t1, t2; Term t1, t2;
char s[2]; char s[2];
s[1] = '\0'; s[1] = '\0';
s[0] = GLOBAL_CharConversionTable[i]; s[0] = GLOBAL_CharConversionTable[i];
t1 = MkAtomTerm(Yap_LookupAtom(s)); t1 = MkAtomTerm(Yap_LookupAtom(s));
out = MkPairTerm(t1, out); out = MkPairTerm(t1, out);
s[0] = i; s[0] = i;
t2 = MkAtomTerm(Yap_LookupAtom(s)); t2 = MkAtomTerm(Yap_LookupAtom(s));
out = MkPairTerm(t2, out); out = MkPairTerm(t2, out);
}
} }
}
return (Yap_unify(ARG1, out)); return (Yap_unify(ARG1, out));
} }
void Yap_InitChtypes(void) { void Yap_InitChtypes(void) {
CACHE_REGS CACHE_REGS
LOCAL_encoding = DefaultEncoding(); LOCAL_encoding = DefaultEncoding();
Yap_InitCPred("$change_type_of_char", 2, p_change_type_of_char, Yap_InitCPred("$change_type_of_char", 2, p_change_type_of_char,
SafePredFlag | SyncPredFlag | HiddenPredFlag); SafePredFlag | SyncPredFlag | HiddenPredFlag);
Yap_InitCPred("toupper", 2, toupper2, SafePredFlag); Yap_InitCPred("toupper", 2, toupper2, SafePredFlag);