3
/* The main engine for reading formatted data
5
** Written by Kiem-Phong Vo.
8
#define MAXWIDTH (int)(((uint)~0)>>1) /* max amount to scan */
11
static char* setclass(reg char* form, reg char* accept)
13
static char* setclass(form,accept)
14
reg char* form; /* format string */
15
reg char* accept; /* accepted characters are set to 1 */
20
if((fmt = *form++) == '^')
21
{ /* we want the complement of this set */
27
for(c = 0; c <= SF_MAXCHAR; ++c)
30
if(fmt == ']' || fmt == '-')
31
{ /* special first char */
36
for(; fmt != ']'; fmt = *form++)
42
if(fmt != '-' || form[0] == ']' || form[-2] > form[0])
44
else for(c = form[-2]+1; c < form[0]; ++c)
52
static void _sfbuf(Sfio_t* f, int* rs)
54
static void _sfbuf(f, rs)
59
if(f->next >= f->endb)
60
{ if(*rs > 0) /* try peeking for a share stream if possible */
62
if(SFFILBUF(f,-1) > 0)
66
*rs = -1; /* can't peek, back to normal reads */
73
int sfvscanf(Sfio_t* f, reg const char* form, va_list args)
75
int sfvscanf(f,form,args)
76
Sfio_t* f; /* file to be scanned */
77
reg char* form; /* scanning format */
81
reg uchar *d, *endd, *data;
82
reg int inp, shift, base, width;
84
int fmt, flags, dot, n_assign, v, n, n_input;
86
char accept[SF_MAXDIGITS];
97
Void_t* value; /* location to assign scanned value */
102
#define SFBUF(f) (_sfbuf(f,&rs), (data = d = f->next), (endd = f->endb) )
103
#define SFLEN(f) (d-data)
104
#define SFEND(f) ((n_input += d-data), \
105
(rs > 0 ? SFREAD(f,(Void_t*)data,d-data) : ((f->next = d), 0)) )
106
#define SFGETC(f,c) ((c) = (d < endd || (SFEND(f), SFBUF(f), d < endd)) ? \
108
#define SFUNGETC(f,c) (--d)
115
if(f->mode != SF_READ && _sfmode(f,SF_READ,0) < 0)
119
rs = (f->extent < 0 && (f->flags&SF_SHARE)) ? 1 : 0;
121
SFCVINIT(); /* initialize conversion tables */
124
n_assign = n_input = 0;
128
fmstk = NIL(Fmt_t*); ft = NIL(Sffmt_t*);
130
fp = NIL(Fmtpos_t*); argn = -1;
131
oform = (char*)form; va_copy(oargs,args);
134
while((fmt = *form++) )
137
{ if(fmt != '\n' || !(f->flags&SF_LINE))
140
{ if(SFGETC(f,inp) < 0 || inp == fmt)
142
else if(!isspace(inp))
150
if(SFGETC(f,inp) != fmt)
168
{ flags = SFFMT_SKIP;
173
/* matching some pattern */
174
base = 10; size = -1;
176
t_str = NIL(char*); n_str = 0;
177
value = NIL(Void_t*);
180
loop_flags: /* LOOP FOR FLAGS, WIDTH, BASE, TYPE */
181
switch((fmt = *form++) )
183
case LEFTP : /* get the type which is enclosed in balanced () */
188
case 0 : /* not balanceable, retract */
193
case LEFTP : /* increasing nested level */
196
case RIGHTP : /* decreasing nested level */
200
n_str = (form-1) - t_str;
202
{ t_str = (*_Sffmtintf)(t_str+1,&n);
210
else n = FP_SET(-1,argn);
213
{ t_str = fp[n].argv.s;
214
n_str = fp[n].ft.size;
216
else if(ft && ft->extf )
217
{ FMTSET(ft, form,args,
221
(f,(Void_t*)&argv,ft);
224
if(!(ft->flags&SFFMT_VALUE) )
226
if((t_str = argv.s) &&
227
(n_str = (int)ft->size) < 0)
228
n_str = strlen(t_str);
232
if((t_str = va_arg(args,char*)) )
233
n_str = strlen(t_str);
240
case '#' : /* alternative format */
241
flags |= SFFMT_ALTER;
244
case '.' : /* width & base */
250
else if(*form == '*')
251
{ form = (*_Sffmtintf)(form+1,&n);
254
if(!fp && !(fp = (*_Sffmtposf)(f,oform,oargs,1)) )
258
else n = FP_SET(-1,argn);
262
else if(ft && ft->extf )
263
{ FMTSET(ft, form,args, '.',dot, 0, 0,0,0,
265
if((*ft->extf)(f, (Void_t*)(&argv), ft) < 0)
267
if(ft->flags&SFFMT_VALUE)
269
else v = (dot <= 2) ? va_arg(args,int) : 0;
271
else v = (dot <= 2) ? va_arg(args,int) : 0;
276
else goto loop_flags;
278
case '0' : case '1' : case '2' : case '3' : case '4' :
279
case '5' : case '6' : case '7' : case '8' : case '9' :
281
for(v = fmt-'0'; isdigit(*form); ++form)
282
v = v*10 + (*form - '0');
286
if(!fp && !(fp = (*_Sffmtposf)(f,oform,oargs,1)) )
293
if(dot == 0 || dot == 1)
299
case 'I' : /* object size */
301
flags = (flags & ~SFFMT_TYPES) | SFFMT_IFLAG;
303
{ for(n = *form; isdigit(n); n = *++form)
304
size = size*10 + (n - '0');
306
else if(*form == '*')
307
{ form = (*_Sffmtintf)(form+1,&n);
310
if(!fp && !(fp = (*_Sffmtposf)(f,oform,oargs,1)))
314
else n = FP_SET(-1,argn);
316
if(fp) /* use position list */
318
else if(ft && ft->extf )
319
{ FMTSET(ft, form,args, 'I',sizeof(int), 0, 0,0,0,
321
if((*ft->extf)(f, (Void_t*)(&argv), ft) < 0)
323
if(ft->flags&SFFMT_VALUE)
325
else size = va_arg(args,int);
327
else size = va_arg(args,int);
333
flags &= ~SFFMT_TYPES;
336
flags |= SFFMT_LLONG;
338
else flags |= SFFMT_LONG;
342
flags &= ~SFFMT_TYPES;
345
flags |= SFFMT_SSHORT;
347
else flags |= SFFMT_SHORT;
351
flags = (flags&~SFFMT_TYPES) | SFFMT_LDOUBLE;
355
flags = (flags&~SFFMT_TYPES) | SFFMT_JFLAG;
359
flags = (flags&~SFFMT_TYPES) | SFFMT_ZFLAG;
363
flags = (flags&~SFFMT_TYPES) | SFFMT_TFLAG;
367
/* set object size */
368
if(flags & (SFFMT_TYPES & ~SFFMT_IFLAG) )
369
{ if((_Sftype[fmt]&(SFFMT_INT|SFFMT_UINT)) || fmt == 'n')
370
{ size = (flags&SFFMT_LLONG) ? sizeof(Sflong_t) :
371
(flags&SFFMT_LONG) ? sizeof(long) :
372
(flags&SFFMT_SHORT) ? sizeof(short) :
373
(flags&SFFMT_SSHORT) ? sizeof(char) :
374
(flags&SFFMT_JFLAG) ? sizeof(Sflong_t) :
375
(flags&SFFMT_TFLAG) ? sizeof(ptrdiff_t) :
376
(flags&SFFMT_ZFLAG) ? sizeof(size_t) :
379
else if(_Sftype[fmt]&SFFMT_FLOAT)
380
{ size = (flags&SFFMT_LDOUBLE) ? sizeof(Sfdouble_t) :
381
(flags&(SFFMT_LONG|SFFMT_LLONG)) ?
386
argp = FP_SET(argp,argn);
388
{ if(!(fp[argp].ft.flags&SFFMT_SKIP) )
390
value = fp[argp].argv.vp;
391
size = fp[argp].ft.size;
392
if(ft && ft->extf && fp[argp].ft.fmt != fp[argp].fmt)
393
fmt = fp[argp].ft.fmt;
395
else flags |= SFFMT_SKIP;
397
else if(ft && ft->extf)
398
{ FMTSET(ft, form,args, fmt, size,flags, width,0,base, t_str,n_str);
399
SFEND(f); SFOPEN(f,0);
400
v = (*ft->extf)(f, (Void_t*)&argv, ft);
401
SFLOCK(f,0); SFBUF(f);
405
else if(v == 0) /* extf did not use input stream */
406
{ FMTGET(ft, form,args, fmt, size, flags, width,n,base);
407
if((ft->flags&SFFMT_VALUE) && !(ft->flags&SFFMT_SKIP) )
410
else /* v > 0: number of input bytes consumed */
412
if(!(ft->flags&SFFMT_SKIP) )
418
if(_Sftype[fmt] == 0) /* unknown pattern */
421
/* get the address to assign value */
422
if(!value && !(flags&SFFMT_SKIP) )
423
value = va_arg(args,Void_t*);
428
fp = (*_Sffmtposf)(f,oform,oargs,1);
431
if(!(argv.ft = va_arg(args,Sffmt_t*)) )
433
if(!argv.ft->form && ft ) /* change extension functions */
435
(*ft->eventf)(f,SF_DPOP,(Void_t*)form,ft) < 0)
437
fmstk->ft = ft = argv.ft;
439
else /* stack a new environment */
440
{ if(!(fm = (Fmt_t*)malloc(sizeof(Fmt_t))) )
444
{ fm->form = (char*)form;
445
va_copy(fm->args,args);
448
va_copy(fm->oargs,oargs);
452
form = argv.ft->form;
453
va_copy(args,argv.ft->args);
457
else fm->form = NIL(char*);
459
fm->eventf = argv.ft->eventf;
468
if(fmt == 'n') /* return length of consumed input */
470
#if !_ast_intmax_long
471
if(FMTCMP(size,Sflong_t,Sflong_t))
472
*((Sflong_t*)value) = (Sflong_t)(n_input+SFLEN(f));
475
if(sizeof(long) > sizeof(int) && FMTCMP(size,long,Sflong_t))
476
*((long*)value) = (long)(n_input+SFLEN(f));
477
else if(sizeof(short) < sizeof(int) &&
478
FMTCMP(size,short,Sflong_t))
479
*((short*)value) = (short)(n_input+SFLEN(f));
480
else if(size == sizeof(char))
481
*((char*)value) = (char)(n_input+SFLEN(f));
482
else *((int*)value) = (int)(n_input+SFLEN(f));
486
/* if get here, start scanning input */
488
width = fmt == 'c' ? 1 : MAXWIDTH;
490
/* define the first input character */
491
if(fmt == 'c' || fmt == '[')
494
{ do { SFGETC(f,inp); }
495
while(isspace(inp)) /* skip starting blanks */
501
if(_Sftype[fmt] == SFFMT_FLOAT)
503
reg int dot, exponent;
506
if(width >= SF_MAXDIGITS)
507
width = SF_MAXDIGITS-1;
513
{ /* too many dots */
518
else if(inp == 'e' || inp == 'E')
523
if(--width <= 0 || SFGETC(f,inp) < 0 ||
524
(inp != '-' && inp != '+' && !isdigit(inp)) )
528
else if(inp == '-' || inp == '+')
529
{ /* too many signs */
536
} while(--width > 0 && SFGETC(f,inp) >= 0);
540
#if !_ast_fltmax_double
541
if(FMTCMP(size,Sfdouble_t,Sfdouble_t))
542
argv.ld = _sfstrtod(accept,NIL(char**));
545
argv.d = (double)strtod(accept,NIL(char**));
550
#if !_ast_fltmax_double
551
if(FMTCMP(size,Sfdouble_t,Sfdouble_t))
552
*((Sfdouble_t*)value) = argv.ld;
555
if(FMTCMP(size,double,Sfdouble_t))
556
*((double*)value) = argv.d;
557
else *((float*)value) = (float)argv.d;
560
else if(_Sftype[fmt] == SFFMT_UINT || fmt == 'p')
567
else if(_Sftype[fmt] == SFFMT_INT)
569
if(inp == '-' || inp == '+')
571
flags |= SFFMT_MINUS;
572
while(--width > 0 && SFGETC(f,inp) >= 0)
581
else if(fmt == 'x' || fmt == 'p')
583
else if(fmt == 'i' && inp == '0') /* self-described data */
585
if(width > 1) /* peek to see if it's a base-16 */
586
{ if(SFGETC(f,inp) >= 0)
587
{ if(inp == 'x' || inp == 'X')
598
{ sp = (char*)_Sfcv36;
604
if(inp == '0' && --width > 0)
605
{ /* skip leading 0x or 0X */
606
if(SFGETC(f,inp) >= 0 &&
607
(inp == 'x' || inp == 'X') && --width > 0)
610
if(inp >= 0 && sp[inp] < 16)
614
{ /* fast base 10 conversion */
615
if(inp < '0' || inp > '9')
621
{ argv.lu = (argv.lu<<3) + (argv.lu<<1) + (inp-'0');
622
} while(--width > 0 &&
623
SFGETC(f,inp) >= '0' && inp <= '9');
625
if(fmt == 'i' && inp == '#' && !(flags&SFFMT_ALTER) )
626
{ base = (int)argv.lu;
627
if(base < 2 || base > SF_RADIX)
630
sp = base <= 36 ? (char*)_Sfcv36 : (char*)_Sfcv64;
632
SFGETC(f,inp) >= 0 && sp[inp] < base)
638
sp = base <= 36 ? (char*)_Sfcv36 : (char*)_Sfcv64;
639
if(base < 2 || base > SF_RADIX || sp[inp] >= base)
644
base_conv: /* check for power of 2 conversions */
645
if((base & ~(base-1)) == base)
647
shift = base < 4 ? 1 : 2;
649
shift = base < 16 ? 3 : 4;
650
else shift = base < 64 ? 5 : 6;
653
{ argv.lu = (argv.lu << shift) + sp[inp];
654
} while(--width > 0 &&
655
SFGETC(f,inp) >= 0 && sp[inp] < base);
659
{ argv.lu = (argv.lu * base) + sp[inp];
660
} while(--width > 0 &&
661
SFGETC(f,inp) >= 0 && sp[inp] < base);
665
if(flags&SFFMT_MINUS)
673
*((Void_t**)value) = (Void_t*)((ulong)argv.lu);
675
*((Void_t**)value) = (Void_t*)((uint)argv.lu);
677
#if !_ast_intmax_long
678
else if(FMTCMP(size,Sflong_t,Sflong_t))
679
*((Sflong_t*)value) = argv.ll;
681
else if(sizeof(long) > sizeof(int) &&
682
FMTCMP(size,long,Sflong_t))
683
{ if(fmt == 'd' || fmt == 'i')
684
*((long*)value) = (long)argv.ll;
685
else *((ulong*)value) = (ulong)argv.lu;
687
else if(sizeof(short) < sizeof(int) &&
688
FMTCMP(size,short,Sflong_t))
689
{ if(fmt == 'd' || fmt == 'i')
690
*((short*)value) = (short)argv.ll;
691
else *((ushort*)value) = (ushort)argv.lu;
693
else if(size == sizeof(char) )
694
{ if(fmt == 'd' || fmt == 'i')
695
*((char*)value) = (char)argv.ll;
696
else *((uchar*)value) = (uchar)argv.lu;
699
{ if(fmt == 'd' || fmt == 'i')
700
*((int*)value) = (int)argv.ll;
701
else *((uint*)value) = (uint)argv.lu;
705
else if(fmt == 's' || fmt == 'c' || fmt == '[')
709
{ argv.s = (char*)value;
722
} while(--width > 0 && SFGETC(f,inp) >= 0);
726
{ if((n += 1) <= size)
728
} while(--width > 0 && SFGETC(f,inp) >= 0);
730
else /* if(fmt == '[') */
731
{ form = setclass((char*)form,accept);
734
{ if(n > 0 || (flags&SFFMT_ALTER) )
743
} while(--width > 0 && SFGETC(f,inp) >= 0);
746
if(value && (n > 0 || fmt == '[') )
748
if(fmt != 'c' && size >= 0)
753
if(width > 0 && inp >= 0)
762
while((fm = fmstk) ) /* pop the format stack and continue */
764
{ if(!form || !form[0])
765
(*fm->eventf)(f,SF_FINAL,NIL(Void_t*),ft);
766
else if((*fm->eventf)(f,SF_DPOP,(Void_t*)form,ft) < 0)
771
if((form = fm->form) )
772
{ va_copy(args, fm->args);
774
va_copy(oargs,fm->oargs);
789
(*fm->eventf)(f,SF_FINAL,NIL(Void_t*),fm->ft);
797
if(n_assign == 0 && inp < 0)
800
SFMTXRETURN(f,n_assign);