| 1 | /************************************************* |
| 2 | * Exim - an Internet mail transport agent * |
| 3 | *************************************************/ |
| 4 | |
| 5 | /* Copyright (c) University of Cambridge 1995 - 2018 */ |
| 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 |
| 29 | if there is no / followed by trailing digits, *maskptr is set 0 |
| 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 |
| 37 | string_is_ip_address(const uschar *s, int *maskptr) |
| 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) |
| 46 | { |
| 47 | const uschar *ss = s + Ustrlen(s); |
| 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 | |
| 129 | if (!v4end) |
| 130 | return (*s == 0 || *s == '%' || |
| 131 | (*s == '/' && maskptr != NULL && *maskptr != 0))? yield : 0; |
| 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 | long n; |
| 139 | uschar * end; |
| 140 | |
| 141 | if (i != 0 && *s++ != '.') return 0; |
| 142 | n = strtol(CCS s, CSS &end, 10); |
| 143 | if (n > 255 || n < 0 || end <= s || end > s+3) return 0; |
| 144 | s = end; |
| 145 | } |
| 146 | |
| 147 | return !*s || (*s == '/' && maskptr && *maskptr != 0) ? yield : 0; |
| 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 | { |
| 171 | if (size == 0) Ustrcpy(buffer, " "); |
| 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 | |
| 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 |
| 230 | string_interpret_escape(const uschar **pp) |
| 231 | { |
| 232 | #ifdef COMPILE_UTILITY |
| 233 | const uschar *hex_digits= CUS"0123456789abcdef"; |
| 234 | #endif |
| 235 | int ch; |
| 236 | const uschar *p = *pp; |
| 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 | { |
| 250 | case 'b': ch = '\b'; break; |
| 251 | case 'f': ch = '\f'; break; |
| 252 | case 'n': ch = '\n'; break; |
| 253 | case 'r': ch = '\r'; break; |
| 254 | case 't': ch = '\t'; break; |
| 255 | case 'v': ch = '\v'; break; |
| 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 | } |
| 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 | |
| 290 | const uschar * |
| 291 | string_printing2(const uschar *s, BOOL allow_tab) |
| 292 | { |
| 293 | int nonprintcount = 0; |
| 294 | int length = 0; |
| 295 | const uschar *t = s; |
| 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 | |
| 310 | ss = store_get(length + nonprintcount * 3 + 1); |
| 311 | |
| 312 | /* Copy everything, escaping non printers. */ |
| 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; |
| 337 | return ss; |
| 338 | } |
| 339 | #endif /* COMPILE_UTILITY */ |
| 340 | |
| 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 | { |
| 380 | *q++ = string_interpret_escape((const uschar **)&p); |
| 381 | p++; |
| 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 | |
| 405 | return ss; |
| 406 | } |
| 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 * |
| 422 | string_copy(const uschar *s) |
| 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 * |
| 443 | string_copy_malloc(const uschar *s) |
| 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 * |
| 463 | string_copylc(const uschar *s) |
| 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 * |
| 489 | string_copyn(const uschar *s, int n) |
| 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 | |
| 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 | |
| 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 * |
| 645 | string_dequote(const uschar **sptr) |
| 646 | { |
| 647 | const uschar *s = *sptr; |
| 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 * |
| 716 | string_sprintf(const char *format, ...) |
| 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, |
| 723 | "string_sprintf expansion was longer than " SIZE_T_FMT |
| 724 | "; format string was (%s)\nexpansion started '%.32s'", |
| 725 | sizeof(buffer), format, buffer); |
| 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 |
| 746 | strncmpic(const uschar *s, const uschar *t, int n) |
| 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 |
| 770 | strcmpic(const uschar *s, const uschar *t) |
| 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 | |
| 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. |
| 846 | This is used for some cases when called via functions that sometimes |
| 847 | plough through lists, and sometimes are given single items.) |
| 848 | |
| 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. |
| 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 * |
| 875 | string_nextinlist(const uschar **listptr, int *separator, uschar *buffer, int buflen) |
| 876 | { |
| 877 | int sep = *separator; |
| 878 | const uschar *s = *listptr; |
| 879 | BOOL sep_is_special; |
| 880 | |
| 881 | if (s == NULL) return NULL; |
| 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. */ |
| 891 | |
| 892 | if (sep <= 0) |
| 893 | { |
| 894 | if (*s == '<' && (ispunct(s[1]) || iscntrl(s[1]))) |
| 895 | { |
| 896 | sep = s[1]; |
| 897 | s += 2; |
| 898 | while (isspace(*s) && *s != sep) s++; |
| 899 | } |
| 900 | else |
| 901 | { |
| 902 | sep = (sep == 0)? ':' : -sep; |
| 903 | } |
| 904 | *separator = sep; |
| 905 | } |
| 906 | |
| 907 | /* An empty string has no list elements */ |
| 908 | |
| 909 | if (*s == 0) return NULL; |
| 910 | |
| 911 | /* Note whether whether or not the separator is an iscntrl() character. */ |
| 912 | |
| 913 | sep_is_special = iscntrl(sep); |
| 914 | |
| 915 | /* Handle the case when a buffer is provided. */ |
| 916 | |
| 917 | if (buffer) |
| 918 | { |
| 919 | int p = 0; |
| 920 | for (; *s != 0; s++) |
| 921 | { |
| 922 | if (*s == sep && (*(++s) != sep || sep_is_special)) break; |
| 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 | { |
| 933 | const uschar *ss; |
| 934 | gstring * g = NULL; |
| 935 | |
| 936 | /* We know that *s != 0 at this point. However, it might be pointing to a |
| 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. */ |
| 940 | |
| 941 | if (*s == sep) |
| 942 | { |
| 943 | s++; |
| 944 | if (*s != sep || sep_is_special) |
| 945 | { |
| 946 | *listptr = s; |
| 947 | return string_copy(US""); |
| 948 | } |
| 949 | } |
| 950 | |
| 951 | /* Not an empty string; the first character is guaranteed to be a data |
| 952 | character. */ |
| 953 | |
| 954 | for (;;) |
| 955 | { |
| 956 | for (ss = s + 1; *ss != 0 && *ss != sep; ss++) ; |
| 957 | g = string_catn(g, s, ss-s); |
| 958 | s = ss; |
| 959 | if (*s == 0 || *(++s) != sep || sep_is_special) break; |
| 960 | } |
| 961 | while (g->ptr > 0 && isspace(g->s[g->ptr-1])) g->ptr--; |
| 962 | buffer = string_from_gstring(g); |
| 963 | gstring_reset_unused(g); |
| 964 | } |
| 965 | |
| 966 | /* Update the current pointer and return the new string */ |
| 967 | |
| 968 | *listptr = s; |
| 969 | return buffer; |
| 970 | } |
| 971 | |
| 972 | |
| 973 | static const uschar * |
| 974 | Ustrnchr(const uschar * s, int c, unsigned * len) |
| 975 | { |
| 976 | unsigned siz = *len; |
| 977 | while (siz) |
| 978 | { |
| 979 | if (!*s) return NULL; |
| 980 | if (*s == c) |
| 981 | { |
| 982 | *len = siz; |
| 983 | return s; |
| 984 | } |
| 985 | s++; |
| 986 | siz--; |
| 987 | } |
| 988 | return NULL; |
| 989 | } |
| 990 | |
| 991 | |
| 992 | /************************************************ |
| 993 | * Add element to separated list * |
| 994 | ************************************************/ |
| 995 | /* This function is used to build a list, returning an allocated null-terminated |
| 996 | growable string. The given element has any embedded separator characters |
| 997 | doubled. |
| 998 | |
| 999 | Despite having the same growable-string interface as string_cat() the list is |
| 1000 | always returned null-terminated. |
| 1001 | |
| 1002 | Arguments: |
| 1003 | list expanding-string for the list that is being built, or NULL |
| 1004 | if this is a new list that has no contents yet |
| 1005 | sep list separator character |
| 1006 | ele new element to be appended to the list |
| 1007 | |
| 1008 | Returns: pointer to the start of the list, changed if copied for expansion. |
| 1009 | */ |
| 1010 | |
| 1011 | gstring * |
| 1012 | string_append_listele(gstring * list, uschar sep, const uschar * ele) |
| 1013 | { |
| 1014 | uschar * sp; |
| 1015 | |
| 1016 | if (list && list->ptr) |
| 1017 | list = string_catn(list, &sep, 1); |
| 1018 | |
| 1019 | while((sp = Ustrchr(ele, sep))) |
| 1020 | { |
| 1021 | list = string_catn(list, ele, sp-ele+1); |
| 1022 | list = string_catn(list, &sep, 1); |
| 1023 | ele = sp+1; |
| 1024 | } |
| 1025 | list = string_cat(list, ele); |
| 1026 | (void) string_from_gstring(list); |
| 1027 | return list; |
| 1028 | } |
| 1029 | |
| 1030 | |
| 1031 | gstring * |
| 1032 | string_append_listele_n(gstring * list, uschar sep, const uschar * ele, |
| 1033 | unsigned len) |
| 1034 | { |
| 1035 | const uschar * sp; |
| 1036 | |
| 1037 | if (list && list->ptr) |
| 1038 | list = string_catn(list, &sep, 1); |
| 1039 | |
| 1040 | while((sp = Ustrnchr(ele, sep, &len))) |
| 1041 | { |
| 1042 | list = string_catn(list, ele, sp-ele+1); |
| 1043 | list = string_catn(list, &sep, 1); |
| 1044 | ele = sp+1; |
| 1045 | len--; |
| 1046 | } |
| 1047 | list = string_catn(list, ele, len); |
| 1048 | (void) string_from_gstring(list); |
| 1049 | return list; |
| 1050 | } |
| 1051 | |
| 1052 | |
| 1053 | |
| 1054 | /* A slightly-bogus listmaker utility; the separator is a string so |
| 1055 | can be multiple chars - there is no checking for the element content |
| 1056 | containing any of the separator. */ |
| 1057 | |
| 1058 | gstring * |
| 1059 | string_append2_listele_n(gstring * list, const uschar * sepstr, |
| 1060 | const uschar * ele, unsigned len) |
| 1061 | { |
| 1062 | if (list && list->ptr) |
| 1063 | list = string_cat(list, sepstr); |
| 1064 | |
| 1065 | list = string_catn(list, ele, len); |
| 1066 | (void) string_from_gstring(list); |
| 1067 | return list; |
| 1068 | } |
| 1069 | |
| 1070 | |
| 1071 | |
| 1072 | /************************************************/ |
| 1073 | /* Create a growable-string with some preassigned space */ |
| 1074 | |
| 1075 | gstring * |
| 1076 | string_get(unsigned size) |
| 1077 | { |
| 1078 | gstring * g = store_get(sizeof(gstring) + size); |
| 1079 | g->size = size; |
| 1080 | g->ptr = 0; |
| 1081 | g->s = US(g + 1); |
| 1082 | return g; |
| 1083 | } |
| 1084 | |
| 1085 | /* NUL-terminate the C string in the growable-string, and return it. */ |
| 1086 | |
| 1087 | uschar * |
| 1088 | string_from_gstring(gstring * g) |
| 1089 | { |
| 1090 | if (!g) return NULL; |
| 1091 | g->s[g->ptr] = '\0'; |
| 1092 | return g->s; |
| 1093 | } |
| 1094 | |
| 1095 | void |
| 1096 | gstring_reset_unused(gstring * g) |
| 1097 | { |
| 1098 | store_reset(g->s + (g->size = g->ptr + 1)); |
| 1099 | } |
| 1100 | |
| 1101 | /************************************************* |
| 1102 | * Add chars to string * |
| 1103 | *************************************************/ |
| 1104 | |
| 1105 | /* Arguments: |
| 1106 | g the grawable-string |
| 1107 | p current end of data |
| 1108 | count amount to grow by |
| 1109 | */ |
| 1110 | |
| 1111 | static void |
| 1112 | gstring_grow(gstring * g, int p, int count) |
| 1113 | { |
| 1114 | int oldsize = g->size; |
| 1115 | |
| 1116 | /* Mostly, string_cat() is used to build small strings of a few hundred |
| 1117 | characters at most. There are times, however, when the strings are very much |
| 1118 | longer (for example, a lookup that returns a vast number of alias addresses). |
| 1119 | To try to keep things reasonable, we use increments whose size depends on the |
| 1120 | existing length of the string. */ |
| 1121 | |
| 1122 | unsigned inc = oldsize < 4096 ? 127 : 1023; |
| 1123 | g->size = ((p + count + inc) & ~inc) + 1; |
| 1124 | |
| 1125 | /* Try to extend an existing allocation. If the result of calling |
| 1126 | store_extend() is false, either there isn't room in the current memory block, |
| 1127 | or this string is not the top item on the dynamic store stack. We then have |
| 1128 | to get a new chunk of store and copy the old string. When building large |
| 1129 | strings, it is helpful to call store_release() on the old string, to release |
| 1130 | memory blocks that have become empty. (The block will be freed if the string |
| 1131 | is at its start.) However, we can do this only if we know that the old string |
| 1132 | was the last item on the dynamic memory stack. This is the case if it matches |
| 1133 | store_last_get. */ |
| 1134 | |
| 1135 | if (!store_extend(g->s, oldsize, g->size)) |
| 1136 | g->s = store_newblock(g->s, g->size, p); |
| 1137 | } |
| 1138 | |
| 1139 | |
| 1140 | |
| 1141 | /* This function is used when building up strings of unknown length. Room is |
| 1142 | always left for a terminating zero to be added to the string that is being |
| 1143 | built. This function does not require the string that is being added to be NUL |
| 1144 | terminated, because the number of characters to add is given explicitly. It is |
| 1145 | sometimes called to extract parts of other strings. |
| 1146 | |
| 1147 | Arguments: |
| 1148 | string points to the start of the string that is being built, or NULL |
| 1149 | if this is a new string that has no contents yet |
| 1150 | s points to characters to add |
| 1151 | count count of characters to add; must not exceed the length of s, if s |
| 1152 | is a C string. |
| 1153 | |
| 1154 | Returns: pointer to the start of the string, changed if copied for expansion. |
| 1155 | Note that a NUL is not added, though space is left for one. This is |
| 1156 | because string_cat() is often called multiple times to build up a |
| 1157 | string - there's no point adding the NUL till the end. |
| 1158 | |
| 1159 | */ |
| 1160 | /* coverity[+alloc] */ |
| 1161 | |
| 1162 | gstring * |
| 1163 | string_catn(gstring * g, const uschar *s, int count) |
| 1164 | { |
| 1165 | int p; |
| 1166 | |
| 1167 | if (!g) |
| 1168 | { |
| 1169 | unsigned inc = count < 4096 ? 127 : 1023; |
| 1170 | unsigned size = ((count + inc) & ~inc) + 1; |
| 1171 | g = string_get(size); |
| 1172 | } |
| 1173 | |
| 1174 | p = g->ptr; |
| 1175 | if (p + count >= g->size) |
| 1176 | gstring_grow(g, p, count); |
| 1177 | |
| 1178 | /* Because we always specify the exact number of characters to copy, we can |
| 1179 | use memcpy(), which is likely to be more efficient than strncopy() because the |
| 1180 | latter has to check for zero bytes. */ |
| 1181 | |
| 1182 | memcpy(g->s + p, s, count); |
| 1183 | g->ptr = p + count; |
| 1184 | return g; |
| 1185 | } |
| 1186 | |
| 1187 | |
| 1188 | gstring * |
| 1189 | string_cat(gstring *string, const uschar *s) |
| 1190 | { |
| 1191 | return string_catn(string, s, Ustrlen(s)); |
| 1192 | } |
| 1193 | |
| 1194 | |
| 1195 | |
| 1196 | /************************************************* |
| 1197 | * Append strings to another string * |
| 1198 | *************************************************/ |
| 1199 | |
| 1200 | /* This function can be used to build a string from many other strings. |
| 1201 | It calls string_cat() to do the dirty work. |
| 1202 | |
| 1203 | Arguments: |
| 1204 | string expanding-string that is being built, or NULL |
| 1205 | if this is a new string that has no contents yet |
| 1206 | count the number of strings to append |
| 1207 | ... "count" uschar* arguments, which must be valid zero-terminated |
| 1208 | C strings |
| 1209 | |
| 1210 | Returns: pointer to the start of the string, changed if copied for expansion. |
| 1211 | The string is not zero-terminated - see string_cat() above. |
| 1212 | */ |
| 1213 | |
| 1214 | __inline__ gstring * |
| 1215 | string_append(gstring *string, int count, ...) |
| 1216 | { |
| 1217 | va_list ap; |
| 1218 | |
| 1219 | va_start(ap, count); |
| 1220 | while (count-- > 0) |
| 1221 | { |
| 1222 | uschar *t = va_arg(ap, uschar *); |
| 1223 | string = string_cat(string, t); |
| 1224 | } |
| 1225 | va_end(ap); |
| 1226 | |
| 1227 | return string; |
| 1228 | } |
| 1229 | #endif |
| 1230 | |
| 1231 | |
| 1232 | |
| 1233 | /************************************************* |
| 1234 | * Format a string with length checks * |
| 1235 | *************************************************/ |
| 1236 | |
| 1237 | /* This function is used to format a string with checking of the length of the |
| 1238 | output for all conversions. It protects Exim from absent-mindedness when |
| 1239 | calling functions like debug_printf and string_sprintf, and elsewhere. There |
| 1240 | are two different entry points to what is actually the same function, depending |
| 1241 | on whether the variable length list of data arguments are given explicitly or |
| 1242 | as a va_list item. |
| 1243 | |
| 1244 | The formats are the usual printf() ones, with some omissions (never used) and |
| 1245 | three additions for strings: %S forces lower case, %T forces upper case, and |
| 1246 | %#s or %#S prints nothing for a NULL string. Without the # "NULL" is printed |
| 1247 | (useful in debugging). There is also the addition of %D and %M, which insert |
| 1248 | the date in the form used for datestamped log files. |
| 1249 | |
| 1250 | Arguments: |
| 1251 | buffer a buffer in which to put the formatted string |
| 1252 | buflen the length of the buffer |
| 1253 | format the format string - deliberately char * and not uschar * |
| 1254 | ... or ap variable list of supplementary arguments |
| 1255 | |
| 1256 | Returns: TRUE if the result fitted in the buffer |
| 1257 | */ |
| 1258 | |
| 1259 | BOOL |
| 1260 | string_format(uschar *buffer, int buflen, const char *format, ...) |
| 1261 | { |
| 1262 | BOOL yield; |
| 1263 | va_list ap; |
| 1264 | va_start(ap, format); |
| 1265 | yield = string_vformat(buffer, buflen, format, ap); |
| 1266 | va_end(ap); |
| 1267 | return yield; |
| 1268 | } |
| 1269 | |
| 1270 | |
| 1271 | BOOL |
| 1272 | string_vformat(uschar *buffer, int buflen, const char *format, va_list ap) |
| 1273 | { |
| 1274 | /* We assume numbered ascending order, C does not guarantee that */ |
| 1275 | enum { L_NORMAL=1, L_SHORT=2, L_LONG=3, L_LONGLONG=4, L_LONGDOUBLE=5, L_SIZE=6 }; |
| 1276 | |
| 1277 | BOOL yield = TRUE; |
| 1278 | int width, precision; |
| 1279 | const char *fp = format; /* Deliberately not unsigned */ |
| 1280 | uschar *p = buffer; |
| 1281 | uschar *last = buffer + buflen - 1; |
| 1282 | |
| 1283 | string_datestamp_offset = -1; /* Datestamp not inserted */ |
| 1284 | string_datestamp_length = 0; /* Datestamp not inserted */ |
| 1285 | string_datestamp_type = 0; /* Datestamp not inserted */ |
| 1286 | |
| 1287 | /* Scan the format and handle the insertions */ |
| 1288 | |
| 1289 | while (*fp != 0) |
| 1290 | { |
| 1291 | int length = L_NORMAL; |
| 1292 | int *nptr; |
| 1293 | int slen; |
| 1294 | const char *null = "NULL"; /* ) These variables */ |
| 1295 | const char *item_start, *s; /* ) are deliberately */ |
| 1296 | char newformat[16]; /* ) not unsigned */ |
| 1297 | |
| 1298 | /* Non-% characters just get copied verbatim */ |
| 1299 | |
| 1300 | if (*fp != '%') |
| 1301 | { |
| 1302 | if (p >= last) { yield = FALSE; break; } |
| 1303 | *p++ = (uschar)*fp++; |
| 1304 | continue; |
| 1305 | } |
| 1306 | |
| 1307 | /* Deal with % characters. Pick off the width and precision, for checking |
| 1308 | strings, skipping over the flag and modifier characters. */ |
| 1309 | |
| 1310 | item_start = fp; |
| 1311 | width = precision = -1; |
| 1312 | |
| 1313 | if (strchr("-+ #0", *(++fp)) != NULL) |
| 1314 | { |
| 1315 | if (*fp == '#') null = ""; |
| 1316 | fp++; |
| 1317 | } |
| 1318 | |
| 1319 | if (isdigit((uschar)*fp)) |
| 1320 | { |
| 1321 | width = *fp++ - '0'; |
| 1322 | while (isdigit((uschar)*fp)) width = width * 10 + *fp++ - '0'; |
| 1323 | } |
| 1324 | else if (*fp == '*') |
| 1325 | { |
| 1326 | width = va_arg(ap, int); |
| 1327 | fp++; |
| 1328 | } |
| 1329 | |
| 1330 | if (*fp == '.') |
| 1331 | { |
| 1332 | if (*(++fp) == '*') |
| 1333 | { |
| 1334 | precision = va_arg(ap, int); |
| 1335 | fp++; |
| 1336 | } |
| 1337 | else |
| 1338 | { |
| 1339 | precision = 0; |
| 1340 | while (isdigit((uschar)*fp)) |
| 1341 | precision = precision*10 + *fp++ - '0'; |
| 1342 | } |
| 1343 | } |
| 1344 | |
| 1345 | /* Skip over 'h', 'L', 'l', 'll' and 'z', remembering the item length */ |
| 1346 | |
| 1347 | if (*fp == 'h') |
| 1348 | { fp++; length = L_SHORT; } |
| 1349 | else if (*fp == 'L') |
| 1350 | { fp++; length = L_LONGDOUBLE; } |
| 1351 | else if (*fp == 'l') |
| 1352 | { |
| 1353 | if (fp[1] == 'l') |
| 1354 | { |
| 1355 | fp += 2; |
| 1356 | length = L_LONGLONG; |
| 1357 | } |
| 1358 | else |
| 1359 | { |
| 1360 | fp++; |
| 1361 | length = L_LONG; |
| 1362 | } |
| 1363 | } |
| 1364 | else if (*fp == 'z') |
| 1365 | { fp++; length = L_SIZE; } |
| 1366 | |
| 1367 | /* Handle each specific format type. */ |
| 1368 | |
| 1369 | switch (*fp++) |
| 1370 | { |
| 1371 | case 'n': |
| 1372 | nptr = va_arg(ap, int *); |
| 1373 | *nptr = p - buffer; |
| 1374 | break; |
| 1375 | |
| 1376 | case 'd': |
| 1377 | case 'o': |
| 1378 | case 'u': |
| 1379 | case 'x': |
| 1380 | case 'X': |
| 1381 | if (p >= last - ((length > L_LONG)? 24 : 12)) |
| 1382 | { yield = FALSE; goto END_FORMAT; } |
| 1383 | strncpy(newformat, item_start, fp - item_start); |
| 1384 | newformat[fp - item_start] = 0; |
| 1385 | |
| 1386 | /* Short int is promoted to int when passing through ..., so we must use |
| 1387 | int for va_arg(). */ |
| 1388 | |
| 1389 | switch(length) |
| 1390 | { |
| 1391 | case L_SHORT: |
| 1392 | case L_NORMAL: p += sprintf(CS p, newformat, va_arg(ap, int)); break; |
| 1393 | case L_LONG: p += sprintf(CS p, newformat, va_arg(ap, long int)); break; |
| 1394 | case L_LONGLONG: p += sprintf(CS p, newformat, va_arg(ap, LONGLONG_T)); break; |
| 1395 | case L_SIZE: p += sprintf(CS p, newformat, va_arg(ap, size_t)); break; |
| 1396 | } |
| 1397 | break; |
| 1398 | |
| 1399 | case 'p': |
| 1400 | { |
| 1401 | void * ptr; |
| 1402 | if (p >= last - 24) { yield = FALSE; goto END_FORMAT; } |
| 1403 | /* sprintf() saying "(nil)" for a null pointer seems unreliable. |
| 1404 | Handle it explicitly. */ |
| 1405 | if ((ptr = va_arg(ap, void *))) |
| 1406 | { |
| 1407 | strncpy(newformat, item_start, fp - item_start); |
| 1408 | newformat[fp - item_start] = 0; |
| 1409 | p += sprintf(CS p, newformat, ptr); |
| 1410 | } |
| 1411 | else |
| 1412 | p += sprintf(CS p, "(nil)"); |
| 1413 | } |
| 1414 | break; |
| 1415 | |
| 1416 | /* %f format is inherently insecure if the numbers that it may be |
| 1417 | handed are unknown (e.g. 1e300). However, in Exim, %f is used for |
| 1418 | printing load averages, and these are actually stored as integers |
| 1419 | (load average * 1000) so the size of the numbers is constrained. |
| 1420 | It is also used for formatting sending rates, where the simplicity |
| 1421 | of the format prevents overflow. */ |
| 1422 | |
| 1423 | case 'f': |
| 1424 | case 'e': |
| 1425 | case 'E': |
| 1426 | case 'g': |
| 1427 | case 'G': |
| 1428 | if (precision < 0) precision = 6; |
| 1429 | if (p >= last - precision - 8) { yield = FALSE; goto END_FORMAT; } |
| 1430 | strncpy(newformat, item_start, fp - item_start); |
| 1431 | newformat[fp-item_start] = 0; |
| 1432 | if (length == L_LONGDOUBLE) |
| 1433 | p += sprintf(CS p, newformat, va_arg(ap, long double)); |
| 1434 | else |
| 1435 | p += sprintf(CS p, newformat, va_arg(ap, double)); |
| 1436 | break; |
| 1437 | |
| 1438 | /* String types */ |
| 1439 | |
| 1440 | case '%': |
| 1441 | if (p >= last) { yield = FALSE; goto END_FORMAT; } |
| 1442 | *p++ = '%'; |
| 1443 | break; |
| 1444 | |
| 1445 | case 'c': |
| 1446 | if (p >= last) { yield = FALSE; goto END_FORMAT; } |
| 1447 | *p++ = va_arg(ap, int); |
| 1448 | break; |
| 1449 | |
| 1450 | case 'D': /* Insert daily datestamp for log file names */ |
| 1451 | s = CS tod_stamp(tod_log_datestamp_daily); |
| 1452 | string_datestamp_offset = p - buffer; /* Passed back via global */ |
| 1453 | string_datestamp_length = Ustrlen(s); /* Passed back via global */ |
| 1454 | string_datestamp_type = tod_log_datestamp_daily; |
| 1455 | slen = string_datestamp_length; |
| 1456 | goto INSERT_STRING; |
| 1457 | |
| 1458 | case 'M': /* Insert monthly datestamp for log file names */ |
| 1459 | s = CS tod_stamp(tod_log_datestamp_monthly); |
| 1460 | string_datestamp_offset = p - buffer; /* Passed back via global */ |
| 1461 | string_datestamp_length = Ustrlen(s); /* Passed back via global */ |
| 1462 | string_datestamp_type = tod_log_datestamp_monthly; |
| 1463 | slen = string_datestamp_length; |
| 1464 | goto INSERT_STRING; |
| 1465 | |
| 1466 | case 's': |
| 1467 | case 'S': /* Forces *lower* case */ |
| 1468 | case 'T': /* Forces *upper* case */ |
| 1469 | s = va_arg(ap, char *); |
| 1470 | |
| 1471 | if (s == NULL) s = null; |
| 1472 | slen = Ustrlen(s); |
| 1473 | |
| 1474 | INSERT_STRING: /* Come to from %D or %M above */ |
| 1475 | |
| 1476 | /* If the width is specified, check that there is a precision |
| 1477 | set; if not, set it to the width to prevent overruns of long |
| 1478 | strings. */ |
| 1479 | |
| 1480 | if (width >= 0) |
| 1481 | { |
| 1482 | if (precision < 0) precision = width; |
| 1483 | } |
| 1484 | |
| 1485 | /* If a width is not specified and the precision is specified, set |
| 1486 | the width to the precision, or the string length if shorted. */ |
| 1487 | |
| 1488 | else if (precision >= 0) |
| 1489 | { |
| 1490 | width = (precision < slen)? precision : slen; |
| 1491 | } |
| 1492 | |
| 1493 | /* If neither are specified, set them both to the string length. */ |
| 1494 | |
| 1495 | else width = precision = slen; |
| 1496 | |
| 1497 | /* Check string space, and add the string to the buffer if ok. If |
| 1498 | not OK, add part of the string (debugging uses this to show as |
| 1499 | much as possible). */ |
| 1500 | |
| 1501 | if (p == last) |
| 1502 | { |
| 1503 | yield = FALSE; |
| 1504 | goto END_FORMAT; |
| 1505 | } |
| 1506 | if (p >= last - width) |
| 1507 | { |
| 1508 | yield = FALSE; |
| 1509 | width = precision = last - p - 1; |
| 1510 | if (width < 0) width = 0; |
| 1511 | if (precision < 0) precision = 0; |
| 1512 | } |
| 1513 | sprintf(CS p, "%*.*s", width, precision, s); |
| 1514 | if (fp[-1] == 'S') |
| 1515 | while (*p) { *p = tolower(*p); p++; } |
| 1516 | else if (fp[-1] == 'T') |
| 1517 | while (*p) { *p = toupper(*p); p++; } |
| 1518 | else |
| 1519 | while (*p) p++; |
| 1520 | if (!yield) goto END_FORMAT; |
| 1521 | break; |
| 1522 | |
| 1523 | /* Some things are never used in Exim; also catches junk. */ |
| 1524 | |
| 1525 | default: |
| 1526 | strncpy(newformat, item_start, fp - item_start); |
| 1527 | newformat[fp-item_start] = 0; |
| 1528 | log_write(0, LOG_MAIN|LOG_PANIC_DIE, "string_format: unsupported type " |
| 1529 | "in \"%s\" in \"%s\"", newformat, format); |
| 1530 | break; |
| 1531 | } |
| 1532 | } |
| 1533 | |
| 1534 | /* Ensure string is complete; return TRUE if got to the end of the format */ |
| 1535 | |
| 1536 | END_FORMAT: |
| 1537 | |
| 1538 | *p = 0; |
| 1539 | return yield; |
| 1540 | } |
| 1541 | |
| 1542 | |
| 1543 | |
| 1544 | #ifndef COMPILE_UTILITY |
| 1545 | /************************************************* |
| 1546 | * Generate an "open failed" message * |
| 1547 | *************************************************/ |
| 1548 | |
| 1549 | /* This function creates a message after failure to open a file. It includes a |
| 1550 | string supplied as data, adds the strerror() text, and if the failure was |
| 1551 | "Permission denied", reads and includes the euid and egid. |
| 1552 | |
| 1553 | Arguments: |
| 1554 | eno the value of errno after the failure |
| 1555 | format a text format string - deliberately not uschar * |
| 1556 | ... arguments for the format string |
| 1557 | |
| 1558 | Returns: a message, in dynamic store |
| 1559 | */ |
| 1560 | |
| 1561 | uschar * |
| 1562 | string_open_failed(int eno, const char *format, ...) |
| 1563 | { |
| 1564 | va_list ap; |
| 1565 | uschar buffer[1024]; |
| 1566 | |
| 1567 | Ustrcpy(buffer, "failed to open "); |
| 1568 | va_start(ap, format); |
| 1569 | |
| 1570 | /* Use the checked formatting routine to ensure that the buffer |
| 1571 | does not overflow. It should not, since this is called only for internally |
| 1572 | specified messages. If it does, the message just gets truncated, and there |
| 1573 | doesn't seem much we can do about that. */ |
| 1574 | |
| 1575 | (void)string_vformat(buffer+15, sizeof(buffer) - 15, format, ap); |
| 1576 | va_end(ap); |
| 1577 | |
| 1578 | return (eno == EACCES)? |
| 1579 | string_sprintf("%s: %s (euid=%ld egid=%ld)", buffer, strerror(eno), |
| 1580 | (long int)geteuid(), (long int)getegid()) : |
| 1581 | string_sprintf("%s: %s", buffer, strerror(eno)); |
| 1582 | } |
| 1583 | #endif /* COMPILE_UTILITY */ |
| 1584 | |
| 1585 | |
| 1586 | |
| 1587 | |
| 1588 | |
| 1589 | #ifndef COMPILE_UTILITY |
| 1590 | /* qsort(3), currently used to sort the environment variables |
| 1591 | for -bP environment output, needs a function to compare two pointers to string |
| 1592 | pointers. Here it is. */ |
| 1593 | |
| 1594 | int |
| 1595 | string_compare_by_pointer(const void *a, const void *b) |
| 1596 | { |
| 1597 | return Ustrcmp(* CUSS a, * CUSS b); |
| 1598 | } |
| 1599 | #endif /* COMPILE_UTILITY */ |
| 1600 | |
| 1601 | |
| 1602 | |
| 1603 | /************************************************* |
| 1604 | ************************************************** |
| 1605 | * Stand-alone test program * |
| 1606 | ************************************************** |
| 1607 | *************************************************/ |
| 1608 | |
| 1609 | #ifdef STAND_ALONE |
| 1610 | int main(void) |
| 1611 | { |
| 1612 | uschar buffer[256]; |
| 1613 | |
| 1614 | printf("Testing is_ip_address\n"); |
| 1615 | |
| 1616 | while (fgets(CS buffer, sizeof(buffer), stdin) != NULL) |
| 1617 | { |
| 1618 | int offset; |
| 1619 | buffer[Ustrlen(buffer) - 1] = 0; |
| 1620 | printf("%d\n", string_is_ip_address(buffer, NULL)); |
| 1621 | printf("%d %d %s\n", string_is_ip_address(buffer, &offset), offset, buffer); |
| 1622 | } |
| 1623 | |
| 1624 | printf("Testing string_nextinlist\n"); |
| 1625 | |
| 1626 | while (fgets(CS buffer, sizeof(buffer), stdin) != NULL) |
| 1627 | { |
| 1628 | uschar *list = buffer; |
| 1629 | uschar *lp1, *lp2; |
| 1630 | uschar item[256]; |
| 1631 | int sep1 = 0; |
| 1632 | int sep2 = 0; |
| 1633 | |
| 1634 | if (*list == '<') |
| 1635 | { |
| 1636 | sep1 = sep2 = list[1]; |
| 1637 | list += 2; |
| 1638 | } |
| 1639 | |
| 1640 | lp1 = lp2 = list; |
| 1641 | for (;;) |
| 1642 | { |
| 1643 | uschar *item1 = string_nextinlist(&lp1, &sep1, item, sizeof(item)); |
| 1644 | uschar *item2 = string_nextinlist(&lp2, &sep2, NULL, 0); |
| 1645 | |
| 1646 | if (item1 == NULL && item2 == NULL) break; |
| 1647 | if (item == NULL || item2 == NULL || Ustrcmp(item1, item2) != 0) |
| 1648 | { |
| 1649 | printf("***ERROR\nitem1=\"%s\"\nitem2=\"%s\"\n", |
| 1650 | (item1 == NULL)? "NULL" : CS item1, |
| 1651 | (item2 == NULL)? "NULL" : CS item2); |
| 1652 | break; |
| 1653 | } |
| 1654 | else printf(" \"%s\"\n", CS item1); |
| 1655 | } |
| 1656 | } |
| 1657 | |
| 1658 | /* This is a horrible lash-up, but it serves its purpose. */ |
| 1659 | |
| 1660 | printf("Testing string_format\n"); |
| 1661 | |
| 1662 | while (fgets(CS buffer, sizeof(buffer), stdin) != NULL) |
| 1663 | { |
| 1664 | void *args[3]; |
| 1665 | long long llargs[3]; |
| 1666 | double dargs[3]; |
| 1667 | int dflag = 0; |
| 1668 | int llflag = 0; |
| 1669 | int n = 0; |
| 1670 | int count; |
| 1671 | int countset = 0; |
| 1672 | uschar format[256]; |
| 1673 | uschar outbuf[256]; |
| 1674 | uschar *s; |
| 1675 | buffer[Ustrlen(buffer) - 1] = 0; |
| 1676 | |
| 1677 | s = Ustrchr(buffer, ','); |
| 1678 | if (s == NULL) s = buffer + Ustrlen(buffer); |
| 1679 | |
| 1680 | Ustrncpy(format, buffer, s - buffer); |
| 1681 | format[s-buffer] = 0; |
| 1682 | |
| 1683 | if (*s == ',') s++; |
| 1684 | |
| 1685 | while (*s != 0) |
| 1686 | { |
| 1687 | uschar *ss = s; |
| 1688 | s = Ustrchr(ss, ','); |
| 1689 | if (s == NULL) s = ss + Ustrlen(ss); |
| 1690 | |
| 1691 | if (isdigit(*ss)) |
| 1692 | { |
| 1693 | Ustrncpy(outbuf, ss, s-ss); |
| 1694 | if (Ustrchr(outbuf, '.') != NULL) |
| 1695 | { |
| 1696 | dflag = 1; |
| 1697 | dargs[n++] = Ustrtod(outbuf, NULL); |
| 1698 | } |
| 1699 | else if (Ustrstr(outbuf, "ll") != NULL) |
| 1700 | { |
| 1701 | llflag = 1; |
| 1702 | llargs[n++] = strtoull(CS outbuf, NULL, 10); |
| 1703 | } |
| 1704 | else |
| 1705 | { |
| 1706 | args[n++] = (void *)Uatoi(outbuf); |
| 1707 | } |
| 1708 | } |
| 1709 | |
| 1710 | else if (Ustrcmp(ss, "*") == 0) |
| 1711 | { |
| 1712 | args[n++] = (void *)(&count); |
| 1713 | countset = 1; |
| 1714 | } |
| 1715 | |
| 1716 | else |
| 1717 | { |
| 1718 | uschar *sss = malloc(s - ss + 1); |
| 1719 | Ustrncpy(sss, ss, s-ss); |
| 1720 | args[n++] = sss; |
| 1721 | } |
| 1722 | |
| 1723 | if (*s == ',') s++; |
| 1724 | } |
| 1725 | |
| 1726 | if (!dflag && !llflag) |
| 1727 | printf("%s\n", string_format(outbuf, sizeof(outbuf), CS format, |
| 1728 | args[0], args[1], args[2])? "True" : "False"); |
| 1729 | |
| 1730 | else if (dflag) |
| 1731 | printf("%s\n", string_format(outbuf, sizeof(outbuf), CS format, |
| 1732 | dargs[0], dargs[1], dargs[2])? "True" : "False"); |
| 1733 | |
| 1734 | else printf("%s\n", string_format(outbuf, sizeof(outbuf), CS format, |
| 1735 | llargs[0], llargs[1], llargs[2])? "True" : "False"); |
| 1736 | |
| 1737 | printf("%s\n", CS outbuf); |
| 1738 | if (countset) printf("count=%d\n", count); |
| 1739 | } |
| 1740 | |
| 1741 | return 0; |
| 1742 | } |
| 1743 | #endif |
| 1744 | |
| 1745 | /* End of string.c */ |