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

warnings and stream read wchar simplification

Browse Source
This commit is contained in:
Vitor Santos Costa 2016-08-01 19:17:56 -05:00
parent a82a72fc14
commit 304489c74f
6 changed files with 394 additions and 328 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
os/format.c
View File

@ -982,8 +982,7 @@ static Int doformat(volatile Term otail, volatile Term oargs,
goto do_format_control_sequence_error;
t = targs[targ++];
yhandle_t slf = Yap_StartSlots();
Yap_plwrite(t, GLOBAL_Stream + sno, 0,
Handle_vars_f | To_heap_f,
Yap_plwrite(t, GLOBAL_Stream + sno, 0, Handle_vars_f | To_heap_f,
GLOBAL_MaxPriority);
Yap_CloseSlots(slf);
break;
@ -1200,7 +1199,7 @@ static Int with_output_to(USES_REGS1) {
Term tin = Deref(ARG1);
Functor f;
bool out;
bool mem_stream = false;
bool my_mem_stream;
yhandle_t hdl = Yap_PushHandle(tin);
if (IsVarTerm(tin)) {
Yap_Error(INSTANTIATION_ERROR, tin, "with_output_to/3");
@ -1215,6 +1214,7 @@ static Int with_output_to(USES_REGS1) {
/* needs to change LOCAL_c_output_stream for write */
output_stream = Yap_CheckStream(ARG1, Output_Stream_f, "format/3");
my_mem_stream = false;
f = NIL;
}
if (output_stream == -1) {
return false;
@ -1223,7 +1223,7 @@ static Int with_output_to(USES_REGS1) {
out = Yap_Execute(Deref(ARG2) PASS_REGS);
LOCK(GLOBAL_Stream[output_stream].streamlock);
LOCAL_c_output_stream = old_out;
if (mem_stream) {
if (my_mem_stream) {
Term tat;
Term inp = Yap_GetFromHandle(hdl);
if (out) {

100
os/getw.h
View File

@ -2,9 +2,9 @@
/// compose a wide char from a sequence of getchars
/// this is a slow lane routine, called if no specialised code
/// isavailable.
static int GETW(int sno) {
extern int get_wchar(int sno) {
StreamDesc *st = GLOBAL_Stream + sno;
int ch = GETC();
int ch = st->stream_getc(sno);
if (ch == -1)
return post_process_weof(st);
@ -31,7 +31,7 @@ static int GETW(int sno) {
buf[0] = ch;
int n = 1;
while ((out = mbrtowc(&wch, buf, 1, &(mbstate))) != 1) {
int ch = buf[0] = GETC();
int ch = buf[0] = st->stream_getc(sno);
n++;
if (ch == -1)
return post_process_weof(st);
@ -49,7 +49,7 @@ static int GETW(int sno) {
// if ((ch - 0xc2) > (0xf4-0xc2)) return UTF8PROC_ERROR_INVALIDUTF8;
if (ch < 0xe0) { // 2-byte sequence
// Must have valid continuation character
int c1 = buf[0] = GETC();
int c1 = buf[0] = st->stream_getc(sno);
if (c1 == -1)
return post_process_weof(st);
// if (!utf_cont(*str)) return UTF8PROC_ERROR_INVALIDUTF8;
@ -62,22 +62,22 @@ static int GETW(int sno) {
// Check for surrogate chars
// if (ch == 0xed && *str > 0x9f)
// return UTF8PROC_ERROR_INVALIDUTF8;
int c1 = GETC();
int c1 = st->stream_getc(sno);
if (c1 == -1)
return post_process_weof(st);
int c2 = GETC();
int c2 = st->stream_getc(sno);
if (c2 == -1)
return post_process_weof(st);
wch = ((ch & 0xf) << 12) | ((c1 & 0x3f) << 6) | (c2 & 0x3f);
return post_process_read_wchar(wch, 3, st);
} else {
int c1 = GETC();
int c1 = st->stream_getc(sno);
if (c1 == -1)
return post_process_weof(st);
int c2 = GETC();
int c2 = st->stream_getc(sno);
if (c2 == -1)
return post_process_weof(st);
int c3 = GETC();
int c3 = st->stream_getc(sno);
if (c3 == -1)
return post_process_weof(st);
wch = ((ch & 7) << 18) | ((c1 & 0x3f) << 12) | ((c2 & 0x3f) << 6) |
@ -89,15 +89,15 @@ static int GETW(int sno) {
// little-endian: start with big shot
{
int wch;
int c1 = GETC();
int c1 = st->stream_getc(sno);
if (c1 == -1)
return post_process_weof(st);
wch = (c1 << 8) + ch;
if (wch >= 0xd800 && wch < 0xdc00) {
int c2 = GETC();
int c2 = st->stream_getc(sno);
if (c2 == -1)
return post_process_weof(st);
int c3 = GETC();
int c3 = st->stream_getc(sno);
if (c3 == -1)
return post_process_weof(st);
wch = wch + (((c3 << 8) + c2) << wch) + SURROGATE_OFFSET;
@ -110,15 +110,15 @@ static int GETW(int sno) {
// little-endian: start with big shot
{
int wch;
int c1 = GETC();
int c1 = st->stream_getc(sno);
if (c1 == -1)
return post_process_weof(st);
wch = (c1) + (ch << 8);
if (wch >= 0xd800 && wch < 0xdc00) {
int c3 = GETC();
int c3 = st->stream_getc(sno);
if (c3 == -1)
return post_process_weof(st);
int c2 = GETC();
int c2 = st->stream_getc(sno);
if (c2 == -1)
return post_process_weof(st);
wch = (((c3 << 8) + c2) << 10) + wch + SURROGATE_OFFSET;
@ -131,7 +131,7 @@ static int GETW(int sno) {
// little-endian: start with big shot
{
int wch;
int c1 = GETC();
int c1 = st->stream_getc(sno);
if (c1 == -1)
return post_process_weof(st);
wch = (c1) + (ch << 8);
@ -142,7 +142,7 @@ static int GETW(int sno) {
// little-endian: start with big shot
{
int wch;
int c1 = GETC();
int c1 = st->stream_getc(sno);
if (c1 == -1)
return post_process_weof(st);
wch = (c1 << 8) + ch;
@ -155,19 +155,19 @@ static int GETW(int sno) {
{
int wch = ch;
{
int c1 = GETC();
int c1 = st->stream_getc(sno);
if (c1 == -1)
return post_process_weof(st);
wch = wch + c1;
}
{
int c1 = GETC();
int c1 = st->stream_getc(sno);
if (c1 == -1)
return post_process_weof(st);
wch = (wch << 8) + c1;
}
{
int c1 = GETC();
int c1 = st->stream_getc(sno);
if (c1 == -1)
return post_process_weof(st);
wch = (wch << 8) + c1;
@ -179,19 +179,19 @@ static int GETW(int sno) {
{
int wch = ch;
{
int c1 = GETC();
int c1 = st->stream_getc(sno);
if (c1 == -1)
return post_process_weof(st);
wch += c1 << 8;
}
{
int c1 = GETC();
int c1 = st->stream_getc(sno);
if (c1 == -1)
return post_process_weof(st);
wch += c1 << 16;
}
{
int c1 = GETC();
int c1 = st->stream_getc(sno);
if (c1 == -1)
return post_process_weof(st);
wch += c1 << 24;
@ -204,3 +204,57 @@ static int GETW(int sno) {
return -1;
}
}
extern int get_wchar_UTF8(int sno) {
StreamDesc *st = GLOBAL_Stream + sno;
int ch = st->stream_getc(sno);
if (ch == -1)
return post_process_weof(st);
else {
int wch;
unsigned char buf[8];
if (ch < 0x80) {
return post_process_read_wchar(ch, 1, st);
}
// if ((ch - 0xc2) > (0xf4-0xc2)) return UTF8PROC_ERROR_INVALIDUTF8;
if (ch < 0xe0) { // 2-byte sequence
// Must have valid continuation character
int c1 = buf[0] = st->stream_getc(sno);
if (c1 == -1)
return post_process_weof(st);
// if (!utf_cont(*str)) return UTF8PROC_ERROR_INVALIDUTF8;
wch = ((ch & 0x1f) << 6) | (c1 & 0x3f);
return post_process_read_wchar(wch, 2, st);
}
if (ch < 0xf0) { // 3-byte sequence
// if ((str + 1 >= end) || !utf_cont(*str) || !utf_cont(str[1]))
// return UTF8PROC_ERROR_INVALIDUTF8;
// Check for surrogate chars
// if (ch == 0xed && *str > 0x9f)
// return UTF8PROC_ERROR_INVALIDUTF8;
int c1 = st->stream_getc(sno);
if (c1 == -1)
return post_process_weof(st);
int c2 = st->stream_getc(sno);
if (c2 == -1)
return post_process_weof(st);
wch = ((ch & 0xf) << 12) | ((c1 & 0x3f) << 6) | (c2 & 0x3f);
return post_process_read_wchar(wch, 3, st);
} else {
int c1 = st->stream_getc(sno);
if (c1 == -1)
return post_process_weof(st);
int c2 = st->stream_getc(sno);
if (c2 == -1)
return post_process_weof(st);
int c3 = st->stream_getc(sno);
if (c3 == -1)
return post_process_weof(st);
wch = ((ch & 7) << 18) | ((c1 & 0x3f) << 12) | ((c2 & 0x3f) << 6) |
(c3 & 0x3f);
return post_process_read_wchar(wch, 4, st);
}
}
}

20
os/iopreds.c
View File

@ -97,16 +97,9 @@ static char SccsId[] = "%W% %G%";
#define S_ISDIR(x) (((x)&_S_IFDIR) == _S_IFDIR)
#endif
#endif
#include "iopreds.h"
//#define GETW get_wchar_from_FILE
//#endif
#define GETC() fgetwc(st->file)
#include "getw.h"
#undef GETW
#undef GETC
#define GETW get_wchar
#define GETC() st->stream_getc(sno)
#include "iopreds.h"
#include "getw.h"
static int get_wchar_from_file(int);
@ -254,7 +247,7 @@ static void unix_upd_stream_info(StreamDesc *s) {
void Yap_DefaultStreamOps(StreamDesc *st) {
CACHE_REGS
st->stream_wputc = put_wchar;
st->stream_wgetc = get_wchar;
st->stream_wgetc = get_wchar_UTF8;
st->stream_putc = FilePutc;
st->stream_getc = PlGetc;
if (st->status & (Promptable_Stream_f)) {
@ -664,10 +657,9 @@ int PlGetc(int sno) {
}
// layered version
static int get_wchar__(int sno) { return fgetwc(GLOBAL_Stream[sno].file); }
static int get_wchar_from_file(int sno) {
return post_process_read_wchar(get_wchar__(sno), 1, GLOBAL_Stream + sno);
static inline int get_wchar_from_file(int sno) {
return post_process_read_wchar(fgetwc(GLOBAL_Stream[sno].file), 1,
GLOBAL_Stream + sno);
}
#ifndef MB_LEN_MAX

73
os/readterm.c
View File

@ -264,58 +264,76 @@ static Term syntax_error(TokEntry *errtok, int sno, Term cmod) {
t0[0] = TermNil;
}
ts[0] = Yap_MkApplTerm(Yap_MkFunctor(AtomAtom, 1), 1, t0);
} break;
case QuasiQuotes_tok: {
}
break;
case QuasiQuotes_tok:
{
Term t0[2];
t0[0] = MkAtomTerm(Yap_LookupAtom("<QQ>"));
ts[0] = Yap_MkApplTerm(Yap_MkFunctor(AtomAtom, 1), 1, t0);
} break;
case WQuasiQuotes_tok: {
}
break;
case WQuasiQuotes_tok:
{
Term t0[2];
t0[0] = MkAtomTerm(Yap_LookupAtom("<WideQQ>"));
ts[0] = Yap_MkApplTerm(Yap_MkFunctor(AtomAtom, 1), 1, t0);
} break;
}
break;
case Number_tok:
ts[0] = Yap_MkApplTerm(Yap_MkFunctor(AtomNumber, 1), 1, &(tok->TokInfo));
break;
case Var_tok: {
case Var_tok:
{
Term t[2];
VarEntry *varinfo = (VarEntry *)info;
t[0] = MkIntTerm(0);
t[1] = Yap_CharsToString(varinfo->VarRep, ENC_ISO_LATIN1 PASS_REGS);
ts[0] = Yap_MkApplTerm(Yap_MkFunctor(AtomGVar, 2), 2, t);
} break;
case String_tok: {
}
break;
case String_tok:
{
Term t0 = Yap_CharsToTDQ((char *)info, cmod, ENC_ISO_LATIN1 PASS_REGS);
if (!t0) {
return 0;
}
ts[0] = Yap_MkApplTerm(Yap_MkFunctor(AtomString, 1), 1, &t0);
} break;
case WString_tok: {
}
break;
case WString_tok:
{
Term t0 = Yap_WCharsToTDQ((wchar_t *)info, cmod PASS_REGS);
if (!t0)
return 0;
ts[0] = Yap_MkApplTerm(Yap_MkFunctor(AtomString, 1), 1, &t0);
} break;
case BQString_tok: {
}
break;
case BQString_tok:
{
Term t0 = Yap_CharsToTBQ((char *)info, cmod, ENC_ISO_LATIN1 PASS_REGS);
ts[0] = Yap_MkApplTerm(Yap_MkFunctor(AtomString, 1), 1, &t0);
} break;
case WBQString_tok: {
}
break;
case WBQString_tok:
{
Term t0 = Yap_WCharsToTBQ((wchar_t *)info, cmod PASS_REGS);
ts[0] = Yap_MkApplTerm(Yap_MkFunctor(AtomString, 1), 1, &t0);
} break;
case Error_tok: {
}
break;
case Error_tok:
{
ts[0] = MkAtomTerm(AtomError);
} break;
}
break;
case eot_tok:
endline = MkIntegerTerm(tok->TokPos);
ts[0] = MkAtomTerm(Yap_LookupAtom("EOT"));
break;
case Ponctuation_tok: {
case Ponctuation_tok:
{
char s[2];
s[1] = '\0';
if ((info) == 'l') {
@ -326,9 +344,13 @@ static Term syntax_error(TokEntry *errtok, int sno, Term cmod) {
ts[0] = MkAtomTerm(Yap_LookupAtom(s));
}
}
if (tok->TokNext) {
tok = tok->TokNext;
if (!tok)
} else {
endline = MkIntegerTerm(tok->TokPos);
tok = NULL;
break;
}
*tailp = MkPairTerm(ts[0], TermNil);
tailp = RepPair(*tailp) + 1;
}
@ -939,7 +961,8 @@ Term Yap_read_term(int inp_stream, Term opts, int nargs) {
case YAP_PARSING_ERROR:
state = parseError(&re, &fe, inp_stream);
break;
case YAP_PARSING_FINISHED: {
case YAP_PARSING_FINISHED:
{
CACHE_REGS
bool done;
if (fe.reading_clause)
@ -1294,7 +1317,6 @@ X_API Term Yap_StringToTerm(const char *s, size_t len, encoding_t *encp,
return rval;
}
/**
* @pred read_term_from_atom( +_Atom_ , - _T_ , + _Options_
*
@ -1397,8 +1419,7 @@ static Int string_to_term(USES_REGS1) {
len = strlen_utf8((const unsigned char *)s);
}
encoding_t enc = ENC_ISO_UTF8;
rc = Yap_StringToTerm(s, len, &enc,
1200, &ARG3);
rc = Yap_StringToTerm(s, len, &enc, 1200, &ARG3);
if (!rc)
return false;
return Yap_unify(rc, ARG2);
@ -1420,8 +1441,7 @@ static Int atomic_to_term(USES_REGS1) {
len = strlen_utf8((unsigned char *)s);
}
encoding_t enc = ENC_ISO_UTF8;
rc = Yap_StringToTerm(s, len, &enc,
1200, &ARG3);
rc = Yap_StringToTerm(s, len, &enc, 1200, &ARG3);
if (!rc)
return false;
return Yap_unify(rc, ARG2);
@ -1443,8 +1463,7 @@ static Int atom_to_term(USES_REGS1) {
len = strlen_utf8((const unsigned char *)s);
}
encoding_t enc = ENC_ISO_UTF8;
rc = Yap_StringToTerm(s, len, &enc,
1200, &ARG3);
rc = Yap_StringToTerm(s, len, &enc, 1200, &ARG3);
if (!rc)
return false;
return Yap_unify(rc, ARG2);

7
os/streams.c
View File

@ -248,11 +248,12 @@ has_reposition(int sno,
}
char *Yap_guessFileName(FILE *file, int sno, char *nameb, size_t max) {
size_t maxs = max(255, max);
if (!nameb) {
nameb = malloc(max(256, max));
nameb = malloc(maxs + 1);
}
if (!file) {
strcpy(nameb, "memory buffer");
strncpy(nameb, "memory buffer", maxs);
return nameb;
}
int f = fileno(file);
@ -275,7 +276,7 @@ char *Yap_guessFileName(FILE *file, int sno, char *nameb, size_t max) {
return NULL;
else {
int i;
unsigned char *ptr = nameb;
unsigned char *ptr = (unsigned char *)nameb;
for (i = 0; i < strlen(path); i++)
ptr += put_utf8(ptr, path[i]);
*ptr = '\0';

2
os/sysbits.c
View File

@ -1181,7 +1181,7 @@ const char *Yap_findFile(const char *isource, const char *idef,
YAP_file_type_t ftype, bool expand_root, bool in_lib) {
char *save_buffer = NULL;
const char *root, *source = isource;
const char *root = iroot, *source = isource;
int rc = FAIL_RESTORE;
int try
= 0;