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 "sbr/error.h"
  10 #include "h/addrsbr.h"
  11 #include "h/fmt_scan.h"
  12 #include "h/scansbr.h"
  13 #include "h/tws.h"
  14 #include "h/utils.h"
  15 #include "sbr/terminal.h"
  16
  17 static struct format *fmt;
  18 static struct comp *datecomp;           /* pntr to "date" comp             */
  19 static struct comp *bodycomp;           /* pntr to "body" pseudo-comp      *
  20                                          * (if referenced)                 */
  21 static int ncomps = 0;                  /* # of interesting components     */
  22 static char **compbuffers = 0;          /* buffers for component text      */
  23 static struct comp **used_buf = 0;      /* stack for comp that use buffers */
  24
  25 static int dat[5];                      /* aux. data for format routine    */
  26
  27 static m_getfld_state_t gstate;         /* for accessor functions below    */
  28
  29 #define DIEWRERR() adios (scnmsg, "write error on")
  30
  31 #define PUTC(c) \
  32     if (putc((c), scnout) == EOF) \
  33         DIEWRERR();
  34
  35 #define FPUTS(buf) {\
  36                 if (fputs(buf,scnout) == EOF)\
  37                     DIEWRERR();\
  38                 }
  39
  40 /* outnum determines how the input from inb is copied.  If positive then
  41  * it is the number of the message to create, e.g. for inc(1), and all
  42  * of the email is copied into that message, with some tweaks.  If 0,
  43  * e.g. `scan 42', then reading inb can dubiously stop after a whole
  44  * buffer of body, even though this might not be enough to fulfill the
  45  * scan format and width.  Or if -1 then no copy is being created, but
  46  * all of inb must be read because the next message must be found, e.g.
  47  * `scan -file foo.mbox'. */
  48
  49 int
  50 scan (FILE *inb, int innum, int outnum, char *nfs, int width, int curflg,
  51       int unseen, char *folder, long size, int noisy, charstring_t *scanl)
  52 {
  53     static bool deja_vu;
  54     static int tty_width;
  55     int i, compnum, encrypted, state;
  56     char *cp, *tmpbuf, *startbody, **nxtbuf;
  57     char *saved_c_text = NULL;
  58     struct comp *cptr;
  59     struct comp **savecomp;
  60     char *scnmsg = NULL;
  61     FILE *scnout = NULL;
  62     char name[NAMESZ];
  63     int bufsz;
  64     static int rlwidth, slwidth;
  65
  66     /* first-time only initialization, which will always happen the
  67        way the code is now, with callers initializing *scanl to NULL.
  68        scanl used to be a global. */
  69     if (! *scanl) {
  70         if (width == -1) {
  71             if (!deja_vu) {
  72                 deja_vu = true;
  73                 tty_width = sc_width();
  74             }
  75
  76             width = max(tty_width, WIDTH / 2);
  77         } else if (width == 0) {
  78             /* Unlimited width. */
  79             width = INT_MAX;
  80         }
  81         dat[3] = slwidth = width;
  82         *scanl = charstring_create (min(width, NMH_BUFSIZ));
  83         if (outnum)
  84             umask(~m_gmprot());
  85
  86         /* Compile format string */
  87         ncomps = fmt_compile (nfs, &fmt, 1) + 2;
  88
  89         bodycomp = fmt_findcomp("body");
  90         datecomp = fmt_findcomp("date");
  91         cptr = fmt_findcomp("folder");
  92         if (cptr && folder)
  93             cptr->c_text = mh_xstrdup(folder);
  94         if (fmt_addcompentry("encrypted")) {
  95                 ncomps++;
  96         }
  97         cptr =  fmt_findcomp("dtimenow");
  98         if (cptr)
  99             cptr->c_text = getcpy(dtimenow (0));
 100
 101         /*
 102          * In other programs I got rid of this complicated buffer switching,
 103          * but since scan reads lots of messages at once and this complicated
 104          * memory management, I decided to keep it; otherwise there was
 105          * the potential for a lot of malloc() and free()s, and I could
 106          * see the malloc() pool really getting fragmented.  Maybe it
 107          * wouldn't be an issue in practice; perhaps this will get
 108          * revisited someday.
 109          *
 110          * So, some notes for what's going on:
 111          *
 112          * nxtbuf is an array of pointers that contains malloc()'d buffers
 113          * to hold our component text.  used_buf is an array of struct comp
 114          * pointers that holds pointers to component structures we found while
 115          * processing a message.
 116          *
 117          * We read in the message with m_getfld(), using "tmpbuf" as our
 118          * input buffer.  tmpbuf is set at the start of message processing
 119          * to the first buffer in our buffer pool (nxtbuf).
 120          *
 121          * Every time we find a component we care about, we set that component's
 122          * text buffer to the current value of tmpbuf, and then switch tmpbuf
 123          * to the next buffer in our pool.  We also add that component to
 124          * our used_buf pool.
 125          *
 126          * When we're done, we go back and zero out all of the component
 127          * text buffer pointers that we saved in used_buf.
 128          *
 129          * Note that this means c_text memory is NOT owned by the fmt_module
 130          * and it's our responsibility to free it.
 131          */
 132
 133         nxtbuf = compbuffers = mh_xcalloc(ncomps, sizeof *nxtbuf);
 134         used_buf = mh_xcalloc(ncomps + 1, sizeof *used_buf);
 135         used_buf += ncomps+1; *--used_buf = 0;
 136         rlwidth = NMH_BUFSIZ;
 137         for (i = ncomps; i--; )
 138             *nxtbuf++ = mh_xmalloc(rlwidth);
 139     }
 140
 141     /*
 142      * each-message initialization
 143      */
 144     nxtbuf = compbuffers;
 145     savecomp = used_buf;
 146     tmpbuf = *nxtbuf++;
 147     startbody = NULL;
 148     dat[0] = innum ? innum : outnum;
 149     dat[1] = curflg;
 150     dat[4] = unseen;
 151
 152     /*
 153      * Get the first field.  If the message is non-empty
 154      * and we're doing an "inc", open the output file.
 155      */
 156     bufsz = rlwidth;
 157     m_getfld_state_reset (&gstate);
 158     if ((state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) == FILEEOF) {
 159         if (ferror(inb)) {
 160             advise("read", "unable to"); /* "read error" */
 161             return SCNFAT;
 162         }
 163         return SCNEOF;
 164     }
 165
 166     if (outnum > 0) {
 167         scnmsg = m_name (outnum);
 168         if (*scnmsg == '?')             /* msg num out of range */
 169             return SCNNUM;
 170         if ((scnout = fopen (scnmsg, "w")) == NULL)
 171             adios (scnmsg, "unable to write");
 172     }
 173
 174     /* scan - main loop */
 175     for (compnum = 1; ;
 176         bufsz = rlwidth, state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb)) {
 177         switch (state) {
 178             case FLD:
 179             case FLDPLUS:
 180                 compnum++;
 181                 if (scnout) {
 182                     FPUTS (name);
 183                     PUTC(':');
 184                     FPUTS (tmpbuf);
 185                 }
 186                 /*
 187                  * if we're interested in this component, save a pointer
 188                  * to the component text, then start using our next free
 189                  * buffer as the component temp buffer (buffer switching
 190                  * saves an extra copy of the component text).
 191                  */
 192                 if ((cptr = fmt_findcasecomp(name))) {
 193                     if (! cptr->c_text) {
 194                         cptr->c_text = tmpbuf;
 195                         for (cp = tmpbuf + strlen (tmpbuf) - 1;
 196                                         cp >= tmpbuf; cp--)
 197                             if (isspace ((unsigned char) *cp))
 198                                 *cp = 0;
 199                             else
 200                                 break;
 201                         *--savecomp = cptr;
 202                         tmpbuf = *nxtbuf++;
 203                     }
 204                 }
 205
 206                 while (state == FLDPLUS) {
 207                     bufsz = rlwidth;
 208                     state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb);
 209                     if (scnout)
 210                         FPUTS (tmpbuf);
 211                 }
 212                 break;
 213
 214             case BODY:
 215                 /*
 216                  * A slight hack ... if we have less than rlwidth characters
 217                  * in the buffer, call m_getfld again.
 218                  */
 219
 220                 if ((i = strlen(tmpbuf)) < rlwidth) {
 221                     bufsz = rlwidth - i;
 222                     state = m_getfld (&gstate, name, tmpbuf + i, &bufsz, inb);
 223                 }
 224
 225                 if (outnum == 0) {
 226                     state = FILEEOF; /* stop now if scan cmd */
 227                     if (bodycomp && startbody == NULL)
 228                         startbody = tmpbuf;
 229                     goto finished;
 230                 }
 231                 if (scnout) {
 232                     PUTC('\n');
 233                     FPUTS (tmpbuf);
 234                 }
 235                 /*
 236                  * The previous code here used to call m_getfld() using
 237                  * pointers to the underlying output stdio buffers to
 238                  * avoid the extra copy.  Tests by Markus Schnalke show
 239                  * no noticeable performance loss on larger mailboxes
 240                  * if we incur an extra copy, and messing around with
 241                  * internal stdio buffers is becoming more and more
 242                  * unportable as times go on.  So from now on just deal
 243                  * with the overhead of an extra copy.
 244                  *
 245                  * Subtle change - with the previous code tmpbuf wasn't
 246                  * used, so we could reuse it for the {body} component.
 247                  * Now since we're using tmpbuf as our read buffer we
 248                  * need to save the beginning of the body for later.
 249                  * See the above (and below) use of startbody.
 250                  */
 251 body:;
 252                 if (bodycomp && startbody == NULL) {
 253                     startbody = tmpbuf;
 254                     tmpbuf = *nxtbuf++;
 255                 }
 256
 257                 while (state == BODY) {
 258                     bufsz = rlwidth;
 259                     state = m_getfld (&gstate, name, tmpbuf, &bufsz, inb);
 260                     if (scnout)
 261                         FPUTS(tmpbuf);
 262                 }
 263                 goto finished;
 264
 265             case LENERR:
 266             case FMTERR:
 267                 if (innum)
 268                     fprintf (stderr, "??Format error (message %d) in ",
 269                              outnum ? outnum : innum);
 270                 else
 271                     fprintf (stderr, "??Format error in ");
 272
 273                 fprintf (stderr, "component %d\n", compnum);
 274
 275                 if (scnout) {
 276                     FPUTS ("\n\nBAD MSG:\n");
 277                     FPUTS (name);
 278                     PUTC('\n');
 279                     state = BODY;
 280                     goto body;
 281                 }
 282                 goto finished;
 283
 284             case FILEEOF:
 285                 goto finished;
 286
 287             default:
 288                 die("getfld() returned %d", state);
 289         }
 290     }
 291
 292     /*
 293      * format and output the scan line.
 294      */
 295 finished:
 296     if (ferror(inb)) {
 297         advise("read", "unable to"); /* "read error" */
 298         return SCNFAT;
 299     }
 300
 301     /* Save and restore buffer so we don't trash our dynamic pool! */
 302     if (bodycomp) {
 303         saved_c_text = bodycomp->c_text;
 304         bodycomp->c_text = startbody;
 305     }
 306
 307     if (size)
 308         dat[2] = size;
 309     else if (scnout)
 310     {
 311         dat[2] = ftell(scnout);
 312         if (dat[2] == EOF) DIEWRERR();
 313     }
 314
 315     if ((datecomp && !datecomp->c_text) || (!size && !outnum)) {
 316         struct stat st;
 317
 318         fstat (fileno(inb), &st);
 319         if (!size && !outnum)
 320             dat[2] = st.st_size;
 321         if (datecomp) {
 322             if (! datecomp->c_text) {
 323                 if (datecomp->c_tws == NULL)
 324                     NEW0(datecomp->c_tws);
 325                 *datecomp->c_tws = *dlocaltime ((time_t *) &st.st_mtime);
 326                 datecomp->c_flags |= CF_DATEFAB|CF_TRUE;
 327             } else {
 328                 datecomp->c_flags &= ~CF_DATEFAB;
 329             }
 330         }
 331     }
 332
 333     fmt_scan (fmt, *scanl, slwidth, dat, NULL);
 334
 335     if (bodycomp)
 336         bodycomp->c_text = saved_c_text;
 337
 338     if (noisy)
 339         fputs (charstring_buffer (*scanl), stdout);
 340
 341     cptr = fmt_findcomp ("encrypted");
 342     encrypted = cptr && cptr->c_text;
 343
 344     /* return dynamically allocated buffers to pool */
 345     while ((cptr = *savecomp++)) {
 346         cptr->c_text = NULL;
 347     }
 348
 349     if (scnout && (ferror(scnout) || fclose (scnout) == EOF))
 350         DIEWRERR();
 351
 352     return state != FILEEOF ? SCNERR : encrypted ? SCNENC : SCNMSG;
 353 }
 354
 355
 356 /* The following two functions allow access to the global gstate above. */
 357 void
 358 scan_finished(void)
 359 {
 360     m_getfld_state_destroy (&gstate);
 361 }
 362
 363 void
 364 scan_detect_mbox_style (FILE *f)
 365 {
 366     m_unknown (&gstate, f);
 367 }