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 PUTC(c) \
  30     if (putc((c), scnout) == EOF) \
  31         DIEWRERR();
  32
  33 #define FPUTS(buf) {\
  34                 if (fputs(buf,scnout) == EOF)\
  35                     DIEWRERR();\
  36                 }
  37
  38 /* outnum determines how the input from inb is copied.  If positive then
  39  * it is the number of the message to create, e.g. for inc(1), and all
  40  * of the email is copied into that message, with some tweaks.  If 0,
  41  * e.g. `scan 42', then reading inb can dubiously stop after a whole
  42  * buffer of body, even though this might not be enough to fulfill the
  43  * scan format and width.  Or if -1 then no copy is being created, but
  44  * all of inb must be read because the next message must be found, e.g.
  45  * `scan -file foo.mbox'. */
  46
  47 int
  48 scan (FILE *inb, int innum, int outnum, char *nfs, int width, int curflg,
  49       int unseen, char *folder, long size, int noisy, charstring_t *scanl)
  50 {
  51     int i, compnum, encrypted, state;
  52     char *cp, *tmpbuf, *startbody, **nxtbuf;
  53     char *saved_c_text = NULL;
  54     struct comp *cptr;
  55     struct comp **savecomp;
  56     char *scnmsg = NULL;
  57     FILE *scnout = NULL;
  58     char name[NAMESZ];
  59     int bufsz;
  60     static int rlwidth, slwidth;
  61
  62     /* first-time only initialization, which will always happen the
  63        way the code is now, with callers initializing *scanl to NULL.
  64        scanl used to be a global. */
  65     if (! *scanl) {
  66         if (width == -1) {
  67             /* Default:  width of the terminal, but at least WIDTH/2. */
  68             if ((width = sc_width ()) < WIDTH/2)
  69                 width = WIDTH/2;
  70         } else if (width == 0) {
  71             /* Unlimited width. */
  72             width = INT_MAX;
  73         }
  74         dat[3] = slwidth = width;
  75         *scanl = charstring_create (min(width, NMH_BUFSIZ));
  76         if (outnum)
  77             umask(~m_gmprot());
  78
  79         /* Compile format string */
  80         ncomps = fmt_compile (nfs, &fmt, 1) + 2;
  81
  82         bodycomp = fmt_findcomp("body");
  83         datecomp = fmt_findcomp("date");
  84         cptr = fmt_findcomp("folder");
  85         if (cptr && folder)
  86             cptr->c_text = mh_xstrdup(folder);
  87         if (fmt_addcompentry("encrypted")) {
  88                 ncomps++;
  89         }
  90         cptr =  fmt_findcomp("dtimenow");
  91         if (cptr)
  92             cptr->c_text = getcpy(dtimenow (0));
  93
  94         /*
  95          * In other programs I got rid of this complicated buffer switching,
  96          * but since scan reads lots of messages at once and this complicated
  97          * memory management, I decided to keep it; otherwise there was
  98          * the potential for a lot of malloc() and free()s, and I could
  99          * see the malloc() pool really getting fragmented.  Maybe it
 100          * wouldn't be an issue in practice; perhaps this will get
 101          * revisited someday.
 102          *
 103          * So, some notes for what's going on:
 104          *
 105          * nxtbuf is an array of pointers that contains malloc()'d buffers
 106          * to hold our component text.  used_buf is an array of struct comp
 107          * pointers that holds pointers to component structures we found while
 108          * processing a message.
 109          *
 110          * We read in the message with m_getfld(), using "tmpbuf" as our
 111          * input buffer.  tmpbuf is set at the start of message processing
 112          * to the first buffer in our buffer pool (nxtbuf).
 113          *
 114          * Every time we find a component we care about, we set that component's
 115          * text buffer to the current value of tmpbuf, and then switch tmpbuf
 116          * to the next buffer in our pool.  We also add that component to
 117          * our used_buf pool.
 118          *
 119          * When we're done, we go back and zero out all of the component
 120          * text buffer pointers that we saved in used_buf.
 121          *
 122          * Note that this means c_text memory is NOT owned by the fmt_module
 123          * and it's our responsibility to free it.
 124          */
 125
 126         nxtbuf = compbuffers = mh_xcalloc(ncomps, sizeof *nxtbuf);
 127         used_buf = mh_xcalloc(ncomps + 1, sizeof *used_buf);
 128         used_buf += ncomps+1; *--used_buf = 0;
 129         rlwidth = NMH_BUFSIZ;
 130         for (i = ncomps; i--; )
 131             *nxtbuf++ = mh_xmalloc(rlwidth);
 132     }
 133
 134     /*
 135      * each-message initialization
 136      */
 137     nxtbuf = compbuffers;
 138     savecomp = used_buf;
 139     tmpbuf = *nxtbuf++;
 140     startbody = NULL;
 141     dat[0] = innum ? innum : outnum;
 142     dat[1] = curflg;
 143     dat[4] = unseen;
 144
 145     /*
 146      * Get the first field.  If the message is non-empty
 147      * and we're doing an "inc", open the output file.
 148      */
 149     bufsz = rlwidth;
 150     m_getfld_state_reset (&gstate);
 151     if ((state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) == FILEEOF) {
 152         if (ferror(inb)) {
 153             advise("read", "unable to"); /* "read error" */
 154             return SCNFAT;
 155         }
 156         return SCNEOF;
 157     }
 158
 159     if (outnum > 0) {
 160         scnmsg = m_name (outnum);
 161         if (*scnmsg == '?')             /* msg num out of range */
 162             return SCNNUM;
 163         if ((scnout = fopen (scnmsg, "w")) == NULL)
 164             adios (scnmsg, "unable to write");
 165     }
 166
 167     /* scan - main loop */
 168     for (compnum = 1; ;
 169         bufsz = rlwidth, state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) {
 170         switch (state) {
 171             case FLD:
 172             case FLDPLUS:
 173                 compnum++;
 174                 if (scnout) {
 175                     FPUTS (name);
 176                     PUTC(':');
 177                     FPUTS (tmpbuf);
 178                 }
 179                 /*
 180                  * if we're interested in this component, save a pointer
 181                  * to the component text, then start using our next free
 182                  * buffer as the component temp buffer (buffer switching
 183                  * saves an extra copy of the component text).
 184                  */
 185                 if ((cptr = fmt_findcasecomp(name))) {
 186                     if (! cptr->c_text) {
 187                         cptr->c_text = tmpbuf;
 188                         for (cp = tmpbuf + strlen (tmpbuf) - 1;
 189                                         cp >= tmpbuf; cp--)
 190                             if (isspace ((unsigned char) *cp))
 191                                 *cp = 0;
 192                             else
 193                                 break;
 194                         *--savecomp = cptr;
 195                         tmpbuf = *nxtbuf++;
 196                     }
 197                 }
 198
 199                 while (state == FLDPLUS) {
 200                     bufsz = rlwidth;
 201                     state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb);
 202                     if (scnout)
 203                         FPUTS (tmpbuf);
 204                 }
 205                 break;
 206
 207             case BODY:
 208                 /*
 209                  * A slight hack ... if we have less than rlwidth characters
 210                  * in the buffer, call m_getfld again.
 211                  */
 212
 213                 if ((i = strlen(tmpbuf)) < rlwidth) {
 214                     bufsz = rlwidth - i;
 215                     state = m_getfld (&gstate, name, tmpbuf + i, &bufsz, inb);
 216                 }
 217
 218                 if (outnum == 0) {
 219                     state = FILEEOF; /* stop now if scan cmd */
 220                     if (bodycomp && startbody == NULL)
 221                         startbody = tmpbuf;
 222                     goto finished;
 223                 }
 224                 if (scnout) {
 225                     PUTC('\n');
 226                     FPUTS (tmpbuf);
 227                 }
 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                     if (scnout)
 254                         FPUTS(tmpbuf);
 255                 }
 256                 goto finished;
 257
 258             case LENERR:
 259             case FMTERR:
 260                 if (innum)
 261                     fprintf (stderr, "??Format error (message %d) in ",
 262                              outnum ? outnum : innum);
 263                 else
 264                     fprintf (stderr, "??Format error in ");
 265
 266                 fprintf (stderr, "component %d\n", compnum);
 267
 268                 if (scnout) {
 269                     FPUTS ("\n\nBAD MSG:\n");
 270                     FPUTS (name);
 271                     PUTC('\n');
 272                     state = BODY;
 273                     goto body;
 274                 }
 275                 goto finished;
 276
 277             case FILEEOF:
 278                 goto finished;
 279
 280             default:
 281                 adios (NULL, "getfld() returned %d", state);
 282         }
 283     }
 284
 285     /*
 286      * format and output the scan line.
 287      */
 288 finished:
 289     if (ferror(inb)) {
 290         advise("read", "unable to"); /* "read error" */
 291         return SCNFAT;
 292     }
 293
 294     /* Save and restore buffer so we don't trash our dynamic pool! */
 295     if (bodycomp) {
 296         saved_c_text = bodycomp->c_text;
 297         bodycomp->c_text = startbody;
 298     }
 299
 300     if (size)
 301         dat[2] = size;
 302     else if (scnout)
 303     {
 304         dat[2] = ftell(scnout);
 305         if (dat[2] == EOF) DIEWRERR();
 306     }
 307
 308     if ((datecomp && !datecomp->c_text) || (!size && !outnum)) {
 309         struct stat st;
 310
 311         fstat (fileno(inb), &st);
 312         if (!size && !outnum)
 313             dat[2] = st.st_size;
 314         if (datecomp) {
 315             if (! datecomp->c_text) {
 316                 if (datecomp->c_tws == NULL)
 317                     NEW0(datecomp->c_tws);
 318                 *datecomp->c_tws = *dlocaltime ((time_t *) &st.st_mtime);
 319                 datecomp->c_flags |= CF_DATEFAB|CF_TRUE;
 320             } else {
 321                 datecomp->c_flags &= ~CF_DATEFAB;
 322             }
 323         }
 324     }
 325
 326     fmt_scan (fmt, *scanl, slwidth, dat, NULL);
 327
 328     if (bodycomp)
 329         bodycomp->c_text = saved_c_text;
 330
 331     if (noisy)
 332         fputs (charstring_buffer (*scanl), stdout);
 333
 334     cptr = fmt_findcomp ("encrypted");
 335     encrypted = cptr && cptr->c_text;
 336
 337     /* return dynamically allocated buffers to pool */
 338     while ((cptr = *savecomp++)) {
 339         cptr->c_text = NULL;
 340     }
 341
 342     if (scnout && (ferror(scnout) || fclose (scnout) == EOF))
 343         DIEWRERR();
 344
 345     return (state != FILEEOF ? SCNERR : encrypted ? SCNENC : SCNMSG);
 346 }
 347
 348
 349 /* The following two functions allow access to the global gstate above. */
 350 void
 351 scan_finished(void) {
 352     m_getfld_state_destroy (&gstate);
 353 }
 354
 355 void
 356 scan_detect_mbox_style (FILE *f) {
 357     m_unknown (&gstate, f);
 358 }