Pileus Git - ~andy/fetchmail/blob - rfc822.c

   1 /*
   2  * rfc822.c -- code for slicing and dicing RFC822 mail headers
   3  *
   4  * Copyright 1997 by Eric S. Raymond
   5  * For license terms, see the file COPYING in this directory.
   6  */
   7
   8 #include  <stdio.h>
   9 #include  <ctype.h>
  10 #include  <string.h>
  11 #if defined(STDC_HEADERS)
  12 #include  <stdlib.h>
  13 #endif
  14
  15 #include "config.h"
  16 #include "fetchmail.h"
  17 #include "i18n.h"
  18
  19 #define HEADER_END(p)   ((p)[0] == '\n' && ((p)[1] != ' ' && (p)[1] != '\t' && (p)[1] != '\0'))
  20
  21 #ifdef TESTMAIN
  22 static int verbose;
  23 char *program_name = "rfc822";
  24 #endif /* TESTMAIN */
  25
  26 unsigned char *reply_hack(buf, host)
  27 /* hack message headers so replies will work properly */
  28 unsigned char *buf;             /* header to be hacked */
  29 const unsigned char *host;      /* server hostname */
  30 {
  31     unsigned char *from, *cp, last_nws = '\0', *parens_from = NULL;
  32     int parendepth, state, has_bare_name_part, has_host_part;
  33 #ifndef TESTMAIN
  34     int addresscount = 1;
  35 #endif /* TESTMAIN */
  36
  37     if (strncasecmp("From:", buf, 5)
  38         && strncasecmp("To:", buf, 3)
  39         && strncasecmp("Reply-To:", buf, 9)
  40         && strncasecmp("Return-Path:", buf, 12)
  41         && strncasecmp("Cc:", buf, 3)
  42         && strncasecmp("Bcc:", buf, 4)
  43         && strncasecmp("Resent-From:", buf, 12)
  44         && strncasecmp("Resent-To:", buf, 10)
  45         && strncasecmp("Resent-Cc:", buf, 10)
  46         && strncasecmp("Resent-Bcc:", buf, 11)
  47         && strncasecmp("Apparently-From:", buf, 16)
  48         && strncasecmp("Apparently-To:", buf, 14)
  49         && strncasecmp("Sender:", buf, 7)
  50         && strncasecmp("Resent-Sender:", buf, 14)
  51        ) {
  52         return(buf);
  53     }
  54
  55 #ifndef TESTMAIN
  56     if (outlevel >= O_DEBUG)
  57         report_build(stdout, _("About to rewrite %s"), buf);
  58
  59     /* make room to hack the address; buf must be malloced */
  60     for (cp = buf; *cp; cp++)
  61         if (*cp == ',' || isspace(*cp))
  62             addresscount++;
  63     buf = (unsigned char *)xrealloc(buf, strlen(buf) + addresscount * strlen(host) + 1);
  64 #endif /* TESTMAIN */
  65
  66     /*
  67      * This is going to foo up on some ill-formed addresses.
  68      * Note that we don't rewrite the fake address <> in order to
  69      * avoid screwing up bounce suppression with a null Return-Path.
  70      */
  71
  72     parendepth = state = 0;
  73     has_host_part = has_bare_name_part = FALSE;
  74     for (from = buf; *from; from++)
  75     {
  76 #ifdef TESTMAIN
  77         if (verbose)
  78         {
  79             printf("state %d: %s", state, buf);
  80             printf("%*s^\n", from - buf + 10, " ");
  81         }
  82 #endif /* TESTMAIN */
  83         if (state != 2)
  84         {
  85             if (*from == '(')
  86                 ++parendepth;
  87             else if (*from == ')')
  88                 --parendepth;
  89         }
  90
  91         if (!parendepth && !has_host_part)
  92             switch (state)
  93             {
  94             case 0:     /* before header colon */
  95                 if (*from == ':')
  96                     state = 1;
  97                 break;
  98
  99             case 1:     /* we've seen the colon, we're looking for addresses */
 100                 if (!isspace(*from))
 101                     last_nws = *from;
 102                 if (*from == '<')
 103                     state = 3;
 104                 else if (*from == '@')
 105                     has_host_part = TRUE;
 106                 else if (*from == '"')
 107                     state = 2;
 108                 /*
 109                  * Not expanding on last non-WS == ';' deals with groupnames,
 110                  * an obscure misfeature described in sections
 111                  * 6.1, 6.2.6, and A.1.5 of the RFC822 standard.
 112                  */
 113                 else if ((*from == ',' || HEADER_END(from))
 114                          && has_bare_name_part
 115                          && !has_host_part
 116                          && last_nws != ';')
 117                 {
 118                     int hostlen;
 119                     unsigned char *p;
 120
 121                     p = from;
 122                     if (parens_from)
 123                         from = parens_from;
 124                     while (isspace(*from) || (*from == ','))
 125                         --from;
 126                     from++;
 127                     hostlen = strlen(host);
 128                     for (cp = from + strlen(from); cp >= from; --cp)
 129                         cp[hostlen+1] = *cp;
 130                     *from++ = '@';
 131                     memcpy(from, host, hostlen);
 132                     from = p + hostlen + 1;
 133                     has_host_part = TRUE;
 134                 }
 135                 else if (from[1] == '('
 136                          && has_bare_name_part
 137                          && !has_host_part
 138                          && last_nws != ';' && last_nws != ')')
 139                 {
 140                     parens_from = from;
 141                 }
 142                 else if (!isspace(*from))
 143                     has_bare_name_part = TRUE;
 144                 break;
 145
 146             case 2:     /* we're in a string */
 147                 if (*from == '"')
 148                     state = 1;
 149                 break;
 150
 151             case 3:     /* we're in a <>-enclosed address */
 152                 if (*from == '@')
 153                     has_host_part = TRUE;
 154                 else if (*from == '>' && from[-1] != '<')
 155                 {
 156                     state = 1;
 157                     if (!has_host_part)
 158                     {
 159                         int hostlen;
 160
 161                         hostlen = strlen(host);
 162                         for (cp = from + strlen(from); cp >= from; --cp)
 163                             cp[hostlen+1] = *cp;
 164                         *from++ = '@';
 165                         memcpy(from, host, hostlen);
 166                         from += hostlen;
 167                         has_host_part = TRUE;
 168                     }
 169                 }
 170                 break;
 171             }
 172
 173         /*
 174          * If we passed a comma, reset everything.
 175          */
 176         if (from[-1] == ',' && !parendepth) {
 177           has_host_part = has_bare_name_part = FALSE;
 178           parens_from = NULL;
 179         }
 180     }
 181
 182 #ifndef TESTMAIN
 183     if (outlevel >= O_DEBUG)
 184         report_complete(stdout, _("Rewritten version is %s\n"), buf);
 185 #endif /* TESTMAIN */
 186     return(buf);
 187 }
 188
 189 unsigned char *nxtaddr(hdr)
 190 /* parse addresses in succession out of a specified RFC822 header */
 191 const unsigned char *hdr;       /* header to be parsed, NUL to continue previous hdr */
 192 {
 193     static unsigned char address[POPBUFSIZE+1];
 194     static int tp;
 195     static const unsigned char *hp;
 196     static int  state, oldstate;
 197 #ifdef TESTMAIN
 198     static const unsigned char *orighdr;
 199 #endif /* TESTMAIN */
 200     int parendepth = 0;
 201
 202 #define START_HDR       0       /* before header colon */
 203 #define SKIP_JUNK       1       /* skip whitespace, \n, and junk */
 204 #define BARE_ADDRESS    2       /* collecting address without delimiters */
 205 #define INSIDE_DQUOTE   3       /* inside double quotes */
 206 #define INSIDE_PARENS   4       /* inside parentheses */
 207 #define INSIDE_BRACKETS 5       /* inside bracketed address */
 208 #define ENDIT_ALL       6       /* after last address */
 209
 210 #define NEXTTP()        ((tp < sizeof(address)-1) ? tp++ : tp)
 211
 212     if (hdr)
 213     {
 214         hp = hdr;
 215         state = START_HDR;
 216 #ifdef TESTMAIN
 217         orighdr = hdr;
 218 #endif /* TESTMAIN */
 219         tp = 0;
 220     }
 221
 222     for (; *hp; hp++)
 223     {
 224 #ifdef TESTMAIN
 225         if (verbose)
 226         {
 227             printf("state %d: %s", state, orighdr);
 228             printf("%*s^\n", hp - orighdr + 10, " ");
 229         }
 230 #endif /* TESTMAIN */
 231
 232         if (state == ENDIT_ALL)         /* after last address */
 233             return(NULL);
 234         else if (HEADER_END(hp))
 235         {
 236             state = ENDIT_ALL;
 237             if (tp)
 238             {
 239                 while (isspace(address[--tp]))
 240                     continue;
 241                 address[++tp] = '\0';
 242                 tp = 0;
 243                 return (address);
 244             }
 245             return((unsigned char *)NULL);
 246         }
 247         else if (*hp == '\\')           /* handle RFC822 escaping */
 248         {
 249             if (state != INSIDE_PARENS)
 250             {
 251                 address[NEXTTP()] = *hp++;      /* take the escape */
 252                 address[NEXTTP()] = *hp;        /* take following unsigned char */
 253             }
 254         }
 255         else switch (state)
 256         {
 257         case START_HDR:   /* before header colon */
 258             if (*hp == ':')
 259                 state = SKIP_JUNK;
 260             break;
 261
 262         case SKIP_JUNK:         /* looking for address start */
 263             if (*hp == '"')     /* quoted string */
 264             {
 265                 oldstate = SKIP_JUNK;
 266                 state = INSIDE_DQUOTE;
 267                 address[NEXTTP()] = *hp;
 268             }
 269             else if (*hp == '(')        /* address comment -- ignore */
 270             {
 271                 parendepth = 1;
 272                 oldstate = SKIP_JUNK;
 273                 state = INSIDE_PARENS;
 274             }
 275             else if (*hp == '<')        /* begin <address> */
 276             {
 277                 state = INSIDE_BRACKETS;
 278                 tp = 0;
 279             }
 280             else if (*hp != ',' && !isspace(*hp))
 281             {
 282                 --hp;
 283                 state = BARE_ADDRESS;
 284             }
 285             break;
 286
 287         case BARE_ADDRESS:      /* collecting address without delimiters */
 288             if (*hp == ',')     /* end of address */
 289             {
 290                 if (tp)
 291                 {
 292                     address[NEXTTP()] = '\0';
 293                     state = SKIP_JUNK;
 294                     tp = 0;
 295                     return(address);
 296                 }
 297             }
 298             else if (*hp == '(')        /* beginning of comment */
 299             {
 300                 parendepth = 1;
 301                 oldstate = BARE_ADDRESS;
 302                 state = INSIDE_PARENS;
 303             }
 304             else if (*hp == '<')        /* beginning of real address */
 305             {
 306                 state = INSIDE_BRACKETS;
 307                 tp = 0;
 308             }
 309             else if (!isspace(*hp))     /* just take it, ignoring whitespace */
 310                 address[NEXTTP()] = *hp;
 311             break;
 312
 313         case INSIDE_DQUOTE:     /* we're in a quoted string, copy verbatim */
 314             if (*hp != '"')
 315                 address[NEXTTP()] = *hp;
 316             else
 317             {
 318                 address[NEXTTP()] = *hp;
 319                 state = oldstate;
 320             }
 321             break;
 322
 323         case INSIDE_PARENS:     /* we're in a parenthesized comment, ignore */
 324             if (*hp == '(')
 325                 ++parendepth;
 326             else if (*hp == ')')
 327                 --parendepth;
 328             if (parendepth == 0)
 329                 state = oldstate;
 330             break;
 331
 332         case INSIDE_BRACKETS:   /* possible <>-enclosed address */
 333             if (*hp == '>')     /* end of address */
 334             {
 335                 address[NEXTTP()] = '\0';
 336                 state = SKIP_JUNK;
 337                 ++hp;
 338                 tp = 0;
 339                 return(address);
 340             }
 341             else if (*hp == '<')        /* nested <> */
 342                 tp = 0;
 343             else if (*hp == '"')        /* quoted address */
 344             {
 345                 address[NEXTTP()] = *hp;
 346                 oldstate = INSIDE_BRACKETS;
 347                 state = INSIDE_DQUOTE;
 348             }
 349             else                        /* just copy address */
 350                 address[NEXTTP()] = *hp;
 351             break;
 352         }
 353     }
 354
 355     return(NULL);
 356 }
 357
 358 #ifdef TESTMAIN
 359 static void parsebuf(unsigned char *longbuf, int reply)
 360 {
 361     unsigned char       *cp;
 362
 363     if (reply)
 364     {
 365         reply_hack(longbuf, "HOSTNAME.NET");
 366         printf("Rewritten buffer: %s", longbuf);
 367     }
 368     else
 369         if ((cp = nxtaddr(longbuf)) != (unsigned char *)NULL)
 370             do {
 371                 printf("\t-> \"%s\"\n", cp);
 372             } while
 373                 ((cp = nxtaddr((unsigned char *)NULL)) != (unsigned char *)NULL);
 374 }
 375
 376
 377
 378 main(int argc, char *argv[])
 379 {
 380     unsigned char       buf[MSGBUFSIZE], longbuf[BUFSIZ];
 381     int                 ch, reply;
 382
 383     verbose = reply = FALSE;
 384     while ((ch = getopt(argc, argv, "rv")) != EOF)
 385         switch(ch)
 386         {
 387         case 'r':
 388             reply = TRUE;
 389             break;
 390
 391         case 'v':
 392             verbose = TRUE;
 393             break;
 394         }
 395
 396     while (fgets(buf, sizeof(buf)-1, stdin))
 397     {
 398         if (buf[0] == ' ' || buf[0] == '\t')
 399             strcat(longbuf, buf);
 400         else if (!strncasecmp("From: ", buf, 6)
 401                     || !strncasecmp("To: ", buf, 4)
 402                     || !strncasecmp("Reply-", buf, 6)
 403                     || !strncasecmp("Cc: ", buf, 4)
 404                     || !strncasecmp("Bcc: ", buf, 5))
 405             strcpy(longbuf, buf);
 406         else if (longbuf[0])
 407         {
 408             if (verbose)
 409                 fputs(longbuf, stdout);
 410             parsebuf(longbuf, reply);
 411             longbuf[0] = '\0';
 412         }
 413     }
 414     if (longbuf[0])
 415     {
 416         if (verbose)
 417             fputs(longbuf, stdout);
 418         parsebuf(longbuf, reply);
 419     }
 420 }
 421 #endif /* TESTMAIN */
 422
 423 /* rfc822.c end */