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