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 /*
  18  * Buffer size for content part of header fields.  We want this
  19  * to be large enough so that we don't do a lot of extra FLDPLUS
  20  * calls on m_getfld but small enough so that we don't snarf
  21  * the entire message body when we're only going to display 30
  22  * characters of it.
  23  */
  24 #define SBUFSIZ 512
  25
  26 static struct format *fmt;
  27 static struct comp *datecomp;           /* pntr to "date" comp             */
  28 static struct comp *bodycomp;           /* pntr to "body" pseudo-comp      *
  29                                          * (if referenced)                 */
  30 static int ncomps = 0;                  /* # of interesting components     */
  31 static char **compbuffers = 0;          /* buffers for component text      */
  32 static struct comp **used_buf = 0;      /* stack for comp that use buffers */
  33
  34 static int dat[5];                      /* aux. data for format routine    */
  35
  36 static m_getfld_state_t gstate;         /* for accessor functions below    */
  37
  38 #define DIEWRERR() adios (scnmsg, "write error on")
  39
  40 #define FPUTS(buf) {\
  41                 if (fputs(buf,scnout) == EOF)\
  42                     DIEWRERR();\
  43                 }
  44
  45 int
  46 scan (FILE *inb, int innum, int outnum, char *nfs, int width, int curflg,
  47       int unseen, char *folder, long size, int noisy, charstring_t *scanl)
  48 {
  49     int i, compnum, encrypted, state;
  50     char *cp, *tmpbuf, *startbody, **nxtbuf;
  51     char *saved_c_text = NULL;
  52     struct comp *cptr;
  53     struct comp **savecomp;
  54     char *scnmsg = NULL;
  55     FILE *scnout = NULL;
  56     char name[NAMESZ];
  57     int bufsz;
  58     static int rlwidth, slwidth;
  59
  60     /* first-time only initialization, which will always happen the
  61        way the code is now, with callers initializing *scanl to NULL.
  62        scanl used to be a global. */
  63     if (! *scanl) {
  64         if (width == -1) {
  65             /* Default:  width of the terminal, but at least WIDTH/2. */
  66             if ((width = sc_width ()) < WIDTH/2)
  67                 width = WIDTH/2;
  68         } else if (width == 0) {
  69             /* Unlimited width. */
  70             width = INT_MAX;
  71         }
  72         dat[3] = slwidth = width;
  73         *scanl = charstring_create (width < NMH_BUFSIZ ? width : NMH_BUFSIZ);
  74         if (outnum)
  75             umask(~m_gmprot());
  76
  77         /* Compile format string */
  78         ncomps = fmt_compile (nfs, &fmt, 1) + 2;
  79
  80         bodycomp = fmt_findcomp("body");
  81         datecomp = fmt_findcomp("date");
  82         cptr = fmt_findcomp("folder");
  83         if (cptr && folder)
  84             cptr->c_text = mh_xstrdup(folder);
  85         if (fmt_addcompentry("encrypted")) {
  86                 ncomps++;
  87         }
  88         cptr =  fmt_findcomp("dtimenow");
  89         if (cptr)
  90             cptr->c_text = getcpy(dtimenow (0));
  91
  92         /*
  93          * In other programs I got rid of this complicated buffer switching,
  94          * but since scan reads lots of messages at once and this complicated
  95          * memory management, I decided to keep it; otherwise there was
  96          * the potential for a lot of malloc() and free()s, and I could
  97          * see the malloc() pool really getting fragmented.  Maybe it
  98          * wouldn't be an issue in practice; perhaps this will get
  99          * revisited someday.
 100          *
 101          * So, some notes for what's going on:
 102          *
 103          * nxtbuf is an array of pointers that contains malloc()'d buffers
 104          * to hold our component text.  used_buf is an array of struct comp
 105          * pointers that holds pointers to component structures we found while
 106          * processing a message.
 107          *
 108          * We read in the message with m_getfld(), using "tmpbuf" as our
 109          * input buffer.  tmpbuf is set at the start of message processing
 110          * to the first buffer in our buffer pool (nxtbuf).
 111          *
 112          * Every time we find a component we care about, we set that component's
 113          * text buffer to the current value of tmpbuf, and then switch tmpbuf
 114          * to the next buffer in our pool.  We also add that component to
 115          * our used_buf pool.
 116          *
 117          * When we're done, we go back and zero out all of the component
 118          * text buffer pointers that we saved in used_buf.
 119          *
 120          * Note that this means c_text memory is NOT owned by the fmt_module
 121          * and it's our responsibility to free it.
 122          */
 123
 124         nxtbuf = compbuffers = mh_xcalloc(ncomps, sizeof *nxtbuf);
 125         used_buf = mh_xcalloc(ncomps + 1, sizeof *used_buf);
 126         used_buf += ncomps+1; *--used_buf = 0;
 127         rlwidth = bodycomp && (width > SBUFSIZ)
 128             ? min (width, NMH_BUFSIZ)
 129             : SBUFSIZ;
 130         for (i = ncomps; i--; )
 131             *nxtbuf++ = mh_xmalloc(rlwidth);
 132     }
 133
 134     /*
 135      * each-message initialization
 136      */
 137     nxtbuf = compbuffers;
 138     savecomp = used_buf;
 139     tmpbuf = *nxtbuf++;
 140     startbody = NULL;
 141     dat[0] = innum ? innum : outnum;
 142     dat[1] = curflg;
 143     dat[4] = unseen;
 144
 145     /*
 146      * Get the first field.  If the message is non-empty
 147      * and we're doing an "inc", open the output file.
 148      */
 149     bufsz = rlwidth;
 150     m_getfld_state_reset (&gstate);
 151     if ((state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) == FILEEOF) {
 152         if (ferror(inb)) {
 153             advise("read", "unable to"); /* "read error" */
 154             return SCNFAT;
 155         }
 156         return SCNEOF;
 157     }
 158
 159     if (outnum) {
 160         if (outnum > 0) {
 161             scnmsg = m_name (outnum);
 162             if (*scnmsg == '?')         /* msg num out of range */
 163                 return SCNNUM;
 164         } else {
 165             scnmsg = "/dev/null";
 166         }
 167         if ((scnout = fopen (scnmsg, "w")) == NULL)
 168             adios (scnmsg, "unable to write");
 169     }
 170
 171     /* scan - main loop */
 172     for (compnum = 1; ;
 173         bufsz = rlwidth, state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) {
 174         switch (state) {
 175             case FLD:
 176             case FLDPLUS:
 177                 compnum++;
 178                 if (outnum) {
 179                     FPUTS (name);
 180                     if ( putc (':', scnout) == EOF) DIEWRERR();
 181                     FPUTS (tmpbuf);
 182                 }
 183                 /*
 184                  * if we're interested in this component, save a pointer
 185                  * to the component text, then start using our next free
 186                  * buffer as the component temp buffer (buffer switching
 187                  * saves an extra copy of the component text).
 188                  */
 189                 if ((cptr = fmt_findcasecomp(name))) {
 190                     if (! cptr->c_text) {
 191                         cptr->c_text = tmpbuf;
 192                         for (cp = tmpbuf + strlen (tmpbuf) - 1;
 193                                         cp >= tmpbuf; cp--)
 194                             if (isspace ((unsigned char) *cp))
 195                                 *cp = 0;
 196                             else
 197                                 break;
 198                         *--savecomp = cptr;
 199                         tmpbuf = *nxtbuf++;
 200                     }
 201                 }
 202
 203                 while (state == FLDPLUS) {
 204                     bufsz = rlwidth;
 205                     state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb);
 206                     if (outnum)
 207                         FPUTS (tmpbuf);
 208                 }
 209                 break;
 210
 211             case BODY:
 212                 /*
 213                  * A slight hack ... if we have less than rlwidth characters
 214                  * in the buffer, call m_getfld again.
 215                  */
 216
 217                 if ((i = strlen(tmpbuf)) < rlwidth) {
 218                     bufsz = rlwidth - i;
 219                     state = m_getfld (&gstate, name, tmpbuf + i, &bufsz, inb);
 220                 }
 221
 222                 if (! outnum) {
 223                     state = FILEEOF; /* stop now if scan cmd */
 224                     if (bodycomp && startbody == NULL)
 225                         startbody = tmpbuf;
 226                     goto finished;
 227                 }
 228                 if (putc ('\n', scnout) == EOF) DIEWRERR();
 229                 FPUTS (tmpbuf);
 230                 /*
 231                  * The previous code here used to call m_getfld() using
 232                  * pointers to the underlying output stdio buffers to
 233                  * avoid the extra copy.  Tests by Markus Schnalke show
 234                  * no noticeable performance loss on larger mailboxes
 235                  * if we incur an extra copy, and messing around with
 236                  * internal stdio buffers is becoming more and more
 237                  * unportable as times go on.  So from now on just deal
 238                  * with the overhead of an extra copy.
 239                  *
 240                  * Subtle change - with the previous code tmpbuf wasn't
 241                  * used, so we could reuse it for the {body} component.
 242                  * Now since we're using tmpbuf as our read buffer we
 243                  * need to save the beginning of the body for later.
 244                  * See the above (and below) use of startbody.
 245                  */
 246 body:;
 247                 if (bodycomp && startbody == NULL) {
 248                     startbody = tmpbuf;
 249                     tmpbuf = *nxtbuf++;
 250                 }
 251
 252                 while (state == BODY) {
 253                     bufsz = rlwidth;
 254                     state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb);
 255                     FPUTS(tmpbuf);
 256                 }
 257                 goto finished;
 258
 259             case LENERR:
 260             case FMTERR:
 261                 if (innum)
 262                     fprintf (stderr, "??Format error (message %d) in ",
 263                              outnum ? outnum : innum);
 264                 else
 265                     fprintf (stderr, "??Format error in ");
 266
 267                 fprintf (stderr, "component %d\n", compnum);
 268
 269                 if (outnum) {
 270                     FPUTS ("\n\nBAD MSG:\n");
 271                     FPUTS (name);
 272                     if (putc ('\n', scnout) == EOF) DIEWRERR();
 273                     state = BODY;
 274                     goto body;
 275                 }
 276                 /* FALLTHRU */
 277
 278             case FILEEOF:
 279                 goto finished;
 280
 281             default:
 282                 adios (NULL, "getfld() returned %d", state);
 283         }
 284     }
 285
 286     /*
 287      * format and output the scan line.
 288      */
 289 finished:
 290     if (ferror(inb)) {
 291         advise("read", "unable to"); /* "read error" */
 292         return SCNFAT;
 293     }
 294
 295     /* Save and restore buffer so we don't trash our dynamic pool! */
 296     if (bodycomp) {
 297         saved_c_text = bodycomp->c_text;
 298         bodycomp->c_text = startbody;
 299     }
 300
 301     if (size)
 302         dat[2] = size;
 303     else if (outnum > 0)
 304     {
 305         dat[2] = ftell(scnout);
 306         if (dat[2] == EOF) DIEWRERR();
 307     }
 308
 309     if ((datecomp && !datecomp->c_text) || (!size && !outnum)) {
 310         struct stat st;
 311
 312         fstat (fileno(inb), &st);
 313         if (!size && !outnum)
 314             dat[2] = st.st_size;
 315         if (datecomp) {
 316             if (! datecomp->c_text) {
 317                 if (datecomp->c_tws == NULL)
 318                     NEW0(datecomp->c_tws);
 319                 *datecomp->c_tws = *dlocaltime ((time_t *) &st.st_mtime);
 320                 datecomp->c_flags |= CF_DATEFAB|CF_TRUE;
 321             } else {
 322                 datecomp->c_flags &= ~CF_DATEFAB;
 323             }
 324         }
 325     }
 326
 327     fmt_scan (fmt, *scanl, slwidth, dat, NULL);
 328
 329     if (bodycomp)
 330         bodycomp->c_text = saved_c_text;
 331
 332     if (noisy)
 333         fputs (charstring_buffer (*scanl), stdout);
 334
 335     cptr = fmt_findcomp ("encrypted");
 336     encrypted = cptr && cptr->c_text;
 337
 338     /* return dynamically allocated buffers to pool */
 339     while ((cptr = *savecomp++)) {
 340         cptr->c_text = NULL;
 341     }
 342
 343     if (outnum && (ferror(scnout) || fclose (scnout) == EOF))
 344         DIEWRERR();
 345
 346     return (state != FILEEOF ? SCNERR : encrypted ? SCNENC : SCNMSG);
 347 }
 348
 349
 350 /* The following two functions allow access to the global gstate above. */
 351 void
 352 scan_finished () {
 353     m_getfld_state_destroy (&gstate);
 354 }
 355
 356 void
 357 scan_detect_mbox_style (FILE *f) {
 358     m_unknown (&gstate, f);
 359 }