diplodocus.org Git - nmh/blob - uip/scansbr.c

   1
   2 /*
   3  * scansbr.c -- routines to help scan along...
   4  *
   5  * This code is Copyright (c) 2002, by the authors of nmh.  See the
   6  * COPYRIGHT file in the root directory of the nmh distribution for
   7  * complete copyright information.
   8  */
   9
  10 #include <h/mh.h>
  11 #include <h/addrsbr.h>
  12 #include <h/fmt_scan.h>
  13 #include <h/scansbr.h>
  14 #include <h/tws.h>
  15 #include <h/utils.h>
  16
  17 #define MAXSCANL 256            /* longest possible scan line */
  18
  19 /*
  20  * Buffer size for content part of header fields.  We want this
  21  * to be large enough so that we don't do a lot of extra FLDPLUS
  22  * calls on m_getfld but small enough so that we don't snarf
  23  * the entire message body when we're only going to display 30
  24  * characters of it.
  25  */
  26 #define SBUFSIZ 512
  27
  28 static struct format *fmt;
  29 static struct comp *datecomp;           /* pntr to "date" comp             */
  30 static struct comp *bodycomp;           /* pntr to "body" pseudo-comp      *
  31                                          * (if referenced)                 */
  32 static int ncomps = 0;                  /* # of interesting components     */
  33 static char **compbuffers = 0;          /* buffers for component text      */
  34 static struct comp **used_buf = 0;      /* stack for comp that use buffers */
  35
  36 static int dat[5];                      /* aux. data for format routine    */
  37
  38 char *scanl = 0;                        /* text of most recent scanline    */
  39 m_getfld_state_t gstate;                /* for access by msh */
  40
  41 #define DIEWRERR() adios (scnmsg, "write error on")
  42
  43 #define FPUTS(buf) {\
  44                 if (mh_fputs(buf,scnout) == EOF)\
  45                     DIEWRERR();\
  46                 }
  47
  48 /*
  49  * prototypes
  50  */
  51 static int mh_fputs(char *, FILE *);
  52
  53 #ifdef MULTIBYTE_SUPPORT
  54 #define SCAN_CHARWIDTH MB_CUR_MAX
  55 #else
  56 #define SCAN_CHARWIDTH 1
  57 #endif
  58
  59 int
  60 scan (FILE *inb, int innum, int outnum, char *nfs, int width, int curflg,
  61       int unseen, char *folder, long size, int noisy)
  62 {
  63     int i, compnum, encrypted, state;
  64     unsigned char *cp, *tmpbuf, *startbody;
  65     char **nxtbuf;
  66     char *saved_c_text = NULL;
  67     struct comp *cptr;
  68     struct comp **savecomp;
  69     char *scnmsg = NULL;
  70     FILE *scnout = NULL;
  71     char name[NAMESZ];
  72     int bufsz;
  73     static int rlwidth, slwidth;
  74     static size_t scanl_size;
  75
  76     /* first-time only initialization */
  77     if (!scanl) {
  78         if (width == 0) {
  79             if ((width = sc_width ()) < WIDTH/2)
  80                 width = WIDTH/2;
  81             else if (width > MAXSCANL)
  82                 width = MAXSCANL;
  83         }
  84         dat[3] = slwidth = width;
  85         /* Arbitrarily allocate 20 * slwidth to provide room for lots
  86            of escape sequences. */
  87         scanl_size = SCAN_CHARWIDTH * (20 * slwidth + 2);
  88         scanl = (char *) mh_xmalloc (scanl_size);
  89         if (outnum)
  90             umask(~m_gmprot());
  91
  92         /* Compile format string */
  93         ncomps = fmt_compile (nfs, &fmt, 1) + 2;
  94
  95         bodycomp = fmt_findcomp("body");
  96         datecomp = fmt_findcomp("date");
  97         cptr = fmt_findcomp("folder");
  98         if (cptr && folder)
  99             cptr->c_text = getcpy(folder);
 100         if (fmt_addcompentry("encrypted")) {
 101                 ncomps++;
 102         }
 103         cptr =  fmt_findcomp("dtimenow");
 104         if (cptr)
 105             cptr->c_text = getcpy(dtimenow (0));
 106
 107         /*
 108          * In other programs I got rid of this complicated buffer switching,
 109          * but since scan reads lots of messages at once and this complicated
 110          * memory management, I decided to keep it; otherwise there was
 111          * the potential for a lot of malloc() and free()s, and I could
 112          * see the malloc() pool really getting fragmented.  Maybe it
 113          * wouldn't be an issue in practice; perhaps this will get
 114          * revisited someday.
 115          *
 116          * So, some notes for what's going on:
 117          *
 118          * nxtbuf is an array of pointers that contains malloc()'d buffers
 119          * to hold our component text.  used_buf is an array of struct comp
 120          * pointers that holds pointers to component structures we found while
 121          * processing a message.
 122          *
 123          * We read in the message with m_getfld(), using "tmpbuf" as our
 124          * input buffer.  tmpbuf is set at the start of message processing
 125          * to the first buffer in our buffer pool (nxtbuf).
 126          *
 127          * Every time we find a component we care about, we set that component's
 128          * text buffer to the current value of tmpbuf, and then switch tmpbuf
 129          * to the next buffer in our pool.  We also add that component to
 130          * our used_buf pool.
 131          *
 132          * When we're done, we go back and zero out all of the component
 133          * text buffer pointers that we saved in used_buf.
 134          *
 135          * Note that this means c_text memory is NOT owned by the fmt_module
 136          * and it's our responsibility to free it.
 137          */
 138
 139         nxtbuf = compbuffers = (char **) calloc((size_t) ncomps, sizeof(char *));
 140         if (nxtbuf == NULL)
 141             adios (NULL, "unable to allocate component buffers");
 142         used_buf = (struct comp **) calloc((size_t) (ncomps+1),
 143             sizeof(struct comp *));
 144         if (used_buf == NULL)
 145             adios (NULL, "unable to allocate component buffer stack");
 146         used_buf += ncomps+1; *--used_buf = 0;
 147         rlwidth = bodycomp && (width > SBUFSIZ) ? width : SBUFSIZ;
 148         for (i = ncomps; i--; )
 149             *nxtbuf++ = mh_xmalloc(rlwidth);
 150     }
 151
 152     /*
 153      * each-message initialization
 154      */
 155     nxtbuf = compbuffers;
 156     savecomp = used_buf;
 157     tmpbuf = *nxtbuf++;
 158     startbody = NULL;
 159     dat[0] = innum ? innum : outnum;
 160     dat[1] = curflg;
 161     dat[4] = unseen;
 162
 163     /*
 164      * Get the first field.  If the message is non-empty
 165      * and we're doing an "inc", open the output file.
 166      */
 167     bufsz = rlwidth;
 168     m_getfld_state_reset (&gstate);
 169     if ((state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) == FILEEOF) {
 170         if (ferror(inb)) {
 171             advise("read", "unable to"); /* "read error" */
 172             return SCNFAT;
 173         } else {
 174             return SCNEOF;
 175         }
 176     }
 177
 178     if (outnum) {
 179         if (outnum > 0) {
 180             scnmsg = m_name (outnum);
 181             if (*scnmsg == '?')         /* msg num out of range */
 182                 return SCNNUM;
 183         } else {
 184             scnmsg = "/dev/null";
 185         }
 186         if ((scnout = fopen (scnmsg, "w")) == NULL)
 187             adios (scnmsg, "unable to write");
 188     }
 189
 190     /* scan - main loop */
 191     for (compnum = 1; ;
 192         bufsz = rlwidth, state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) {
 193         switch (state) {
 194             case FLD:
 195             case FLDPLUS:
 196                 compnum++;
 197                 if (outnum) {
 198                     FPUTS (name);
 199                     if ( putc (':', scnout) == EOF) DIEWRERR();
 200                     FPUTS (tmpbuf);
 201                 }
 202                 /*
 203                  * if we're interested in this component, save a pointer
 204                  * to the component text, then start using our next free
 205                  * buffer as the component temp buffer (buffer switching
 206                  * saves an extra copy of the component text).
 207                  */
 208                 if ((cptr = fmt_findcasecomp(name))) {
 209                     if (! cptr->c_text) {
 210                         cptr->c_text = tmpbuf;
 211                         for (cp = tmpbuf + strlen (tmpbuf) - 1;
 212                                         cp >= tmpbuf; cp--)
 213                             if (isspace (*cp))
 214                                 *cp = 0;
 215                             else
 216                                 break;
 217                         *--savecomp = cptr;
 218                         tmpbuf = *nxtbuf++;
 219                     }
 220                 }
 221
 222                 while (state == FLDPLUS) {
 223                     bufsz = rlwidth;
 224                     state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb);
 225                     if (outnum)
 226                         FPUTS (tmpbuf);
 227                 }
 228                 break;
 229
 230             case BODY:
 231                 compnum = -1;
 232                 /*
 233                  * A slight hack ... if we have less than rlwidth characters
 234                  * in the buffer, call m_getfld again.
 235                  */
 236
 237                 if ((i = strlen(tmpbuf)) < rlwidth) {
 238                     bufsz = rlwidth - i;
 239                     state = m_getfld (&gstate, name, tmpbuf + i, &bufsz, inb);
 240                 }
 241
 242                 if (! outnum) {
 243                     state = FILEEOF; /* stop now if scan cmd */
 244                     if (bodycomp && startbody == NULL)
 245                         startbody = tmpbuf;
 246                     goto finished;
 247                 }
 248                 if (putc ('\n', scnout) == EOF) DIEWRERR();
 249                 FPUTS (tmpbuf);
 250                 /*
 251                  * The previous code here used to call m_getfld() using
 252                  * pointers to the underlying output stdio buffers to
 253                  * avoid the extra copy.  Tests by Markus Schnalke show
 254                  * no noticable performance loss on larger mailboxes
 255                  * if we incur an extra copy, and messing around with
 256                  * internal stdio buffers is becoming more and more
 257                  * unportable as times go on.  So from now on just deal
 258                  * with the overhead of an extra copy.
 259                  *
 260                  * Subtle change - with the previous code tmpbuf wasn't
 261                  * used, so we could reuse it for the {body} component.
 262                  * Now since we're using tmpbuf as our read buffer we
 263                  * need to save the beginning of the body for later.
 264                  * See the above (and below) use of startbody.
 265                  */
 266 body:;
 267                 if (bodycomp && startbody == NULL) {
 268                     startbody = tmpbuf;
 269                     tmpbuf = *nxtbuf++;
 270                 }
 271
 272                 while (state == BODY) {
 273                     bufsz = rlwidth;
 274                     state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb);
 275                     FPUTS(tmpbuf);
 276                 }
 277                 goto finished;
 278
 279             case LENERR:
 280             case FMTERR:
 281                 fprintf (stderr,
 282                         innum ? "??Format error (message %d) in "
 283                               : "??Format error in ",
 284                         outnum ? outnum : innum);
 285                 fprintf (stderr, "component %d\n", compnum);
 286
 287                 if (outnum) {
 288                     FPUTS ("\n\nBAD MSG:\n");
 289                     FPUTS (name);
 290                     if (putc ('\n', scnout) == EOF) DIEWRERR();
 291                     state = BODY;
 292                     goto body;
 293                 }
 294                 /* fall through */
 295
 296             case FILEEOF:
 297                 goto finished;
 298
 299             default:
 300                 adios (NULL, "getfld() returned %d", state);
 301         }
 302     }
 303
 304     /*
 305      * format and output the scan line.
 306      */
 307 finished:
 308     if (ferror(inb)) {
 309         advise("read", "unable to"); /* "read error" */
 310         return SCNFAT;
 311     }
 312
 313     /* Save and restore buffer so we don't trash our dynamic pool! */
 314     if (bodycomp) {
 315         saved_c_text = bodycomp->c_text;
 316         bodycomp->c_text = startbody;
 317     }
 318
 319     if (size)
 320         dat[2] = size;
 321     else if (outnum > 0)
 322     {
 323         dat[2] = ftell(scnout);
 324         if (dat[2] == EOF) DIEWRERR();
 325     }
 326
 327     if ((datecomp && !datecomp->c_text) || (!size && !outnum)) {
 328         struct stat st;
 329
 330         fstat (fileno(inb), &st);
 331         if (!size && !outnum)
 332             dat[2] = st.st_size;
 333         if (datecomp) {
 334             if (! datecomp->c_text) {
 335                 if (datecomp->c_tws == NULL)
 336                     datecomp->c_tws = (struct tws *)
 337                         calloc((size_t) 1, sizeof(*datecomp->c_tws));
 338                 if (datecomp->c_tws == NULL)
 339                     adios (NULL, "unable to allocate tws buffer");
 340                 *datecomp->c_tws = *dlocaltime ((time_t *) &st.st_mtime);
 341                 datecomp->c_flags |= CF_DATEFAB|CF_TRUE;
 342             } else {
 343                 datecomp->c_flags &= ~CF_DATEFAB;
 344             }
 345         }
 346     }
 347
 348     fmt_scan (fmt, scanl, scanl_size, slwidth, dat);
 349
 350     if (bodycomp)
 351         bodycomp->c_text = saved_c_text;
 352
 353     if (noisy)
 354         fputs (scanl, stdout);
 355
 356     cptr = fmt_findcomp ("encrypted");
 357     encrypted = cptr && cptr->c_text;
 358
 359     /* return dynamically allocated buffers to pool */
 360     while ((cptr = *savecomp++)) {
 361         cptr->c_text = NULL;
 362     }
 363
 364     if (outnum && (ferror(scnout) || fclose (scnout) == EOF))
 365         DIEWRERR();
 366
 367     return (state != FILEEOF ? SCNERR : encrypted ? SCNENC : SCNMSG);
 368 }
 369
 370
 371 static int
 372 mh_fputs(char *s, FILE *stream)
 373 {
 374     char c;
 375
 376     while ((c = *s++))
 377         if (putc (c,stream) == EOF )
 378             return(EOF);
 379     return (0);
 380 }
 381
 382 /* The following three functions allow access to the global gstate above. */
 383 void
 384 scan_finished () {
 385     m_getfld_state_destroy (&gstate);
 386 }
 387
 388 void
 389 scan_detect_mbox_style (FILE *f) {
 390     m_unknown (&gstate, f);
 391 }
 392
 393 void
 394 scan_eom_action (int (*action)()) {
 395     m_eomsbr (gstate, action);
 396 }
 397
 398 void
 399 scan_reset_m_getfld_state () {
 400     m_getfld_state_reset (&gstate);
 401 }