diplodocus.org Git - nmh/blob - uip/sortm.c

   1
   2 /*
   3  * sortm.c -- sort messages in a folder by date/time
   4  *
   5  * $Id$
   6  *
   7  * This code is Copyright (c) 2002, by the authors of nmh.  See the
   8  * COPYRIGHT file in the root directory of the nmh distribution for
   9  * complete copyright information.
  10  */
  11
  12 #include <h/mh.h>
  13 #include <h/tws.h>
  14 #include <h/utils.h>
  15
  16 static struct swit switches[] = {
  17 #define DATESW                 0
  18      { "datefield field", 0 },
  19 #define TEXTSW                 1
  20      { "textfield field", 0 },
  21 #define NSUBJSW                2
  22      { "notextfield", 0 },
  23 #define SUBJSW                 3
  24      { "subject", -3 },            /* backward-compatibility */
  25 #define LIMSW                  4
  26      { "limit days", 0 },
  27 #define NLIMSW                 5
  28      { "nolimit", 0 },
  29 #define VERBSW                 6
  30      { "verbose", 0 },
  31 #define NVERBSW                7
  32      { "noverbose", 0 },
  33 #define VERSIONSW              8
  34      { "version", 0 },
  35 #define HELPSW                 9
  36      { "help", 0 },
  37      { NULL, 0 }
  38 };
  39
  40 struct smsg {
  41     int s_msg;
  42     time_t s_clock;
  43     char *s_subj;
  44 };
  45
  46 static struct smsg *smsgs;
  47 int nmsgs;
  48
  49 char *subjsort = (char *) 0;    /* sort on subject if != 0 */
  50 unsigned long datelimit = 0;
  51 int submajor = 0;               /* if true, sort on subject-major */
  52 int verbose;
  53
  54 /* This keeps compiler happy on calls to qsort */
  55 typedef int (*qsort_comp) (const void *, const void *);
  56
  57 /*
  58  * static prototypes
  59  */
  60 static int read_hdrs (struct msgs *, char *);
  61 static int get_fields (char *, int, struct smsg *);
  62 static int dsort (struct smsg **, struct smsg **);
  63 static int subsort (struct smsg **, struct smsg **);
  64 static int txtsort (struct smsg **, struct smsg **);
  65 static void rename_chain (struct msgs *, struct smsg **, int, int);
  66 static void rename_msgs (struct msgs *, struct smsg **);
  67
  68
  69 int
  70 main (int argc, char **argv)
  71 {
  72     int i, msgnum;
  73     unsigned char *cp;
  74     char *maildir, *datesw = NULL;
  75     char *folder = NULL, buf[BUFSIZ], **argp;
  76     char **arguments;
  77     struct msgs_array msgs = { 0, 0, NULL };
  78     struct msgs *mp;
  79     struct smsg **dlist;
  80
  81 #ifdef LOCALE
  82     setlocale(LC_ALL, "");
  83 #endif
  84     invo_name = r1bindex (argv[0], '/');
  85
  86     /* read user profile/context */
  87     context_read();
  88
  89     arguments = getarguments (invo_name, argc, argv, 1);
  90     argp = arguments;
  91
  92     /*
  93      * Parse arguments
  94      */
  95     while ((cp = *argp++)) {
  96         if (*cp == '-') {
  97             switch (smatch (++cp, switches)) {
  98             case AMBIGSW:
  99                 ambigsw (cp, switches);
 100                 done (1);
 101             case UNKWNSW:
 102                 adios (NULL, "-%s unknown", cp);
 103
 104             case HELPSW:
 105                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 106                         invo_name);
 107                 print_help (buf, switches, 1);
 108                 done (1);
 109             case VERSIONSW:
 110                 print_version(invo_name);
 111                 done (1);
 112
 113             case DATESW:
 114                 if (datesw)
 115                     adios (NULL, "only one date field at a time");
 116                 if (!(datesw = *argp++) || *datesw == '-')
 117                     adios (NULL, "missing argument to %s", argp[-2]);
 118                 continue;
 119
 120             case TEXTSW:
 121                 if (subjsort)
 122                     adios (NULL, "only one text field at a time");
 123                 if (!(subjsort = *argp++) || *subjsort == '-')
 124                     adios (NULL, "missing argument to %s", argp[-2]);
 125                 continue;
 126
 127             case SUBJSW:
 128                 subjsort = "subject";
 129                 continue;
 130             case NSUBJSW:
 131                 subjsort = (char *)0;
 132                 continue;
 133
 134             case LIMSW:
 135                 if (!(cp = *argp++) || *cp == '-')
 136                         adios (NULL, "missing argument to %s", argp[-2]);
 137                 while (*cp == '0')
 138                     cp++;               /* skip any leading zeros */
 139                 if (!*cp) {             /* hit end of string */
 140                     submajor++;         /* sort subject-major */
 141                     continue;
 142                 }
 143                 if (!isdigit(*cp) || !(datelimit = atoi(cp)))
 144                     adios (NULL, "impossible limit %s", cp);
 145                 datelimit *= 60*60*24;
 146                 continue;
 147             case NLIMSW:
 148                 submajor = 0;   /* use date-major, but */
 149                 datelimit = 0;  /* use no limit */
 150                 continue;
 151
 152             case VERBSW:
 153                 verbose++;
 154                 continue;
 155             case NVERBSW:
 156                 verbose = 0;
 157                 continue;
 158             }
 159         }
 160         if (*cp == '+' || *cp == '@') {
 161             if (folder)
 162                 adios (NULL, "only one folder at a time!");
 163             else
 164                 folder = pluspath (cp);
 165         } else
 166                 app_msgarg(&msgs, cp);
 167     }
 168
 169     if (!context_find ("path"))
 170         free (path ("./", TFOLDER));
 171     if (!msgs.size)
 172         app_msgarg(&msgs, "all");
 173     if (!datesw)
 174         datesw = "date";
 175     if (!folder)
 176         folder = getfolder (1);
 177     maildir = m_maildir (folder);
 178
 179     if (chdir (maildir) == NOTOK)
 180         adios (maildir, "unable to change directory to");
 181
 182     /* read folder and create message structure */
 183     if (!(mp = folder_read (folder)))
 184         adios (NULL, "unable to read folder %s", folder);
 185
 186     /* check for empty folder */
 187     if (mp->nummsg == 0)
 188         adios (NULL, "no messages in %s", folder);
 189
 190     /* parse all the message ranges/sequences and set SELECTED */
 191     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 192         if (!m_convert (mp, msgs.msgs[msgnum]))
 193             done (1);
 194     seq_setprev (mp);   /* set the previous sequence */
 195
 196     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 197         adios (NULL, "no messages to sort");
 198
 199     /*
 200      * sort a list of pointers to our "messages to be sorted".
 201      */
 202     dlist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 203     for (i = 0; i < nmsgs; i++)
 204         dlist[i] = &smsgs[i];
 205     dlist[nmsgs] = 0;
 206
 207     if (verbose) {      /* announce what we're doing */
 208         if (subjsort)
 209             printf ("sorting by %s-major %s-minor\n",
 210                 submajor ? subjsort : datesw,
 211                 submajor ? datesw : subjsort);
 212         else
 213             printf ("sorting by datefield %s\n", datesw);
 214     }
 215
 216     /* first sort by date, or by subject-major, date-minor */
 217     qsort ((char *) dlist, nmsgs, sizeof(*dlist),
 218             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 219
 220     /*
 221      * if we're sorting on subject, we need another list
 222      * in subject order, then a merge pass to collate the
 223      * two sorts.
 224      */
 225     if (!submajor && subjsort) {        /* already date sorted */
 226         struct smsg **slist, **flist;
 227         register struct smsg ***il, **fp, **dp;
 228
 229         slist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 230         memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 231         qsort((char *)slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 232
 233         /*
 234          * make an inversion list so we can quickly find
 235          * the collection of messages with the same subj
 236          * given a message number.
 237          */
 238         il = (struct smsg ***) calloc (mp->hghsel+1, sizeof(*il));
 239         if (! il)
 240             adios (NULL, "couldn't allocate msg list");
 241         for (i = 0; i < nmsgs; i++)
 242             il[slist[i]->s_msg] = &slist[i];
 243         /*
 244          * make up the final list, chronological but with
 245          * all the same subjects grouped together.
 246          */
 247         flist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 248         fp = flist;
 249         for (dp = dlist; *dp;) {
 250             register struct smsg **s = il[(*dp++)->s_msg];
 251
 252             /* see if we already did this guy */
 253             if (! s)
 254                 continue;
 255
 256             *fp++ = *s++;
 257             /*
 258              * take the next message(s) if there is one,
 259              * its subject isn't null and its subject
 260              * is the same as this one and it's not too
 261              * far away in time.
 262              */
 263             while (*s && (*s)->s_subj[0] &&
 264                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 265                    (datelimit == 0 ||
 266                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 267                 il[(*s)->s_msg] = 0;
 268                 *fp++ = *s++;
 269             }
 270         }
 271         *fp = 0;
 272         free (slist);
 273         free (dlist);
 274         dlist = flist;
 275     }
 276
 277     /*
 278      * At this point, dlist is a sorted array of pointers to smsg structures,
 279      * each of which contains a message number.
 280      */
 281
 282     rename_msgs (mp, dlist);
 283
 284     context_replace (pfolder, folder);  /* update current folder         */
 285     seq_save (mp);                      /* synchronize message sequences */
 286     context_save ();                    /* save the context file         */
 287     folder_free (mp);                   /* free folder/message structure */
 288     return done (0);
 289 }
 290
 291 static int
 292 read_hdrs (struct msgs *mp, char *datesw)
 293 {
 294     int msgnum;
 295     struct tws tb;
 296     register struct smsg *s;
 297
 298     twscopy (&tb, dlocaltimenow ());
 299
 300     smsgs = (struct smsg *)
 301         calloc ((size_t) (mp->hghsel - mp->lowsel + 2),
 302             sizeof(*smsgs));
 303     if (smsgs == NULL)
 304         adios (NULL, "unable to allocate sort storage");
 305
 306     s = smsgs;
 307     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 308         if (is_selected(mp, msgnum)) {
 309             if (get_fields (datesw, msgnum, s)) {
 310                 s->s_msg = msgnum;
 311                 s++;
 312             }
 313         }
 314     }
 315     s->s_msg = 0;
 316     return(s - smsgs);
 317 }
 318
 319
 320 /*
 321  * Parse the message and get the data or subject field,
 322  * if needed.
 323  */
 324
 325 static int
 326 get_fields (char *datesw, int msg, struct smsg *smsg)
 327 {
 328     register int state;
 329     int compnum;
 330     char *msgnam, buf[BUFSIZ], nam[NAMESZ];
 331     register struct tws *tw;
 332     register char *datecomp = NULL, *subjcomp = NULL;
 333     register FILE *in;
 334
 335     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 336         admonish (msgnam, "unable to read message");
 337         return (0);
 338     }
 339     for (compnum = 1, state = FLD;;) {
 340         switch (state = m_getfld (state, nam, buf, sizeof(buf), in)) {
 341         case FLD:
 342         case FLDEOF:
 343         case FLDPLUS:
 344             compnum++;
 345             if (!mh_strcasecmp (nam, datesw)) {
 346                 datecomp = add (buf, datecomp);
 347                 while (state == FLDPLUS) {
 348                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 349                     datecomp = add (buf, datecomp);
 350                 }
 351                 if (!subjsort || subjcomp)
 352                     break;
 353             } else if (subjsort && !mh_strcasecmp (nam, subjsort)) {
 354                 subjcomp = add (buf, subjcomp);
 355                 while (state == FLDPLUS) {
 356                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 357                     subjcomp = add (buf, subjcomp);
 358                 }
 359                 if (datecomp)
 360                     break;
 361             } else {
 362                 /* just flush this guy */
 363                 while (state == FLDPLUS)
 364                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 365             }
 366             continue;
 367
 368         case BODY:
 369         case BODYEOF:
 370         case FILEEOF:
 371             break;
 372
 373         case LENERR:
 374         case FMTERR:
 375             if (state == LENERR || state == FMTERR)
 376                 admonish (NULL, "format error in message %d (header #%d)",
 377                       msg, compnum);
 378             if (datecomp)
 379                 free (datecomp);
 380             if (subjcomp)
 381                 free (subjcomp);
 382             fclose (in);
 383             return (0);
 384
 385         default:
 386             adios (NULL, "internal error -- you lose");
 387         }
 388         break;
 389     }
 390
 391     /*
 392      * If no date component, then use the modification
 393      * time of the file as its date
 394      */
 395     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 396         struct stat st;
 397
 398         admonish (NULL, "can't parse %s field in message %d", datesw, msg);
 399         fstat (fileno (in), &st);
 400         smsg->s_clock = st.st_mtime;
 401     } else {
 402         smsg->s_clock = dmktime (tw);
 403     }
 404
 405     if (subjsort) {
 406         if (subjcomp) {
 407             /*
 408              * try to make the subject "canonical": delete
 409              * leading "re:", everything but letters & smash
 410              * letters to lower case.
 411              */
 412             register char  *cp, *cp2;
 413             register unsigned char c;
 414
 415             cp = subjcomp;
 416             cp2 = subjcomp;
 417             if (strcmp (subjsort, "subject") == 0) {
 418                 while ((c = *cp)) {
 419                     if (! isspace(c)) {
 420                         if(uprf(cp, "re:"))
 421                             cp += 2;
 422                         else
 423                             break;
 424                     }
 425                     cp++;
 426                 }
 427             }
 428
 429             while ((c = *cp++)) {
 430                 if (isalnum(c))
 431                     *cp2++ = isupper(c) ? tolower(c) : c;
 432             }
 433
 434             *cp2 = '\0';
 435         }
 436         else
 437             subjcomp = "";
 438
 439         smsg->s_subj = subjcomp;
 440     }
 441     fclose (in);
 442     if (datecomp)
 443         free (datecomp);
 444
 445     return (1);
 446 }
 447
 448 /*
 449  * sort on dates.
 450  */
 451 static int
 452 dsort (struct smsg **a, struct smsg **b)
 453 {
 454     if ((*a)->s_clock < (*b)->s_clock)
 455         return (-1);
 456     else if ((*a)->s_clock > (*b)->s_clock)
 457         return (1);
 458     else if ((*a)->s_msg < (*b)->s_msg)
 459         return (-1);
 460     else
 461         return (1);
 462 }
 463
 464 /*
 465  * sort on subjects.
 466  */
 467 static int
 468 subsort (struct smsg **a, struct smsg **b)
 469 {
 470     register int i;
 471
 472     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 473         return (i);
 474
 475     return (dsort (a, b));
 476 }
 477
 478 static int
 479 txtsort (struct smsg **a, struct smsg **b)
 480 {
 481     register int i;
 482
 483     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 484         return (i);
 485     else if ((*a)->s_msg < (*b)->s_msg)
 486         return (-1);
 487     else
 488         return (1);
 489 }
 490
 491 static void
 492 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 493 {
 494     int nxt, old, new;
 495     char *newname, oldname[BUFSIZ];
 496     char newbuf[MAXPATHLEN + 1];
 497
 498     for (;;) {
 499         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 500         mlist[msg] = (struct smsg *)0;
 501         old = smsgs[nxt].s_msg;
 502         new = smsgs[msg].s_msg;
 503         strncpy (oldname, m_name (old), sizeof(oldname));
 504         newname = m_name (new);
 505         if (verbose)
 506             printf ("message %d becomes message %d\n", old, new);
 507
 508         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 509         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 510         ext_hook("ref-hook", oldname, newbuf);
 511
 512         if (rename (oldname, newname) == NOTOK)
 513             adios (newname, "unable to rename %s to", oldname);
 514
 515         copy_msg_flags (mp, new, old);
 516         if (mp->curmsg == old)
 517             seq_setcur (mp, new);
 518
 519         if (nxt == endmsg)
 520             break;
 521
 522         msg = nxt;
 523     }
 524 /*      if (nxt != endmsg); */
 525 /*      rename_chain (mp, mlist, nxt, endmsg); */
 526 }
 527
 528 static void
 529 rename_msgs (struct msgs *mp, struct smsg **mlist)
 530 {
 531     int i, j, old, new;
 532     seqset_t tmpset;
 533     char f1[BUFSIZ], tmpfil[BUFSIZ];
 534     char newbuf[MAXPATHLEN + 1];
 535     struct smsg *sp;
 536
 537     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 538
 539     for (i = 0; i < nmsgs; i++) {
 540         if (! (sp = mlist[i]))
 541             continue;   /* did this one */
 542
 543         j = sp - smsgs;
 544         if (j == i)
 545             continue;   /* this one doesn't move */
 546
 547         /*
 548          * the guy that was msg j is about to become msg i.
 549          * rename 'j' to make a hole, then recursively rename
 550          * guys to fill up the hole.
 551          */
 552         old = smsgs[j].s_msg;
 553         new = smsgs[i].s_msg;
 554         strncpy (f1, m_name (old), sizeof(f1));
 555
 556         if (verbose)
 557             printf ("renaming message chain from %d to %d\n", old, new);
 558
 559         /*
 560          *      Run the external hook to refile the old message as the
 561          *      temporary message number that is off of the end of the
 562          *      messages in the folder.
 563          */
 564
 565         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 566         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 567         ext_hook("ref-hook", f1, newbuf);
 568
 569         if (rename (f1, tmpfil) == NOTOK)
 570             adios (tmpfil, "unable to rename %s to ", f1);
 571
 572         get_msg_flags (mp, &tmpset, old);
 573
 574         rename_chain (mp, mlist, j, i);
 575
 576         /*
 577          *      Run the external hook to refile the temorary message number
 578          *      to the real place.
 579          */
 580
 581         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 582         ext_hook("ref-hook", newbuf, f1);
 583
 584         if (rename (tmpfil, m_name(new)) == NOTOK)
 585             adios (m_name(new), "unable to rename %s to", tmpfil);
 586
 587         set_msg_flags (mp, &tmpset, new);
 588         mp->msgflags |= SEQMOD;
 589     }
 590 }