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 #include "sbr/terminal.h"
  15
  16 static struct format *fmt;
  17 static struct comp *datecomp;           /* pntr to "date" comp             */
  18 static struct comp *bodycomp;           /* pntr to "body" pseudo-comp      *
  19                                          * (if referenced)                 */
  20 static int ncomps = 0;                  /* # of interesting components     */
  21 static char **compbuffers = 0;          /* buffers for component text      */
  22 static struct comp **used_buf = 0;      /* stack for comp that use buffers */
  23
  24 static int dat[5];                      /* aux. data for format routine    */
  25
  26 static m_getfld_state_t gstate;         /* for accessor functions below    */
  27
  28 #define DIEWRERR() adios (scnmsg, "write error on")
  29
  30 #define PUTC(c) \
  31     if (putc((c), scnout) == EOF) \
  32         DIEWRERR();
  33
  34 #define FPUTS(buf) {\
  35                 if (fputs(buf,scnout) == EOF)\
  36                     DIEWRERR();\
  37                 }
  38
  39 /* outnum determines how the input from inb is copied.  If positive then
  40  * it is the number of the message to create, e.g. for inc(1), and all
  41  * of the email is copied into that message, with some tweaks.  If 0,
  42  * e.g. `scan 42', then reading inb can dubiously stop after a whole
  43  * buffer of body, even though this might not be enough to fulfill the
  44  * scan format and width.  Or if -1 then no copy is being created, but
  45  * all of inb must be read because the next message must be found, e.g.
  46  * `scan -file foo.mbox'. */
  47
  48 int
  49 scan (FILE *inb, int innum, int outnum, char *nfs, int width, int curflg,
  50       int unseen, char *folder, long size, int noisy, charstring_t *scanl)
  51 {
  52     static bool deja_vu;
  53     static int tty_width;
  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             if (!deja_vu) {
  71                 deja_vu = true;
  72                 tty_width = sc_width();
  73             }
  74
  75             width = max(tty_width, WIDTH / 2);
  76         } else if (width == 0) {
  77             /* Unlimited width. */
  78             width = INT_MAX;
  79         }
  80         dat[3] = slwidth = width;
  81         *scanl = charstring_create (min(width, NMH_BUFSIZ));
  82         if (outnum)
  83             umask(~m_gmprot());
  84
  85         /* Compile format string */
  86         ncomps = fmt_compile (nfs, &fmt, 1) + 2;
  87
  88         bodycomp = fmt_findcomp("body");
  89         datecomp = fmt_findcomp("date");
  90         cptr = fmt_findcomp("folder");
  91         if (cptr && folder)
  92             cptr->c_text = mh_xstrdup(folder);
  93         if (fmt_addcompentry("encrypted")) {
  94                 ncomps++;
  95         }
  96         cptr =  fmt_findcomp("dtimenow");
  97         if (cptr)
  98             cptr->c_text = getcpy(dtimenow (0));
  99
 100         /*
 101          * In other programs I got rid of this complicated buffer switching,
 102          * but since scan reads lots of messages at once and this complicated
 103          * memory management, I decided to keep it; otherwise there was
 104          * the potential for a lot of malloc() and free()s, and I could
 105          * see the malloc() pool really getting fragmented.  Maybe it
 106          * wouldn't be an issue in practice; perhaps this will get
 107          * revisited someday.
 108          *
 109          * So, some notes for what's going on:
 110          *
 111          * nxtbuf is an array of pointers that contains malloc()'d buffers
 112          * to hold our component text.  used_buf is an array of struct comp
 113          * pointers that holds pointers to component structures we found while
 114          * processing a message.
 115          *
 116          * We read in the message with m_getfld(), using "tmpbuf" as our
 117          * input buffer.  tmpbuf is set at the start of message processing
 118          * to the first buffer in our buffer pool (nxtbuf).
 119          *
 120          * Every time we find a component we care about, we set that component's
 121          * text buffer to the current value of tmpbuf, and then switch tmpbuf
 122          * to the next buffer in our pool.  We also add that component to
 123          * our used_buf pool.
 124          *
 125          * When we're done, we go back and zero out all of the component
 126          * text buffer pointers that we saved in used_buf.
 127          *
 128          * Note that this means c_text memory is NOT owned by the fmt_module
 129          * and it's our responsibility to free it.
 130          */
 131
 132         nxtbuf = compbuffers = mh_xcalloc(ncomps, sizeof *nxtbuf);
 133         used_buf = mh_xcalloc(ncomps + 1, sizeof *used_buf);
 134         used_buf += ncomps+1; *--used_buf = 0;
 135         rlwidth = NMH_BUFSIZ;
 136         for (i = ncomps; i--; )
 137             *nxtbuf++ = mh_xmalloc(rlwidth);
 138     }
 139
 140     /*
 141      * each-message initialization
 142      */
 143     nxtbuf = compbuffers;
 144     savecomp = used_buf;
 145     tmpbuf = *nxtbuf++;
 146     startbody = NULL;
 147     dat[0] = innum ? innum : outnum;
 148     dat[1] = curflg;
 149     dat[4] = unseen;
 150
 151     /*
 152      * Get the first field.  If the message is non-empty
 153      * and we're doing an "inc", open the output file.
 154      */
 155     bufsz = rlwidth;
 156     m_getfld_state_reset (&gstate);
 157     if ((state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) == FILEEOF) {
 158         if (ferror(inb)) {
 159             advise("read", "unable to"); /* "read error" */
 160             return SCNFAT;
 161         }
 162         return SCNEOF;
 163     }
 164
 165     if (outnum > 0) {
 166         scnmsg = m_name (outnum);
 167         if (*scnmsg == '?')             /* msg num out of range */
 168             return SCNNUM;
 169         if ((scnout = fopen (scnmsg, "w")) == NULL)
 170             adios (scnmsg, "unable to write");
 171     }
 172
 173     /* scan - main loop */
 174     for (compnum = 1; ;
 175         bufsz = rlwidth, state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) {
 176         switch (state) {
 177             case FLD:
 178             case FLDPLUS:
 179                 compnum++;
 180                 if (scnout) {
 181                     FPUTS (name);
 182                     PUTC(':');
 183                     FPUTS (tmpbuf);
 184                 }
 185                 /*
 186                  * if we're interested in this component, save a pointer
 187                  * to the component text, then start using our next free
 188                  * buffer as the component temp buffer (buffer switching
 189                  * saves an extra copy of the component text).
 190                  */
 191                 if ((cptr = fmt_findcasecomp(name))) {
 192                     if (! cptr->c_text) {
 193                         cptr->c_text = tmpbuf;
 194                         for (cp = tmpbuf + strlen (tmpbuf) - 1;
 195                                         cp >= tmpbuf; cp--)
 196                             if (isspace ((unsigned char) *cp))
 197                                 *cp = 0;
 198                             else
 199                                 break;
 200                         *--savecomp = cptr;
 201                         tmpbuf = *nxtbuf++;
 202                     }
 203                 }
 204
 205                 while (state == FLDPLUS) {
 206                     bufsz = rlwidth;
 207                     state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb);
 208                     if (scnout)
 209                         FPUTS (tmpbuf);
 210                 }
 211                 break;
 212
 213             case BODY:
 214                 /*
 215                  * A slight hack ... if we have less than rlwidth characters
 216                  * in the buffer, call m_getfld again.
 217                  */
 218
 219                 if ((i = strlen(tmpbuf)) < rlwidth) {
 220                     bufsz = rlwidth - i;
 221                     state = m_getfld (&gstate, name, tmpbuf + i, &bufsz, inb);
 222                 }
 223
 224                 if (outnum == 0) {
 225                     state = FILEEOF; /* stop now if scan cmd */
 226                     if (bodycomp && startbody == NULL)
 227                         startbody = tmpbuf;
 228                     goto finished;
 229                 }
 230                 if (scnout) {
 231                     PUTC('\n');
 232                     FPUTS (tmpbuf);
 233                 }
 234                 /*
 235                  * The previous code here used to call m_getfld() using
 236                  * pointers to the underlying output stdio buffers to
 237                  * avoid the extra copy.  Tests by Markus Schnalke show
 238                  * no noticeable performance loss on larger mailboxes
 239                  * if we incur an extra copy, and messing around with
 240                  * internal stdio buffers is becoming more and more
 241                  * unportable as times go on.  So from now on just deal
 242                  * with the overhead of an extra copy.
 243                  *
 244                  * Subtle change - with the previous code tmpbuf wasn't
 245                  * used, so we could reuse it for the {body} component.
 246                  * Now since we're using tmpbuf as our read buffer we
 247                  * need to save the beginning of the body for later.
 248                  * See the above (and below) use of startbody.
 249                  */
 250 body:;
 251                 if (bodycomp && startbody == NULL) {
 252                     startbody = tmpbuf;
 253                     tmpbuf = *nxtbuf++;
 254                 }
 255
 256                 while (state == BODY) {
 257                     bufsz = rlwidth;
 258                     state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb);
 259                     if (scnout)
 260                         FPUTS(tmpbuf);
 261                 }
 262                 goto finished;
 263
 264             case LENERR:
 265             case FMTERR:
 266                 if (innum)
 267                     fprintf (stderr, "??Format error (message %d) in ",
 268                              outnum ? outnum : innum);
 269                 else
 270                     fprintf (stderr, "??Format error in ");
 271
 272                 fprintf (stderr, "component %d\n", compnum);
 273
 274                 if (scnout) {
 275                     FPUTS ("\n\nBAD MSG:\n");
 276                     FPUTS (name);
 277                     PUTC('\n');
 278                     state = BODY;
 279                     goto body;
 280                 }
 281                 goto finished;
 282
 283             case FILEEOF:
 284                 goto finished;
 285
 286             default:
 287                 adios (NULL, "getfld() returned %d", state);
 288         }
 289     }
 290
 291     /*
 292      * format and output the scan line.
 293      */
 294 finished:
 295     if (ferror(inb)) {
 296         advise("read", "unable to"); /* "read error" */
 297         return SCNFAT;
 298     }
 299
 300     /* Save and restore buffer so we don't trash our dynamic pool! */
 301     if (bodycomp) {
 302         saved_c_text = bodycomp->c_text;
 303         bodycomp->c_text = startbody;
 304     }
 305
 306     if (size)
 307         dat[2] = size;
 308     else if (scnout)
 309     {
 310         dat[2] = ftell(scnout);
 311         if (dat[2] == EOF) DIEWRERR();
 312     }
 313
 314     if ((datecomp && !datecomp->c_text) || (!size && !outnum)) {
 315         struct stat st;
 316
 317         fstat (fileno(inb), &st);
 318         if (!size && !outnum)
 319             dat[2] = st.st_size;
 320         if (datecomp) {
 321             if (! datecomp->c_text) {
 322                 if (datecomp->c_tws == NULL)
 323                     NEW0(datecomp->c_tws);
 324                 *datecomp->c_tws = *dlocaltime ((time_t *) &st.st_mtime);
 325                 datecomp->c_flags |= CF_DATEFAB|CF_TRUE;
 326             } else {
 327                 datecomp->c_flags &= ~CF_DATEFAB;
 328             }
 329         }
 330     }
 331
 332     fmt_scan (fmt, *scanl, slwidth, dat, NULL);
 333
 334     if (bodycomp)
 335         bodycomp->c_text = saved_c_text;
 336
 337     if (noisy)
 338         fputs (charstring_buffer (*scanl), stdout);
 339
 340     cptr = fmt_findcomp ("encrypted");
 341     encrypted = cptr && cptr->c_text;
 342
 343     /* return dynamically allocated buffers to pool */
 344     while ((cptr = *savecomp++)) {
 345         cptr->c_text = NULL;
 346     }
 347
 348     if (scnout && (ferror(scnout) || fclose (scnout) == EOF))
 349         DIEWRERR();
 350
 351     return (state != FILEEOF ? SCNERR : encrypted ? SCNENC : SCNMSG);
 352 }
 353
 354
 355 /* The following two functions allow access to the global gstate above. */
 356 void
 357 scan_finished(void) {
 358     m_getfld_state_destroy (&gstate);
 359 }
 360
 361 void
 362 scan_detect_mbox_style (FILE *f) {
 363     m_unknown (&gstate, f);
 364 }