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);
  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 = getcpy(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 = (char **) calloc((size_t) ncomps, sizeof(char *));
 130         if (nxtbuf == NULL)
 131             adios (NULL, "unable to allocate component buffers");
 132         used_buf = (struct comp **) calloc((size_t) (ncomps+1),
 133             sizeof(struct comp *));
 134         if (used_buf == NULL)
 135             adios (NULL, "unable to allocate component buffer stack");
 136         used_buf += ncomps+1; *--used_buf = 0;
 137         rlwidth = bodycomp && (width > SBUFSIZ) ? min (width, BUFSIZ) : SBUFSIZ;
 138         for (i = ncomps; i--; )
 139             *nxtbuf++ = mh_xmalloc(rlwidth);
 140     }
 141
 142     /*
 143      * each-message initialization
 144      */
 145     nxtbuf = compbuffers;
 146     savecomp = used_buf;
 147     tmpbuf = *nxtbuf++;
 148     startbody = NULL;
 149     dat[0] = innum ? innum : outnum;
 150     dat[1] = curflg;
 151     dat[4] = unseen;
 152
 153     /*
 154      * Get the first field.  If the message is non-empty
 155      * and we're doing an "inc", open the output file.
 156      */
 157     bufsz = rlwidth;
 158     m_getfld_state_reset (&gstate);
 159     if ((state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) == FILEEOF) {
 160         if (ferror(inb)) {
 161             advise("read", "unable to"); /* "read error" */
 162             return SCNFAT;
 163         } else {
 164             return SCNEOF;
 165         }
 166     }
 167
 168     if (outnum) {
 169         if (outnum > 0) {
 170             scnmsg = m_name (outnum);
 171             if (*scnmsg == '?')         /* msg num out of range */
 172                 return SCNNUM;
 173         } else {
 174             scnmsg = "/dev/null";
 175         }
 176         if ((scnout = fopen (scnmsg, "w")) == NULL)
 177             adios (scnmsg, "unable to write");
 178     }
 179
 180     /* scan - main loop */
 181     for (compnum = 1; ;
 182         bufsz = rlwidth, state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) {
 183         switch (state) {
 184             case FLD:
 185             case FLDPLUS:
 186                 compnum++;
 187                 if (outnum) {
 188                     FPUTS (name);
 189                     if ( putc (':', scnout) == EOF) DIEWRERR();
 190                     FPUTS (tmpbuf);
 191                 }
 192                 /*
 193                  * if we're interested in this component, save a pointer
 194                  * to the component text, then start using our next free
 195                  * buffer as the component temp buffer (buffer switching
 196                  * saves an extra copy of the component text).
 197                  */
 198                 if ((cptr = fmt_findcasecomp(name))) {
 199                     if (! cptr->c_text) {
 200                         cptr->c_text = tmpbuf;
 201                         for (cp = tmpbuf + strlen (tmpbuf) - 1;
 202                                         cp >= tmpbuf; cp--)
 203                             if (isspace ((unsigned char) *cp))
 204                                 *cp = 0;
 205                             else
 206                                 break;
 207                         *--savecomp = cptr;
 208                         tmpbuf = *nxtbuf++;
 209                     }
 210                 }
 211
 212                 while (state == FLDPLUS) {
 213                     bufsz = rlwidth;
 214                     state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb);
 215                     if (outnum)
 216                         FPUTS (tmpbuf);
 217                 }
 218                 break;
 219
 220             case BODY:
 221                 compnum = -1;
 222                 /*
 223                  * A slight hack ... if we have less than rlwidth characters
 224                  * in the buffer, call m_getfld again.
 225                  */
 226
 227                 if ((i = strlen(tmpbuf)) < rlwidth) {
 228                     bufsz = rlwidth - i;
 229                     state = m_getfld (&gstate, name, tmpbuf + i, &bufsz, inb);
 230                 }
 231
 232                 if (! outnum) {
 233                     state = FILEEOF; /* stop now if scan cmd */
 234                     if (bodycomp && startbody == NULL)
 235                         startbody = tmpbuf;
 236                     goto finished;
 237                 }
 238                 if (putc ('\n', scnout) == EOF) DIEWRERR();
 239                 FPUTS (tmpbuf);
 240                 /*
 241                  * The previous code here used to call m_getfld() using
 242                  * pointers to the underlying output stdio buffers to
 243                  * avoid the extra copy.  Tests by Markus Schnalke show
 244                  * no noticable performance loss on larger mailboxes
 245                  * if we incur an extra copy, and messing around with
 246                  * internal stdio buffers is becoming more and more
 247                  * unportable as times go on.  So from now on just deal
 248                  * with the overhead of an extra copy.
 249                  *
 250                  * Subtle change - with the previous code tmpbuf wasn't
 251                  * used, so we could reuse it for the {body} component.
 252                  * Now since we're using tmpbuf as our read buffer we
 253                  * need to save the beginning of the body for later.
 254                  * See the above (and below) use of startbody.
 255                  */
 256 body:;
 257                 if (bodycomp && startbody == NULL) {
 258                     startbody = tmpbuf;
 259                     tmpbuf = *nxtbuf++;
 260                 }
 261
 262                 while (state == BODY) {
 263                     bufsz = rlwidth;
 264                     state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb);
 265                     FPUTS(tmpbuf);
 266                 }
 267                 goto finished;
 268
 269             case LENERR:
 270             case FMTERR:
 271                 if (innum)
 272                     fprintf (stderr, "??Format error (message %d) in ",
 273                              outnum ? outnum : innum);
 274                 else
 275                     fprintf (stderr, "??Format error in ");
 276
 277                 fprintf (stderr, "component %d\n", compnum);
 278
 279                 if (outnum) {
 280                     FPUTS ("\n\nBAD MSG:\n");
 281                     FPUTS (name);
 282                     if (putc ('\n', scnout) == EOF) DIEWRERR();
 283                     state = BODY;
 284                     goto body;
 285                 }
 286                 /* fall through */
 287
 288             case FILEEOF:
 289                 goto finished;
 290
 291             default:
 292                 adios (NULL, "getfld() returned %d", state);
 293         }
 294     }
 295
 296     /*
 297      * format and output the scan line.
 298      */
 299 finished:
 300     if (ferror(inb)) {
 301         advise("read", "unable to"); /* "read error" */
 302         return SCNFAT;
 303     }
 304
 305     /* Save and restore buffer so we don't trash our dynamic pool! */
 306     if (bodycomp) {
 307         saved_c_text = bodycomp->c_text;
 308         bodycomp->c_text = startbody;
 309     }
 310
 311     if (size)
 312         dat[2] = size;
 313     else if (outnum > 0)
 314     {
 315         dat[2] = ftell(scnout);
 316         if (dat[2] == EOF) DIEWRERR();
 317     }
 318
 319     if ((datecomp && !datecomp->c_text) || (!size && !outnum)) {
 320         struct stat st;
 321
 322         fstat (fileno(inb), &st);
 323         if (!size && !outnum)
 324             dat[2] = st.st_size;
 325         if (datecomp) {
 326             if (! datecomp->c_text) {
 327                 if (datecomp->c_tws == NULL)
 328                     datecomp->c_tws = (struct tws *)
 329                         calloc((size_t) 1, sizeof(*datecomp->c_tws));
 330                 if (datecomp->c_tws == NULL)
 331                     adios (NULL, "unable to allocate tws buffer");
 332                 *datecomp->c_tws = *dlocaltime ((time_t *) &st.st_mtime);
 333                 datecomp->c_flags |= CF_DATEFAB|CF_TRUE;
 334             } else {
 335                 datecomp->c_flags &= ~CF_DATEFAB;
 336             }
 337         }
 338     }
 339
 340     fmt_scan (fmt, *scanl, slwidth, dat, NULL);
 341
 342     if (bodycomp)
 343         bodycomp->c_text = saved_c_text;
 344
 345     if (noisy)
 346         fputs (charstring_buffer (*scanl), stdout);
 347
 348     cptr = fmt_findcomp ("encrypted");
 349     encrypted = cptr && cptr->c_text;
 350
 351     /* return dynamically allocated buffers to pool */
 352     while ((cptr = *savecomp++)) {
 353         cptr->c_text = NULL;
 354     }
 355
 356     if (outnum && (ferror(scnout) || fclose (scnout) == EOF))
 357         DIEWRERR();
 358
 359     return (state != FILEEOF ? SCNERR : encrypted ? SCNENC : SCNMSG);
 360 }
 361
 362
 363 static int
 364 mh_fputs(char *s, FILE *stream)
 365 {
 366     char c;
 367
 368     while ((c = *s++))
 369         if (putc (c,stream) == EOF )
 370             return(EOF);
 371     return (0);
 372 }
 373
 374 /* The following two functions allow access to the global gstate above. */
 375 void
 376 scan_finished () {
 377     m_getfld_state_destroy (&gstate);
 378 }
 379
 380 void
 381 scan_detect_mbox_style (FILE *f) {
 382     m_unknown (&gstate, f);
 383 }