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

   1 /* scansbr.c -- routines to help scan along...
   2  *
   3  * This code is Copyright (c) 2002, by the authors of nmh.  See the
   4  * COPYRIGHT file in the root directory of the nmh distribution for
   5  * complete copyright information.
   6  */
   7
   8 #include <h/mh.h>
   9 #include <h/addrsbr.h>
  10 #include <h/fmt_scan.h>
  11 #include <h/scansbr.h>
  12 #include <h/tws.h>
  13 #include <h/utils.h>
  14
  15 static struct format *fmt;
  16 static struct comp *datecomp;           /* pntr to "date" comp             */
  17 static struct comp *bodycomp;           /* pntr to "body" pseudo-comp      *
  18                                          * (if referenced)                 */
  19 static int ncomps = 0;                  /* # of interesting components     */
  20 static char **compbuffers = 0;          /* buffers for component text      */
  21 static struct comp **used_buf = 0;      /* stack for comp that use buffers */
  22
  23 static int dat[5];                      /* aux. data for format routine    */
  24
  25 static m_getfld_state_t gstate;         /* for accessor functions below    */
  26
  27 #define DIEWRERR() adios (scnmsg, "write error on")
  28
  29 #define FPUTS(buf) {\
  30                 if (fputs(buf,scnout) == EOF)\
  31                     DIEWRERR();\
  32                 }
  33
  34 int
  35 scan (FILE *inb, int innum, int outnum, char *nfs, int width, int curflg,
  36       int unseen, char *folder, long size, int noisy, charstring_t *scanl)
  37 {
  38     int i, compnum, encrypted, state;
  39     char *cp, *tmpbuf, *startbody, **nxtbuf;
  40     char *saved_c_text = NULL;
  41     struct comp *cptr;
  42     struct comp **savecomp;
  43     char *scnmsg = NULL;
  44     FILE *scnout = NULL;
  45     char name[NAMESZ];
  46     int bufsz;
  47     static int rlwidth, slwidth;
  48
  49     /* first-time only initialization, which will always happen the
  50        way the code is now, with callers initializing *scanl to NULL.
  51        scanl used to be a global. */
  52     if (! *scanl) {
  53         if (width == -1) {
  54             /* Default:  width of the terminal, but at least WIDTH/2. */
  55             if ((width = sc_width ()) < WIDTH/2)
  56                 width = WIDTH/2;
  57         } else if (width == 0) {
  58             /* Unlimited width. */
  59             width = INT_MAX;
  60         }
  61         dat[3] = slwidth = width;
  62         *scanl = charstring_create (min(width, NMH_BUFSIZ));
  63         if (outnum)
  64             umask(~m_gmprot());
  65
  66         /* Compile format string */
  67         ncomps = fmt_compile (nfs, &fmt, 1) + 2;
  68
  69         bodycomp = fmt_findcomp("body");
  70         datecomp = fmt_findcomp("date");
  71         cptr = fmt_findcomp("folder");
  72         if (cptr && folder)
  73             cptr->c_text = mh_xstrdup(folder);
  74         if (fmt_addcompentry("encrypted")) {
  75                 ncomps++;
  76         }
  77         cptr =  fmt_findcomp("dtimenow");
  78         if (cptr)
  79             cptr->c_text = getcpy(dtimenow (0));
  80
  81         /*
  82          * In other programs I got rid of this complicated buffer switching,
  83          * but since scan reads lots of messages at once and this complicated
  84          * memory management, I decided to keep it; otherwise there was
  85          * the potential for a lot of malloc() and free()s, and I could
  86          * see the malloc() pool really getting fragmented.  Maybe it
  87          * wouldn't be an issue in practice; perhaps this will get
  88          * revisited someday.
  89          *
  90          * So, some notes for what's going on:
  91          *
  92          * nxtbuf is an array of pointers that contains malloc()'d buffers
  93          * to hold our component text.  used_buf is an array of struct comp
  94          * pointers that holds pointers to component structures we found while
  95          * processing a message.
  96          *
  97          * We read in the message with m_getfld(), using "tmpbuf" as our
  98          * input buffer.  tmpbuf is set at the start of message processing
  99          * to the first buffer in our buffer pool (nxtbuf).
 100          *
 101          * Every time we find a component we care about, we set that component's
 102          * text buffer to the current value of tmpbuf, and then switch tmpbuf
 103          * to the next buffer in our pool.  We also add that component to
 104          * our used_buf pool.
 105          *
 106          * When we're done, we go back and zero out all of the component
 107          * text buffer pointers that we saved in used_buf.
 108          *
 109          * Note that this means c_text memory is NOT owned by the fmt_module
 110          * and it's our responsibility to free it.
 111          */
 112
 113         nxtbuf = compbuffers = mh_xcalloc(ncomps, sizeof *nxtbuf);
 114         used_buf = mh_xcalloc(ncomps + 1, sizeof *used_buf);
 115         used_buf += ncomps+1; *--used_buf = 0;
 116         rlwidth = NMH_BUFSIZ;
 117         for (i = ncomps; i--; )
 118             *nxtbuf++ = mh_xmalloc(rlwidth);
 119     }
 120
 121     /*
 122      * each-message initialization
 123      */
 124     nxtbuf = compbuffers;
 125     savecomp = used_buf;
 126     tmpbuf = *nxtbuf++;
 127     startbody = NULL;
 128     dat[0] = innum ? innum : outnum;
 129     dat[1] = curflg;
 130     dat[4] = unseen;
 131
 132     /*
 133      * Get the first field.  If the message is non-empty
 134      * and we're doing an "inc", open the output file.
 135      */
 136     bufsz = rlwidth;
 137     m_getfld_state_reset (&gstate);
 138     if ((state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) == FILEEOF) {
 139         if (ferror(inb)) {
 140             advise("read", "unable to"); /* "read error" */
 141             return SCNFAT;
 142         }
 143         return SCNEOF;
 144     }
 145
 146     if (outnum) {
 147         if (outnum > 0) {
 148             scnmsg = m_name (outnum);
 149             if (*scnmsg == '?')         /* msg num out of range */
 150                 return SCNNUM;
 151         } else {
 152             scnmsg = "/dev/null";
 153         }
 154         if ((scnout = fopen (scnmsg, "w")) == NULL)
 155             adios (scnmsg, "unable to write");
 156     }
 157
 158     /* scan - main loop */
 159     for (compnum = 1; ;
 160         bufsz = rlwidth, state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) {
 161         switch (state) {
 162             case FLD:
 163             case FLDPLUS:
 164                 compnum++;
 165                 if (outnum) {
 166                     FPUTS (name);
 167                     if ( putc (':', scnout) == EOF) DIEWRERR();
 168                     FPUTS (tmpbuf);
 169                 }
 170                 /*
 171                  * if we're interested in this component, save a pointer
 172                  * to the component text, then start using our next free
 173                  * buffer as the component temp buffer (buffer switching
 174                  * saves an extra copy of the component text).
 175                  */
 176                 if ((cptr = fmt_findcasecomp(name))) {
 177                     if (! cptr->c_text) {
 178                         cptr->c_text = tmpbuf;
 179                         for (cp = tmpbuf + strlen (tmpbuf) - 1;
 180                                         cp >= tmpbuf; cp--)
 181                             if (isspace ((unsigned char) *cp))
 182                                 *cp = 0;
 183                             else
 184                                 break;
 185                         *--savecomp = cptr;
 186                         tmpbuf = *nxtbuf++;
 187                     }
 188                 }
 189
 190                 while (state == FLDPLUS) {
 191                     bufsz = rlwidth;
 192                     state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb);
 193                     if (outnum)
 194                         FPUTS (tmpbuf);
 195                 }
 196                 break;
 197
 198             case BODY:
 199                 /*
 200                  * A slight hack ... if we have less than rlwidth characters
 201                  * in the buffer, call m_getfld again.
 202                  */
 203
 204                 if ((i = strlen(tmpbuf)) < rlwidth) {
 205                     bufsz = rlwidth - i;
 206                     state = m_getfld (&gstate, name, tmpbuf + i, &bufsz, inb);
 207                 }
 208
 209                 if (! outnum) {
 210                     state = FILEEOF; /* stop now if scan cmd */
 211                     if (bodycomp && startbody == NULL)
 212                         startbody = tmpbuf;
 213                     goto finished;
 214                 }
 215                 if (putc ('\n', scnout) == EOF) DIEWRERR();
 216                 FPUTS (tmpbuf);
 217                 /*
 218                  * The previous code here used to call m_getfld() using
 219                  * pointers to the underlying output stdio buffers to
 220                  * avoid the extra copy.  Tests by Markus Schnalke show
 221                  * no noticeable performance loss on larger mailboxes
 222                  * if we incur an extra copy, and messing around with
 223                  * internal stdio buffers is becoming more and more
 224                  * unportable as times go on.  So from now on just deal
 225                  * with the overhead of an extra copy.
 226                  *
 227                  * Subtle change - with the previous code tmpbuf wasn't
 228                  * used, so we could reuse it for the {body} component.
 229                  * Now since we're using tmpbuf as our read buffer we
 230                  * need to save the beginning of the body for later.
 231                  * See the above (and below) use of startbody.
 232                  */
 233 body:;
 234                 if (bodycomp && startbody == NULL) {
 235                     startbody = tmpbuf;
 236                     tmpbuf = *nxtbuf++;
 237                 }
 238
 239                 while (state == BODY) {
 240                     bufsz = rlwidth;
 241                     state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb);
 242                     FPUTS(tmpbuf);
 243                 }
 244                 goto finished;
 245
 246             case LENERR:
 247             case FMTERR:
 248                 if (innum)
 249                     fprintf (stderr, "??Format error (message %d) in ",
 250                              outnum ? outnum : innum);
 251                 else
 252                     fprintf (stderr, "??Format error in ");
 253
 254                 fprintf (stderr, "component %d\n", compnum);
 255
 256                 if (outnum) {
 257                     FPUTS ("\n\nBAD MSG:\n");
 258                     FPUTS (name);
 259                     if (putc ('\n', scnout) == EOF) DIEWRERR();
 260                     state = BODY;
 261                     goto body;
 262                 }
 263                 goto finished;
 264
 265             case FILEEOF:
 266                 goto finished;
 267
 268             default:
 269                 adios (NULL, "getfld() returned %d", state);
 270         }
 271     }
 272
 273     /*
 274      * format and output the scan line.
 275      */
 276 finished:
 277     if (ferror(inb)) {
 278         advise("read", "unable to"); /* "read error" */
 279         return SCNFAT;
 280     }
 281
 282     /* Save and restore buffer so we don't trash our dynamic pool! */
 283     if (bodycomp) {
 284         saved_c_text = bodycomp->c_text;
 285         bodycomp->c_text = startbody;
 286     }
 287
 288     if (size)
 289         dat[2] = size;
 290     else if (outnum > 0)
 291     {
 292         dat[2] = ftell(scnout);
 293         if (dat[2] == EOF) DIEWRERR();
 294     }
 295
 296     if ((datecomp && !datecomp->c_text) || (!size && !outnum)) {
 297         struct stat st;
 298
 299         fstat (fileno(inb), &st);
 300         if (!size && !outnum)
 301             dat[2] = st.st_size;
 302         if (datecomp) {
 303             if (! datecomp->c_text) {
 304                 if (datecomp->c_tws == NULL)
 305                     NEW0(datecomp->c_tws);
 306                 *datecomp->c_tws = *dlocaltime ((time_t *) &st.st_mtime);
 307                 datecomp->c_flags |= CF_DATEFAB|CF_TRUE;
 308             } else {
 309                 datecomp->c_flags &= ~CF_DATEFAB;
 310             }
 311         }
 312     }
 313
 314     fmt_scan (fmt, *scanl, slwidth, dat, NULL);
 315
 316     if (bodycomp)
 317         bodycomp->c_text = saved_c_text;
 318
 319     if (noisy)
 320         fputs (charstring_buffer (*scanl), stdout);
 321
 322     cptr = fmt_findcomp ("encrypted");
 323     encrypted = cptr && cptr->c_text;
 324
 325     /* return dynamically allocated buffers to pool */
 326     while ((cptr = *savecomp++)) {
 327         cptr->c_text = NULL;
 328     }
 329
 330     if (outnum && (ferror(scnout) || fclose (scnout) == EOF))
 331         DIEWRERR();
 332
 333     return (state != FILEEOF ? SCNERR : encrypted ? SCNENC : SCNMSG);
 334 }
 335
 336
 337 /* The following two functions allow access to the global gstate above. */
 338 void
 339 scan_finished(void) {
 340     m_getfld_state_destroy (&gstate);
 341 }
 342
 343 void
 344 scan_detect_mbox_style (FILE *f) {
 345     m_unknown (&gstate, f);
 346 }