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)
 138             ? min (width, NMH_BUFSIZ)
 139             : SBUFSIZ;
 140         for (i = ncomps; i--; )
 141             *nxtbuf++ = mh_xmalloc(rlwidth);
 142     }
 143
 144     /*
 145      * each-message initialization
 146      */
 147     nxtbuf = compbuffers;
 148     savecomp = used_buf;
 149     tmpbuf = *nxtbuf++;
 150     startbody = NULL;
 151     dat[0] = innum ? innum : outnum;
 152     dat[1] = curflg;
 153     dat[4] = unseen;
 154
 155     /*
 156      * Get the first field.  If the message is non-empty
 157      * and we're doing an "inc", open the output file.
 158      */
 159     bufsz = rlwidth;
 160     m_getfld_state_reset (&gstate);
 161     if ((state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) == FILEEOF) {
 162         if (ferror(inb)) {
 163             advise("read", "unable to"); /* "read error" */
 164             return SCNFAT;
 165         } else {
 166             return SCNEOF;
 167         }
 168     }
 169
 170     if (outnum) {
 171         if (outnum > 0) {
 172             scnmsg = m_name (outnum);
 173             if (*scnmsg == '?')         /* msg num out of range */
 174                 return SCNNUM;
 175         } else {
 176             scnmsg = "/dev/null";
 177         }
 178         if ((scnout = fopen (scnmsg, "w")) == NULL)
 179             adios (scnmsg, "unable to write");
 180     }
 181
 182     /* scan - main loop */
 183     for (compnum = 1; ;
 184         bufsz = rlwidth, state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) {
 185         switch (state) {
 186             case FLD:
 187             case FLDPLUS:
 188                 compnum++;
 189                 if (outnum) {
 190                     FPUTS (name);
 191                     if ( putc (':', scnout) == EOF) DIEWRERR();
 192                     FPUTS (tmpbuf);
 193                 }
 194                 /*
 195                  * if we're interested in this component, save a pointer
 196                  * to the component text, then start using our next free
 197                  * buffer as the component temp buffer (buffer switching
 198                  * saves an extra copy of the component text).
 199                  */
 200                 if ((cptr = fmt_findcasecomp(name))) {
 201                     if (! cptr->c_text) {
 202                         cptr->c_text = tmpbuf;
 203                         for (cp = tmpbuf + strlen (tmpbuf) - 1;
 204                                         cp >= tmpbuf; cp--)
 205                             if (isspace ((unsigned char) *cp))
 206                                 *cp = 0;
 207                             else
 208                                 break;
 209                         *--savecomp = cptr;
 210                         tmpbuf = *nxtbuf++;
 211                     }
 212                 }
 213
 214                 while (state == FLDPLUS) {
 215                     bufsz = rlwidth;
 216                     state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb);
 217                     if (outnum)
 218                         FPUTS (tmpbuf);
 219                 }
 220                 break;
 221
 222             case BODY:
 223                 compnum = -1;
 224                 /*
 225                  * A slight hack ... if we have less than rlwidth characters
 226                  * in the buffer, call m_getfld again.
 227                  */
 228
 229                 if ((i = strlen(tmpbuf)) < rlwidth) {
 230                     bufsz = rlwidth - i;
 231                     state = m_getfld (&gstate, name, tmpbuf + i, &bufsz, inb);
 232                 }
 233
 234                 if (! outnum) {
 235                     state = FILEEOF; /* stop now if scan cmd */
 236                     if (bodycomp && startbody == NULL)
 237                         startbody = tmpbuf;
 238                     goto finished;
 239                 }
 240                 if (putc ('\n', scnout) == EOF) DIEWRERR();
 241                 FPUTS (tmpbuf);
 242                 /*
 243                  * The previous code here used to call m_getfld() using
 244                  * pointers to the underlying output stdio buffers to
 245                  * avoid the extra copy.  Tests by Markus Schnalke show
 246                  * no noticable performance loss on larger mailboxes
 247                  * if we incur an extra copy, and messing around with
 248                  * internal stdio buffers is becoming more and more
 249                  * unportable as times go on.  So from now on just deal
 250                  * with the overhead of an extra copy.
 251                  *
 252                  * Subtle change - with the previous code tmpbuf wasn't
 253                  * used, so we could reuse it for the {body} component.
 254                  * Now since we're using tmpbuf as our read buffer we
 255                  * need to save the beginning of the body for later.
 256                  * See the above (and below) use of startbody.
 257                  */
 258 body:;
 259                 if (bodycomp && startbody == NULL) {
 260                     startbody = tmpbuf;
 261                     tmpbuf = *nxtbuf++;
 262                 }
 263
 264                 while (state == BODY) {
 265                     bufsz = rlwidth;
 266                     state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb);
 267                     FPUTS(tmpbuf);
 268                 }
 269                 goto finished;
 270
 271             case LENERR:
 272             case FMTERR:
 273                 if (innum)
 274                     fprintf (stderr, "??Format error (message %d) in ",
 275                              outnum ? outnum : innum);
 276                 else
 277                     fprintf (stderr, "??Format error in ");
 278
 279                 fprintf (stderr, "component %d\n", compnum);
 280
 281                 if (outnum) {
 282                     FPUTS ("\n\nBAD MSG:\n");
 283                     FPUTS (name);
 284                     if (putc ('\n', scnout) == EOF) DIEWRERR();
 285                     state = BODY;
 286                     goto body;
 287                 }
 288                 /* fall through */
 289
 290             case FILEEOF:
 291                 goto finished;
 292
 293             default:
 294                 adios (NULL, "getfld() returned %d", state);
 295         }
 296     }
 297
 298     /*
 299      * format and output the scan line.
 300      */
 301 finished:
 302     if (ferror(inb)) {
 303         advise("read", "unable to"); /* "read error" */
 304         return SCNFAT;
 305     }
 306
 307     /* Save and restore buffer so we don't trash our dynamic pool! */
 308     if (bodycomp) {
 309         saved_c_text = bodycomp->c_text;
 310         bodycomp->c_text = startbody;
 311     }
 312
 313     if (size)
 314         dat[2] = size;
 315     else if (outnum > 0)
 316     {
 317         dat[2] = ftell(scnout);
 318         if (dat[2] == EOF) DIEWRERR();
 319     }
 320
 321     if ((datecomp && !datecomp->c_text) || (!size && !outnum)) {
 322         struct stat st;
 323
 324         fstat (fileno(inb), &st);
 325         if (!size && !outnum)
 326             dat[2] = st.st_size;
 327         if (datecomp) {
 328             if (! datecomp->c_text) {
 329                 if (datecomp->c_tws == NULL)
 330                     datecomp->c_tws = (struct tws *)
 331                         calloc((size_t) 1, sizeof(*datecomp->c_tws));
 332                 if (datecomp->c_tws == NULL)
 333                     adios (NULL, "unable to allocate tws buffer");
 334                 *datecomp->c_tws = *dlocaltime ((time_t *) &st.st_mtime);
 335                 datecomp->c_flags |= CF_DATEFAB|CF_TRUE;
 336             } else {
 337                 datecomp->c_flags &= ~CF_DATEFAB;
 338             }
 339         }
 340     }
 341
 342     fmt_scan (fmt, *scanl, slwidth, dat, NULL);
 343
 344     if (bodycomp)
 345         bodycomp->c_text = saved_c_text;
 346
 347     if (noisy)
 348         fputs (charstring_buffer (*scanl), stdout);
 349
 350     cptr = fmt_findcomp ("encrypted");
 351     encrypted = cptr && cptr->c_text;
 352
 353     /* return dynamically allocated buffers to pool */
 354     while ((cptr = *savecomp++)) {
 355         cptr->c_text = NULL;
 356     }
 357
 358     if (outnum && (ferror(scnout) || fclose (scnout) == EOF))
 359         DIEWRERR();
 360
 361     return (state != FILEEOF ? SCNERR : encrypted ? SCNENC : SCNMSG);
 362 }
 363
 364
 365 static int
 366 mh_fputs(char *s, FILE *stream)
 367 {
 368     char c;
 369
 370     while ((c = *s++))
 371         if (putc (c,stream) == EOF )
 372             return(EOF);
 373     return (0);
 374 }
 375
 376 /* The following two functions allow access to the global gstate above. */
 377 void
 378 scan_finished () {
 379     m_getfld_state_destroy (&gstate);
 380 }
 381
 382 void
 383 scan_detect_mbox_style (FILE *f) {
 384     m_unknown (&gstate, f);
 385 }