Commit | Line | Data |
---|---|---|
059ec3d9 PH |
1 | /************************************************* |
2 | * Exim - an Internet mail transport agent * | |
3 | *************************************************/ | |
4 | ||
5a66c31b | 5 | /* Copyright (c) University of Cambridge 1995 - 2014 */ |
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 | ||
45 | if (maskptr != NULL) | |
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 | ||
82 | if (*s == 0 || *s == '%' || *s == '/') return had_double_colon? yield : 0; | |
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 | { | |
138 | if (i != 0 && *s++ != '.') return 0; | |
139 | if (!isdigit(*s++)) return 0; | |
140 | if (isdigit(*s) && isdigit(*(++s))) s++; | |
141 | } | |
142 | ||
1688f43b PH |
143 | return (*s == 0 || (*s == '/' && maskptr != NULL && *maskptr != 0))? |
144 | yield : 0; | |
059ec3d9 PH |
145 | } |
146 | #endif /* COMPILE_UTILITY */ | |
147 | ||
148 | ||
149 | /************************************************* | |
150 | * Format message size * | |
151 | *************************************************/ | |
152 | ||
153 | /* Convert a message size in bytes to printing form, rounding | |
154 | according to the magnitude of the number. A value of zero causes | |
155 | a string of spaces to be returned. | |
156 | ||
157 | Arguments: | |
158 | size the message size in bytes | |
159 | buffer where to put the answer | |
160 | ||
161 | Returns: pointer to the buffer | |
162 | a string of exactly 5 characters is normally returned | |
163 | */ | |
164 | ||
165 | uschar * | |
166 | string_format_size(int size, uschar *buffer) | |
167 | { | |
45500060 | 168 | if (size == 0) Ustrcpy(buffer, " "); |
059ec3d9 PH |
169 | else if (size < 1024) sprintf(CS buffer, "%5d", size); |
170 | else if (size < 10*1024) | |
171 | sprintf(CS buffer, "%4.1fK", (double)size / 1024.0); | |
172 | else if (size < 1024*1024) | |
173 | sprintf(CS buffer, "%4dK", (size + 512)/1024); | |
174 | else if (size < 10*1024*1024) | |
175 | sprintf(CS buffer, "%4.1fM", (double)size / (1024.0 * 1024.0)); | |
176 | else | |
177 | sprintf(CS buffer, "%4dM", (size + 512 * 1024)/(1024*1024)); | |
178 | return buffer; | |
179 | } | |
180 | ||
181 | ||
182 | ||
183 | #ifndef COMPILE_UTILITY | |
184 | /************************************************* | |
185 | * Convert a number to base 62 format * | |
186 | *************************************************/ | |
187 | ||
188 | /* Convert a long integer into an ASCII base 62 string. For Cygwin the value of | |
189 | BASE_62 is actually 36. Always return exactly 6 characters plus zero, in a | |
190 | static area. | |
191 | ||
192 | Argument: a long integer | |
193 | Returns: pointer to base 62 string | |
194 | */ | |
195 | ||
196 | uschar * | |
197 | string_base62(unsigned long int value) | |
198 | { | |
199 | static uschar yield[7]; | |
200 | uschar *p = yield + sizeof(yield) - 1; | |
201 | *p = 0; | |
202 | while (p > yield) | |
203 | { | |
204 | *(--p) = base62_chars[value % BASE_62]; | |
205 | value /= BASE_62; | |
206 | } | |
207 | return yield; | |
208 | } | |
209 | #endif /* COMPILE_UTILITY */ | |
210 | ||
211 | ||
212 | ||
059ec3d9 PH |
213 | /************************************************* |
214 | * Interpret escape sequence * | |
215 | *************************************************/ | |
216 | ||
217 | /* This function is called from several places where escape sequences are to be | |
218 | interpreted in strings. | |
219 | ||
220 | Arguments: | |
221 | pp points a pointer to the initiating "\" in the string; | |
222 | the pointer gets updated to point to the final character | |
223 | Returns: the value of the character escape | |
224 | */ | |
225 | ||
226 | int | |
227 | string_interpret_escape(uschar **pp) | |
228 | { | |
3fb3c68d JH |
229 | #ifdef COMPILE_UTILITY |
230 | const uschar *hex_digits= CUS"0123456789abcdef"; | |
231 | #endif | |
059ec3d9 PH |
232 | int ch; |
233 | uschar *p = *pp; | |
234 | ch = *(++p); | |
235 | if (isdigit(ch) && ch != '8' && ch != '9') | |
236 | { | |
237 | ch -= '0'; | |
238 | if (isdigit(p[1]) && p[1] != '8' && p[1] != '9') | |
239 | { | |
240 | ch = ch * 8 + *(++p) - '0'; | |
241 | if (isdigit(p[1]) && p[1] != '8' && p[1] != '9') | |
242 | ch = ch * 8 + *(++p) - '0'; | |
243 | } | |
244 | } | |
245 | else switch(ch) | |
246 | { | |
c7396ac5 PP |
247 | case 'b': ch = '\b'; break; |
248 | case 'f': ch = '\f'; break; | |
059ec3d9 PH |
249 | case 'n': ch = '\n'; break; |
250 | case 'r': ch = '\r'; break; | |
251 | case 't': ch = '\t'; break; | |
c7396ac5 | 252 | case 'v': ch = '\v'; break; |
059ec3d9 PH |
253 | case 'x': |
254 | ch = 0; | |
255 | if (isxdigit(p[1])) | |
256 | { | |
257 | ch = ch * 16 + | |
258 | Ustrchr(hex_digits, tolower(*(++p))) - hex_digits; | |
259 | if (isxdigit(p[1])) ch = ch * 16 + | |
260 | Ustrchr(hex_digits, tolower(*(++p))) - hex_digits; | |
261 | } | |
262 | break; | |
263 | } | |
264 | *pp = p; | |
265 | return ch; | |
266 | } | |
059ec3d9 PH |
267 | |
268 | ||
269 | ||
270 | #ifndef COMPILE_UTILITY | |
271 | /************************************************* | |
272 | * Ensure string is printable * | |
273 | *************************************************/ | |
274 | ||
275 | /* This function is called for critical strings. It checks for any | |
276 | non-printing characters, and if any are found, it makes a new copy | |
277 | of the string with suitable escape sequences. It is most often called by the | |
278 | macro string_printing(), which sets allow_tab TRUE. | |
279 | ||
280 | Arguments: | |
281 | s the input string | |
282 | allow_tab TRUE to allow tab as a printing character | |
283 | ||
284 | Returns: string with non-printers encoded as printing sequences | |
285 | */ | |
286 | ||
287 | uschar * | |
288 | string_printing2(uschar *s, BOOL allow_tab) | |
289 | { | |
290 | int nonprintcount = 0; | |
291 | int length = 0; | |
292 | uschar *t = s; | |
293 | uschar *ss, *tt; | |
294 | ||
295 | while (*t != 0) | |
296 | { | |
297 | int c = *t++; | |
298 | if (!mac_isprint(c) || (!allow_tab && c == '\t')) nonprintcount++; | |
299 | length++; | |
300 | } | |
301 | ||
302 | if (nonprintcount == 0) return s; | |
303 | ||
304 | /* Get a new block of store guaranteed big enough to hold the | |
305 | expanded string. */ | |
306 | ||
36719342 | 307 | ss = store_get(length + nonprintcount * 3 + 1); |
059ec3d9 PH |
308 | |
309 | /* Copy everying, escaping non printers. */ | |
310 | ||
311 | t = s; | |
312 | tt = ss; | |
313 | ||
314 | while (*t != 0) | |
315 | { | |
316 | int c = *t; | |
317 | if (mac_isprint(c) && (allow_tab || c != '\t')) *tt++ = *t++; else | |
318 | { | |
319 | *tt++ = '\\'; | |
320 | switch (*t) | |
321 | { | |
322 | case '\n': *tt++ = 'n'; break; | |
323 | case '\r': *tt++ = 'r'; break; | |
324 | case '\b': *tt++ = 'b'; break; | |
325 | case '\v': *tt++ = 'v'; break; | |
326 | case '\f': *tt++ = 'f'; break; | |
327 | case '\t': *tt++ = 't'; break; | |
328 | default: sprintf(CS tt, "%03o", *t); tt += 3; break; | |
329 | } | |
330 | t++; | |
331 | } | |
332 | } | |
333 | *tt = 0; | |
c7396ac5 PP |
334 | return ss; |
335 | } | |
79fe97d8 PP |
336 | #endif /* COMPILE_UTILITY */ |
337 | ||
c7396ac5 PP |
338 | /************************************************* |
339 | * Undo printing escapes in string * | |
340 | *************************************************/ | |
341 | ||
342 | /* This function is the reverse of string_printing2. It searches for | |
343 | backslash characters and if any are found, it makes a new copy of the | |
344 | string with escape sequences parsed. Otherwise it returns the original | |
345 | string. | |
346 | ||
347 | Arguments: | |
348 | s the input string | |
349 | ||
350 | Returns: string with printing escapes parsed back | |
351 | */ | |
352 | ||
353 | uschar * | |
354 | string_unprinting(uschar *s) | |
355 | { | |
356 | uschar *p, *q, *r, *ss; | |
357 | int len, off; | |
358 | ||
359 | p = Ustrchr(s, '\\'); | |
360 | if (!p) return s; | |
361 | ||
362 | len = Ustrlen(s) + 1; | |
363 | ss = store_get(len); | |
364 | ||
365 | q = ss; | |
366 | off = p - s; | |
367 | if (off) | |
368 | { | |
369 | memcpy(q, s, off); | |
370 | q += off; | |
371 | } | |
372 | ||
373 | while (*p) | |
374 | { | |
375 | if (*p == '\\') | |
376 | { | |
823ad74f JH |
377 | *q++ = string_interpret_escape(&p); |
378 | p++; | |
c7396ac5 PP |
379 | } |
380 | else | |
381 | { | |
382 | r = Ustrchr(p, '\\'); | |
383 | if (!r) | |
384 | { | |
385 | off = Ustrlen(p); | |
386 | memcpy(q, p, off); | |
387 | p += off; | |
388 | q += off; | |
389 | break; | |
390 | } | |
391 | else | |
392 | { | |
393 | off = r - p; | |
394 | memcpy(q, p, off); | |
395 | q += off; | |
396 | p = r; | |
397 | } | |
398 | } | |
399 | } | |
400 | *q = '\0'; | |
401 | ||
059ec3d9 PH |
402 | return ss; |
403 | } | |
059ec3d9 PH |
404 | |
405 | ||
406 | ||
407 | ||
408 | /************************************************* | |
409 | * Copy and save string * | |
410 | *************************************************/ | |
411 | ||
412 | /* This function assumes that memcpy() is faster than strcpy(). | |
413 | ||
414 | Argument: string to copy | |
415 | Returns: copy of string in new store | |
416 | */ | |
417 | ||
418 | uschar * | |
3f0945ff | 419 | string_copy(const uschar *s) |
059ec3d9 PH |
420 | { |
421 | int len = Ustrlen(s) + 1; | |
422 | uschar *ss = store_get(len); | |
423 | memcpy(ss, s, len); | |
424 | return ss; | |
425 | } | |
426 | ||
427 | ||
428 | ||
429 | /************************************************* | |
430 | * Copy and save string in malloc'd store * | |
431 | *************************************************/ | |
432 | ||
433 | /* This function assumes that memcpy() is faster than strcpy(). | |
434 | ||
435 | Argument: string to copy | |
436 | Returns: copy of string in new store | |
437 | */ | |
438 | ||
439 | uschar * | |
440 | string_copy_malloc(uschar *s) | |
441 | { | |
442 | int len = Ustrlen(s) + 1; | |
443 | uschar *ss = store_malloc(len); | |
444 | memcpy(ss, s, len); | |
445 | return ss; | |
446 | } | |
447 | ||
448 | ||
449 | ||
450 | /************************************************* | |
451 | * Copy, lowercase and save string * | |
452 | *************************************************/ | |
453 | ||
454 | /* | |
455 | Argument: string to copy | |
456 | Returns: copy of string in new store, with letters lowercased | |
457 | */ | |
458 | ||
459 | uschar * | |
1dc92d5a | 460 | string_copylc(const uschar *s) |
059ec3d9 PH |
461 | { |
462 | uschar *ss = store_get(Ustrlen(s) + 1); | |
463 | uschar *p = ss; | |
464 | while (*s != 0) *p++ = tolower(*s++); | |
465 | *p = 0; | |
466 | return ss; | |
467 | } | |
468 | ||
469 | ||
470 | ||
471 | /************************************************* | |
472 | * Copy and save string, given length * | |
473 | *************************************************/ | |
474 | ||
475 | /* It is assumed the data contains no zeros. A zero is added | |
476 | onto the end. | |
477 | ||
478 | Arguments: | |
479 | s string to copy | |
480 | n number of characters | |
481 | ||
482 | Returns: copy of string in new store | |
483 | */ | |
484 | ||
485 | uschar * | |
1dc92d5a | 486 | string_copyn(const uschar *s, int n) |
059ec3d9 PH |
487 | { |
488 | uschar *ss = store_get(n + 1); | |
489 | Ustrncpy(ss, s, n); | |
490 | ss[n] = 0; | |
491 | return ss; | |
492 | } | |
493 | ||
494 | ||
495 | /************************************************* | |
496 | * Copy, lowercase, and save string, given length * | |
497 | *************************************************/ | |
498 | ||
499 | /* It is assumed the data contains no zeros. A zero is added | |
500 | onto the end. | |
501 | ||
502 | Arguments: | |
503 | s string to copy | |
504 | n number of characters | |
505 | ||
506 | Returns: copy of string in new store, with letters lowercased | |
507 | */ | |
508 | ||
509 | uschar * | |
510 | string_copynlc(uschar *s, int n) | |
511 | { | |
512 | uschar *ss = store_get(n + 1); | |
513 | uschar *p = ss; | |
514 | while (n-- > 0) *p++ = tolower(*s++); | |
515 | *p = 0; | |
516 | return ss; | |
517 | } | |
518 | ||
519 | ||
520 | ||
e28326d8 PH |
521 | /************************************************* |
522 | * Copy string if long, inserting newlines * | |
523 | *************************************************/ | |
524 | ||
525 | /* If the given string is longer than 75 characters, it is copied, and within | |
526 | the copy, certain space characters are converted into newlines. | |
527 | ||
528 | Argument: pointer to the string | |
529 | Returns: pointer to the possibly altered string | |
530 | */ | |
531 | ||
532 | uschar * | |
533 | string_split_message(uschar *msg) | |
534 | { | |
535 | uschar *s, *ss; | |
536 | ||
537 | if (msg == NULL || Ustrlen(msg) <= 75) return msg; | |
538 | s = ss = msg = string_copy(msg); | |
539 | ||
540 | for (;;) | |
541 | { | |
542 | int i = 0; | |
543 | while (i < 75 && *ss != 0 && *ss != '\n') ss++, i++; | |
544 | if (*ss == 0) break; | |
545 | if (*ss == '\n') | |
546 | s = ++ss; | |
547 | else | |
548 | { | |
549 | uschar *t = ss + 1; | |
550 | uschar *tt = NULL; | |
551 | while (--t > s + 35) | |
552 | { | |
553 | if (*t == ' ') | |
554 | { | |
555 | if (t[-1] == ':') { tt = t; break; } | |
556 | if (tt == NULL) tt = t; | |
557 | } | |
558 | } | |
559 | ||
560 | if (tt == NULL) /* Can't split behind - try ahead */ | |
561 | { | |
562 | t = ss + 1; | |
563 | while (*t != 0) | |
564 | { | |
565 | if (*t == ' ' || *t == '\n') | |
566 | { tt = t; break; } | |
567 | t++; | |
568 | } | |
569 | } | |
570 | ||
571 | if (tt == NULL) break; /* Can't find anywhere to split */ | |
572 | *tt = '\n'; | |
573 | s = ss = tt+1; | |
574 | } | |
575 | } | |
576 | ||
577 | return msg; | |
578 | } | |
579 | ||
580 | ||
581 | ||
059ec3d9 PH |
582 | /************************************************* |
583 | * Copy returned DNS domain name, de-escaping * | |
584 | *************************************************/ | |
585 | ||
586 | /* If a domain name contains top-bit characters, some resolvers return | |
587 | the fully qualified name with those characters turned into escapes. The | |
588 | convention is a backslash followed by _decimal_ digits. We convert these | |
589 | back into the original binary values. This will be relevant when | |
590 | allow_utf8_domains is set true and UTF-8 characters are used in domain | |
591 | names. Backslash can also be used to escape other characters, though we | |
592 | shouldn't come across them in domain names. | |
593 | ||
594 | Argument: the domain name string | |
595 | Returns: copy of string in new store, de-escaped | |
596 | */ | |
597 | ||
598 | uschar * | |
599 | string_copy_dnsdomain(uschar *s) | |
600 | { | |
601 | uschar *yield; | |
602 | uschar *ss = yield = store_get(Ustrlen(s) + 1); | |
603 | ||
604 | while (*s != 0) | |
605 | { | |
606 | if (*s != '\\') | |
607 | { | |
608 | *ss++ = *s++; | |
609 | } | |
610 | else if (isdigit(s[1])) | |
611 | { | |
612 | *ss++ = (s[1] - '0')*100 + (s[2] - '0')*10 + s[3] - '0'; | |
613 | s += 4; | |
614 | } | |
615 | else if (*(++s) != 0) | |
616 | { | |
617 | *ss++ = *s++; | |
618 | } | |
619 | } | |
620 | ||
621 | *ss = 0; | |
622 | return yield; | |
623 | } | |
624 | ||
625 | ||
626 | #ifndef COMPILE_UTILITY | |
627 | /************************************************* | |
628 | * Copy space-terminated or quoted string * | |
629 | *************************************************/ | |
630 | ||
631 | /* This function copies from a string until its end, or until whitespace is | |
632 | encountered, unless the string begins with a double quote, in which case the | |
633 | terminating quote is sought, and escaping within the string is done. The length | |
634 | of a de-quoted string can be no longer than the original, since escaping always | |
635 | turns n characters into 1 character. | |
636 | ||
637 | Argument: pointer to the pointer to the first character, which gets updated | |
638 | Returns: the new string | |
639 | */ | |
640 | ||
641 | uschar * | |
642 | string_dequote(uschar **sptr) | |
643 | { | |
644 | uschar *s = *sptr; | |
645 | uschar *t, *yield; | |
646 | ||
647 | /* First find the end of the string */ | |
648 | ||
649 | if (*s != '\"') | |
650 | { | |
651 | while (*s != 0 && !isspace(*s)) s++; | |
652 | } | |
653 | else | |
654 | { | |
655 | s++; | |
656 | while (*s != 0 && *s != '\"') | |
657 | { | |
658 | if (*s == '\\') (void)string_interpret_escape(&s); | |
659 | s++; | |
660 | } | |
661 | if (*s != 0) s++; | |
662 | } | |
663 | ||
664 | /* Get enough store to copy into */ | |
665 | ||
666 | t = yield = store_get(s - *sptr + 1); | |
667 | s = *sptr; | |
668 | ||
669 | /* Do the copy */ | |
670 | ||
671 | if (*s != '\"') | |
672 | { | |
673 | while (*s != 0 && !isspace(*s)) *t++ = *s++; | |
674 | } | |
675 | else | |
676 | { | |
677 | s++; | |
678 | while (*s != 0 && *s != '\"') | |
679 | { | |
680 | if (*s == '\\') *t++ = string_interpret_escape(&s); | |
681 | else *t++ = *s; | |
682 | s++; | |
683 | } | |
684 | if (*s != 0) s++; | |
685 | } | |
686 | ||
687 | /* Update the pointer and return the terminated copy */ | |
688 | ||
689 | *sptr = s; | |
690 | *t = 0; | |
691 | return yield; | |
692 | } | |
693 | #endif /* COMPILE_UTILITY */ | |
694 | ||
695 | ||
696 | ||
697 | /************************************************* | |
698 | * Format a string and save it * | |
699 | *************************************************/ | |
700 | ||
701 | /* The formatting is done by string_format, which checks the length of | |
702 | everything. | |
703 | ||
704 | Arguments: | |
705 | format a printf() format - deliberately char * rather than uschar * | |
706 | because it will most usually be a literal string | |
707 | ... arguments for format | |
708 | ||
709 | Returns: pointer to fresh piece of store containing sprintf'ed string | |
710 | */ | |
711 | ||
712 | uschar * | |
1ba28e2b | 713 | string_sprintf(const char *format, ...) |
059ec3d9 PH |
714 | { |
715 | va_list ap; | |
716 | uschar buffer[STRING_SPRINTF_BUFFER_SIZE]; | |
717 | va_start(ap, format); | |
718 | if (!string_vformat(buffer, sizeof(buffer), format, ap)) | |
719 | log_write(0, LOG_MAIN|LOG_PANIC_DIE, | |
a37a8eec TL |
720 | "string_sprintf expansion was longer than " SIZE_T_FMT " (%s)", |
721 | sizeof(buffer), format); | |
059ec3d9 PH |
722 | va_end(ap); |
723 | return string_copy(buffer); | |
724 | } | |
725 | ||
726 | ||
727 | ||
728 | /************************************************* | |
729 | * Case-independent strncmp() function * | |
730 | *************************************************/ | |
731 | ||
732 | /* | |
733 | Arguments: | |
734 | s first string | |
735 | t second string | |
736 | n number of characters to compare | |
737 | ||
738 | Returns: < 0, = 0, or > 0, according to the comparison | |
739 | */ | |
740 | ||
741 | int | |
1ba28e2b | 742 | strncmpic(const uschar *s, const uschar *t, int n) |
059ec3d9 PH |
743 | { |
744 | while (n--) | |
745 | { | |
746 | int c = tolower(*s++) - tolower(*t++); | |
747 | if (c) return c; | |
748 | } | |
749 | return 0; | |
750 | } | |
751 | ||
752 | ||
753 | /************************************************* | |
754 | * Case-independent strcmp() function * | |
755 | *************************************************/ | |
756 | ||
757 | /* | |
758 | Arguments: | |
759 | s first string | |
760 | t second string | |
761 | ||
762 | Returns: < 0, = 0, or > 0, according to the comparison | |
763 | */ | |
764 | ||
765 | int | |
1ba28e2b | 766 | strcmpic(const uschar *s, const uschar *t) |
059ec3d9 PH |
767 | { |
768 | while (*s != 0) | |
769 | { | |
770 | int c = tolower(*s++) - tolower(*t++); | |
771 | if (c != 0) return c; | |
772 | } | |
773 | return *t; | |
774 | } | |
775 | ||
776 | ||
777 | /************************************************* | |
778 | * Case-independent strstr() function * | |
779 | *************************************************/ | |
780 | ||
781 | /* The third argument specifies whether whitespace is required | |
782 | to follow the matched string. | |
783 | ||
784 | Arguments: | |
785 | s string to search | |
786 | t substring to search for | |
787 | space_follows if TRUE, match only if whitespace follows | |
788 | ||
789 | Returns: pointer to substring in string, or NULL if not found | |
790 | */ | |
791 | ||
792 | uschar * | |
793 | strstric(uschar *s, uschar *t, BOOL space_follows) | |
794 | { | |
795 | uschar *p = t; | |
796 | uschar *yield = NULL; | |
797 | int cl = tolower(*p); | |
798 | int cu = toupper(*p); | |
799 | ||
800 | while (*s) | |
801 | { | |
802 | if (*s == cl || *s == cu) | |
803 | { | |
804 | if (yield == NULL) yield = s; | |
805 | if (*(++p) == 0) | |
806 | { | |
807 | if (!space_follows || s[1] == ' ' || s[1] == '\n' ) return yield; | |
808 | yield = NULL; | |
809 | p = t; | |
810 | } | |
811 | cl = tolower(*p); | |
812 | cu = toupper(*p); | |
813 | s++; | |
814 | } | |
815 | else if (yield != NULL) | |
816 | { | |
817 | yield = NULL; | |
818 | p = t; | |
819 | cl = tolower(*p); | |
820 | cu = toupper(*p); | |
821 | } | |
822 | else s++; | |
823 | } | |
824 | return NULL; | |
825 | } | |
826 | ||
827 | ||
828 | ||
829 | #ifndef COMPILE_UTILITY | |
830 | /************************************************* | |
831 | * Get next string from separated list * | |
832 | *************************************************/ | |
833 | ||
834 | /* Leading and trailing space is removed from each item. The separator in the | |
835 | list is controlled by the int pointed to by the separator argument as follows: | |
836 | ||
ec95d1a6 PH |
837 | If the value is > 0 it is used as the separator. This is typically used for |
838 | sublists such as slash-separated options. The value is always a printing | |
839 | character. | |
840 | ||
841 | (If the value is actually > UCHAR_MAX there is only one item in the list. | |
059ec3d9 PH |
842 | This is used for some cases when called via functions that sometimes |
843 | plough through lists, and sometimes are given single items.) | |
059ec3d9 | 844 | |
ec95d1a6 PH |
845 | If the value is <= 0, the string is inspected for a leading <x, where x is an |
846 | ispunct() or an iscntrl() character. If found, x is used as the separator. If | |
847 | not found: | |
848 | ||
849 | (a) if separator == 0, ':' is used | |
850 | (b) if separator <0, -separator is used | |
851 | ||
852 | In all cases the value of the separator that is used is written back to the | |
853 | int so that it is used on subsequent calls as we progress through the list. | |
854 | ||
855 | A literal ispunct() separator can be represented in an item by doubling, but | |
856 | there is no way to include an iscntrl() separator as part of the data. | |
059ec3d9 PH |
857 | |
858 | Arguments: | |
859 | listptr points to a pointer to the current start of the list; the | |
860 | pointer gets updated to point after the end of the next item | |
861 | separator a pointer to the separator character in an int (see above) | |
862 | buffer where to put a copy of the next string in the list; or | |
863 | NULL if the next string is returned in new memory | |
864 | buflen when buffer is not NULL, the size of buffer; otherwise ignored | |
865 | ||
866 | Returns: pointer to buffer, containing the next substring, | |
867 | or NULL if no more substrings | |
868 | */ | |
869 | ||
870 | uschar * | |
871 | string_nextinlist(uschar **listptr, int *separator, uschar *buffer, int buflen) | |
872 | { | |
059ec3d9 PH |
873 | register int sep = *separator; |
874 | register uschar *s = *listptr; | |
ec95d1a6 | 875 | BOOL sep_is_special; |
059ec3d9 PH |
876 | |
877 | if (s == NULL) return NULL; | |
ec95d1a6 PH |
878 | |
879 | /* This allows for a fixed specified separator to be an iscntrl() character, | |
880 | but at the time of implementation, this is never the case. However, it's best | |
881 | to be conservative. */ | |
882 | ||
883 | while (isspace(*s) && *s != sep) s++; | |
884 | ||
885 | /* A change of separator is permitted, so look for a leading '<' followed by an | |
886 | allowed character. */ | |
059ec3d9 PH |
887 | |
888 | if (sep <= 0) | |
889 | { | |
ec95d1a6 | 890 | if (*s == '<' && (ispunct(s[1]) || iscntrl(s[1]))) |
059ec3d9 PH |
891 | { |
892 | sep = s[1]; | |
893 | s += 2; | |
ec95d1a6 | 894 | while (isspace(*s) && *s != sep) s++; |
059ec3d9 PH |
895 | } |
896 | else | |
897 | { | |
898 | sep = (sep == 0)? ':' : -sep; | |
899 | } | |
900 | *separator = sep; | |
901 | } | |
902 | ||
ec95d1a6 PH |
903 | /* An empty string has no list elements */ |
904 | ||
059ec3d9 PH |
905 | if (*s == 0) return NULL; |
906 | ||
ec95d1a6 PH |
907 | /* Note whether whether or not the separator is an iscntrl() character. */ |
908 | ||
909 | sep_is_special = iscntrl(sep); | |
910 | ||
059ec3d9 PH |
911 | /* Handle the case when a buffer is provided. */ |
912 | ||
913 | if (buffer != NULL) | |
914 | { | |
ec95d1a6 | 915 | register int p = 0; |
059ec3d9 PH |
916 | for (; *s != 0; s++) |
917 | { | |
ec95d1a6 | 918 | if (*s == sep && (*(++s) != sep || sep_is_special)) break; |
059ec3d9 PH |
919 | if (p < buflen - 1) buffer[p++] = *s; |
920 | } | |
921 | while (p > 0 && isspace(buffer[p-1])) p--; | |
922 | buffer[p] = 0; | |
923 | } | |
924 | ||
925 | /* Handle the case when a buffer is not provided. */ | |
926 | ||
927 | else | |
928 | { | |
ec95d1a6 PH |
929 | int size = 0; |
930 | int ptr = 0; | |
931 | uschar *ss; | |
932 | ||
059ec3d9 | 933 | /* We know that *s != 0 at this point. However, it might be pointing to a |
ec95d1a6 PH |
934 | separator, which could indicate an empty string, or (if an ispunct() |
935 | character) could be doubled to indicate a separator character as data at the | |
936 | start of a string. Avoid getting working memory for an empty item. */ | |
059ec3d9 PH |
937 | |
938 | if (*s == sep) | |
939 | { | |
940 | s++; | |
ec95d1a6 PH |
941 | if (*s != sep || sep_is_special) |
942 | { | |
943 | *listptr = s; | |
944 | return string_copy(US""); | |
945 | } | |
059ec3d9 PH |
946 | } |
947 | ||
ec95d1a6 PH |
948 | /* Not an empty string; the first character is guaranteed to be a data |
949 | character. */ | |
950 | ||
951 | for (;;) | |
059ec3d9 | 952 | { |
ec95d1a6 PH |
953 | for (ss = s + 1; *ss != 0 && *ss != sep; ss++); |
954 | buffer = string_cat(buffer, &size, &ptr, s, ss-s); | |
955 | s = ss; | |
956 | if (*s == 0 || *(++s) != sep || sep_is_special) break; | |
059ec3d9 | 957 | } |
ec95d1a6 PH |
958 | while (ptr > 0 && isspace(buffer[ptr-1])) ptr--; |
959 | buffer[ptr] = 0; | |
059ec3d9 PH |
960 | } |
961 | ||
962 | /* Update the current pointer and return the new string */ | |
963 | ||
964 | *listptr = s; | |
965 | return buffer; | |
966 | } | |
967 | #endif /* COMPILE_UTILITY */ | |
968 | ||
969 | ||
76146973 JH |
970 | #ifndef COMPILE_UTILITY |
971 | /************************************************ | |
972 | * Add element to seperated list * | |
973 | ************************************************/ | |
974 | /* This function is used to build a list, returning | |
975 | an allocated null-terminated growable string. The | |
976 | given element has any embedded seperator characters | |
977 | doubled. | |
978 | ||
979 | Arguments: | |
980 | list points to the start of the list that is being built, or NULL | |
981 | if this is a new list that has no contents yet | |
982 | sep list seperator charactoer | |
983 | ele new lement to be appended to the list | |
984 | ||
985 | Returns: pointer to the start of the list, changed if copied for expansion. | |
986 | */ | |
987 | ||
988 | uschar * | |
989 | string_append_listele(uschar * list, uschar sep, const uschar * ele) | |
990 | { | |
991 | uschar * new = NULL; | |
992 | int sz = 0, off = 0; | |
993 | uschar * sp; | |
994 | ||
995 | if (list) | |
996 | { | |
997 | new = string_cat(new, &sz, &off, list, Ustrlen(list)); | |
998 | new = string_cat(new, &sz, &off, &sep, 1); | |
999 | } | |
1000 | ||
e3dd1d67 | 1001 | while((sp = Ustrchr(ele, sep))) |
76146973 JH |
1002 | { |
1003 | new = string_cat(new, &sz, &off, ele, sp-ele+1); | |
1004 | new = string_cat(new, &sz, &off, &sep, 1); | |
1005 | ele = sp+1; | |
1006 | } | |
1007 | new = string_cat(new, &sz, &off, ele, Ustrlen(ele)); | |
1008 | new[off] = '\0'; | |
1009 | return new; | |
1010 | } | |
1011 | #endif /* COMPILE_UTILITY */ | |
1012 | ||
1013 | ||
059ec3d9 PH |
1014 | |
1015 | #ifndef COMPILE_UTILITY | |
1016 | /************************************************* | |
1017 | * Add chars to string * | |
1018 | *************************************************/ | |
1019 | ||
1020 | /* This function is used when building up strings of unknown length. Room is | |
1021 | always left for a terminating zero to be added to the string that is being | |
1022 | built. This function does not require the string that is being added to be NUL | |
1023 | terminated, because the number of characters to add is given explicitly. It is | |
1024 | sometimes called to extract parts of other strings. | |
1025 | ||
1026 | Arguments: | |
1027 | string points to the start of the string that is being built, or NULL | |
1028 | if this is a new string that has no contents yet | |
1029 | size points to a variable that holds the current capacity of the memory | |
1030 | block (updated if changed) | |
1031 | ptr points to a variable that holds the offset at which to add | |
1032 | characters, updated to the new offset | |
1033 | s points to characters to add | |
1034 | count count of characters to add; must not exceed the length of s, if s | |
1035 | is a C string | |
1036 | ||
1037 | If string is given as NULL, *size and *ptr should both be zero. | |
1038 | ||
1039 | Returns: pointer to the start of the string, changed if copied for expansion. | |
1040 | Note that a NUL is not added, though space is left for one. This is | |
1041 | because string_cat() is often called multiple times to build up a | |
1042 | string - there's no point adding the NUL till the end. | |
1043 | */ | |
1044 | ||
1045 | uschar * | |
1046 | string_cat(uschar *string, int *size, int *ptr, const uschar *s, int count) | |
1047 | { | |
1048 | int p = *ptr; | |
1049 | ||
1050 | if (p + count >= *size) | |
1051 | { | |
1052 | int oldsize = *size; | |
1053 | ||
1054 | /* Mostly, string_cat() is used to build small strings of a few hundred | |
1055 | characters at most. There are times, however, when the strings are very much | |
1056 | longer (for example, a lookup that returns a vast number of alias addresses). | |
1057 | To try to keep things reasonable, we use increments whose size depends on the | |
1058 | existing length of the string. */ | |
1059 | ||
1060 | int inc = (oldsize < 4096)? 100 : 1024; | |
1061 | while (*size <= p + count) *size += inc; | |
1062 | ||
1063 | /* New string */ | |
1064 | ||
1065 | if (string == NULL) string = store_get(*size); | |
1066 | ||
1067 | /* Try to extend an existing allocation. If the result of calling | |
1068 | store_extend() is false, either there isn't room in the current memory block, | |
1069 | or this string is not the top item on the dynamic store stack. We then have | |
1070 | to get a new chunk of store and copy the old string. When building large | |
1071 | strings, it is helpful to call store_release() on the old string, to release | |
1072 | memory blocks that have become empty. (The block will be freed if the string | |
1073 | is at its start.) However, we can do this only if we know that the old string | |
1074 | was the last item on the dynamic memory stack. This is the case if it matches | |
1075 | store_last_get. */ | |
1076 | ||
1077 | else if (!store_extend(string, oldsize, *size)) | |
1078 | { | |
1079 | BOOL release_ok = store_last_get[store_pool] == string; | |
1080 | uschar *newstring = store_get(*size); | |
1081 | memcpy(newstring, string, p); | |
1082 | if (release_ok) store_release(string); | |
1083 | string = newstring; | |
1084 | } | |
1085 | } | |
1086 | ||
1087 | /* Because we always specify the exact number of characters to copy, we can | |
1088 | use memcpy(), which is likely to be more efficient than strncopy() because the | |
1089 | latter has to check for zero bytes. */ | |
1090 | ||
1091 | memcpy(string + p, s, count); | |
1092 | *ptr = p + count; | |
1093 | return string; | |
1094 | } | |
1095 | #endif /* COMPILE_UTILITY */ | |
1096 | ||
1097 | ||
1098 | ||
1099 | #ifndef COMPILE_UTILITY | |
1100 | /************************************************* | |
1101 | * Append strings to another string * | |
1102 | *************************************************/ | |
1103 | ||
1104 | /* This function can be used to build a string from many other strings. | |
1105 | It calls string_cat() to do the dirty work. | |
1106 | ||
1107 | Arguments: | |
1108 | string points to the start of the string that is being built, or NULL | |
1109 | if this is a new string that has no contents yet | |
1110 | size points to a variable that holds the current capacity of the memory | |
1111 | block (updated if changed) | |
1112 | ptr points to a variable that holds the offset at which to add | |
1113 | characters, updated to the new offset | |
1114 | count the number of strings to append | |
1115 | ... "count" uschar* arguments, which must be valid zero-terminated | |
1116 | C strings | |
1117 | ||
1118 | Returns: pointer to the start of the string, changed if copied for expansion. | |
1119 | The string is not zero-terminated - see string_cat() above. | |
1120 | */ | |
1121 | ||
1122 | uschar * | |
1123 | string_append(uschar *string, int *size, int *ptr, int count, ...) | |
1124 | { | |
1125 | va_list ap; | |
1126 | int i; | |
1127 | ||
1128 | va_start(ap, count); | |
1129 | for (i = 0; i < count; i++) | |
1130 | { | |
1131 | uschar *t = va_arg(ap, uschar *); | |
1132 | string = string_cat(string, size, ptr, t, Ustrlen(t)); | |
1133 | } | |
1134 | va_end(ap); | |
1135 | ||
1136 | return string; | |
1137 | } | |
1138 | #endif | |
1139 | ||
1140 | ||
1141 | ||
1142 | /************************************************* | |
1143 | * Format a string with length checks * | |
1144 | *************************************************/ | |
1145 | ||
1146 | /* This function is used to format a string with checking of the length of the | |
1147 | output for all conversions. It protects Exim from absent-mindedness when | |
1148 | calling functions like debug_printf and string_sprintf, and elsewhere. There | |
1149 | are two different entry points to what is actually the same function, depending | |
1150 | on whether the variable length list of data arguments are given explicitly or | |
1151 | as a va_list item. | |
1152 | ||
1153 | The formats are the usual printf() ones, with some omissions (never used) and | |
0d7eb84a PH |
1154 | two additions for strings: %S forces lower case, and %#s or %#S prints nothing |
1155 | for a NULL string. Without the # "NULL" is printed (useful in debugging). There | |
f1e5fef5 | 1156 | is also the addition of %D and %M, which insert the date in the form used for |
059ec3d9 PH |
1157 | datestamped log files. |
1158 | ||
1159 | Arguments: | |
1160 | buffer a buffer in which to put the formatted string | |
1161 | buflen the length of the buffer | |
1162 | format the format string - deliberately char * and not uschar * | |
1163 | ... or ap variable list of supplementary arguments | |
1164 | ||
1165 | Returns: TRUE if the result fitted in the buffer | |
1166 | */ | |
1167 | ||
1168 | BOOL | |
1ba28e2b | 1169 | string_format(uschar *buffer, int buflen, const char *format, ...) |
059ec3d9 PH |
1170 | { |
1171 | BOOL yield; | |
1172 | va_list ap; | |
1173 | va_start(ap, format); | |
1174 | yield = string_vformat(buffer, buflen, format, ap); | |
1175 | va_end(ap); | |
1176 | return yield; | |
1177 | } | |
1178 | ||
1179 | ||
1180 | BOOL | |
1ba28e2b | 1181 | string_vformat(uschar *buffer, int buflen, const char *format, va_list ap) |
059ec3d9 | 1182 | { |
91a246f6 PP |
1183 | /* We assume numbered ascending order, C does not guarantee that */ |
1184 | enum { L_NORMAL=1, L_SHORT=2, L_LONG=3, L_LONGLONG=4, L_LONGDOUBLE=5, L_SIZE=6 }; | |
b1c749bb | 1185 | |
059ec3d9 PH |
1186 | BOOL yield = TRUE; |
1187 | int width, precision; | |
1ba28e2b | 1188 | const char *fp = format; /* Deliberately not unsigned */ |
059ec3d9 PH |
1189 | uschar *p = buffer; |
1190 | uschar *last = buffer + buflen - 1; | |
1191 | ||
1192 | string_datestamp_offset = -1; /* Datestamp not inserted */ | |
f1e5fef5 PP |
1193 | string_datestamp_length = 0; /* Datestamp not inserted */ |
1194 | string_datestamp_type = 0; /* Datestamp not inserted */ | |
059ec3d9 PH |
1195 | |
1196 | /* Scan the format and handle the insertions */ | |
1197 | ||
1198 | while (*fp != 0) | |
1199 | { | |
b1c749bb | 1200 | int length = L_NORMAL; |
059ec3d9 PH |
1201 | int *nptr; |
1202 | int slen; | |
1ba28e2b PP |
1203 | const char *null = "NULL"; /* ) These variables */ |
1204 | const char *item_start, *s; /* ) are deliberately */ | |
059ec3d9 PH |
1205 | char newformat[16]; /* ) not unsigned */ |
1206 | ||
1207 | /* Non-% characters just get copied verbatim */ | |
1208 | ||
1209 | if (*fp != '%') | |
1210 | { | |
1211 | if (p >= last) { yield = FALSE; break; } | |
1212 | *p++ = (uschar)*fp++; | |
1213 | continue; | |
1214 | } | |
1215 | ||
1216 | /* Deal with % characters. Pick off the width and precision, for checking | |
1217 | strings, skipping over the flag and modifier characters. */ | |
1218 | ||
1219 | item_start = fp; | |
1220 | width = precision = -1; | |
1221 | ||
1222 | if (strchr("-+ #0", *(++fp)) != NULL) | |
1223 | { | |
1224 | if (*fp == '#') null = ""; | |
1225 | fp++; | |
1226 | } | |
1227 | ||
1228 | if (isdigit((uschar)*fp)) | |
1229 | { | |
1230 | width = *fp++ - '0'; | |
1231 | while (isdigit((uschar)*fp)) width = width * 10 + *fp++ - '0'; | |
1232 | } | |
1233 | else if (*fp == '*') | |
1234 | { | |
1235 | width = va_arg(ap, int); | |
1236 | fp++; | |
1237 | } | |
1238 | ||
1239 | if (*fp == '.') | |
1240 | { | |
1241 | if (*(++fp) == '*') | |
1242 | { | |
1243 | precision = va_arg(ap, int); | |
1244 | fp++; | |
1245 | } | |
1246 | else | |
1247 | { | |
1248 | precision = 0; | |
1249 | while (isdigit((uschar)*fp)) | |
1250 | precision = precision*10 + *fp++ - '0'; | |
1251 | } | |
1252 | } | |
1253 | ||
91a246f6 | 1254 | /* Skip over 'h', 'L', 'l', 'll' and 'z', remembering the item length */ |
b1c749bb PH |
1255 | |
1256 | if (*fp == 'h') | |
1257 | { fp++; length = L_SHORT; } | |
1258 | else if (*fp == 'L') | |
1259 | { fp++; length = L_LONGDOUBLE; } | |
1260 | else if (*fp == 'l') | |
1261 | { | |
1262 | if (fp[1] == 'l') | |
1263 | { | |
1264 | fp += 2; | |
1265 | length = L_LONGLONG; | |
1266 | } | |
1267 | else | |
1268 | { | |
1269 | fp++; | |
1270 | length = L_LONG; | |
1271 | } | |
1272 | } | |
91a246f6 PP |
1273 | else if (*fp == 'z') |
1274 | { fp++; length = L_SIZE; } | |
059ec3d9 PH |
1275 | |
1276 | /* Handle each specific format type. */ | |
1277 | ||
1278 | switch (*fp++) | |
1279 | { | |
1280 | case 'n': | |
1281 | nptr = va_arg(ap, int *); | |
1282 | *nptr = p - buffer; | |
1283 | break; | |
1284 | ||
1285 | case 'd': | |
1286 | case 'o': | |
1287 | case 'u': | |
1288 | case 'x': | |
1289 | case 'X': | |
1549ea3b PH |
1290 | if (p >= last - ((length > L_LONG)? 24 : 12)) |
1291 | { yield = FALSE; goto END_FORMAT; } | |
059ec3d9 PH |
1292 | strncpy(newformat, item_start, fp - item_start); |
1293 | newformat[fp - item_start] = 0; | |
b1c749bb PH |
1294 | |
1295 | /* Short int is promoted to int when passing through ..., so we must use | |
1296 | int for va_arg(). */ | |
1297 | ||
1298 | switch(length) | |
1299 | { | |
1300 | case L_SHORT: | |
1301 | case L_NORMAL: sprintf(CS p, newformat, va_arg(ap, int)); break; | |
1302 | case L_LONG: sprintf(CS p, newformat, va_arg(ap, long int)); break; | |
c6c2dc1d | 1303 | case L_LONGLONG: sprintf(CS p, newformat, va_arg(ap, LONGLONG_T)); break; |
91a246f6 | 1304 | case L_SIZE: sprintf(CS p, newformat, va_arg(ap, size_t)); break; |
b1c749bb | 1305 | } |
059ec3d9 PH |
1306 | while (*p) p++; |
1307 | break; | |
1308 | ||
1309 | case 'p': | |
1310 | if (p >= last - 24) { yield = FALSE; goto END_FORMAT; } | |
1311 | strncpy(newformat, item_start, fp - item_start); | |
1312 | newformat[fp - item_start] = 0; | |
1313 | sprintf(CS p, newformat, va_arg(ap, void *)); | |
1314 | while (*p) p++; | |
1315 | break; | |
1316 | ||
1317 | /* %f format is inherently insecure if the numbers that it may be | |
870f6ba8 TF |
1318 | handed are unknown (e.g. 1e300). However, in Exim, %f is used for |
1319 | printing load averages, and these are actually stored as integers | |
1320 | (load average * 1000) so the size of the numbers is constrained. | |
1321 | It is also used for formatting sending rates, where the simplicity | |
1322 | of the format prevents overflow. */ | |
059ec3d9 PH |
1323 | |
1324 | case 'f': | |
1325 | case 'e': | |
1326 | case 'E': | |
1327 | case 'g': | |
1328 | case 'G': | |
1329 | if (precision < 0) precision = 6; | |
1330 | if (p >= last - precision - 8) { yield = FALSE; goto END_FORMAT; } | |
1331 | strncpy(newformat, item_start, fp - item_start); | |
1332 | newformat[fp-item_start] = 0; | |
b1c749bb PH |
1333 | if (length == L_LONGDOUBLE) |
1334 | sprintf(CS p, newformat, va_arg(ap, long double)); | |
1335 | else | |
1336 | sprintf(CS p, newformat, va_arg(ap, double)); | |
059ec3d9 PH |
1337 | while (*p) p++; |
1338 | break; | |
1339 | ||
1340 | /* String types */ | |
1341 | ||
1342 | case '%': | |
1343 | if (p >= last) { yield = FALSE; goto END_FORMAT; } | |
1344 | *p++ = '%'; | |
1345 | break; | |
1346 | ||
1347 | case 'c': | |
1348 | if (p >= last) { yield = FALSE; goto END_FORMAT; } | |
1349 | *p++ = va_arg(ap, int); | |
1350 | break; | |
1351 | ||
f1e5fef5 PP |
1352 | case 'D': /* Insert daily datestamp for log file names */ |
1353 | s = CS tod_stamp(tod_log_datestamp_daily); | |
059ec3d9 | 1354 | string_datestamp_offset = p - buffer; /* Passed back via global */ |
f1e5fef5 PP |
1355 | string_datestamp_length = Ustrlen(s); /* Passed back via global */ |
1356 | string_datestamp_type = tod_log_datestamp_daily; | |
1357 | slen = string_datestamp_length; | |
1358 | goto INSERT_STRING; | |
1359 | ||
1360 | case 'M': /* Insert monthly datestamp for log file names */ | |
1361 | s = CS tod_stamp(tod_log_datestamp_monthly); | |
1362 | string_datestamp_offset = p - buffer; /* Passed back via global */ | |
1363 | string_datestamp_length = Ustrlen(s); /* Passed back via global */ | |
1364 | string_datestamp_type = tod_log_datestamp_monthly; | |
1365 | slen = string_datestamp_length; | |
059ec3d9 PH |
1366 | goto INSERT_STRING; |
1367 | ||
1368 | case 's': | |
1369 | case 'S': /* Forces *lower* case */ | |
1370 | s = va_arg(ap, char *); | |
1371 | ||
059ec3d9 PH |
1372 | if (s == NULL) s = null; |
1373 | slen = Ustrlen(s); | |
1374 | ||
f1e5fef5 PP |
1375 | INSERT_STRING: /* Come to from %D or %M above */ |
1376 | ||
059ec3d9 PH |
1377 | /* If the width is specified, check that there is a precision |
1378 | set; if not, set it to the width to prevent overruns of long | |
1379 | strings. */ | |
1380 | ||
1381 | if (width >= 0) | |
1382 | { | |
1383 | if (precision < 0) precision = width; | |
1384 | } | |
1385 | ||
1386 | /* If a width is not specified and the precision is specified, set | |
1387 | the width to the precision, or the string length if shorted. */ | |
1388 | ||
1389 | else if (precision >= 0) | |
1390 | { | |
1391 | width = (precision < slen)? precision : slen; | |
1392 | } | |
1393 | ||
1394 | /* If neither are specified, set them both to the string length. */ | |
1395 | ||
1396 | else width = precision = slen; | |
1397 | ||
1398 | /* Check string space, and add the string to the buffer if ok. If | |
1399 | not OK, add part of the string (debugging uses this to show as | |
1400 | much as possible). */ | |
1401 | ||
24c929a2 NM |
1402 | if (p == last) |
1403 | { | |
1404 | yield = FALSE; | |
1405 | goto END_FORMAT; | |
1406 | } | |
059ec3d9 PH |
1407 | if (p >= last - width) |
1408 | { | |
1409 | yield = FALSE; | |
1410 | width = precision = last - p - 1; | |
24c929a2 NM |
1411 | if (width < 0) width = 0; |
1412 | if (precision < 0) precision = 0; | |
059ec3d9 PH |
1413 | } |
1414 | sprintf(CS p, "%*.*s", width, precision, s); | |
1415 | if (fp[-1] == 'S') | |
1416 | while (*p) { *p = tolower(*p); p++; } | |
1417 | else | |
1418 | while (*p) p++; | |
1419 | if (!yield) goto END_FORMAT; | |
1420 | break; | |
1421 | ||
1422 | /* Some things are never used in Exim; also catches junk. */ | |
1423 | ||
1424 | default: | |
1425 | strncpy(newformat, item_start, fp - item_start); | |
1426 | newformat[fp-item_start] = 0; | |
1427 | log_write(0, LOG_MAIN|LOG_PANIC_DIE, "string_format: unsupported type " | |
1428 | "in \"%s\" in \"%s\"", newformat, format); | |
1429 | break; | |
1430 | } | |
1431 | } | |
1432 | ||
1433 | /* Ensure string is complete; return TRUE if got to the end of the format */ | |
1434 | ||
1435 | END_FORMAT: | |
1436 | ||
1437 | *p = 0; | |
1438 | return yield; | |
1439 | } | |
1440 | ||
1441 | ||
1442 | ||
1443 | #ifndef COMPILE_UTILITY | |
1444 | /************************************************* | |
1445 | * Generate an "open failed" message * | |
1446 | *************************************************/ | |
1447 | ||
1448 | /* This function creates a message after failure to open a file. It includes a | |
1449 | string supplied as data, adds the strerror() text, and if the failure was | |
1450 | "Permission denied", reads and includes the euid and egid. | |
1451 | ||
1452 | Arguments: | |
1453 | eno the value of errno after the failure | |
1454 | format a text format string - deliberately not uschar * | |
1455 | ... arguments for the format string | |
1456 | ||
1457 | Returns: a message, in dynamic store | |
1458 | */ | |
1459 | ||
1460 | uschar * | |
1ba28e2b | 1461 | string_open_failed(int eno, const char *format, ...) |
059ec3d9 PH |
1462 | { |
1463 | va_list ap; | |
1464 | uschar buffer[1024]; | |
1465 | ||
1466 | Ustrcpy(buffer, "failed to open "); | |
1467 | va_start(ap, format); | |
1468 | ||
1469 | /* Use the checked formatting routine to ensure that the buffer | |
1470 | does not overflow. It should not, since this is called only for internally | |
1471 | specified messages. If it does, the message just gets truncated, and there | |
1472 | doesn't seem much we can do about that. */ | |
1473 | ||
1474 | (void)string_vformat(buffer+15, sizeof(buffer) - 15, format, ap); | |
1475 | ||
1476 | return (eno == EACCES)? | |
1477 | string_sprintf("%s: %s (euid=%ld egid=%ld)", buffer, strerror(eno), | |
1478 | (long int)geteuid(), (long int)getegid()) : | |
1479 | string_sprintf("%s: %s", buffer, strerror(eno)); | |
1480 | } | |
1481 | #endif /* COMPILE_UTILITY */ | |
1482 | ||
1483 | ||
1484 | ||
1485 | #ifndef COMPILE_UTILITY | |
1486 | /************************************************* | |
1487 | * Generate local prt for logging * | |
1488 | *************************************************/ | |
1489 | ||
1490 | /* This function is a subroutine for use in string_log_address() below. | |
1491 | ||
1492 | Arguments: | |
1493 | addr the address being logged | |
1494 | yield the current dynamic buffer pointer | |
1495 | sizeptr points to current size | |
1496 | ptrptr points to current insert pointer | |
1497 | ||
1498 | Returns: the new value of the buffer pointer | |
1499 | */ | |
1500 | ||
1501 | static uschar * | |
1502 | string_get_localpart(address_item *addr, uschar *yield, int *sizeptr, | |
1503 | int *ptrptr) | |
1504 | { | |
1505 | if (testflag(addr, af_include_affixes) && addr->prefix != NULL) | |
1506 | yield = string_cat(yield, sizeptr, ptrptr, addr->prefix, | |
1507 | Ustrlen(addr->prefix)); | |
1508 | yield = string_cat(yield, sizeptr, ptrptr, addr->local_part, | |
1509 | Ustrlen(addr->local_part)); | |
1510 | if (testflag(addr, af_include_affixes) && addr->suffix != NULL) | |
1511 | yield = string_cat(yield, sizeptr, ptrptr, addr->suffix, | |
1512 | Ustrlen(addr->suffix)); | |
1513 | return yield; | |
1514 | } | |
1515 | ||
1516 | ||
1517 | /************************************************* | |
1518 | * Generate log address list * | |
1519 | *************************************************/ | |
1520 | ||
1521 | /* This function generates a list consisting of an address and its parents, for | |
1522 | use in logging lines. For saved onetime aliased addresses, the onetime parent | |
1523 | field is used. If the address was delivered by a transport with rcpt_include_ | |
1524 | affixes set, the af_include_affixes bit will be set in the address. In that | |
1525 | case, we include the affixes here too. | |
1526 | ||
1527 | Arguments: | |
1528 | addr bottom (ultimate) address | |
1529 | all_parents if TRUE, include all parents | |
1530 | success TRUE for successful delivery | |
1531 | ||
1532 | Returns: a string in dynamic store | |
1533 | */ | |
1534 | ||
1535 | uschar * | |
1536 | string_log_address(address_item *addr, BOOL all_parents, BOOL success) | |
1537 | { | |
1538 | int size = 64; | |
1539 | int ptr = 0; | |
1540 | BOOL add_topaddr = TRUE; | |
1541 | uschar *yield = store_get(size); | |
1542 | address_item *topaddr; | |
1543 | ||
1544 | /* Find the ultimate parent */ | |
1545 | ||
1546 | for (topaddr = addr; topaddr->parent != NULL; topaddr = topaddr->parent); | |
1547 | ||
1548 | /* We start with just the local part for pipe, file, and reply deliveries, and | |
1549 | for successful local deliveries from routers that have the log_as_local flag | |
1550 | set. File deliveries from filters can be specified as non-absolute paths in | |
1551 | cases where the transport is goin to complete the path. If there is an error | |
1552 | before this happens (expansion failure) the local part will not be updated, and | |
1553 | so won't necessarily look like a path. Add extra text for this case. */ | |
1554 | ||
1555 | if (testflag(addr, af_pfr) || | |
1556 | (success && | |
1557 | addr->router != NULL && addr->router->log_as_local && | |
1558 | addr->transport != NULL && addr->transport->info->local)) | |
1559 | { | |
1560 | if (testflag(addr, af_file) && addr->local_part[0] != '/') | |
1561 | yield = string_cat(yield, &size, &ptr, CUS"save ", 5); | |
1562 | yield = string_get_localpart(addr, yield, &size, &ptr); | |
1563 | } | |
1564 | ||
1565 | /* Other deliveries start with the full address. It we have split it into local | |
1566 | part and domain, use those fields. Some early failures can happen before the | |
1567 | splitting is done; in those cases use the original field. */ | |
1568 | ||
1569 | else | |
1570 | { | |
1571 | if (addr->local_part != NULL) | |
1572 | { | |
1573 | yield = string_get_localpart(addr, yield, &size, &ptr); | |
1574 | yield = string_cat(yield, &size, &ptr, US"@", 1); | |
1575 | yield = string_cat(yield, &size, &ptr, addr->domain, | |
1576 | Ustrlen(addr->domain) ); | |
1577 | } | |
1578 | else | |
1579 | { | |
1580 | yield = string_cat(yield, &size, &ptr, addr->address, Ustrlen(addr->address)); | |
1581 | } | |
1582 | yield[ptr] = 0; | |
1583 | ||
1584 | /* If the address we are going to print is the same as the top address, | |
1585 | and all parents are not being included, don't add on the top address. First | |
1586 | of all, do a caseless comparison; if this succeeds, do a caseful comparison | |
1587 | on the local parts. */ | |
1588 | ||
1589 | if (strcmpic(yield, topaddr->address) == 0 && | |
1590 | Ustrncmp(yield, topaddr->address, Ustrchr(yield, '@') - yield) == 0 && | |
1591 | addr->onetime_parent == NULL && | |
1592 | (!all_parents || addr->parent == NULL || addr->parent == topaddr)) | |
1593 | add_topaddr = FALSE; | |
1594 | } | |
1595 | ||
1596 | /* If all parents are requested, or this is a local pipe/file/reply, and | |
1597 | there is at least one intermediate parent, show it in brackets, and continue | |
1598 | with all of them if all are wanted. */ | |
1599 | ||
1600 | if ((all_parents || testflag(addr, af_pfr)) && | |
1601 | addr->parent != NULL && | |
1602 | addr->parent != topaddr) | |
1603 | { | |
1604 | uschar *s = US" ("; | |
1605 | address_item *addr2; | |
1606 | for (addr2 = addr->parent; addr2 != topaddr; addr2 = addr2->parent) | |
1607 | { | |
1608 | yield = string_cat(yield, &size, &ptr, s, 2); | |
1609 | yield = string_cat(yield, &size, &ptr, addr2->address, Ustrlen(addr2->address)); | |
1610 | if (!all_parents) break; | |
1611 | s = US", "; | |
1612 | } | |
1613 | yield = string_cat(yield, &size, &ptr, US")", 1); | |
1614 | } | |
1615 | ||
1616 | /* Add the top address if it is required */ | |
1617 | ||
1618 | if (add_topaddr) | |
1619 | { | |
1620 | yield = string_cat(yield, &size, &ptr, US" <", 2); | |
1621 | ||
1622 | if (addr->onetime_parent == NULL) | |
1623 | yield = string_cat(yield, &size, &ptr, topaddr->address, | |
1624 | Ustrlen(topaddr->address)); | |
1625 | else | |
1626 | yield = string_cat(yield, &size, &ptr, addr->onetime_parent, | |
1627 | Ustrlen(addr->onetime_parent)); | |
1628 | ||
1629 | yield = string_cat(yield, &size, &ptr, US">", 1); | |
1630 | } | |
1631 | ||
1632 | yield[ptr] = 0; /* string_cat() leaves space */ | |
1633 | return yield; | |
1634 | } | |
1635 | #endif /* COMPILE_UTILITY */ | |
1636 | ||
1637 | ||
1638 | ||
1639 | ||
1640 | ||
1641 | /************************************************* | |
1642 | ************************************************** | |
1643 | * Stand-alone test program * | |
1644 | ************************************************** | |
1645 | *************************************************/ | |
1646 | ||
1647 | #ifdef STAND_ALONE | |
1648 | int main(void) | |
1649 | { | |
1650 | uschar buffer[256]; | |
1651 | ||
1652 | printf("Testing is_ip_address\n"); | |
1653 | ||
1654 | while (fgets(CS buffer, sizeof(buffer), stdin) != NULL) | |
1655 | { | |
1656 | int offset; | |
1657 | buffer[Ustrlen(buffer) - 1] = 0; | |
1658 | printf("%d\n", string_is_ip_address(buffer, NULL)); | |
1659 | printf("%d %d %s\n", string_is_ip_address(buffer, &offset), offset, buffer); | |
1660 | } | |
1661 | ||
1662 | printf("Testing string_nextinlist\n"); | |
1663 | ||
1664 | while (fgets(CS buffer, sizeof(buffer), stdin) != NULL) | |
1665 | { | |
1666 | uschar *list = buffer; | |
1667 | uschar *lp1, *lp2; | |
1668 | uschar item[256]; | |
1669 | int sep1 = 0; | |
1670 | int sep2 = 0; | |
1671 | ||
1672 | if (*list == '<') | |
1673 | { | |
1674 | sep1 = sep2 = list[1]; | |
1675 | list += 2; | |
1676 | } | |
1677 | ||
1678 | lp1 = lp2 = list; | |
1679 | for (;;) | |
1680 | { | |
1681 | uschar *item1 = string_nextinlist(&lp1, &sep1, item, sizeof(item)); | |
1682 | uschar *item2 = string_nextinlist(&lp2, &sep2, NULL, 0); | |
1683 | ||
1684 | if (item1 == NULL && item2 == NULL) break; | |
1685 | if (item == NULL || item2 == NULL || Ustrcmp(item1, item2) != 0) | |
1686 | { | |
1687 | printf("***ERROR\nitem1=\"%s\"\nitem2=\"%s\"\n", | |
1688 | (item1 == NULL)? "NULL" : CS item1, | |
1689 | (item2 == NULL)? "NULL" : CS item2); | |
1690 | break; | |
1691 | } | |
1692 | else printf(" \"%s\"\n", CS item1); | |
1693 | } | |
1694 | } | |
1695 | ||
1696 | /* This is a horrible lash-up, but it serves its purpose. */ | |
1697 | ||
1698 | printf("Testing string_format\n"); | |
1699 | ||
1700 | while (fgets(CS buffer, sizeof(buffer), stdin) != NULL) | |
1701 | { | |
1702 | void *args[3]; | |
ed72ace5 | 1703 | long long llargs[3]; |
059ec3d9 PH |
1704 | double dargs[3]; |
1705 | int dflag = 0; | |
ed72ace5 | 1706 | int llflag = 0; |
059ec3d9 PH |
1707 | int n = 0; |
1708 | int count; | |
1709 | int countset = 0; | |
1710 | uschar format[256]; | |
1711 | uschar outbuf[256]; | |
1712 | uschar *s; | |
1713 | buffer[Ustrlen(buffer) - 1] = 0; | |
1714 | ||
1715 | s = Ustrchr(buffer, ','); | |
1716 | if (s == NULL) s = buffer + Ustrlen(buffer); | |
1717 | ||
1718 | Ustrncpy(format, buffer, s - buffer); | |
1719 | format[s-buffer] = 0; | |
1720 | ||
1721 | if (*s == ',') s++; | |
1722 | ||
1723 | while (*s != 0) | |
1724 | { | |
1725 | uschar *ss = s; | |
1726 | s = Ustrchr(ss, ','); | |
1727 | if (s == NULL) s = ss + Ustrlen(ss); | |
1728 | ||
1729 | if (isdigit(*ss)) | |
1730 | { | |
1731 | Ustrncpy(outbuf, ss, s-ss); | |
1732 | if (Ustrchr(outbuf, '.') != NULL) | |
1733 | { | |
1734 | dflag = 1; | |
1735 | dargs[n++] = Ustrtod(outbuf, NULL); | |
1736 | } | |
ed72ace5 PH |
1737 | else if (Ustrstr(outbuf, "ll") != NULL) |
1738 | { | |
1739 | llflag = 1; | |
1740 | llargs[n++] = strtoull(CS outbuf, NULL, 10); | |
1741 | } | |
059ec3d9 PH |
1742 | else |
1743 | { | |
1744 | args[n++] = (void *)Uatoi(outbuf); | |
1745 | } | |
1746 | } | |
1747 | ||
1748 | else if (Ustrcmp(ss, "*") == 0) | |
1749 | { | |
1750 | args[n++] = (void *)(&count); | |
1751 | countset = 1; | |
1752 | } | |
1753 | ||
1754 | else | |
1755 | { | |
1756 | uschar *sss = malloc(s - ss + 1); | |
1757 | Ustrncpy(sss, ss, s-ss); | |
1758 | args[n++] = sss; | |
1759 | } | |
1760 | ||
1761 | if (*s == ',') s++; | |
1762 | } | |
1763 | ||
ed72ace5 PH |
1764 | if (!dflag && !llflag) |
1765 | printf("%s\n", string_format(outbuf, sizeof(outbuf), CS format, | |
1766 | args[0], args[1], args[2])? "True" : "False"); | |
1767 | ||
1768 | else if (dflag) | |
1769 | printf("%s\n", string_format(outbuf, sizeof(outbuf), CS format, | |
1770 | dargs[0], dargs[1], dargs[2])? "True" : "False"); | |
059ec3d9 PH |
1771 | |
1772 | else printf("%s\n", string_format(outbuf, sizeof(outbuf), CS format, | |
ed72ace5 | 1773 | llargs[0], llargs[1], llargs[2])? "True" : "False"); |
059ec3d9 PH |
1774 | |
1775 | printf("%s\n", CS outbuf); | |
1776 | if (countset) printf("count=%d\n", count); | |
1777 | } | |
1778 | ||
1779 | return 0; | |
1780 | } | |
1781 | #endif | |
1782 | ||
1783 | /* End of string.c */ |