1 # include <stdio.h> 2 # include <ctype.h> 3 # include "postbox.h" 4 5 static char SccsId[] = "@(#)parseaddr.c 3.2 03/07/81"; 6 7 /* 8 ** PARSE -- Parse an address 9 ** 10 ** Parses an address and breaks it up into three parts: a 11 ** net to transmit the message on, the host to transmit it 12 ** to, and a user on that host. These are loaded into an 13 ** ADDRESS header with the values squirreled away if necessary. 14 ** The "user" part may not be a real user; the process may 15 ** just reoccur on that machine. For example, on a machine 16 ** with an arpanet connection, the address 17 ** csvax.bill@berkeley 18 ** will break up to a "user" of 'csvax.bill' and a host 19 ** of 'berkeley' -- to be transmitted over the arpanet. 20 ** 21 ** Parameters: 22 ** addr -- the address to parse. 23 ** a -- a pointer to the address descriptor buffer. 24 ** If NULL, a header will be created. 25 ** copyf -- determines what shall be copied: 26 ** -1 -- don't copy anything. The printname 27 ** (q_paddr) is just addr, and the 28 ** user & host are allocated internally 29 ** to parse. 30 ** 0 -- copy out the parsed user & host, but 31 ** don't copy the printname. 32 ** +1 -- copy everything. 33 ** 34 ** Returns: 35 ** A pointer to the address descriptor header (`a' if 36 ** `a' is non-NULL). 37 ** NULL on error. 38 ** 39 ** Side Effects: 40 ** none 41 ** 42 ** Called By: 43 ** main 44 ** sendto 45 ** alias 46 ** savemail 47 */ 48 49 # define DELIMCHARS "()<>@!.,;:\\\" \t\r\n" /* word delimiters */ 50 # define SPACESUB ('.'|0200) /* substitution for <lwsp> */ 51 52 ADDRESS * 53 parse(addr, a, copyf) 54 char *addr; 55 register ADDRESS *a; 56 int copyf; 57 { 58 register char *p; 59 register struct parsetab *t; 60 extern struct parsetab ParseTab[]; 61 static char buf[MAXNAME]; 62 register char c; 63 register char *q; 64 bool got_one; 65 extern char *prescan(); 66 extern char *xalloc(); 67 extern char *newstr(); 68 char **pvp; 69 70 /* 71 ** Initialize and prescan address. 72 */ 73 74 To = addr; 75 if (prescan(addr, buf, &buf[sizeof buf], '\0') == NULL) 76 return (NULL); 77 78 /* 79 ** Scan parse table. 80 ** Look for the first entry designating a character 81 ** that is contained in the address. 82 ** Arrange for q to point to that character. 83 ** Check to see that there is only one of the char 84 ** if it must be unique. 85 ** Find the last one if the host is on the RHS. 86 ** Insist that the host name is atomic. 87 ** If just doing a map, do the map and then start all 88 ** over. 89 */ 90 91 rescan: 92 got_one = FALSE; 93 for (t = ParseTab; t->p_char != '\0'; t++) 94 { 95 q = NULL; 96 for (p = buf; (c = *p) != '\0'; p++) 97 { 98 /* find the end of this token */ 99 while (isalnum(c) || c == '-' || c == '_') 100 c = *++p; 101 if (c == '\0') 102 break; 103 104 if (c == t->p_char) 105 { 106 got_one = TRUE; 107 108 /* do mapping as appropriate */ 109 if (bitset(P_MAP, t->p_flags)) 110 { 111 *p = t->p_arg[0]; 112 if (bitset(P_ONE, t->p_flags)) 113 goto rescan; 114 else 115 continue; 116 } 117 118 /* arrange for q to point to it */ 119 if (q != NULL && bitset(P_ONE, t->p_flags)) 120 { 121 usrerr("multichar error"); 122 ExitStat = EX_USAGE; 123 return (NULL); 124 } 125 if (q == NULL || bitset(P_HLAST, t->p_flags)) 126 q = p; 127 } 128 else 129 { 130 /* insist that host name is atomic */ 131 if (bitset(P_HLAST, t->p_flags)) 132 q = NULL; 133 else 134 break; 135 } 136 } 137 138 if (q != NULL) 139 break; 140 } 141 142 /* 143 ** If we matched nothing cleanly, but we did match something 144 ** somewhere in the process of scanning, then we have a 145 ** syntax error. This can happen on things like a@b:c where 146 ** @ has a right host and : has a left host. 147 ** 148 ** We also set `q' to the null string, in case someone forgets 149 ** to put the P_MOVE bit in the local mailer entry of the 150 ** configuration table. 151 */ 152 153 if (q == NULL) 154 { 155 q = ""; 156 if (got_one) 157 { 158 usrerr("syntax error"); 159 ExitStat = EX_USAGE; 160 return (NULL); 161 } 162 } 163 164 /* 165 ** Interpret entry. 166 ** t points to the entry for the mailer we will use. 167 ** q points to the significant character. 168 */ 169 170 if (a == NULL) 171 a = (ADDRESS *) xalloc(sizeof *a); 172 if (copyf > 0) 173 a->q_paddr = newstr(addr); 174 else 175 a->q_paddr = addr; 176 a->q_rmailer = t->p_mailer; 177 a->q_mailer = &Mailer[t->p_mailer]; 178 179 if (bitset(P_MOVE, t->p_flags)) 180 { 181 /* send the message to another host & retry */ 182 a->q_host = t->p_arg; 183 if (copyf >= 0) 184 a->q_user = newstr(buf); 185 else 186 a->q_user = buf; 187 } 188 else 189 { 190 /* 191 ** Make local copies of the host & user and then 192 ** transport them out. 193 */ 194 195 *q++ = '\0'; 196 if (bitset(P_HLAST, t->p_flags)) 197 { 198 a->q_host = q; 199 a->q_user = buf; 200 } 201 else 202 { 203 a->q_host = buf; 204 a->q_user = q; 205 } 206 207 /* 208 ** Don't go to the net if already on the target host. 209 ** This is important on the berkeley network, since 210 ** it get confused if we ask to send to ourselves. 211 ** For nets like the ARPANET, we probably will have 212 ** the local list set to NULL to simplify testing. 213 ** The canonical representation of the name is also set 214 ** to be just the local name so the duplicate letter 215 ** suppression algorithm will work. 216 */ 217 218 if ((pvp = a->q_mailer->m_local) != NULL) 219 { 220 while (*pvp != NULL) 221 { 222 auto char buf2[MAXNAME]; 223 224 strcpy(buf2, a->q_host); 225 if (!bitset(P_HST_UPPER, t->p_flags)) 226 makelower(buf2); 227 if (strcmp(*pvp++, buf2) == 0) 228 { 229 strcpy(buf2, a->q_user); 230 p = a->q_paddr; 231 if (parse(buf2, a, -1) == NULL) 232 { 233 To = addr; 234 return (NULL); 235 } 236 To = a->q_paddr = p; 237 break; 238 } 239 } 240 } 241 242 /* make copies if specified */ 243 if (copyf >= 0) 244 { 245 a->q_host = newstr(a->q_host); 246 a->q_user = newstr(a->q_user); 247 } 248 } 249 250 /* 251 ** Do UPPER->lower case mapping unless inhibited. 252 */ 253 254 if (!bitset(P_HST_UPPER, t->p_flags)) 255 makelower(a->q_host); 256 if (!bitset(P_USR_UPPER, t->p_flags)) 257 makelower(a->q_user); 258 259 /* 260 ** Compute return value. 261 */ 262 263 # ifdef DEBUG 264 if (Debug) 265 printf("parse(\"%s\"): host \"%s\" user \"%s\" mailer %d\n", 266 addr, a->q_host, a->q_user, t->p_mailer); 267 # endif DEBUG 268 269 return (a); 270 } 271 /* 272 ** MAKELOWER -- Translate a line into lower case 273 ** 274 ** Parameters: 275 ** p -- the string to translate. If NULL, return is 276 ** immediate. 277 ** 278 ** Returns: 279 ** none. 280 ** 281 ** Side Effects: 282 ** String pointed to by p is translated to lower case. 283 ** 284 ** Called By: 285 ** parse 286 */ 287 288 makelower(p) 289 register char *p; 290 { 291 register char c; 292 293 if (p == NULL) 294 return; 295 for (; (c = *p) != '\0'; p++) 296 if ((c & 0200) == 0 && isupper(c)) 297 *p = c - 'A' + 'a'; 298 } 299 /* 300 ** PRESCAN -- Prescan name and make it canonical 301 ** 302 ** Scans a name and turns it into canonical form. This involves 303 ** deleting blanks, comments (in parentheses), and turning the 304 ** word "at" into an at-sign ("@"). The name is copied as this 305 ** is done; it is legal to copy a name onto itself, since this 306 ** process can only make things smaller. 307 ** 308 ** This routine knows about quoted strings and angle brackets. 309 ** 310 ** There are certain subtleties to this routine. The one that 311 ** comes to mind now is that backslashes on the ends of names 312 ** are silently stripped off; this is intentional. The problem 313 ** is that some versions of sndmsg (like at LBL) set the kill 314 ** character to something other than @ when reading addresses; 315 ** so people type "csvax.eric\@berkeley" -- which screws up the 316 ** berknet mailer. 317 ** 318 ** Parameters: 319 ** addr -- the name to chomp. 320 ** buf -- the buffer to copy it into. 321 ** buflim -- the last usable address in the buffer 322 ** (which will old a null byte). Normally 323 ** &buf[sizeof buf - 1]. 324 ** delim -- the delimiter for the address, normally 325 ** '\0' or ','; \0 is accepted in any case. 326 ** are moving in place; set buflim to high core. 327 ** 328 ** Returns: 329 ** A pointer to the terminator of buf. 330 ** NULL on error. 331 ** 332 ** Side Effects: 333 ** buf gets clobbered. 334 ** 335 ** Called By: 336 ** parse 337 ** maketemp 338 */ 339 340 char * 341 prescan(addr, buf, buflim, delim) 342 char *addr; 343 char *buf; 344 char *buflim; 345 char delim; 346 { 347 register char *p; 348 bool space; 349 bool quotemode; 350 bool bslashmode; 351 bool delimmode; 352 int cmntcnt; 353 int brccnt; 354 register char c; 355 register char *q; 356 extern char *index(); 357 358 space = FALSE; 359 delimmode = TRUE; 360 q = buf; 361 bslashmode = quotemode = FALSE; 362 cmntcnt = brccnt = 0; 363 for (p = addr; (c = *p++) != '\0'; ) 364 { 365 /* chew up special characters */ 366 *q = '\0'; 367 if (bslashmode) 368 { 369 c |= 0200; 370 bslashmode = FALSE; 371 } 372 else if (c == '"') 373 quotemode = !quotemode; 374 else if (c == '\\') 375 { 376 bslashmode++; 377 continue; 378 } 379 else if (quotemode) 380 c |= 0200; 381 else if (c == delim) 382 break; 383 else if (c == '(') 384 { 385 cmntcnt++; 386 continue; 387 } 388 else if (c == ')') 389 { 390 if (cmntcnt <= 0) 391 { 392 usrerr("Unbalanced ')'"); 393 return (NULL); 394 } 395 else 396 { 397 cmntcnt--; 398 continue; 399 } 400 } 401 if (cmntcnt > 0) 402 continue; 403 else if (isascii(c) && isspace(c) && (space || delimmode)) 404 continue; 405 else if (c == '<') 406 { 407 if (brccnt < 0) 408 { 409 usrerr("multiple < spec"); 410 return (NULL); 411 } 412 brccnt++; 413 delimmode = TRUE; 414 space = FALSE; 415 if (brccnt == 1) 416 { 417 /* we prefer using machine readable name */ 418 q = buf; 419 *q = '\0'; 420 continue; 421 } 422 } 423 else if (c == '>') 424 { 425 if (brccnt <= 0) 426 { 427 usrerr("Unbalanced `>'"); 428 return (NULL); 429 } 430 else 431 brccnt--; 432 if (brccnt <= 0) 433 { 434 brccnt = -1; 435 continue; 436 } 437 } 438 439 /* 440 ** Turn "at" into "@", 441 ** but only if "at" is a word. 442 ** By the way, I violate the ARPANET RFC-733 443 ** standard here, by assuming that 'space' delimits 444 ** atoms. I assume that is just a mistake, since 445 ** it violates the spirit of the semantics 446 ** of the document..... 447 */ 448 449 if (delimmode && (c == 'a' || c == 'A') && 450 (p[0] == 't' || p[0] == 'T') && 451 (index(DELIMCHARS, p[1]) != NULL || p[1] <= 040)) 452 { 453 c = '@'; 454 p++; 455 } 456 457 if (delimmode = (index(DELIMCHARS, c) != NULL)) 458 space = FALSE; 459 460 /* if not a space, squirrel it away */ 461 if ((!isascii(c) || !isspace(c)) && brccnt >= 0) 462 { 463 if (q >= buflim-1) 464 { 465 usrerr("Address too long"); 466 return (NULL); 467 } 468 if (space) 469 *q++ = SPACESUB; 470 *q++ = c; 471 } 472 space = isascii(c) && isspace(c); 473 } 474 *q = '\0'; 475 if (c == '\0') 476 p--; 477 if (cmntcnt > 0) 478 usrerr("Unbalanced '('"); 479 else if (quotemode) 480 usrerr("Unbalanced '\"'"); 481 else if (brccnt > 0) 482 usrerr("Unbalanced '<'"); 483 else if (buf[0] != '\0') 484 return (p); 485 return (NULL); 486 } 487