Rev 8687 | 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 | base = 8; |
||
175 | ch = vgetc(save, src); |
||
176 | if (ch == EOF || isspace(ch)) |
||
177 | have_digits++; |
||
178 | else |
||
179 | if (ch == 'x' || ch == 'X') |
||
180 | { |
||
181 | base = 16; |
||
182 | ch = vgetc(save, src); |
||
183 | if (ch == EOF) return EOF; |
||
184 | } |
||
185 | } |
||
186 | *digit = 0; |
||
187 | for (;;) |
||
188 | { |
||
189 | // test ch is valid |
||
190 | if ((isdigit(ch) && base == 10) || |
||
191 | (isdigit(ch) && base == 8 && ch < '8') || |
||
192 | (isxdigit(ch) && base == 16)) |
||
193 | { |
||
194 | if (base == 16) |
||
195 | { |
||
196 | if (ch <= '9') ch-= '0'; |
||
197 | else |
||
198 | if (ch <= 'F') ch = 10 + ch - 'A'; |
||
199 | else |
||
200 | ch = 10 + ch - 'a'; |
||
201 | } |
||
202 | else |
||
203 | ch -= '0'; |
||
204 | *digit = *digit * base + ch; |
||
205 | have_digits++; |
||
206 | ch = vgetc(save, src); |
||
207 | if (ch == EOF || isspace(ch)) break; // ok, just finish num |
||
208 | } |
||
209 | else if (ch == EOF || isspace(ch)) |
||
210 | break; |
||
211 | else |
||
212 | { |
||
213 | vungetc(save, ch, src); |
||
214 | break; |
||
215 | } |
||
216 | } |
||
217 | *digit *= sign; |
||
218 | if (have_digits) |
||
219 | return 1; |
||
220 | else |
||
221 | return -2; |
||
222 | } |
||
223 | |||
224 | |||
225 | |||
8718 | turbocat | 226 | int format_scan(const void *src, const char *fmt, va_list argp, virtual_getc vgetc, virtual_ungetc vungetc) |
8687 | turbocat | 227 | { |
8718 | turbocat | 228 | int i; |
229 | int length; |
||
230 | int fmt1, fmt2; // width, precision |
||
231 | size_t pos, posc; |
||
232 | const char *fmtc; // first point to %, fmtc points to specifier |
||
233 | int ch; |
||
234 | int format_flag; |
||
235 | int flag_long; // 2 = long double or long long int or wchar |
||
236 | int *point_to_n = NULL, nread = 0; |
||
237 | int flags; // parsed flags |
||
238 | int save = 0; |
||
239 | char *arg_str; |
||
8687 | turbocat | 240 | int *arg_int; |
241 | long *arg_long; |
||
242 | long long *arg_longlong; |
||
243 | float *arg_float; |
||
244 | double *arg_double; |
||
245 | long double *arg_longdouble; |
||
246 | long long digit; |
||
247 | long double real; |
||
248 | |||
249 | |||
250 | pos = 0; |
||
251 | while(*fmt) |
||
252 | { |
||
253 | while (*fmt && isspace(*fmt)) fmt++; // skip paces in format str |
||
254 | |||
255 | if (*fmt != '%') // usual char |
||
256 | { |
||
257 | ch = vgetc(&save, src); |
||
258 | if (ch != *fmt++) // char not match format |
||
259 | { |
||
260 | vungetc(&save, ch, src); |
||
261 | break; |
||
262 | } |
||
263 | pos++; |
||
264 | continue; |
||
265 | } |
||
266 | |||
267 | if (*(fmt + 1) == '%') // %% |
||
268 | { |
||
269 | ch = vgetc(&save, src); |
||
270 | if (ch != '%') // char not match format |
||
271 | { |
||
272 | vungetc(&save, ch, src); |
||
273 | break; |
||
274 | } |
||
275 | pos++; |
||
276 | fmt += 2; |
||
277 | continue; |
||
278 | } |
||
279 | //checking to containg format in the string |
||
280 | fmtc = fmt; |
||
281 | posc = pos; |
||
282 | |||
283 | flags = 0; |
||
284 | format_flag = 0; |
||
285 | flag_long = 0; // 2 = long double or long long int or wchar |
||
286 | |||
287 | while(*fmtc != '\0' && !format_flag) // searching end of format |
||
288 | { |
||
289 | fmtc++; posc++; |
||
290 | switch( *fmtc ) |
||
291 | { |
||
292 | case 'a': |
||
293 | format_flag = 1; |
||
294 | flags |= flag_unsigned; |
||
295 | break; |
||
296 | case 'c': case 'd': case 'i': case 'e': case 'f': case 'g': case 's': case 'n': |
||
297 | format_flag = 1; |
||
298 | break; |
||
299 | case 'l': |
||
300 | flag_long = flag_long ? 2 : 1; // ll.eq.L |
||
301 | break; |
||
302 | case 'L': |
||
303 | flag_long = 2; |
||
304 | break; |
||
305 | case 'o': case 'u': case 'x': case 'p': |
||
306 | format_flag = 1; |
||
307 | flags |= flag_unsigned; |
||
308 | break; |
||
309 | case '*': case '.': // just skip |
||
310 | break; |
||
311 | default: |
||
312 | if(isdigit(*fmtc)) break; |
||
313 | goto exit_me; // non format char found - user error |
||
314 | } |
||
315 | } |
||
316 | |||
317 | if (format_flag == 0) |
||
318 | { |
||
319 | goto exit_me; // format char not found - user error |
||
320 | } |
||
321 | |||
322 | fmt1 = 0; |
||
323 | fmt2 = 0; |
||
324 | if (posc - pos > 1) // try to read width, precision |
||
325 | { |
||
326 | fmt++; |
||
327 | for(i = pos + 1; i < posc; i++) |
||
328 | { |
||
329 | switch(*fmt) |
||
330 | { |
||
331 | case '0': |
||
332 | if(fmt1 == 0 && (flags & flag_point) == 0) flags |= flag_lead_zeros; |
||
333 | case '1': case '2': case '3': case '4': |
||
334 | case '5': case '6': case '7': case '8': case '9': |
||
335 | if ((flags & flag_point) == 0) |
||
336 | fmt1 = fmt1 * 10 + (*fmt -'0'); |
||
337 | else |
||
338 | fmt2 = fmt2 * 10 + (*fmt -'0'); |
||
339 | break; |
||
340 | case '*': // ignoring |
||
341 | break; |
||
342 | case '.': |
||
343 | flags |= flag_point; |
||
344 | break; |
||
345 | case 'l': case 'L': // valid chars - skip |
||
346 | break; |
||
347 | default: // must be error |
||
348 | goto exit_me; // format char not found - user error |
||
349 | } |
||
350 | fmt++; |
||
351 | } |
||
352 | } |
||
353 | |||
354 | // do real work - format arguments values |
||
355 | // skip input spaces |
||
356 | do { |
||
357 | ch = vgetc(&save, src); |
||
358 | if (ch == EOF) goto exit_me; |
||
359 | } while (isspace(ch)); |
||
360 | |||
361 | switch(*fmtc) |
||
362 | { |
||
363 | case 'n': |
||
364 | point_to_n = va_arg(argp, int*); |
||
365 | vungetc(&save, ch, src); |
||
366 | break; |
||
367 | case 'c': // read width chars, ever spaces |
||
368 | arg_str = va_arg(argp, char*); |
||
369 | if (fmt1 == 0) length = 1; |
||
370 | else length = fmt1; |
||
371 | for (i = 0; i < length;) |
||
372 | { |
||
373 | *arg_str++ = ch; i++; |
||
374 | ch = vgetc(&save, src); |
||
375 | if (ch == EOF) break; |
||
376 | } |
||
377 | if (i < length) goto exit_me; // not enough chars |
||
378 | break; |
||
379 | case 's': |
||
380 | arg_str = va_arg(argp, char*); |
||
8718 | turbocat | 381 | if (fmt1 == 0) length = 4095; // max string scan 4096 |
8687 | turbocat | 382 | else length = fmt1; |
383 | for (i = 0; i < length; i++) |
||
384 | { |
||
385 | *arg_str++ = ch; |
||
386 | ch = vgetc(&save, src); |
||
387 | if (ch == EOF || isspace(ch)) break; // ok, just finish string |
||
388 | } |
||
389 | *arg_str++ = '\0'; |
||
390 | break; |
||
391 | case 'd': case 'i': case 'u': |
||
392 | case 'o': case 'p': case 'x': |
||
8718 | turbocat | 393 | i = try_parse_int(&digit, ch, src, &save, vgetc, vungetc); |
8687 | turbocat | 394 | if (i < 0) goto exit_me; |
395 | |||
396 | if (flag_long == 0) { arg_int = va_arg(argp, int*); *arg_int = (int)digit; } else |
||
397 | if (flag_long == 1) { arg_long = va_arg(argp, long*); *arg_long = (long)digit; } else |
||
398 | if (flag_long == 2) { arg_longlong = va_arg(argp, long long*); *arg_longlong = digit; } |
||
399 | break; |
||
400 | case 'a': case 'A': case 'f': case 'F': |
||
401 | case 'e': case 'E': |
||
402 | case 'g': case 'G': |
||
8718 | turbocat | 403 | i = try_parse_real(&real, ch, src, &save, vgetc, vungetc); |
8687 | turbocat | 404 | if (i < 0) goto exit_me; |
405 | |||
406 | if (flag_long == 0) { arg_float = va_arg(argp, float*); *arg_float = (float)real; } else |
||
407 | if (flag_long == 1) { arg_double = va_arg(argp, double*); *arg_double = (double)real; } else |
||
408 | if (flag_long == 2) { arg_longdouble = va_arg(argp, long double*); *arg_longdouble = real; } |
||
409 | break; |
||
410 | } |
||
411 | |||
412 | fmt = fmtc + 1; |
||
413 | nread++; |
||
414 | } |
||
415 | exit_me: |
||
416 | if (point_to_n) *point_to_n = nread; |
||
417 | return nread; |
||
418 | }>>>>>>=>=>> |