Commit | Line | Data |
---|---|---|
059ec3d9 PH |
1 | /************************************************* |
2 | * Exim - an Internet mail transport agent * | |
3 | *************************************************/ | |
4 | ||
f9ba5e22 | 5 | /* Copyright (c) University of Cambridge 1995 - 2018 */ |
059ec3d9 PH |
6 | /* See the file NOTICE for conditions of use and distribution. */ |
7 | ||
8 | /* Miscellaneous string-handling functions. Some are not required for | |
9 | utilities and tests, and are cut out by the COMPILE_UTILITY macro. */ | |
10 | ||
11 | ||
12 | #include "exim.h" | |
13 | ||
14 | ||
15 | #ifndef COMPILE_UTILITY | |
16 | /************************************************* | |
17 | * Test for IP address * | |
18 | *************************************************/ | |
19 | ||
20 | /* This used just to be a regular expression, but with IPv6 things are a bit | |
21 | more complicated. If the address contains a colon, it is assumed to be a v6 | |
22 | address (assuming HAVE_IPV6 is set). If a mask is permitted and one is present, | |
23 | and maskptr is not NULL, its offset is placed there. | |
24 | ||
25 | Arguments: | |
26 | s a string | |
27 | maskptr NULL if no mask is permitted to follow | |
28 | otherwise, points to an int where the offset of '/' is placed | |
1688f43b | 29 | if there is no / followed by trailing digits, *maskptr is set 0 |
059ec3d9 PH |
30 | |
31 | Returns: 0 if the string is not a textual representation of an IP address | |
32 | 4 if it is an IPv4 address | |
33 | 6 if it is an IPv6 address | |
34 | */ | |
35 | ||
36 | int | |
b1f8e4f8 | 37 | string_is_ip_address(const uschar *s, int *maskptr) |
059ec3d9 PH |
38 | { |
39 | int i; | |
40 | int yield = 4; | |
41 | ||
42 | /* If an optional mask is permitted, check for it. If found, pass back the | |
43 | offset. */ | |
44 | ||
8d909960 | 45 | if (maskptr) |
059ec3d9 | 46 | { |
b1f8e4f8 | 47 | const uschar *ss = s + Ustrlen(s); |
059ec3d9 PH |
48 | *maskptr = 0; |
49 | if (s != ss && isdigit(*(--ss))) | |
50 | { | |
51 | while (ss > s && isdigit(ss[-1])) ss--; | |
52 | if (ss > s && *(--ss) == '/') *maskptr = ss - s; | |
53 | } | |
54 | } | |
55 | ||
56 | /* A colon anywhere in the string => IPv6 address */ | |
57 | ||
58 | if (Ustrchr(s, ':') != NULL) | |
59 | { | |
60 | BOOL had_double_colon = FALSE; | |
61 | BOOL v4end = FALSE; | |
62 | int count = 0; | |
63 | ||
64 | yield = 6; | |
65 | ||
66 | /* An IPv6 address must start with hex digit or double colon. A single | |
67 | colon is invalid. */ | |
68 | ||
69 | if (*s == ':' && *(++s) != ':') return 0; | |
70 | ||
71 | /* Now read up to 8 components consisting of up to 4 hex digits each. There | |
72 | may be one and only one appearance of double colon, which implies any number | |
73 | of binary zero bits. The number of preceding components is held in count. */ | |
74 | ||
75 | for (count = 0; count < 8; count++) | |
76 | { | |
77 | /* If the end of the string is reached before reading 8 components, the | |
78 | address is valid provided a double colon has been read. This also applies | |
79 | if we hit the / that introduces a mask or the % that introduces the | |
80 | interface specifier (scope id) of a link-local address. */ | |
81 | ||
8d909960 | 82 | if (*s == 0 || *s == '%' || *s == '/') return had_double_colon ? yield : 0; |
059ec3d9 PH |
83 | |
84 | /* If a component starts with an additional colon, we have hit a double | |
85 | colon. This is permitted to appear once only, and counts as at least | |
86 | one component. The final component may be of this form. */ | |
87 | ||
88 | if (*s == ':') | |
89 | { | |
90 | if (had_double_colon) return 0; | |
91 | had_double_colon = TRUE; | |
92 | s++; | |
93 | continue; | |
94 | } | |
95 | ||
96 | /* If the remainder of the string contains a dot but no colons, we | |
97 | can expect a trailing IPv4 address. This is valid if either there has | |
98 | been no double-colon and this is the 7th component (with the IPv4 address | |
99 | being the 7th & 8th components), OR if there has been a double-colon | |
100 | and fewer than 6 components. */ | |
101 | ||
102 | if (Ustrchr(s, ':') == NULL && Ustrchr(s, '.') != NULL) | |
103 | { | |
104 | if ((!had_double_colon && count != 6) || | |
105 | (had_double_colon && count > 6)) return 0; | |
106 | v4end = TRUE; | |
107 | yield = 6; | |
108 | break; | |
109 | } | |
110 | ||
111 | /* Check for at least one and not more than 4 hex digits for this | |
112 | component. */ | |
113 | ||
114 | if (!isxdigit(*s++)) return 0; | |
115 | if (isxdigit(*s) && isxdigit(*(++s)) && isxdigit(*(++s))) s++; | |
116 | ||
117 | /* If the component is terminated by colon and there is more to | |
118 | follow, skip over the colon. If there is no more to follow the address is | |
119 | invalid. */ | |
120 | ||
121 | if (*s == ':' && *(++s) == 0) return 0; | |
122 | } | |
123 | ||
124 | /* If about to handle a trailing IPv4 address, drop through. Otherwise | |
125 | all is well if we are at the end of the string or at the mask or at a percent | |
126 | sign, which introduces the interface specifier (scope id) of a link local | |
127 | address. */ | |
128 | ||
1688f43b PH |
129 | if (!v4end) |
130 | return (*s == 0 || *s == '%' || | |
131 | (*s == '/' && maskptr != NULL && *maskptr != 0))? yield : 0; | |
059ec3d9 PH |
132 | } |
133 | ||
134 | /* Test for IPv4 address, which may be the tail-end of an IPv6 address. */ | |
135 | ||
136 | for (i = 0; i < 4; i++) | |
137 | { | |
8d909960 JH |
138 | long n; |
139 | uschar * end; | |
140 | ||
059ec3d9 | 141 | if (i != 0 && *s++ != '.') return 0; |
8d909960 JH |
142 | n = strtol(CCS s, CSS &end, 10); |
143 | if (n > 255 || n < 0 || end <= s || end > s+3) return 0; | |
144 | s = end; | |
059ec3d9 PH |
145 | } |
146 | ||
8d909960 | 147 | return !*s || (*s == '/' && maskptr && *maskptr != 0) ? yield : 0; |
059ec3d9 PH |
148 | } |
149 | #endif /* COMPILE_UTILITY */ | |
150 | ||
151 | ||
152 | /************************************************* | |
153 | * Format message size * | |
154 | *************************************************/ | |
155 | ||
156 | /* Convert a message size in bytes to printing form, rounding | |
157 | according to the magnitude of the number. A value of zero causes | |
158 | a string of spaces to be returned. | |
159 | ||
160 | Arguments: | |
161 | size the message size in bytes | |
162 | buffer where to put the answer | |
163 | ||
164 | Returns: pointer to the buffer | |
165 | a string of exactly 5 characters is normally returned | |
166 | */ | |
167 | ||
168 | uschar * | |
169 | string_format_size(int size, uschar *buffer) | |
170 | { | |
45500060 | 171 | if (size == 0) Ustrcpy(buffer, " "); |
059ec3d9 PH |
172 | else if (size < 1024) sprintf(CS buffer, "%5d", size); |
173 | else if (size < 10*1024) | |
174 | sprintf(CS buffer, "%4.1fK", (double)size / 1024.0); | |
175 | else if (size < 1024*1024) | |
176 | sprintf(CS buffer, "%4dK", (size + 512)/1024); | |
177 | else if (size < 10*1024*1024) | |
178 | sprintf(CS buffer, "%4.1fM", (double)size / (1024.0 * 1024.0)); | |
179 | else | |
180 | sprintf(CS buffer, "%4dM", (size + 512 * 1024)/(1024*1024)); | |
181 | return buffer; | |
182 | } | |
183 | ||
184 | ||
185 | ||
186 | #ifndef COMPILE_UTILITY | |
187 | /************************************************* | |
188 | * Convert a number to base 62 format * | |
189 | *************************************************/ | |
190 | ||
191 | /* Convert a long integer into an ASCII base 62 string. For Cygwin the value of | |
192 | BASE_62 is actually 36. Always return exactly 6 characters plus zero, in a | |
193 | static area. | |
194 | ||
195 | Argument: a long integer | |
196 | Returns: pointer to base 62 string | |
197 | */ | |
198 | ||
199 | uschar * | |
200 | string_base62(unsigned long int value) | |
201 | { | |
202 | static uschar yield[7]; | |
203 | uschar *p = yield + sizeof(yield) - 1; | |
204 | *p = 0; | |
205 | while (p > yield) | |
206 | { | |
207 | *(--p) = base62_chars[value % BASE_62]; | |
208 | value /= BASE_62; | |
209 | } | |
210 | return yield; | |
211 | } | |
212 | #endif /* COMPILE_UTILITY */ | |
213 | ||
214 | ||
215 | ||
059ec3d9 PH |
216 | /************************************************* |
217 | * Interpret escape sequence * | |
218 | *************************************************/ | |
219 | ||
220 | /* This function is called from several places where escape sequences are to be | |
221 | interpreted in strings. | |
222 | ||
223 | Arguments: | |
224 | pp points a pointer to the initiating "\" in the string; | |
225 | the pointer gets updated to point to the final character | |
226 | Returns: the value of the character escape | |
227 | */ | |
228 | ||
229 | int | |
55414b25 | 230 | string_interpret_escape(const uschar **pp) |
059ec3d9 | 231 | { |
3fb3c68d JH |
232 | #ifdef COMPILE_UTILITY |
233 | const uschar *hex_digits= CUS"0123456789abcdef"; | |
234 | #endif | |
059ec3d9 | 235 | int ch; |
55414b25 | 236 | const uschar *p = *pp; |
059ec3d9 PH |
237 | ch = *(++p); |
238 | if (isdigit(ch) && ch != '8' && ch != '9') | |
239 | { | |
240 | ch -= '0'; | |
241 | if (isdigit(p[1]) && p[1] != '8' && p[1] != '9') | |
242 | { | |
243 | ch = ch * 8 + *(++p) - '0'; | |
244 | if (isdigit(p[1]) && p[1] != '8' && p[1] != '9') | |
245 | ch = ch * 8 + *(++p) - '0'; | |
246 | } | |
247 | } | |
248 | else switch(ch) | |
249 | { | |
c7396ac5 PP |
250 | case 'b': ch = '\b'; break; |
251 | case 'f': ch = '\f'; break; | |
059ec3d9 PH |
252 | case 'n': ch = '\n'; break; |
253 | case 'r': ch = '\r'; break; | |
254 | case 't': ch = '\t'; break; | |
c7396ac5 | 255 | case 'v': ch = '\v'; break; |
059ec3d9 PH |
256 | case 'x': |
257 | ch = 0; | |
258 | if (isxdigit(p[1])) | |
259 | { | |
260 | ch = ch * 16 + | |
261 | Ustrchr(hex_digits, tolower(*(++p))) - hex_digits; | |
262 | if (isxdigit(p[1])) ch = ch * 16 + | |
263 | Ustrchr(hex_digits, tolower(*(++p))) - hex_digits; | |
264 | } | |
265 | break; | |
266 | } | |
267 | *pp = p; | |
268 | return ch; | |
269 | } | |
059ec3d9 PH |
270 | |
271 | ||
272 | ||
273 | #ifndef COMPILE_UTILITY | |
274 | /************************************************* | |
275 | * Ensure string is printable * | |
276 | *************************************************/ | |
277 | ||
278 | /* This function is called for critical strings. It checks for any | |
279 | non-printing characters, and if any are found, it makes a new copy | |
280 | of the string with suitable escape sequences. It is most often called by the | |
281 | macro string_printing(), which sets allow_tab TRUE. | |
282 | ||
283 | Arguments: | |
284 | s the input string | |
285 | allow_tab TRUE to allow tab as a printing character | |
286 | ||
287 | Returns: string with non-printers encoded as printing sequences | |
288 | */ | |
289 | ||
55414b25 JH |
290 | const uschar * |
291 | string_printing2(const uschar *s, BOOL allow_tab) | |
059ec3d9 PH |
292 | { |
293 | int nonprintcount = 0; | |
294 | int length = 0; | |
55414b25 | 295 | const uschar *t = s; |
059ec3d9 PH |
296 | uschar *ss, *tt; |
297 | ||
298 | while (*t != 0) | |
299 | { | |
300 | int c = *t++; | |
301 | if (!mac_isprint(c) || (!allow_tab && c == '\t')) nonprintcount++; | |
302 | length++; | |
303 | } | |
304 | ||
305 | if (nonprintcount == 0) return s; | |
306 | ||
307 | /* Get a new block of store guaranteed big enough to hold the | |
308 | expanded string. */ | |
309 | ||
36719342 | 310 | ss = store_get(length + nonprintcount * 3 + 1); |
059ec3d9 | 311 | |
4c04137d | 312 | /* Copy everything, escaping non printers. */ |
059ec3d9 PH |
313 | |
314 | t = s; | |
315 | tt = ss; | |
316 | ||
317 | while (*t != 0) | |
318 | { | |
319 | int c = *t; | |
320 | if (mac_isprint(c) && (allow_tab || c != '\t')) *tt++ = *t++; else | |
321 | { | |
322 | *tt++ = '\\'; | |
323 | switch (*t) | |
324 | { | |
325 | case '\n': *tt++ = 'n'; break; | |
326 | case '\r': *tt++ = 'r'; break; | |
327 | case '\b': *tt++ = 'b'; break; | |
328 | case '\v': *tt++ = 'v'; break; | |
329 | case '\f': *tt++ = 'f'; break; | |
330 | case '\t': *tt++ = 't'; break; | |
331 | default: sprintf(CS tt, "%03o", *t); tt += 3; break; | |
332 | } | |
333 | t++; | |
334 | } | |
335 | } | |
336 | *tt = 0; | |
c7396ac5 PP |
337 | return ss; |
338 | } | |
79fe97d8 PP |
339 | #endif /* COMPILE_UTILITY */ |
340 | ||
c7396ac5 PP |
341 | /************************************************* |
342 | * Undo printing escapes in string * | |
343 | *************************************************/ | |
344 | ||
345 | /* This function is the reverse of string_printing2. It searches for | |
346 | backslash characters and if any are found, it makes a new copy of the | |
347 | string with escape sequences parsed. Otherwise it returns the original | |
348 | string. | |
349 | ||
350 | Arguments: | |
351 | s the input string | |
352 | ||
353 | Returns: string with printing escapes parsed back | |
354 | */ | |
355 | ||
356 | uschar * | |
357 | string_unprinting(uschar *s) | |
358 | { | |
359 | uschar *p, *q, *r, *ss; | |
360 | int len, off; | |
361 | ||
362 | p = Ustrchr(s, '\\'); | |
363 | if (!p) return s; | |
364 | ||
365 | len = Ustrlen(s) + 1; | |
366 | ss = store_get(len); | |
367 | ||
368 | q = ss; | |
369 | off = p - s; | |
370 | if (off) | |
371 | { | |
372 | memcpy(q, s, off); | |
373 | q += off; | |
374 | } | |
375 | ||
376 | while (*p) | |
377 | { | |
378 | if (*p == '\\') | |
379 | { | |
55414b25 | 380 | *q++ = string_interpret_escape((const uschar **)&p); |
823ad74f | 381 | p++; |
c7396ac5 PP |
382 | } |
383 | else | |
384 | { | |
385 | r = Ustrchr(p, '\\'); | |
386 | if (!r) | |
387 | { | |
388 | off = Ustrlen(p); | |
389 | memcpy(q, p, off); | |
390 | p += off; | |
391 | q += off; | |
392 | break; | |
393 | } | |
394 | else | |
395 | { | |
396 | off = r - p; | |
397 | memcpy(q, p, off); | |
398 | q += off; | |
399 | p = r; | |
400 | } | |
401 | } | |
402 | } | |
403 | *q = '\0'; | |
404 | ||
059ec3d9 PH |
405 | return ss; |
406 | } | |
059ec3d9 PH |
407 | |
408 | ||
409 | ||
410 | ||
411 | /************************************************* | |
412 | * Copy and save string * | |
413 | *************************************************/ | |
414 | ||
415 | /* This function assumes that memcpy() is faster than strcpy(). | |
416 | ||
417 | Argument: string to copy | |
418 | Returns: copy of string in new store | |
419 | */ | |
420 | ||
421 | uschar * | |
3f0945ff | 422 | string_copy(const uschar *s) |
059ec3d9 PH |
423 | { |
424 | int len = Ustrlen(s) + 1; | |
425 | uschar *ss = store_get(len); | |
426 | memcpy(ss, s, len); | |
427 | return ss; | |
428 | } | |
429 | ||
430 | ||
431 | ||
432 | /************************************************* | |
433 | * Copy and save string in malloc'd store * | |
434 | *************************************************/ | |
435 | ||
436 | /* This function assumes that memcpy() is faster than strcpy(). | |
437 | ||
438 | Argument: string to copy | |
439 | Returns: copy of string in new store | |
440 | */ | |
441 | ||
442 | uschar * | |
55414b25 | 443 | string_copy_malloc(const uschar *s) |
059ec3d9 PH |
444 | { |
445 | int len = Ustrlen(s) + 1; | |
446 | uschar *ss = store_malloc(len); | |
447 | memcpy(ss, s, len); | |
448 | return ss; | |
449 | } | |
450 | ||
451 | ||
452 | ||
453 | /************************************************* | |
454 | * Copy, lowercase and save string * | |
455 | *************************************************/ | |
456 | ||
457 | /* | |
458 | Argument: string to copy | |
459 | Returns: copy of string in new store, with letters lowercased | |
460 | */ | |
461 | ||
462 | uschar * | |
1dc92d5a | 463 | string_copylc(const uschar *s) |
059ec3d9 PH |
464 | { |
465 | uschar *ss = store_get(Ustrlen(s) + 1); | |
466 | uschar *p = ss; | |
467 | while (*s != 0) *p++ = tolower(*s++); | |
468 | *p = 0; | |
469 | return ss; | |
470 | } | |
471 | ||
472 | ||
473 | ||
474 | /************************************************* | |
475 | * Copy and save string, given length * | |
476 | *************************************************/ | |
477 | ||
478 | /* It is assumed the data contains no zeros. A zero is added | |
479 | onto the end. | |
480 | ||
481 | Arguments: | |
482 | s string to copy | |
483 | n number of characters | |
484 | ||
485 | Returns: copy of string in new store | |
486 | */ | |
487 | ||
488 | uschar * | |
1dc92d5a | 489 | string_copyn(const uschar *s, int n) |
059ec3d9 PH |
490 | { |
491 | uschar *ss = store_get(n + 1); | |
492 | Ustrncpy(ss, s, n); | |
493 | ss[n] = 0; | |
494 | return ss; | |
495 | } | |
496 | ||
497 | ||
498 | /************************************************* | |
499 | * Copy, lowercase, and save string, given length * | |
500 | *************************************************/ | |
501 | ||
502 | /* It is assumed the data contains no zeros. A zero is added | |
503 | onto the end. | |
504 | ||
505 | Arguments: | |
506 | s string to copy | |
507 | n number of characters | |
508 | ||
509 | Returns: copy of string in new store, with letters lowercased | |
510 | */ | |
511 | ||
512 | uschar * | |
513 | string_copynlc(uschar *s, int n) | |
514 | { | |
515 | uschar *ss = store_get(n + 1); | |
516 | uschar *p = ss; | |
517 | while (n-- > 0) *p++ = tolower(*s++); | |
518 | *p = 0; | |
519 | return ss; | |
520 | } | |
521 | ||
522 | ||
523 | ||
e28326d8 PH |
524 | /************************************************* |
525 | * Copy string if long, inserting newlines * | |
526 | *************************************************/ | |
527 | ||
528 | /* If the given string is longer than 75 characters, it is copied, and within | |
529 | the copy, certain space characters are converted into newlines. | |
530 | ||
531 | Argument: pointer to the string | |
532 | Returns: pointer to the possibly altered string | |
533 | */ | |
534 | ||
535 | uschar * | |
536 | string_split_message(uschar *msg) | |
537 | { | |
538 | uschar *s, *ss; | |
539 | ||
540 | if (msg == NULL || Ustrlen(msg) <= 75) return msg; | |
541 | s = ss = msg = string_copy(msg); | |
542 | ||
543 | for (;;) | |
544 | { | |
545 | int i = 0; | |
546 | while (i < 75 && *ss != 0 && *ss != '\n') ss++, i++; | |
547 | if (*ss == 0) break; | |
548 | if (*ss == '\n') | |
549 | s = ++ss; | |
550 | else | |
551 | { | |
552 | uschar *t = ss + 1; | |
553 | uschar *tt = NULL; | |
554 | while (--t > s + 35) | |
555 | { | |
556 | if (*t == ' ') | |
557 | { | |
558 | if (t[-1] == ':') { tt = t; break; } | |
559 | if (tt == NULL) tt = t; | |
560 | } | |
561 | } | |
562 | ||
563 | if (tt == NULL) /* Can't split behind - try ahead */ | |
564 | { | |
565 | t = ss + 1; | |
566 | while (*t != 0) | |
567 | { | |
568 | if (*t == ' ' || *t == '\n') | |
569 | { tt = t; break; } | |
570 | t++; | |
571 | } | |
572 | } | |
573 | ||
574 | if (tt == NULL) break; /* Can't find anywhere to split */ | |
575 | *tt = '\n'; | |
576 | s = ss = tt+1; | |
577 | } | |
578 | } | |
579 | ||
580 | return msg; | |
581 | } | |
582 | ||
583 | ||
584 | ||
059ec3d9 PH |
585 | /************************************************* |
586 | * Copy returned DNS domain name, de-escaping * | |
587 | *************************************************/ | |
588 | ||
589 | /* If a domain name contains top-bit characters, some resolvers return | |
590 | the fully qualified name with those characters turned into escapes. The | |
591 | convention is a backslash followed by _decimal_ digits. We convert these | |
592 | back into the original binary values. This will be relevant when | |
593 | allow_utf8_domains is set true and UTF-8 characters are used in domain | |
594 | names. Backslash can also be used to escape other characters, though we | |
595 | shouldn't come across them in domain names. | |
596 | ||
597 | Argument: the domain name string | |
598 | Returns: copy of string in new store, de-escaped | |
599 | */ | |
600 | ||
601 | uschar * | |
602 | string_copy_dnsdomain(uschar *s) | |
603 | { | |
604 | uschar *yield; | |
605 | uschar *ss = yield = store_get(Ustrlen(s) + 1); | |
606 | ||
607 | while (*s != 0) | |
608 | { | |
609 | if (*s != '\\') | |
610 | { | |
611 | *ss++ = *s++; | |
612 | } | |
613 | else if (isdigit(s[1])) | |
614 | { | |
615 | *ss++ = (s[1] - '0')*100 + (s[2] - '0')*10 + s[3] - '0'; | |
616 | s += 4; | |
617 | } | |
618 | else if (*(++s) != 0) | |
619 | { | |
620 | *ss++ = *s++; | |
621 | } | |
622 | } | |
623 | ||
624 | *ss = 0; | |
625 | return yield; | |
626 | } | |
627 | ||
628 | ||
629 | #ifndef COMPILE_UTILITY | |
630 | /************************************************* | |
631 | * Copy space-terminated or quoted string * | |
632 | *************************************************/ | |
633 | ||
634 | /* This function copies from a string until its end, or until whitespace is | |
635 | encountered, unless the string begins with a double quote, in which case the | |
636 | terminating quote is sought, and escaping within the string is done. The length | |
637 | of a de-quoted string can be no longer than the original, since escaping always | |
638 | turns n characters into 1 character. | |
639 | ||
640 | Argument: pointer to the pointer to the first character, which gets updated | |
641 | Returns: the new string | |
642 | */ | |
643 | ||
644 | uschar * | |
55414b25 | 645 | string_dequote(const uschar **sptr) |
059ec3d9 | 646 | { |
55414b25 | 647 | const uschar *s = *sptr; |
059ec3d9 PH |
648 | uschar *t, *yield; |
649 | ||
650 | /* First find the end of the string */ | |
651 | ||
652 | if (*s != '\"') | |
653 | { | |
654 | while (*s != 0 && !isspace(*s)) s++; | |
655 | } | |
656 | else | |
657 | { | |
658 | s++; | |
659 | while (*s != 0 && *s != '\"') | |
660 | { | |
661 | if (*s == '\\') (void)string_interpret_escape(&s); | |
662 | s++; | |
663 | } | |
664 | if (*s != 0) s++; | |
665 | } | |
666 | ||
667 | /* Get enough store to copy into */ | |
668 | ||
669 | t = yield = store_get(s - *sptr + 1); | |
670 | s = *sptr; | |
671 | ||
672 | /* Do the copy */ | |
673 | ||
674 | if (*s != '\"') | |
675 | { | |
676 | while (*s != 0 && !isspace(*s)) *t++ = *s++; | |
677 | } | |
678 | else | |
679 | { | |
680 | s++; | |
681 | while (*s != 0 && *s != '\"') | |
682 | { | |
683 | if (*s == '\\') *t++ = string_interpret_escape(&s); | |
684 | else *t++ = *s; | |
685 | s++; | |
686 | } | |
687 | if (*s != 0) s++; | |
688 | } | |
689 | ||
690 | /* Update the pointer and return the terminated copy */ | |
691 | ||
692 | *sptr = s; | |
693 | *t = 0; | |
694 | return yield; | |
695 | } | |
696 | #endif /* COMPILE_UTILITY */ | |
697 | ||
698 | ||
699 | ||
700 | /************************************************* | |
701 | * Format a string and save it * | |
702 | *************************************************/ | |
703 | ||
704 | /* The formatting is done by string_format, which checks the length of | |
705 | everything. | |
706 | ||
707 | Arguments: | |
708 | format a printf() format - deliberately char * rather than uschar * | |
709 | because it will most usually be a literal string | |
710 | ... arguments for format | |
711 | ||
712 | Returns: pointer to fresh piece of store containing sprintf'ed string | |
713 | */ | |
714 | ||
715 | uschar * | |
1ba28e2b | 716 | string_sprintf(const char *format, ...) |
059ec3d9 PH |
717 | { |
718 | va_list ap; | |
719 | uschar buffer[STRING_SPRINTF_BUFFER_SIZE]; | |
720 | va_start(ap, format); | |
721 | if (!string_vformat(buffer, sizeof(buffer), format, ap)) | |
722 | log_write(0, LOG_MAIN|LOG_PANIC_DIE, | |
7e09992e JH |
723 | "string_sprintf expansion was longer than " SIZE_T_FMT |
724 | "; format string was (%s)\nexpansion started '%.32s'", | |
725 | sizeof(buffer), format, buffer); | |
059ec3d9 PH |
726 | va_end(ap); |
727 | return string_copy(buffer); | |
728 | } | |
729 | ||
730 | ||
731 | ||
732 | /************************************************* | |
733 | * Case-independent strncmp() function * | |
734 | *************************************************/ | |
735 | ||
736 | /* | |
737 | Arguments: | |
738 | s first string | |
739 | t second string | |
740 | n number of characters to compare | |
741 | ||
742 | Returns: < 0, = 0, or > 0, according to the comparison | |
743 | */ | |
744 | ||
745 | int | |
1ba28e2b | 746 | strncmpic(const uschar *s, const uschar *t, int n) |
059ec3d9 PH |
747 | { |
748 | while (n--) | |
749 | { | |
750 | int c = tolower(*s++) - tolower(*t++); | |
751 | if (c) return c; | |
752 | } | |
753 | return 0; | |
754 | } | |
755 | ||
756 | ||
757 | /************************************************* | |
758 | * Case-independent strcmp() function * | |
759 | *************************************************/ | |
760 | ||
761 | /* | |
762 | Arguments: | |
763 | s first string | |
764 | t second string | |
765 | ||
766 | Returns: < 0, = 0, or > 0, according to the comparison | |
767 | */ | |
768 | ||
769 | int | |
1ba28e2b | 770 | strcmpic(const uschar *s, const uschar *t) |
059ec3d9 PH |
771 | { |
772 | while (*s != 0) | |
773 | { | |
774 | int c = tolower(*s++) - tolower(*t++); | |
775 | if (c != 0) return c; | |
776 | } | |
777 | return *t; | |
778 | } | |
779 | ||
780 | ||
781 | /************************************************* | |
782 | * Case-independent strstr() function * | |
783 | *************************************************/ | |
784 | ||
785 | /* The third argument specifies whether whitespace is required | |
786 | to follow the matched string. | |
787 | ||
788 | Arguments: | |
789 | s string to search | |
790 | t substring to search for | |
791 | space_follows if TRUE, match only if whitespace follows | |
792 | ||
793 | Returns: pointer to substring in string, or NULL if not found | |
794 | */ | |
795 | ||
796 | uschar * | |
797 | strstric(uschar *s, uschar *t, BOOL space_follows) | |
798 | { | |
799 | uschar *p = t; | |
800 | uschar *yield = NULL; | |
801 | int cl = tolower(*p); | |
802 | int cu = toupper(*p); | |
803 | ||
804 | while (*s) | |
805 | { | |
806 | if (*s == cl || *s == cu) | |
807 | { | |
808 | if (yield == NULL) yield = s; | |
809 | if (*(++p) == 0) | |
810 | { | |
811 | if (!space_follows || s[1] == ' ' || s[1] == '\n' ) return yield; | |
812 | yield = NULL; | |
813 | p = t; | |
814 | } | |
815 | cl = tolower(*p); | |
816 | cu = toupper(*p); | |
817 | s++; | |
818 | } | |
819 | else if (yield != NULL) | |
820 | { | |
821 | yield = NULL; | |
822 | p = t; | |
823 | cl = tolower(*p); | |
824 | cu = toupper(*p); | |
825 | } | |
826 | else s++; | |
827 | } | |
828 | return NULL; | |
829 | } | |
830 | ||
831 | ||
832 | ||
833 | #ifndef COMPILE_UTILITY | |
834 | /************************************************* | |
835 | * Get next string from separated list * | |
836 | *************************************************/ | |
837 | ||
838 | /* Leading and trailing space is removed from each item. The separator in the | |
839 | list is controlled by the int pointed to by the separator argument as follows: | |
840 | ||
ec95d1a6 PH |
841 | If the value is > 0 it is used as the separator. This is typically used for |
842 | sublists such as slash-separated options. The value is always a printing | |
843 | character. | |
844 | ||
845 | (If the value is actually > UCHAR_MAX there is only one item in the list. | |
059ec3d9 PH |
846 | This is used for some cases when called via functions that sometimes |
847 | plough through lists, and sometimes are given single items.) | |
059ec3d9 | 848 | |
ec95d1a6 PH |
849 | If the value is <= 0, the string is inspected for a leading <x, where x is an |
850 | ispunct() or an iscntrl() character. If found, x is used as the separator. If | |
851 | not found: | |
852 | ||
853 | (a) if separator == 0, ':' is used | |
854 | (b) if separator <0, -separator is used | |
855 | ||
856 | In all cases the value of the separator that is used is written back to the | |
857 | int so that it is used on subsequent calls as we progress through the list. | |
858 | ||
859 | A literal ispunct() separator can be represented in an item by doubling, but | |
860 | there is no way to include an iscntrl() separator as part of the data. | |
059ec3d9 PH |
861 | |
862 | Arguments: | |
863 | listptr points to a pointer to the current start of the list; the | |
864 | pointer gets updated to point after the end of the next item | |
865 | separator a pointer to the separator character in an int (see above) | |
866 | buffer where to put a copy of the next string in the list; or | |
867 | NULL if the next string is returned in new memory | |
868 | buflen when buffer is not NULL, the size of buffer; otherwise ignored | |
869 | ||
870 | Returns: pointer to buffer, containing the next substring, | |
871 | or NULL if no more substrings | |
872 | */ | |
873 | ||
874 | uschar * | |
55414b25 | 875 | string_nextinlist(const uschar **listptr, int *separator, uschar *buffer, int buflen) |
059ec3d9 | 876 | { |
55414b25 JH |
877 | int sep = *separator; |
878 | const uschar *s = *listptr; | |
ec95d1a6 | 879 | BOOL sep_is_special; |
059ec3d9 PH |
880 | |
881 | if (s == NULL) return NULL; | |
ec95d1a6 PH |
882 | |
883 | /* This allows for a fixed specified separator to be an iscntrl() character, | |
884 | but at the time of implementation, this is never the case. However, it's best | |
885 | to be conservative. */ | |
886 | ||
887 | while (isspace(*s) && *s != sep) s++; | |
888 | ||
889 | /* A change of separator is permitted, so look for a leading '<' followed by an | |
890 | allowed character. */ | |
059ec3d9 PH |
891 | |
892 | if (sep <= 0) | |
893 | { | |
ec95d1a6 | 894 | if (*s == '<' && (ispunct(s[1]) || iscntrl(s[1]))) |
059ec3d9 PH |
895 | { |
896 | sep = s[1]; | |
897 | s += 2; | |
ec95d1a6 | 898 | while (isspace(*s) && *s != sep) s++; |
059ec3d9 PH |
899 | } |
900 | else | |
901 | { | |
902 | sep = (sep == 0)? ':' : -sep; | |
903 | } | |
904 | *separator = sep; | |
905 | } | |
906 | ||
ec95d1a6 PH |
907 | /* An empty string has no list elements */ |
908 | ||
059ec3d9 PH |
909 | if (*s == 0) return NULL; |
910 | ||
ec95d1a6 PH |
911 | /* Note whether whether or not the separator is an iscntrl() character. */ |
912 | ||
913 | sep_is_special = iscntrl(sep); | |
914 | ||
059ec3d9 PH |
915 | /* Handle the case when a buffer is provided. */ |
916 | ||
917 | if (buffer != NULL) | |
918 | { | |
d4ff61d1 | 919 | int p = 0; |
059ec3d9 PH |
920 | for (; *s != 0; s++) |
921 | { | |
ec95d1a6 | 922 | if (*s == sep && (*(++s) != sep || sep_is_special)) break; |
059ec3d9 PH |
923 | if (p < buflen - 1) buffer[p++] = *s; |
924 | } | |
925 | while (p > 0 && isspace(buffer[p-1])) p--; | |
926 | buffer[p] = 0; | |
927 | } | |
928 | ||
929 | /* Handle the case when a buffer is not provided. */ | |
930 | ||
931 | else | |
932 | { | |
55414b25 | 933 | const uschar *ss; |
acec9514 | 934 | gstring * g = NULL; |
ec95d1a6 | 935 | |
059ec3d9 | 936 | /* We know that *s != 0 at this point. However, it might be pointing to a |
ec95d1a6 PH |
937 | separator, which could indicate an empty string, or (if an ispunct() |
938 | character) could be doubled to indicate a separator character as data at the | |
939 | start of a string. Avoid getting working memory for an empty item. */ | |
059ec3d9 PH |
940 | |
941 | if (*s == sep) | |
942 | { | |
943 | s++; | |
ec95d1a6 PH |
944 | if (*s != sep || sep_is_special) |
945 | { | |
946 | *listptr = s; | |
947 | return string_copy(US""); | |
948 | } | |
059ec3d9 PH |
949 | } |
950 | ||
ec95d1a6 PH |
951 | /* Not an empty string; the first character is guaranteed to be a data |
952 | character. */ | |
953 | ||
954 | for (;;) | |
059ec3d9 | 955 | { |
acec9514 JH |
956 | for (ss = s + 1; *ss != 0 && *ss != sep; ss++) ; |
957 | g = string_catn(g, s, ss-s); | |
ec95d1a6 PH |
958 | s = ss; |
959 | if (*s == 0 || *(++s) != sep || sep_is_special) break; | |
059ec3d9 | 960 | } |
acec9514 JH |
961 | while (g->ptr > 0 && isspace(g->s[g->ptr-1])) g->ptr--; |
962 | buffer = string_from_gstring(g); | |
059ec3d9 PH |
963 | } |
964 | ||
965 | /* Update the current pointer and return the new string */ | |
966 | ||
967 | *listptr = s; | |
968 | return buffer; | |
969 | } | |
059ec3d9 PH |
970 | |
971 | ||
4226691b JH |
972 | static const uschar * |
973 | Ustrnchr(const uschar * s, int c, unsigned * len) | |
974 | { | |
975 | unsigned siz = *len; | |
976 | while (siz) | |
977 | { | |
978 | if (!*s) return NULL; | |
979 | if (*s == c) | |
980 | { | |
981 | *len = siz; | |
982 | return s; | |
983 | } | |
984 | s++; | |
985 | siz--; | |
986 | } | |
987 | return NULL; | |
988 | } | |
989 | ||
990 | ||
76146973 | 991 | /************************************************ |
1d9ddac9 | 992 | * Add element to separated list * |
76146973 | 993 | ************************************************/ |
4226691b JH |
994 | /* This function is used to build a list, returning an allocated null-terminated |
995 | growable string. The given element has any embedded separator characters | |
76146973 JH |
996 | doubled. |
997 | ||
4226691b JH |
998 | Despite having the same growable-string interface as string_cat() the list is |
999 | always returned null-terminated. | |
1000 | ||
76146973 | 1001 | Arguments: |
acec9514 | 1002 | list expanding-string for the list that is being built, or NULL |
76146973 | 1003 | if this is a new list that has no contents yet |
4c04137d JS |
1004 | sep list separator character |
1005 | ele new element to be appended to the list | |
76146973 JH |
1006 | |
1007 | Returns: pointer to the start of the list, changed if copied for expansion. | |
1008 | */ | |
1009 | ||
acec9514 JH |
1010 | gstring * |
1011 | string_append_listele(gstring * list, uschar sep, const uschar * ele) | |
76146973 | 1012 | { |
76146973 JH |
1013 | uschar * sp; |
1014 | ||
acec9514 JH |
1015 | if (list && list->ptr) |
1016 | list = string_catn(list, &sep, 1); | |
76146973 | 1017 | |
e3dd1d67 | 1018 | while((sp = Ustrchr(ele, sep))) |
76146973 | 1019 | { |
acec9514 JH |
1020 | list = string_catn(list, ele, sp-ele+1); |
1021 | list = string_catn(list, &sep, 1); | |
76146973 JH |
1022 | ele = sp+1; |
1023 | } | |
acec9514 JH |
1024 | list = string_cat(list, ele); |
1025 | (void) string_from_gstring(list); | |
4226691b | 1026 | return list; |
76146973 | 1027 | } |
00ba27c5 JH |
1028 | |
1029 | ||
acec9514 JH |
1030 | gstring * |
1031 | string_append_listele_n(gstring * list, uschar sep, const uschar * ele, | |
1032 | unsigned len) | |
00ba27c5 | 1033 | { |
00ba27c5 JH |
1034 | const uschar * sp; |
1035 | ||
acec9514 JH |
1036 | if (list && list->ptr) |
1037 | list = string_catn(list, &sep, 1); | |
00ba27c5 JH |
1038 | |
1039 | while((sp = Ustrnchr(ele, sep, &len))) | |
1040 | { | |
acec9514 JH |
1041 | list = string_catn(list, ele, sp-ele+1); |
1042 | list = string_catn(list, &sep, 1); | |
00ba27c5 JH |
1043 | ele = sp+1; |
1044 | len--; | |
1045 | } | |
acec9514 JH |
1046 | list = string_catn(list, ele, len); |
1047 | (void) string_from_gstring(list); | |
4226691b | 1048 | return list; |
00ba27c5 | 1049 | } |
76146973 JH |
1050 | |
1051 | ||
059ec3d9 | 1052 | |
acec9514 JH |
1053 | /************************************************/ |
1054 | /* Create a growable-string with some preassigned space */ | |
1055 | ||
1056 | gstring * | |
1057 | string_get(unsigned size) | |
1058 | { | |
1059 | gstring * g = store_get(sizeof(gstring) + size); | |
1060 | g->size = size; | |
1061 | g->ptr = 0; | |
1062 | g->s = US(g + 1); | |
1063 | return g; | |
1064 | } | |
1065 | ||
1066 | /* NUL-terminate the C string in the growable-string, and return it. */ | |
1067 | ||
1068 | uschar * | |
1069 | string_from_gstring(gstring * g) | |
1070 | { | |
1071 | if (!g) return NULL; | |
1072 | g->s[g->ptr] = '\0'; | |
1073 | return g->s; | |
1074 | } | |
1075 | ||
059ec3d9 PH |
1076 | /************************************************* |
1077 | * Add chars to string * | |
1078 | *************************************************/ | |
1079 | ||
938593e9 JH |
1080 | /* Arguments: |
1081 | g the grawable-string | |
1082 | p current end of data | |
1083 | count amount to grow by | |
1084 | */ | |
1085 | ||
1086 | static void | |
acec9514 JH |
1087 | gstring_grow(gstring * g, int p, int count) |
1088 | { | |
1089 | int oldsize = g->size; | |
1090 | ||
1091 | /* Mostly, string_cat() is used to build small strings of a few hundred | |
1092 | characters at most. There are times, however, when the strings are very much | |
1093 | longer (for example, a lookup that returns a vast number of alias addresses). | |
1094 | To try to keep things reasonable, we use increments whose size depends on the | |
1095 | existing length of the string. */ | |
1096 | ||
1097 | unsigned inc = oldsize < 4096 ? 127 : 1023; | |
1098 | g->size = ((p + count + inc) & ~inc) + 1; | |
1099 | ||
1100 | /* Try to extend an existing allocation. If the result of calling | |
1101 | store_extend() is false, either there isn't room in the current memory block, | |
1102 | or this string is not the top item on the dynamic store stack. We then have | |
1103 | to get a new chunk of store and copy the old string. When building large | |
1104 | strings, it is helpful to call store_release() on the old string, to release | |
1105 | memory blocks that have become empty. (The block will be freed if the string | |
1106 | is at its start.) However, we can do this only if we know that the old string | |
1107 | was the last item on the dynamic memory stack. This is the case if it matches | |
1108 | store_last_get. */ | |
1109 | ||
1110 | if (!store_extend(g->s, oldsize, g->size)) | |
459fca58 | 1111 | g->s = store_newblock(g->s, g->size, p); |
acec9514 JH |
1112 | } |
1113 | ||
1114 | ||
1115 | ||
059ec3d9 PH |
1116 | /* This function is used when building up strings of unknown length. Room is |
1117 | always left for a terminating zero to be added to the string that is being | |
1118 | built. This function does not require the string that is being added to be NUL | |
1119 | terminated, because the number of characters to add is given explicitly. It is | |
1120 | sometimes called to extract parts of other strings. | |
1121 | ||
1122 | Arguments: | |
1123 | string points to the start of the string that is being built, or NULL | |
1124 | if this is a new string that has no contents yet | |
059ec3d9 PH |
1125 | s points to characters to add |
1126 | count count of characters to add; must not exceed the length of s, if s | |
42055a33 | 1127 | is a C string. |
059ec3d9 | 1128 | |
059ec3d9 PH |
1129 | Returns: pointer to the start of the string, changed if copied for expansion. |
1130 | Note that a NUL is not added, though space is left for one. This is | |
1131 | because string_cat() is often called multiple times to build up a | |
1132 | string - there's no point adding the NUL till the end. | |
a1b8a755 | 1133 | |
059ec3d9 | 1134 | */ |
96f5fe4c | 1135 | /* coverity[+alloc] */ |
059ec3d9 | 1136 | |
acec9514 JH |
1137 | gstring * |
1138 | string_catn(gstring * g, const uschar *s, int count) | |
059ec3d9 | 1139 | { |
acec9514 | 1140 | int p; |
059ec3d9 | 1141 | |
acec9514 | 1142 | if (!g) |
059ec3d9 | 1143 | { |
acec9514 JH |
1144 | unsigned inc = count < 4096 ? 127 : 1023; |
1145 | unsigned size = ((count + inc) & ~inc) + 1; | |
1146 | g = string_get(size); | |
059ec3d9 PH |
1147 | } |
1148 | ||
acec9514 JH |
1149 | p = g->ptr; |
1150 | if (p + count >= g->size) | |
1151 | gstring_grow(g, p, count); | |
1152 | ||
059ec3d9 PH |
1153 | /* Because we always specify the exact number of characters to copy, we can |
1154 | use memcpy(), which is likely to be more efficient than strncopy() because the | |
acec9514 | 1155 | latter has to check for zero bytes. */ |
059ec3d9 | 1156 | |
acec9514 JH |
1157 | memcpy(g->s + p, s, count); |
1158 | g->ptr = p + count; | |
1159 | return g; | |
059ec3d9 | 1160 | } |
acec9514 JH |
1161 | |
1162 | ||
1163 | gstring * | |
1164 | string_cat(gstring *string, const uschar *s) | |
c2f669a4 | 1165 | { |
acec9514 | 1166 | return string_catn(string, s, Ustrlen(s)); |
c2f669a4 | 1167 | } |
059ec3d9 PH |
1168 | |
1169 | ||
1170 | ||
059ec3d9 PH |
1171 | /************************************************* |
1172 | * Append strings to another string * | |
1173 | *************************************************/ | |
1174 | ||
1175 | /* This function can be used to build a string from many other strings. | |
1176 | It calls string_cat() to do the dirty work. | |
1177 | ||
1178 | Arguments: | |
acec9514 | 1179 | string expanding-string that is being built, or NULL |
059ec3d9 | 1180 | if this is a new string that has no contents yet |
059ec3d9 PH |
1181 | count the number of strings to append |
1182 | ... "count" uschar* arguments, which must be valid zero-terminated | |
1183 | C strings | |
1184 | ||
1185 | Returns: pointer to the start of the string, changed if copied for expansion. | |
1186 | The string is not zero-terminated - see string_cat() above. | |
1187 | */ | |
1188 | ||
acec9514 JH |
1189 | __inline__ gstring * |
1190 | string_append(gstring *string, int count, ...) | |
059ec3d9 PH |
1191 | { |
1192 | va_list ap; | |
059ec3d9 PH |
1193 | |
1194 | va_start(ap, count); | |
acec9514 | 1195 | while (count-- > 0) |
059ec3d9 PH |
1196 | { |
1197 | uschar *t = va_arg(ap, uschar *); | |
acec9514 | 1198 | string = string_cat(string, t); |
059ec3d9 PH |
1199 | } |
1200 | va_end(ap); | |
1201 | ||
1202 | return string; | |
1203 | } | |
1204 | #endif | |
1205 | ||
1206 | ||
1207 | ||
1208 | /************************************************* | |
1209 | * Format a string with length checks * | |
1210 | *************************************************/ | |
1211 | ||
1212 | /* This function is used to format a string with checking of the length of the | |
1213 | output for all conversions. It protects Exim from absent-mindedness when | |
1214 | calling functions like debug_printf and string_sprintf, and elsewhere. There | |
1215 | are two different entry points to what is actually the same function, depending | |
1216 | on whether the variable length list of data arguments are given explicitly or | |
1217 | as a va_list item. | |
1218 | ||
1219 | The formats are the usual printf() ones, with some omissions (never used) and | |
c0b9d3e8 | 1220 | three additions for strings: %S forces lower case, %T forces upper case, and |
acec9514 | 1221 | %#s or %#S prints nothing for a NULL string. Without the # "NULL" is printed |
c0b9d3e8 JH |
1222 | (useful in debugging). There is also the addition of %D and %M, which insert |
1223 | the date in the form used for datestamped log files. | |
059ec3d9 PH |
1224 | |
1225 | Arguments: | |
1226 | buffer a buffer in which to put the formatted string | |
1227 | buflen the length of the buffer | |
1228 | format the format string - deliberately char * and not uschar * | |
1229 | ... or ap variable list of supplementary arguments | |
1230 | ||
1231 | Returns: TRUE if the result fitted in the buffer | |
1232 | */ | |
1233 | ||
1234 | BOOL | |
1ba28e2b | 1235 | string_format(uschar *buffer, int buflen, const char *format, ...) |
059ec3d9 PH |
1236 | { |
1237 | BOOL yield; | |
1238 | va_list ap; | |
1239 | va_start(ap, format); | |
1240 | yield = string_vformat(buffer, buflen, format, ap); | |
1241 | va_end(ap); | |
1242 | return yield; | |
1243 | } | |
1244 | ||
1245 | ||
1246 | BOOL | |
1ba28e2b | 1247 | string_vformat(uschar *buffer, int buflen, const char *format, va_list ap) |
059ec3d9 | 1248 | { |
91a246f6 PP |
1249 | /* We assume numbered ascending order, C does not guarantee that */ |
1250 | enum { L_NORMAL=1, L_SHORT=2, L_LONG=3, L_LONGLONG=4, L_LONGDOUBLE=5, L_SIZE=6 }; | |
b1c749bb | 1251 | |
059ec3d9 PH |
1252 | BOOL yield = TRUE; |
1253 | int width, precision; | |
1ba28e2b | 1254 | const char *fp = format; /* Deliberately not unsigned */ |
059ec3d9 PH |
1255 | uschar *p = buffer; |
1256 | uschar *last = buffer + buflen - 1; | |
1257 | ||
1258 | string_datestamp_offset = -1; /* Datestamp not inserted */ | |
f1e5fef5 PP |
1259 | string_datestamp_length = 0; /* Datestamp not inserted */ |
1260 | string_datestamp_type = 0; /* Datestamp not inserted */ | |
059ec3d9 PH |
1261 | |
1262 | /* Scan the format and handle the insertions */ | |
1263 | ||
1264 | while (*fp != 0) | |
1265 | { | |
b1c749bb | 1266 | int length = L_NORMAL; |
059ec3d9 PH |
1267 | int *nptr; |
1268 | int slen; | |
1ba28e2b PP |
1269 | const char *null = "NULL"; /* ) These variables */ |
1270 | const char *item_start, *s; /* ) are deliberately */ | |
059ec3d9 PH |
1271 | char newformat[16]; /* ) not unsigned */ |
1272 | ||
1273 | /* Non-% characters just get copied verbatim */ | |
1274 | ||
1275 | if (*fp != '%') | |
1276 | { | |
1277 | if (p >= last) { yield = FALSE; break; } | |
1278 | *p++ = (uschar)*fp++; | |
1279 | continue; | |
1280 | } | |
1281 | ||
1282 | /* Deal with % characters. Pick off the width and precision, for checking | |
1283 | strings, skipping over the flag and modifier characters. */ | |
1284 | ||
1285 | item_start = fp; | |
1286 | width = precision = -1; | |
1287 | ||
1288 | if (strchr("-+ #0", *(++fp)) != NULL) | |
1289 | { | |
1290 | if (*fp == '#') null = ""; | |
1291 | fp++; | |
1292 | } | |
1293 | ||
1294 | if (isdigit((uschar)*fp)) | |
1295 | { | |
1296 | width = *fp++ - '0'; | |
1297 | while (isdigit((uschar)*fp)) width = width * 10 + *fp++ - '0'; | |
1298 | } | |
1299 | else if (*fp == '*') | |
1300 | { | |
1301 | width = va_arg(ap, int); | |
1302 | fp++; | |
1303 | } | |
1304 | ||
1305 | if (*fp == '.') | |
1306 | { | |
1307 | if (*(++fp) == '*') | |
1308 | { | |
1309 | precision = va_arg(ap, int); | |
1310 | fp++; | |
1311 | } | |
1312 | else | |
1313 | { | |
1314 | precision = 0; | |
1315 | while (isdigit((uschar)*fp)) | |
1316 | precision = precision*10 + *fp++ - '0'; | |
1317 | } | |
1318 | } | |
1319 | ||
91a246f6 | 1320 | /* Skip over 'h', 'L', 'l', 'll' and 'z', remembering the item length */ |
b1c749bb PH |
1321 | |
1322 | if (*fp == 'h') | |
1323 | { fp++; length = L_SHORT; } | |
1324 | else if (*fp == 'L') | |
1325 | { fp++; length = L_LONGDOUBLE; } | |
1326 | else if (*fp == 'l') | |
1327 | { | |
1328 | if (fp[1] == 'l') | |
1329 | { | |
1330 | fp += 2; | |
1331 | length = L_LONGLONG; | |
1332 | } | |
1333 | else | |
1334 | { | |
1335 | fp++; | |
1336 | length = L_LONG; | |
1337 | } | |
1338 | } | |
91a246f6 PP |
1339 | else if (*fp == 'z') |
1340 | { fp++; length = L_SIZE; } | |
059ec3d9 PH |
1341 | |
1342 | /* Handle each specific format type. */ | |
1343 | ||
1344 | switch (*fp++) | |
1345 | { | |
1346 | case 'n': | |
1347 | nptr = va_arg(ap, int *); | |
1348 | *nptr = p - buffer; | |
1349 | break; | |
1350 | ||
1351 | case 'd': | |
1352 | case 'o': | |
1353 | case 'u': | |
1354 | case 'x': | |
1355 | case 'X': | |
1549ea3b PH |
1356 | if (p >= last - ((length > L_LONG)? 24 : 12)) |
1357 | { yield = FALSE; goto END_FORMAT; } | |
059ec3d9 PH |
1358 | strncpy(newformat, item_start, fp - item_start); |
1359 | newformat[fp - item_start] = 0; | |
b1c749bb PH |
1360 | |
1361 | /* Short int is promoted to int when passing through ..., so we must use | |
1362 | int for va_arg(). */ | |
1363 | ||
1364 | switch(length) | |
1365 | { | |
1366 | case L_SHORT: | |
5976eb99 JH |
1367 | case L_NORMAL: p += sprintf(CS p, newformat, va_arg(ap, int)); break; |
1368 | case L_LONG: p += sprintf(CS p, newformat, va_arg(ap, long int)); break; | |
1369 | case L_LONGLONG: p += sprintf(CS p, newformat, va_arg(ap, LONGLONG_T)); break; | |
1370 | case L_SIZE: p += sprintf(CS p, newformat, va_arg(ap, size_t)); break; | |
b1c749bb | 1371 | } |
059ec3d9 PH |
1372 | break; |
1373 | ||
1374 | case 'p': | |
2e8db779 JH |
1375 | { |
1376 | void * ptr; | |
1377 | if (p >= last - 24) { yield = FALSE; goto END_FORMAT; } | |
81f358da JH |
1378 | /* sprintf() saying "(nil)" for a null pointer seems unreliable. |
1379 | Handle it explicitly. */ | |
2e8db779 JH |
1380 | if ((ptr = va_arg(ap, void *))) |
1381 | { | |
1382 | strncpy(newformat, item_start, fp - item_start); | |
1383 | newformat[fp - item_start] = 0; | |
81f358da | 1384 | p += sprintf(CS p, newformat, ptr); |
2e8db779 JH |
1385 | } |
1386 | else | |
1387 | p += sprintf(CS p, "(nil)"); | |
1388 | } | |
059ec3d9 PH |
1389 | break; |
1390 | ||
1391 | /* %f format is inherently insecure if the numbers that it may be | |
870f6ba8 TF |
1392 | handed are unknown (e.g. 1e300). However, in Exim, %f is used for |
1393 | printing load averages, and these are actually stored as integers | |
1394 | (load average * 1000) so the size of the numbers is constrained. | |
1395 | It is also used for formatting sending rates, where the simplicity | |
1396 | of the format prevents overflow. */ | |
059ec3d9 PH |
1397 | |
1398 | case 'f': | |
1399 | case 'e': | |
1400 | case 'E': | |
1401 | case 'g': | |
1402 | case 'G': | |
1403 | if (precision < 0) precision = 6; | |
1404 | if (p >= last - precision - 8) { yield = FALSE; goto END_FORMAT; } | |
1405 | strncpy(newformat, item_start, fp - item_start); | |
1406 | newformat[fp-item_start] = 0; | |
b1c749bb | 1407 | if (length == L_LONGDOUBLE) |
5976eb99 | 1408 | p += sprintf(CS p, newformat, va_arg(ap, long double)); |
b1c749bb | 1409 | else |
5976eb99 | 1410 | p += sprintf(CS p, newformat, va_arg(ap, double)); |
059ec3d9 PH |
1411 | break; |
1412 | ||
1413 | /* String types */ | |
1414 | ||
1415 | case '%': | |
1416 | if (p >= last) { yield = FALSE; goto END_FORMAT; } | |
1417 | *p++ = '%'; | |
1418 | break; | |
1419 | ||
1420 | case 'c': | |
1421 | if (p >= last) { yield = FALSE; goto END_FORMAT; } | |
1422 | *p++ = va_arg(ap, int); | |
1423 | break; | |
1424 | ||
f1e5fef5 PP |
1425 | case 'D': /* Insert daily datestamp for log file names */ |
1426 | s = CS tod_stamp(tod_log_datestamp_daily); | |
059ec3d9 | 1427 | string_datestamp_offset = p - buffer; /* Passed back via global */ |
f1e5fef5 PP |
1428 | string_datestamp_length = Ustrlen(s); /* Passed back via global */ |
1429 | string_datestamp_type = tod_log_datestamp_daily; | |
1430 | slen = string_datestamp_length; | |
1431 | goto INSERT_STRING; | |
1432 | ||
1433 | case 'M': /* Insert monthly datestamp for log file names */ | |
1434 | s = CS tod_stamp(tod_log_datestamp_monthly); | |
1435 | string_datestamp_offset = p - buffer; /* Passed back via global */ | |
1436 | string_datestamp_length = Ustrlen(s); /* Passed back via global */ | |
1437 | string_datestamp_type = tod_log_datestamp_monthly; | |
1438 | slen = string_datestamp_length; | |
059ec3d9 PH |
1439 | goto INSERT_STRING; |
1440 | ||
1441 | case 's': | |
1442 | case 'S': /* Forces *lower* case */ | |
c0b9d3e8 | 1443 | case 'T': /* Forces *upper* case */ |
059ec3d9 PH |
1444 | s = va_arg(ap, char *); |
1445 | ||
059ec3d9 PH |
1446 | if (s == NULL) s = null; |
1447 | slen = Ustrlen(s); | |
1448 | ||
f1e5fef5 PP |
1449 | INSERT_STRING: /* Come to from %D or %M above */ |
1450 | ||
059ec3d9 PH |
1451 | /* If the width is specified, check that there is a precision |
1452 | set; if not, set it to the width to prevent overruns of long | |
1453 | strings. */ | |
1454 | ||
1455 | if (width >= 0) | |
1456 | { | |
1457 | if (precision < 0) precision = width; | |
1458 | } | |
1459 | ||
1460 | /* If a width is not specified and the precision is specified, set | |
1461 | the width to the precision, or the string length if shorted. */ | |
1462 | ||
1463 | else if (precision >= 0) | |
1464 | { | |
1465 | width = (precision < slen)? precision : slen; | |
1466 | } | |
1467 | ||
1468 | /* If neither are specified, set them both to the string length. */ | |
1469 | ||
1470 | else width = precision = slen; | |
1471 | ||
1472 | /* Check string space, and add the string to the buffer if ok. If | |
1473 | not OK, add part of the string (debugging uses this to show as | |
1474 | much as possible). */ | |
1475 | ||
24c929a2 NM |
1476 | if (p == last) |
1477 | { | |
1478 | yield = FALSE; | |
1479 | goto END_FORMAT; | |
1480 | } | |
059ec3d9 PH |
1481 | if (p >= last - width) |
1482 | { | |
1483 | yield = FALSE; | |
1484 | width = precision = last - p - 1; | |
24c929a2 NM |
1485 | if (width < 0) width = 0; |
1486 | if (precision < 0) precision = 0; | |
059ec3d9 PH |
1487 | } |
1488 | sprintf(CS p, "%*.*s", width, precision, s); | |
1489 | if (fp[-1] == 'S') | |
1490 | while (*p) { *p = tolower(*p); p++; } | |
c0b9d3e8 JH |
1491 | else if (fp[-1] == 'T') |
1492 | while (*p) { *p = toupper(*p); p++; } | |
059ec3d9 PH |
1493 | else |
1494 | while (*p) p++; | |
1495 | if (!yield) goto END_FORMAT; | |
1496 | break; | |
1497 | ||
1498 | /* Some things are never used in Exim; also catches junk. */ | |
1499 | ||
1500 | default: | |
1501 | strncpy(newformat, item_start, fp - item_start); | |
1502 | newformat[fp-item_start] = 0; | |
1503 | log_write(0, LOG_MAIN|LOG_PANIC_DIE, "string_format: unsupported type " | |
1504 | "in \"%s\" in \"%s\"", newformat, format); | |
1505 | break; | |
1506 | } | |
1507 | } | |
1508 | ||
1509 | /* Ensure string is complete; return TRUE if got to the end of the format */ | |
1510 | ||
1511 | END_FORMAT: | |
1512 | ||
1513 | *p = 0; | |
1514 | return yield; | |
1515 | } | |
1516 | ||
1517 | ||
1518 | ||
1519 | #ifndef COMPILE_UTILITY | |
1520 | /************************************************* | |
1521 | * Generate an "open failed" message * | |
1522 | *************************************************/ | |
1523 | ||
1524 | /* This function creates a message after failure to open a file. It includes a | |
1525 | string supplied as data, adds the strerror() text, and if the failure was | |
1526 | "Permission denied", reads and includes the euid and egid. | |
1527 | ||
1528 | Arguments: | |
1529 | eno the value of errno after the failure | |
1530 | format a text format string - deliberately not uschar * | |
1531 | ... arguments for the format string | |
1532 | ||
1533 | Returns: a message, in dynamic store | |
1534 | */ | |
1535 | ||
1536 | uschar * | |
1ba28e2b | 1537 | string_open_failed(int eno, const char *format, ...) |
059ec3d9 PH |
1538 | { |
1539 | va_list ap; | |
1540 | uschar buffer[1024]; | |
1541 | ||
1542 | Ustrcpy(buffer, "failed to open "); | |
1543 | va_start(ap, format); | |
1544 | ||
1545 | /* Use the checked formatting routine to ensure that the buffer | |
1546 | does not overflow. It should not, since this is called only for internally | |
1547 | specified messages. If it does, the message just gets truncated, and there | |
1548 | doesn't seem much we can do about that. */ | |
1549 | ||
1550 | (void)string_vformat(buffer+15, sizeof(buffer) - 15, format, ap); | |
cb570b5e | 1551 | va_end(ap); |
059ec3d9 PH |
1552 | |
1553 | return (eno == EACCES)? | |
1554 | string_sprintf("%s: %s (euid=%ld egid=%ld)", buffer, strerror(eno), | |
1555 | (long int)geteuid(), (long int)getegid()) : | |
1556 | string_sprintf("%s: %s", buffer, strerror(eno)); | |
1557 | } | |
1558 | #endif /* COMPILE_UTILITY */ | |
1559 | ||
1560 | ||
1561 | ||
059ec3d9 PH |
1562 | |
1563 | ||
bc3c7bb7 HSHR |
1564 | #ifndef COMPILE_UTILITY |
1565 | /* qsort(3), currently used to sort the environment variables | |
1566 | for -bP environment output, needs a function to compare two pointers to string | |
1567 | pointers. Here it is. */ | |
1568 | ||
1569 | int | |
84bbb4d8 | 1570 | string_compare_by_pointer(const void *a, const void *b) |
bc3c7bb7 | 1571 | { |
35a5627d | 1572 | return Ustrcmp(* CUSS a, * CUSS b); |
bc3c7bb7 HSHR |
1573 | } |
1574 | #endif /* COMPILE_UTILITY */ | |
059ec3d9 PH |
1575 | |
1576 | ||
1577 | ||
1578 | /************************************************* | |
1579 | ************************************************** | |
1580 | * Stand-alone test program * | |
1581 | ************************************************** | |
1582 | *************************************************/ | |
1583 | ||
1584 | #ifdef STAND_ALONE | |
1585 | int main(void) | |
1586 | { | |
1587 | uschar buffer[256]; | |
1588 | ||
1589 | printf("Testing is_ip_address\n"); | |
1590 | ||
1591 | while (fgets(CS buffer, sizeof(buffer), stdin) != NULL) | |
1592 | { | |
1593 | int offset; | |
1594 | buffer[Ustrlen(buffer) - 1] = 0; | |
1595 | printf("%d\n", string_is_ip_address(buffer, NULL)); | |
1596 | printf("%d %d %s\n", string_is_ip_address(buffer, &offset), offset, buffer); | |
1597 | } | |
1598 | ||
1599 | printf("Testing string_nextinlist\n"); | |
1600 | ||
1601 | while (fgets(CS buffer, sizeof(buffer), stdin) != NULL) | |
1602 | { | |
1603 | uschar *list = buffer; | |
1604 | uschar *lp1, *lp2; | |
1605 | uschar item[256]; | |
1606 | int sep1 = 0; | |
1607 | int sep2 = 0; | |
1608 | ||
1609 | if (*list == '<') | |
1610 | { | |
1611 | sep1 = sep2 = list[1]; | |
1612 | list += 2; | |
1613 | } | |
1614 | ||
1615 | lp1 = lp2 = list; | |
1616 | for (;;) | |
1617 | { | |
1618 | uschar *item1 = string_nextinlist(&lp1, &sep1, item, sizeof(item)); | |
1619 | uschar *item2 = string_nextinlist(&lp2, &sep2, NULL, 0); | |
1620 | ||
1621 | if (item1 == NULL && item2 == NULL) break; | |
1622 | if (item == NULL || item2 == NULL || Ustrcmp(item1, item2) != 0) | |
1623 | { | |
1624 | printf("***ERROR\nitem1=\"%s\"\nitem2=\"%s\"\n", | |
1625 | (item1 == NULL)? "NULL" : CS item1, | |
1626 | (item2 == NULL)? "NULL" : CS item2); | |
1627 | break; | |
1628 | } | |
1629 | else printf(" \"%s\"\n", CS item1); | |
1630 | } | |
1631 | } | |
1632 | ||
1633 | /* This is a horrible lash-up, but it serves its purpose. */ | |
1634 | ||
1635 | printf("Testing string_format\n"); | |
1636 | ||
1637 | while (fgets(CS buffer, sizeof(buffer), stdin) != NULL) | |
1638 | { | |
1639 | void *args[3]; | |
ed72ace5 | 1640 | long long llargs[3]; |
059ec3d9 PH |
1641 | double dargs[3]; |
1642 | int dflag = 0; | |
ed72ace5 | 1643 | int llflag = 0; |
059ec3d9 PH |
1644 | int n = 0; |
1645 | int count; | |
1646 | int countset = 0; | |
1647 | uschar format[256]; | |
1648 | uschar outbuf[256]; | |
1649 | uschar *s; | |
1650 | buffer[Ustrlen(buffer) - 1] = 0; | |
1651 | ||
1652 | s = Ustrchr(buffer, ','); | |
1653 | if (s == NULL) s = buffer + Ustrlen(buffer); | |
1654 | ||
1655 | Ustrncpy(format, buffer, s - buffer); | |
1656 | format[s-buffer] = 0; | |
1657 | ||
1658 | if (*s == ',') s++; | |
1659 | ||
1660 | while (*s != 0) | |
1661 | { | |
1662 | uschar *ss = s; | |
1663 | s = Ustrchr(ss, ','); | |
1664 | if (s == NULL) s = ss + Ustrlen(ss); | |
1665 | ||
1666 | if (isdigit(*ss)) | |
1667 | { | |
1668 | Ustrncpy(outbuf, ss, s-ss); | |
1669 | if (Ustrchr(outbuf, '.') != NULL) | |
1670 | { | |
1671 | dflag = 1; | |
1672 | dargs[n++] = Ustrtod(outbuf, NULL); | |
1673 | } | |
ed72ace5 PH |
1674 | else if (Ustrstr(outbuf, "ll") != NULL) |
1675 | { | |
1676 | llflag = 1; | |
1677 | llargs[n++] = strtoull(CS outbuf, NULL, 10); | |
1678 | } | |
059ec3d9 PH |
1679 | else |
1680 | { | |
1681 | args[n++] = (void *)Uatoi(outbuf); | |
1682 | } | |
1683 | } | |
1684 | ||
1685 | else if (Ustrcmp(ss, "*") == 0) | |
1686 | { | |
1687 | args[n++] = (void *)(&count); | |
1688 | countset = 1; | |
1689 | } | |
1690 | ||
1691 | else | |
1692 | { | |
1693 | uschar *sss = malloc(s - ss + 1); | |
1694 | Ustrncpy(sss, ss, s-ss); | |
1695 | args[n++] = sss; | |
1696 | } | |
1697 | ||
1698 | if (*s == ',') s++; | |
1699 | } | |
1700 | ||
ed72ace5 PH |
1701 | if (!dflag && !llflag) |
1702 | printf("%s\n", string_format(outbuf, sizeof(outbuf), CS format, | |
1703 | args[0], args[1], args[2])? "True" : "False"); | |
1704 | ||
1705 | else if (dflag) | |
1706 | printf("%s\n", string_format(outbuf, sizeof(outbuf), CS format, | |
1707 | dargs[0], dargs[1], dargs[2])? "True" : "False"); | |
059ec3d9 PH |
1708 | |
1709 | else printf("%s\n", string_format(outbuf, sizeof(outbuf), CS format, | |
ed72ace5 | 1710 | llargs[0], llargs[1], llargs[2])? "True" : "False"); |
059ec3d9 PH |
1711 | |
1712 | printf("%s\n", CS outbuf); | |
1713 | if (countset) printf("count=%d\n", count); | |
1714 | } | |
1715 | ||
1716 | return 0; | |
1717 | } | |
1718 | #endif | |
1719 | ||
1720 | /* End of string.c */ |