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