Commit | Line | Data |
---|---|---|
24f98398 | 1 | /* Output like sprintf to a buffer of specified size. |
762b15be EZ |
2 | Also takes args differently: pass one pointer to the end |
3 | of the format string in addition to the format string itself. | |
73b0cd50 | 4 | Copyright (C) 1985, 2001-2011 Free Software Foundation, Inc. |
24f98398 JA |
5 | |
6 | This file is part of GNU Emacs. | |
7 | ||
9ec0b715 | 8 | GNU Emacs is free software: you can redistribute it and/or modify |
24f98398 | 9 | it under the terms of the GNU General Public License as published by |
9ec0b715 GM |
10 | the Free Software Foundation, either version 3 of the License, or |
11 | (at your option) any later version. | |
24f98398 JA |
12 | |
13 | GNU Emacs is distributed in the hope that it will be useful, | |
14 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
15 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
16 | GNU General Public License for more details. | |
17 | ||
18 | You should have received a copy of the GNU General Public License | |
9ec0b715 | 19 | along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>. */ |
24f98398 | 20 | |
762b15be EZ |
21 | /* If you think about replacing this with some similar standard C function of |
22 | the printf family (such as vsnprintf), please note that this function | |
23 | supports the following Emacs-specific features: | |
24 | ||
25 | . For %c conversions, it produces a string with the multibyte representation | |
26 | of the (`int') argument, suitable for display in an Emacs buffer. | |
27 | ||
28 | . For %s and %c, when field width is specified (e.g., %25s), it accounts for | |
29 | the diplay width of each character, according to char-width-table. That | |
30 | is, it does not assume that each character takes one column on display. | |
31 | ||
32 | . If the size of the buffer is not enough to produce the formatted string in | |
33 | its entirety, it makes sure that truncation does not chop the last | |
34 | character in the middle of its multibyte sequence, producing an invalid | |
35 | sequence. | |
36 | ||
37 | . It accepts a pointer to the end of the format string, so the format string | |
38 | could include embedded null characters. | |
39 | ||
40 | . It signals an error if the length of the formatted string is about to | |
41 | overflow MOST_POSITIVE_FIXNUM, to avoid producing strings longer than what | |
42 | Emacs can handle. | |
43 | ||
44 | OTOH, this function supports only a small subset of the standard C formatted | |
45 | output facilities. E.g., %u and %ll are not supported, and precision is | |
825cd63c EZ |
46 | ignored %s and %c conversions. (See below for the detailed documentation of |
47 | what is supported.) However, this is okay, as this function is supposed to | |
48 | be called from `error' and similar functions, and thus does not need to | |
49 | support features beyond those in `Fformat', which is used by `error' on the | |
50 | Lisp level. */ | |
51 | ||
52 | /* This function supports the following %-sequences in the `format' | |
53 | argument: | |
54 | ||
55 | %s means print a string argument. | |
56 | %S is silently treated as %s, for loose compatibility with `Fformat'. | |
57 | %d means print a `signed int' argument in decimal. | |
825cd63c EZ |
58 | %o means print an `unsigned int' argument in octal. |
59 | %x means print an `unsigned int' argument in hex. | |
60 | %e means print a `double' argument in exponential notation. | |
61 | %f means print a `double' argument in decimal-point notation. | |
62 | %g means print a `double' argument in exponential notation | |
63 | or in decimal-point notation, whichever uses fewer characters. | |
64 | %c means print a `signed int' argument as a single character. | |
65 | %% means produce a literal % character. | |
66 | ||
94dcfacf EZ |
67 | A %-sequence may contain optional flag, width, and precision specifiers, and |
68 | a length modifier, as follows: | |
825cd63c | 69 | |
94dcfacf | 70 | %<flags><width><precision><length>character |
825cd63c | 71 | |
b71a1728 PE |
72 | where flags is [+ -0], width is [0-9]+, precision is .[0-9]+, and length |
73 | modifier is empty or l or ll. | |
825cd63c EZ |
74 | |
75 | The + flag character inserts a + before any positive number, while a space | |
94dcfacf EZ |
76 | inserts a space before any positive number; these flags only affect %d, %o, |
77 | %x, %e, %f, and %g sequences. The - and 0 flags affect the width specifier, | |
78 | as described below. For signed numerical arguments only, the ` ' (space) | |
79 | flag causes the result to be prefixed with a space character if it does not | |
80 | start with a sign (+ or -). | |
81 | ||
82 | The l (lower-case letter ell) length modifier is a `long' data type | |
83 | modifier: it is supported for %d, %o, and %x conversions of integral | |
54b8e3f7 | 84 | arguments, must immediately precede the conversion specifier, and means that |
94dcfacf | 85 | the respective argument is to be treated as `long int' or `unsigned long |
54b8e3f7 | 86 | int'. Similarly, ll (two letter ells) means to use `long long int' or |
b71a1728 PE |
87 | `unsigned long long int'; this can be used only on hosts that have |
88 | these two types. The empty length modifier means to use `int' or | |
54b8e3f7 PE |
89 | `unsigned int'. EMACS_INT arguments should use the pI macro, which |
90 | expands to whatever length modifier is needed for the target host. | |
825cd63c EZ |
91 | |
92 | The width specifier supplies a lower limit for the length of the printed | |
93 | representation. The padding, if any, normally goes on the left, but it goes | |
94 | on the right if the - flag is present. The padding character is normally a | |
95 | space, but (for numerical arguments only) it is 0 if the 0 flag is present. | |
96 | The - flag takes precedence over the 0 flag. | |
97 | ||
98 | For %e, %f, and %g sequences, the number after the "." in the precision | |
99 | specifier says how many decimal places to show; if zero, the decimal point | |
100 | itself is omitted. For %s and %S, the precision specifier is ignored. */ | |
24f98398 | 101 | |
6d291527 | 102 | #include <config.h> |
24f98398 JA |
103 | #include <stdio.h> |
104 | #include <ctype.h> | |
d7306fe6 | 105 | #include <setjmp.h> |
24f98398 | 106 | |
cf09633a | 107 | #ifdef STDC_HEADERS |
be65c2f4 PE |
108 | #include <float.h> |
109 | #endif | |
110 | ||
48236137 | 111 | #include <unistd.h> |
48236137 | 112 | |
e6c3da20 | 113 | #include <limits.h> |
e6c3da20 | 114 | |
523e9291 RS |
115 | #include "lisp.h" |
116 | ||
a0ca925c KH |
117 | /* Since we use the macro CHAR_HEAD_P, we have to include this, but |
118 | don't have to include others because CHAR_HEAD_P does not contains | |
119 | another macro. */ | |
83be827a | 120 | #include "character.h" |
a0ca925c | 121 | |
283cdbef PE |
122 | #ifndef SIZE_MAX |
123 | # define SIZE_MAX ((size_t) -1) | |
124 | #endif | |
125 | ||
92bc9a36 DN |
126 | #ifndef DBL_MAX_10_EXP |
127 | #define DBL_MAX_10_EXP 308 /* IEEE double */ | |
128 | #endif | |
129 | ||
f4c730d3 RS |
130 | /* Generate output from a format-spec FORMAT, |
131 | terminated at position FORMAT_END. | |
132 | Output goes in BUFFER, which has room for BUFSIZE chars. | |
133 | If the output does not fit, truncate it to fit. | |
825cd63c EZ |
134 | Returns the number of bytes stored into BUFFER, excluding |
135 | the terminating null byte. Output is always null-terminated. | |
1513af9e RS |
136 | String arguments are passed as C strings. |
137 | Integers are passed as C integers. */ | |
f4c730d3 | 138 | |
e6c3da20 EZ |
139 | size_t |
140 | doprnt (char *buffer, register size_t bufsize, const char *format, | |
a8fe7202 | 141 | const char *format_end, va_list ap) |
24f98398 | 142 | { |
a8fe7202 | 143 | const char *fmt = format; /* Pointer into format string */ |
24f98398 | 144 | register char *bufptr = buffer; /* Pointer into output buffer.. */ |
03383aaf | 145 | |
f4c730d3 | 146 | /* Use this for sprintf unless we need something really big. */ |
be65c2f4 | 147 | char tembuf[DBL_MAX_10_EXP + 100]; |
03383aaf | 148 | |
f4c730d3 | 149 | /* Size of sprintf_buffer. */ |
e6c3da20 | 150 | size_t size_allocated = sizeof (tembuf); |
03383aaf | 151 | |
f4c730d3 RS |
152 | /* Buffer to use for sprintf. Either tembuf or same as BIG_BUFFER. */ |
153 | char *sprintf_buffer = tembuf; | |
03383aaf | 154 | |
f4c730d3 | 155 | /* Buffer we have got with malloc. */ |
e6c3da20 | 156 | char *big_buffer = NULL; |
03383aaf | 157 | |
e6c3da20 | 158 | register size_t tem; |
7469ef5d | 159 | char *string; |
03383aaf BF |
160 | char fixed_buffer[20]; /* Default buffer for small formatting. */ |
161 | char *fmtcpy; | |
24f98398 | 162 | int minlen; |
7469ef5d | 163 | char charbuf[MAX_MULTIBYTE_LENGTH + 1]; /* Used for %c. */ |
825cd63c | 164 | USE_SAFE_ALLOCA; |
24f98398 JA |
165 | |
166 | if (format_end == 0) | |
167 | format_end = format + strlen (format); | |
168 | ||
03383aaf BF |
169 | if ((format_end - format + 1) < sizeof (fixed_buffer)) |
170 | fmtcpy = fixed_buffer; | |
171 | else | |
825cd63c | 172 | SAFE_ALLOCA (fmtcpy, char *, format_end - format + 1); |
03383aaf | 173 | |
24f98398 | 174 | bufsize--; |
03383aaf BF |
175 | |
176 | /* Loop until end of format string or buffer full. */ | |
94dcfacf | 177 | while (fmt < format_end && bufsize > 0) |
24f98398 JA |
178 | { |
179 | if (*fmt == '%') /* Check for a '%' character */ | |
180 | { | |
e6c3da20 EZ |
181 | size_t size_bound = 0; |
182 | EMACS_INT width; /* Columns occupied by STRING on display. */ | |
183 | int long_flag = 0; | |
f4c730d3 | 184 | |
24f98398 | 185 | fmt++; |
d427b66a | 186 | /* Copy this one %-spec into fmtcpy. */ |
7469ef5d | 187 | string = fmtcpy; |
24f98398 | 188 | *string++ = '%'; |
94dcfacf | 189 | while (fmt < format_end) |
24f98398 JA |
190 | { |
191 | *string++ = *fmt; | |
be65c2f4 PE |
192 | if ('0' <= *fmt && *fmt <= '9') |
193 | { | |
194 | /* Get an idea of how much space we might need. | |
195 | This might be a field width or a precision; e.g. | |
196 | %1.1000f and %1000.1f both might need 1000+ bytes. | |
197 | Parse the width or precision, checking for overflow. */ | |
e6c3da20 | 198 | size_t n = *fmt - '0'; |
94dcfacf EZ |
199 | while (fmt < format_end |
200 | && '0' <= fmt[1] && fmt[1] <= '9') | |
be65c2f4 | 201 | { |
f76dee0c PE |
202 | /* Avoid int overflow, because many sprintfs seriously |
203 | mess up with widths or precisions greater than | |
204 | INT_MAX. Avoid size_t overflow, since our counters | |
205 | use size_t. This test is slightly conservative, for | |
206 | speed and simplicity. */ | |
207 | if (n >= min (INT_MAX, SIZE_MAX) / 10) | |
be65c2f4 | 208 | error ("Format width or precision too large"); |
26898943 | 209 | n = n * 10 + fmt[1] - '0'; |
be65c2f4 PE |
210 | *string++ = *++fmt; |
211 | } | |
212 | ||
213 | if (size_bound < n) | |
214 | size_bound = n; | |
215 | } | |
01769a73 | 216 | else if (*fmt == '-' || *fmt == ' ' || *fmt == '.' || *fmt == '+') |
be65c2f4 | 217 | ; |
e6c3da20 EZ |
218 | else if (*fmt == 'l') |
219 | { | |
226be1b0 | 220 | long_flag = 1 + (fmt + 1 < format_end && fmt[1] == 'l'); |
54b8e3f7 | 221 | fmt += long_flag; |
94dcfacf | 222 | break; |
e6c3da20 | 223 | } |
be65c2f4 | 224 | else |
24f98398 JA |
225 | break; |
226 | fmt++; | |
227 | } | |
94dcfacf EZ |
228 | if (fmt > format_end) |
229 | fmt = format_end; | |
24f98398 | 230 | *string = 0; |
03383aaf | 231 | |
be65c2f4 PE |
232 | /* Make the size bound large enough to handle floating point formats |
233 | with large numbers. */ | |
e6c3da20 | 234 | if (size_bound > SIZE_MAX - DBL_MAX_10_EXP - 50) |
be65c2f4 | 235 | error ("Format width or precision too large"); |
01769a73 | 236 | size_bound += DBL_MAX_10_EXP + 50; |
6e951728 | 237 | |
f4c730d3 RS |
238 | /* Make sure we have that much. */ |
239 | if (size_bound > size_allocated) | |
240 | { | |
241 | if (big_buffer) | |
94dcfacf EZ |
242 | xfree (big_buffer); |
243 | big_buffer = (char *) xmalloc (size_bound); | |
f4c730d3 RS |
244 | sprintf_buffer = big_buffer; |
245 | size_allocated = size_bound; | |
246 | } | |
24f98398 JA |
247 | minlen = 0; |
248 | switch (*fmt++) | |
249 | { | |
250 | default: | |
94dcfacf | 251 | error ("Invalid format operation %%%s%c", |
226be1b0 | 252 | "ll" + 2 - long_flag, fmt[-1]); |
24f98398 JA |
253 | |
254 | /* case 'b': */ | |
e6c3da20 | 255 | case 'l': |
24f98398 | 256 | case 'd': |
e6c3da20 EZ |
257 | { |
258 | int i; | |
259 | long l; | |
260 | ||
e810457d PE |
261 | if (1 < long_flag) |
262 | { | |
263 | #ifdef HAVE_LONG_LONG_INT | |
264 | long long ll = va_arg (ap, long long); | |
265 | sprintf (sprintf_buffer, fmtcpy, ll); | |
266 | #else | |
267 | abort (); | |
268 | #endif | |
269 | } | |
270 | else if (long_flag) | |
e6c3da20 EZ |
271 | { |
272 | l = va_arg(ap, long); | |
273 | sprintf (sprintf_buffer, fmtcpy, l); | |
274 | } | |
275 | else | |
276 | { | |
277 | i = va_arg(ap, int); | |
278 | sprintf (sprintf_buffer, fmtcpy, i); | |
279 | } | |
280 | /* Now copy into final output, truncating as necessary. */ | |
281 | string = sprintf_buffer; | |
282 | goto doit; | |
283 | } | |
284 | ||
24f98398 JA |
285 | case 'o': |
286 | case 'x': | |
e6c3da20 EZ |
287 | { |
288 | unsigned u; | |
289 | unsigned long ul; | |
290 | ||
e810457d PE |
291 | if (1 < long_flag) |
292 | { | |
293 | #ifdef HAVE_UNSIGNED_LONG_LONG_INT | |
294 | unsigned long long ull = va_arg (ap, unsigned long long); | |
295 | sprintf (sprintf_buffer, fmtcpy, ull); | |
296 | #else | |
297 | abort (); | |
298 | #endif | |
299 | } | |
300 | else if (long_flag) | |
e6c3da20 EZ |
301 | { |
302 | ul = va_arg(ap, unsigned long); | |
303 | sprintf (sprintf_buffer, fmtcpy, ul); | |
304 | } | |
305 | else | |
306 | { | |
307 | u = va_arg(ap, unsigned); | |
308 | sprintf (sprintf_buffer, fmtcpy, u); | |
309 | } | |
310 | /* Now copy into final output, truncating as necessary. */ | |
311 | string = sprintf_buffer; | |
312 | goto doit; | |
313 | } | |
24f98398 | 314 | |
f4c730d3 RS |
315 | case 'f': |
316 | case 'e': | |
317 | case 'g': | |
318 | { | |
6a8033e1 KR |
319 | double d = va_arg(ap, double); |
320 | sprintf (sprintf_buffer, fmtcpy, d); | |
e6c3da20 | 321 | /* Now copy into final output, truncating as necessary. */ |
7469ef5d | 322 | string = sprintf_buffer; |
f4c730d3 RS |
323 | goto doit; |
324 | } | |
325 | ||
24f98398 JA |
326 | case 'S': |
327 | string[-1] = 's'; | |
328 | case 's': | |
24f98398 JA |
329 | if (fmtcpy[1] != 's') |
330 | minlen = atoi (&fmtcpy[1]); | |
7469ef5d | 331 | string = va_arg (ap, char *); |
e267324c | 332 | tem = strlen (string); |
e6c3da20 EZ |
333 | if (tem > MOST_POSITIVE_FIXNUM) |
334 | error ("String for %%s or %%S format is too long"); | |
a0ca925c | 335 | width = strwidth (string, tem); |
1513af9e RS |
336 | goto doit1; |
337 | ||
24f98398 JA |
338 | /* Copy string into final output, truncating if no room. */ |
339 | doit: | |
a0ca925c | 340 | /* Coming here means STRING contains ASCII only. */ |
e6c3da20 EZ |
341 | tem = strlen (string); |
342 | if (tem > MOST_POSITIVE_FIXNUM) | |
343 | error ("Format width or precision too large"); | |
344 | width = tem; | |
35a65fce | 345 | doit1: |
a0ca925c KH |
346 | /* We have already calculated: |
347 | TEM -- length of STRING, | |
348 | WIDTH -- columns occupied by STRING when displayed, and | |
349 | MINLEN -- minimum columns of the output. */ | |
24f98398 JA |
350 | if (minlen > 0) |
351 | { | |
a0ca925c | 352 | while (minlen > width && bufsize > 0) |
24f98398 JA |
353 | { |
354 | *bufptr++ = ' '; | |
355 | bufsize--; | |
356 | minlen--; | |
357 | } | |
358 | minlen = 0; | |
359 | } | |
360 | if (tem > bufsize) | |
a0ca925c KH |
361 | { |
362 | /* Truncate the string at character boundary. */ | |
363 | tem = bufsize; | |
a50545d9 | 364 | while (!CHAR_HEAD_P (string[tem - 1])) tem--; |
72af86bd | 365 | memcpy (bufptr, string, tem); |
a0ca925c KH |
366 | /* We must calculate WIDTH again. */ |
367 | width = strwidth (bufptr, tem); | |
368 | } | |
369 | else | |
72af86bd | 370 | memcpy (bufptr, string, tem); |
24f98398 JA |
371 | bufptr += tem; |
372 | bufsize -= tem; | |
373 | if (minlen < 0) | |
374 | { | |
a0ca925c | 375 | while (minlen < - width && bufsize > 0) |
24f98398 JA |
376 | { |
377 | *bufptr++ = ' '; | |
378 | bufsize--; | |
379 | minlen++; | |
380 | } | |
381 | minlen = 0; | |
382 | } | |
383 | continue; | |
384 | ||
385 | case 'c': | |
6a8033e1 | 386 | { |
e6c3da20 EZ |
387 | int chr = va_arg(ap, int); |
388 | tem = CHAR_STRING (chr, (unsigned char *) charbuf); | |
6a8033e1 KR |
389 | string = charbuf; |
390 | string[tem] = 0; | |
391 | width = strwidth (string, tem); | |
392 | if (fmtcpy[1] != 'c') | |
393 | minlen = atoi (&fmtcpy[1]); | |
394 | goto doit1; | |
395 | } | |
24f98398 JA |
396 | |
397 | case '%': | |
398 | fmt--; /* Drop thru and this % will be treated as normal */ | |
399 | } | |
400 | } | |
a0ca925c KH |
401 | |
402 | { | |
403 | /* Just some character; Copy it if the whole multi-byte form | |
404 | fit in the buffer. */ | |
405 | char *save_bufptr = bufptr; | |
406 | ||
407 | do { *bufptr++ = *fmt++; } | |
94dcfacf | 408 | while (fmt < format_end && --bufsize > 0 && !CHAR_HEAD_P (*fmt)); |
a50545d9 | 409 | if (!CHAR_HEAD_P (*fmt)) |
a0ca925c | 410 | { |
d178f871 EZ |
411 | /* Truncate, but return value that will signal to caller |
412 | that the buffer was too small. */ | |
413 | *save_bufptr = 0; | |
a0ca925c KH |
414 | break; |
415 | } | |
416 | } | |
24f98398 JA |
417 | }; |
418 | ||
f4c730d3 | 419 | /* If we had to malloc something, free it. */ |
70fdbb46 | 420 | xfree (big_buffer); |
f4c730d3 | 421 | |
e6c3da20 | 422 | *bufptr = 0; /* Make sure our string ends with a '\0' */ |
825cd63c EZ |
423 | |
424 | SAFE_FREE (); | |
24f98398 JA |
425 | return bufptr - buffer; |
426 | } |