Subversion Repositories Kolibri OS

Rev

Rev 8793 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
8687 turbocat 1
/*
2
        function for format read from any source
3
 
4
Siemargl formats as http://www.cplusplus.com/reference/cstdio/scanf/, no wchar though
5
http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap05.html  is used too
6
 
7
todo:
8
[characters], [^characters]
9
-%n nothing scanned, filled only if good result
10
-%d, i, u, o, x, p read similar - detect base by prefix 0 or 0x
11
-%a
12
-can overflow unsigned as signed
13
-radix point always '.', no LOCALEs
14
*/
15
 
8718 turbocat 16
 
8687 turbocat 17
#include 
18
#include 
19
#include 
20
#include 
21
 
8718 turbocat 22
typedef int (*virtual_getc)(void *sp, const void *obj);
23
typedef void (*virtual_ungetc)(void *sp, int c, const void *obj);
24
 
25
enum flags_t
26
{
27
        flag_unsigned   = 0x02,
28
        flag_register   = 0x04,
29
        flag_plus       = 0x08,
30
        flag_left_just  = 0x10,
31
        flag_lead_zeros = 0x20,
32
        flag_space_plus = 0x40,
33
        flag_hash_sign  = 0x80,
34
        flag_point      = 0x100
35
};
36
 
37
int     try_parse_real(long double *real, int ch, const void *src, void *save, virtual_getc vgetc, virtual_ungetc vungetc)
8687 turbocat 38
// returns 1 if OK, -1 == EOF, -2 parse broken
39
{
40
    int sign = 1, have_digits = 0;
41
    long long div;
42
 
43
    if (ch == '+')
44
    {
45
        ch = vgetc(save, src);
46
        if (ch == EOF) return EOF;
47
    } else
48
    if (ch == '-')
49
    {
50
        sign = -1;
51
        ch = vgetc(save, src);
52
        if (ch == EOF) return EOF;
53
    };
54
    *real = 0.0;
55
    for (;;) // mantissa before point
56
    {
57
        // test ch is valid
58
        if (isdigit(ch))
59
        {
60
            *real = *real * 10 + ch - '0';
61
            have_digits++;
62
            ch = vgetc(save, src);
63
            if (ch == EOF || isspace(ch)) break; // ok, just finish num
64
        } else
65
        if (ch == '.' || ch == 'E' || ch == 'e')
66
        {
67
            break; // ok
68
        }
69
        else
70
        {
71
            vungetc(save, ch, src);
72
            break;
73
        }
74
    }
75
    if (ch != '.' && ch != 'E' && ch != 'e') // ok, just integer part
76
    {
77
        *real *= sign;
78
        if (have_digits)
79
            return 1;
80
        else
81
            return -2;
82
    }
83
 
84
    if(ch == '.')
85
    {
86
        ch = vgetc(save, src);
87
        div = 10; // use as divisor
88
        for (;;) // mantissa after point
89
        {
90
            // test ch is valid
91
            if (isdigit(ch))
92
            {
93
                *real += (double)(ch - '0') / div;
94
                div *= 10;
95
                have_digits++;
96
                ch = vgetc(save, src);
97
                if (ch == EOF || isspace(ch)) break; // ok, just finish num
98
            } else
99
            if (ch == 'E' || ch == 'e')
100
            {
101
                break; // ok
102
            }
103
            else
104
            {
105
                vungetc(save, ch, src);
106
                break;
107
            }
108
        }
109
        if (ch != 'E' && ch != 'e')  // ok, real as XX.YY
110
        {
111
            *real *= sign;
112
            if (have_digits)
113
                return 1;
114
            else
115
                return -2;
116
        }
117
    }
118
 
119
    ch = vgetc(save, src);
120
    *real *= sign;
121
    // exponent
122
    sign = 1;
123
    if (ch == '+')
124
    {
125
        ch = vgetc(save, src);
126
        if (ch == EOF) return EOF;
127
    } else
128
    if (ch == '-')
129
    {
130
        sign = -1;
131
        ch = vgetc(save, src);
132
        if (ch == EOF) return EOF;
133
    };
134
    div = 0;
135
    for (;;)
136
    {
137
        // test ch is valid
138
        if (isdigit(ch))
139
        {
140
            div = div * 10 + ch - '0';
141
            ch = vgetc(save, src);
142
            if (ch == EOF || isspace(ch)) break; // ok, just finish num
143
        }
144
        else
145
        {
146
            vungetc(save, ch, src);
147
            break;
148
        }
149
    }
150
    div *= sign;
8718 turbocat 151
    *real *= pow(10, div);
8687 turbocat 152
 
153
    return 1;
154
}
155
 
8718 turbocat 156
int     try_parse_int(long long *digit, int ch, const void *src, void *save, virtual_getc vgetc, virtual_ungetc vungetc)
8687 turbocat 157
{
158
    int sign = 1, base = 10, have_digits = 0;
159
 
160
    if (ch == '+')
161
    {
162
        ch = vgetc(save, src);
163
        if (ch == EOF) return EOF;
164
    } else
165
    if (ch == '-')
166
    {
167
        sign = -1;
168
        ch = vgetc(save, src);
169
        if (ch == EOF) return EOF;
170
    };
171
 
172
    if (ch == '0')  // octal or hex, read next
173
    {
174
        ch = vgetc(save, src);
9529 IgorA 175
        if (ch == 'c' || ch == 'C')
176
            base = 8;
177
        else if (ch == 'x' || ch == 'X')
178
            base = 16;
179
        if (base == 10)
8687 turbocat 180
            have_digits++;
181
        else
182
        {
9529 IgorA 183
            char tch = vgetc(save, src);
184
            if ((base == 8 && isdigit(tch) && tch < '8') ||
185
                (base == 16 && isxdigit(tch)))
186
                ch = tch;
187
            else
188
            {
189
                have_digits++;
190
                //base = 10; // not required: zero is zero with any (base > 1)
191
                vungetc(save, tch, src);
192
            }
8687 turbocat 193
        }
194
    }
195
    *digit = 0;
196
    for (;;)
197
    {
198
        // test ch is valid
199
        if ((isdigit(ch) && base == 10) ||
200
            (isdigit(ch) && base == 8 && ch < '8') ||
201
            (isxdigit(ch) && base == 16))
202
        {
203
            if (base == 16)
204
            {
9529 IgorA 205
                if (ch <= '9') ch -= '0';
8687 turbocat 206
                else
207
                if (ch <= 'F') ch = 10 + ch - 'A';
208
                else
209
                    ch = 10 + ch - 'a';
210
            }
211
            else
212
                ch -= '0';
213
            *digit = *digit * base + ch;
214
            have_digits++;
215
            ch = vgetc(save, src);
216
            if (ch == EOF || isspace(ch)) break; // ok, just finish num
217
        }
218
        else if (ch == EOF || isspace(ch))
219
            break;
220
        else
221
        {
222
            vungetc(save, ch, src);
223
            break;
224
        }
225
    }
226
    *digit *= sign;
227
    if (have_digits)
228
        return 1;
229
    else
230
        return -2;
231
}
232
 
233
 
234
 
8718 turbocat 235
int format_scan(const void *src, const char *fmt, va_list argp, virtual_getc vgetc, virtual_ungetc vungetc)
8687 turbocat 236
{
8718 turbocat 237
    int                     i;
238
    int                     length;
239
    int                     fmt1, fmt2;  // width, precision
240
    size_t                  pos, posc;
241
    const char            *fmtc;  // first point to %, fmtc points to specifier
242
    int                    ch;
243
    int                     format_flag;
244
    int                     flag_long;        // 2 = long double or long long int or wchar
245
    int                    *point_to_n = NULL, nread = 0;
246
    int                     flags;  // parsed flags
247
    int                     save = 0;
248
    char                *arg_str;
8687 turbocat 249
    int	        *arg_int;
250
    long        *arg_long;
251
    long long	*arg_longlong;
252
    float       *arg_float;
253
    double      *arg_double;
254
    long double *arg_longdouble;
255
    long long  digit;
256
    long double real;
257
 
258
 
259
    pos = 0;
260
    while(*fmt)
261
    {
262
        while (*fmt && isspace(*fmt)) fmt++; // skip paces in format str
263
 
264
        if (*fmt != '%')  // usual char
265
        {
266
            ch = vgetc(&save, src);
267
            if (ch != *fmt++) // char not match format
268
            {
269
                vungetc(&save, ch, src);
270
                break;
271
            }
272
            pos++;
273
            continue;
274
        }
275
 
276
        if (*(fmt + 1) == '%') // %%
277
        {
278
            ch = vgetc(&save, src);
279
            if (ch != '%') // char not match format
280
            {
281
                vungetc(&save, ch, src);
282
                break;
283
            }
284
            pos++;
285
            fmt += 2;
286
            continue;
287
        }
288
        //checking to containg format in the string
289
        fmtc = fmt;
290
        posc = pos;
291
 
292
        flags = 0;
293
        format_flag = 0;
294
        flag_long = 0;  // 2 = long double or long long int or wchar
295
 
296
        while(*fmtc != '\0' && !format_flag)    // searching end of format
297
        {
298
                fmtc++; posc++;
299
                switch( *fmtc )
300
                {
301
                case 'a':
302
                    format_flag = 1;
303
                    flags |= flag_unsigned;
304
                    break;
305
                case 'c':   case 'd':   case 'i':   case 'e':   case 'f':   case 'g':   case 's':   case 'n':
306
                    format_flag = 1;
307
                    break;
308
                case 'l':
309
                    flag_long  = flag_long ? 2 : 1;  // ll.eq.L
310
                    break;
311
                case 'L':
312
                    flag_long = 2;
313
                    break;
314
                case 'o':   case 'u':   case 'x':   case 'p':
315
                    format_flag = 1;
316
                    flags |= flag_unsigned;
317
                    break;
318
                case '*':   case '.':  // just skip
319
                    break;
320
                default:
321
                    if(isdigit(*fmtc))  break;
322
                    goto exit_me;  // non format char found - user error
323
                }
324
        }
325
 
326
        if (format_flag == 0)
327
        {
328
            goto exit_me;  // format char not found - user error
329
        }
330
 
331
        fmt1 = 0;
332
        fmt2 = 0;
333
        if (posc - pos > 1)  // try to read width, precision
334
        {
335
            fmt++;
336
            for(i = pos + 1; i < posc; i++)
337
            {
338
                switch(*fmt)
339
                {
340
                case '0':
341
                    if(fmt1 == 0 && (flags & flag_point) == 0)    flags |= flag_lead_zeros;
342
                case '1':   case '2':   case '3':   case '4':
343
                case '5':   case '6':   case '7':   case '8':   case '9':
344
                    if ((flags & flag_point) == 0)
345
                        fmt1 = fmt1 * 10 + (*fmt -'0');
346
                    else
347
                        fmt2 = fmt2 * 10 + (*fmt -'0');
348
                    break;
349
                case '*': // ignoring
350
                    break;
351
                case '.':
352
                    flags |= flag_point;
353
                    break;
354
                case 'l':   case 'L':      // valid chars - skip
355
                    break;
356
                default: // must be error
357
                    goto exit_me;  // format char not found - user error
358
                }
359
                fmt++;
360
            }
361
        }
362
 
363
        // do real work - format arguments values
364
        // skip input spaces
365
        do {
366
            ch = vgetc(&save, src);
367
            if (ch == EOF) goto exit_me;
368
        } while (isspace(ch));
369
 
370
        switch(*fmtc)
371
        {
372
        case 'n':
373
            point_to_n = va_arg(argp, int*);
374
            vungetc(&save, ch, src);
375
            break;
376
        case 'c':  // read width chars, ever spaces
377
            arg_str = va_arg(argp, char*);
378
            if (fmt1 == 0) length = 1;
379
            else length = fmt1;
380
            for (i = 0; i < length;)
381
            {
382
                *arg_str++ = ch; i++;
383
                ch = vgetc(&save, src);
384
                if (ch == EOF) break;
385
            }
386
            if (i < length) goto exit_me; // not enough chars
387
            break;
388
        case 's':
389
            arg_str = va_arg(argp, char*);
8718 turbocat 390
            if (fmt1 == 0) length = 4095;   // max string scan 4096
8687 turbocat 391
            else length = fmt1;
392
            for (i = 0; i < length; i++)
393
            {
394
                *arg_str++ = ch;
395
                ch = vgetc(&save, src);
396
                if (ch == EOF || isspace(ch)) break; // ok, just finish string
397
            }
398
            *arg_str++ = '\0';
399
            break;
400
        case 'd':   case 'i':   case 'u':
401
        case 'o':   case 'p':   case 'x':
8718 turbocat 402
            i = try_parse_int(&digit, ch, src, &save, vgetc, vungetc);
8687 turbocat 403
            if (i < 0) goto exit_me;
404
 
405
            if (flag_long == 0) { arg_int = va_arg(argp, int*); *arg_int = (int)digit; } else
406
            if (flag_long == 1) { arg_long = va_arg(argp, long*); *arg_long = (long)digit; } else
407
            if (flag_long == 2) { arg_longlong = va_arg(argp, long long*); *arg_longlong = digit; }
408
            break;
409
        case 'a':   case 'A':   case 'f':   case 'F':
410
        case 'e':   case 'E':
411
        case 'g':   case 'G':
8718 turbocat 412
            i = try_parse_real(&real, ch, src, &save, vgetc, vungetc);
8687 turbocat 413
            if (i < 0) goto exit_me;
414
 
415
            if (flag_long == 0) { arg_float = va_arg(argp, float*); *arg_float = (float)real; } else
416
            if (flag_long == 1) { arg_double = va_arg(argp, double*); *arg_double = (double)real; } else
417
            if (flag_long == 2) { arg_longdouble = va_arg(argp, long double*); *arg_longdouble = real; }
418
            break;
419
        }
420
 
421
        fmt = fmtc + 1;
422
        nread++;
423
    }
424
exit_me:
425
    if (point_to_n) *point_to_n = nread;
426
    return  nread;
427
}