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 (mh_fputs(buf,scnout) == EOF)\
  42                     DIEWRERR();\
  43                 }
  44
  45 /*
  46  * prototypes
  47  */
  48 static int mh_fputs(char *, FILE *);
  49
  50 int
  51 scan (FILE *inb, int innum, int outnum, char *nfs, int width, int curflg,
  52       int unseen, char *folder, long size, int noisy, charstring_t *scanl)
  53 {
  54     int i, compnum, encrypted, state;
  55     char *cp, *tmpbuf, *startbody, **nxtbuf;
  56     char *saved_c_text = NULL;
  57     struct comp *cptr;
  58     struct comp **savecomp;
  59     char *scnmsg = NULL;
  60     FILE *scnout = NULL;
  61     char name[NAMESZ];
  62     int bufsz;
  63     static int rlwidth, slwidth;
  64
  65     /* first-time only initialization, which will always happen the
  66        way the code is now, with callers initializing *scanl to NULL.
  67        scanl used to be a global. */
  68     if (! *scanl) {
  69         if (width == -1) {
  70             /* Default:  width of the terminal, but at least WIDTH/2. */
  71             if ((width = sc_width ()) < WIDTH/2)
  72                 width = WIDTH/2;
  73         } else if (width == 0) {
  74             /* Unlimited width. */
  75             width = INT_MAX;
  76         }
  77         dat[3] = slwidth = width;
  78         *scanl = charstring_create (width < NMH_BUFSIZ ? width : NMH_BUFSIZ);
  79         if (outnum)
  80             umask(~m_gmprot());
  81
  82         /* Compile format string */
  83         ncomps = fmt_compile (nfs, &fmt, 1) + 2;
  84
  85         bodycomp = fmt_findcomp("body");
  86         datecomp = fmt_findcomp("date");
  87         cptr = fmt_findcomp("folder");
  88         if (cptr && folder)
  89             cptr->c_text = mh_xstrdup(folder);
  90         if (fmt_addcompentry("encrypted")) {
  91                 ncomps++;
  92         }
  93         cptr =  fmt_findcomp("dtimenow");
  94         if (cptr)
  95             cptr->c_text = getcpy(dtimenow (0));
  96
  97         /*
  98          * In other programs I got rid of this complicated buffer switching,
  99          * but since scan reads lots of messages at once and this complicated
 100          * memory management, I decided to keep it; otherwise there was
 101          * the potential for a lot of malloc() and free()s, and I could
 102          * see the malloc() pool really getting fragmented.  Maybe it
 103          * wouldn't be an issue in practice; perhaps this will get
 104          * revisited someday.
 105          *
 106          * So, some notes for what's going on:
 107          *
 108          * nxtbuf is an array of pointers that contains malloc()'d buffers
 109          * to hold our component text.  used_buf is an array of struct comp
 110          * pointers that holds pointers to component structures we found while
 111          * processing a message.
 112          *
 113          * We read in the message with m_getfld(), using "tmpbuf" as our
 114          * input buffer.  tmpbuf is set at the start of message processing
 115          * to the first buffer in our buffer pool (nxtbuf).
 116          *
 117          * Every time we find a component we care about, we set that component's
 118          * text buffer to the current value of tmpbuf, and then switch tmpbuf
 119          * to the next buffer in our pool.  We also add that component to
 120          * our used_buf pool.
 121          *
 122          * When we're done, we go back and zero out all of the component
 123          * text buffer pointers that we saved in used_buf.
 124          *
 125          * Note that this means c_text memory is NOT owned by the fmt_module
 126          * and it's our responsibility to free it.
 127          */
 128
 129         nxtbuf = compbuffers = mh_xcalloc(ncomps, sizeof *nxtbuf);
 130         used_buf = mh_xcalloc(ncomps + 1, sizeof *used_buf);
 131         used_buf += ncomps+1; *--used_buf = 0;
 132         rlwidth = bodycomp && (width > SBUFSIZ)
 133             ? min (width, NMH_BUFSIZ)
 134             : SBUFSIZ;
 135         for (i = ncomps; i--; )
 136             *nxtbuf++ = mh_xmalloc(rlwidth);
 137     }
 138
 139     /*
 140      * each-message initialization
 141      */
 142     nxtbuf = compbuffers;
 143     savecomp = used_buf;
 144     tmpbuf = *nxtbuf++;
 145     startbody = NULL;
 146     dat[0] = innum ? innum : outnum;
 147     dat[1] = curflg;
 148     dat[4] = unseen;
 149
 150     /*
 151      * Get the first field.  If the message is non-empty
 152      * and we're doing an "inc", open the output file.
 153      */
 154     bufsz = rlwidth;
 155     m_getfld_state_reset (&gstate);
 156     if ((state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) == FILEEOF) {
 157         if (ferror(inb)) {
 158             advise("read", "unable to"); /* "read error" */
 159             return SCNFAT;
 160         }
 161         return SCNEOF;
 162     }
 163
 164     if (outnum) {
 165         if (outnum > 0) {
 166             scnmsg = m_name (outnum);
 167             if (*scnmsg == '?')         /* msg num out of range */
 168                 return SCNNUM;
 169         } else {
 170             scnmsg = "/dev/null";
 171         }
 172         if ((scnout = fopen (scnmsg, "w")) == NULL)
 173             adios (scnmsg, "unable to write");
 174     }
 175
 176     /* scan - main loop */
 177     for (compnum = 1; ;
 178         bufsz = rlwidth, state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) {
 179         switch (state) {
 180             case FLD:
 181             case FLDPLUS:
 182                 compnum++;
 183                 if (outnum) {
 184                     FPUTS (name);
 185                     if ( putc (':', scnout) == EOF) DIEWRERR();
 186                     FPUTS (tmpbuf);
 187                 }
 188                 /*
 189                  * if we're interested in this component, save a pointer
 190                  * to the component text, then start using our next free
 191                  * buffer as the component temp buffer (buffer switching
 192                  * saves an extra copy of the component text).
 193                  */
 194                 if ((cptr = fmt_findcasecomp(name))) {
 195                     if (! cptr->c_text) {
 196                         cptr->c_text = tmpbuf;
 197                         for (cp = tmpbuf + strlen (tmpbuf) - 1;
 198                                         cp >= tmpbuf; cp--)
 199                             if (isspace ((unsigned char) *cp))
 200                                 *cp = 0;
 201                             else
 202                                 break;
 203                         *--savecomp = cptr;
 204                         tmpbuf = *nxtbuf++;
 205                     }
 206                 }
 207
 208                 while (state == FLDPLUS) {
 209                     bufsz = rlwidth;
 210                     state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb);
 211                     if (outnum)
 212                         FPUTS (tmpbuf);
 213                 }
 214                 break;
 215
 216             case BODY:
 217                 /*
 218                  * A slight hack ... if we have less than rlwidth characters
 219                  * in the buffer, call m_getfld again.
 220                  */
 221
 222                 if ((i = strlen(tmpbuf)) < rlwidth) {
 223                     bufsz = rlwidth - i;
 224                     state = m_getfld (&gstate, name, tmpbuf + i, &bufsz, inb);
 225                 }
 226
 227                 if (! outnum) {
 228                     state = FILEEOF; /* stop now if scan cmd */
 229                     if (bodycomp && startbody == NULL)
 230                         startbody = tmpbuf;
 231                     goto finished;
 232                 }
 233                 if (putc ('\n', scnout) == EOF) DIEWRERR();
 234                 FPUTS (tmpbuf);
 235                 /*
 236                  * The previous code here used to call m_getfld() using
 237                  * pointers to the underlying output stdio buffers to
 238                  * avoid the extra copy.  Tests by Markus Schnalke show
 239                  * no noticeable performance loss on larger mailboxes
 240                  * if we incur an extra copy, and messing around with
 241                  * internal stdio buffers is becoming more and more
 242                  * unportable as times go on.  So from now on just deal
 243                  * with the overhead of an extra copy.
 244                  *
 245                  * Subtle change - with the previous code tmpbuf wasn't
 246                  * used, so we could reuse it for the {body} component.
 247                  * Now since we're using tmpbuf as our read buffer we
 248                  * need to save the beginning of the body for later.
 249                  * See the above (and below) use of startbody.
 250                  */
 251 body:;
 252                 if (bodycomp && startbody == NULL) {
 253                     startbody = tmpbuf;
 254                     tmpbuf = *nxtbuf++;
 255                 }
 256
 257                 while (state == BODY) {
 258                     bufsz = rlwidth;
 259                     state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb);
 260                     FPUTS(tmpbuf);
 261                 }
 262                 goto finished;
 263
 264             case LENERR:
 265             case FMTERR:
 266                 if (innum)
 267                     fprintf (stderr, "??Format error (message %d) in ",
 268                              outnum ? outnum : innum);
 269                 else
 270                     fprintf (stderr, "??Format error in ");
 271
 272                 fprintf (stderr, "component %d\n", compnum);
 273
 274                 if (outnum) {
 275                     FPUTS ("\n\nBAD MSG:\n");
 276                     FPUTS (name);
 277                     if (putc ('\n', scnout) == EOF) DIEWRERR();
 278                     state = BODY;
 279                     goto body;
 280                 }
 281                 /* fall through */
 282
 283             case FILEEOF:
 284                 goto finished;
 285
 286             default:
 287                 adios (NULL, "getfld() returned %d", state);
 288         }
 289     }
 290
 291     /*
 292      * format and output the scan line.
 293      */
 294 finished:
 295     if (ferror(inb)) {
 296         advise("read", "unable to"); /* "read error" */
 297         return SCNFAT;
 298     }
 299
 300     /* Save and restore buffer so we don't trash our dynamic pool! */
 301     if (bodycomp) {
 302         saved_c_text = bodycomp->c_text;
 303         bodycomp->c_text = startbody;
 304     }
 305
 306     if (size)
 307         dat[2] = size;
 308     else if (outnum > 0)
 309     {
 310         dat[2] = ftell(scnout);
 311         if (dat[2] == EOF) DIEWRERR();
 312     }
 313
 314     if ((datecomp && !datecomp->c_text) || (!size && !outnum)) {
 315         struct stat st;
 316
 317         fstat (fileno(inb), &st);
 318         if (!size && !outnum)
 319             dat[2] = st.st_size;
 320         if (datecomp) {
 321             if (! datecomp->c_text) {
 322                 if (datecomp->c_tws == NULL)
 323                     NEW0(datecomp->c_tws);
 324                 *datecomp->c_tws = *dlocaltime ((time_t *) &st.st_mtime);
 325                 datecomp->c_flags |= CF_DATEFAB|CF_TRUE;
 326             } else {
 327                 datecomp->c_flags &= ~CF_DATEFAB;
 328             }
 329         }
 330     }
 331
 332     fmt_scan (fmt, *scanl, slwidth, dat, NULL);
 333
 334     if (bodycomp)
 335         bodycomp->c_text = saved_c_text;
 336
 337     if (noisy)
 338         fputs (charstring_buffer (*scanl), stdout);
 339
 340     cptr = fmt_findcomp ("encrypted");
 341     encrypted = cptr && cptr->c_text;
 342
 343     /* return dynamically allocated buffers to pool */
 344     while ((cptr = *savecomp++)) {
 345         cptr->c_text = NULL;
 346     }
 347
 348     if (outnum && (ferror(scnout) || fclose (scnout) == EOF))
 349         DIEWRERR();
 350
 351     return (state != FILEEOF ? SCNERR : encrypted ? SCNENC : SCNMSG);
 352 }
 353
 354
 355 static int
 356 mh_fputs(char *s, FILE *stream)
 357 {
 358     char c;
 359
 360     while ((c = *s++))
 361         if (putc (c,stream) == EOF )
 362             return(EOF);
 363     return (0);
 364 }
 365
 366 /* The following two functions allow access to the global gstate above. */
 367 void
 368 scan_finished () {
 369     m_getfld_state_destroy (&gstate);
 370 }
 371
 372 void
 373 scan_detect_mbox_style (FILE *f) {
 374     m_unknown (&gstate, f);
 375 }