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

   1 /*
   2  * sortm.c -- sort messages in a folder by date/time
   3  *
   4  * This code is Copyright (c) 2002, by the authors of nmh.  See the
   5  * COPYRIGHT file in the root directory of the nmh distribution for
   6  * complete copyright information.
   7  */
   8
   9 #include <h/mh.h>
  10 #include <h/tws.h>
  11 #include <h/utils.h>
  12
  13 #define SORTM_SWITCHES \
  14     X("datefield field", 0, DATESW) \
  15     X("textfield field", 0, TEXTSW) \
  16     X("notextfield", 0, NSUBJSW) \
  17     X("subject", -3, SUBJSW) /* backward-compatibility */ \
  18     X("limit days", 0, LIMSW) \
  19     X("nolimit", 0, NLIMSW) \
  20     X("verbose", 0, VERBSW) \
  21     X("noverbose", 0, NVERBSW) \
  22     X("all", 0, ALLMSGS) \
  23     X("noall", 0, NALLMSGS) \
  24     X("check", 0, CHECKSW) \
  25     X("nocheck", 0, NCHECKSW) \
  26     X("version", 0, VERSIONSW) \
  27     X("help", 0, HELPSW) \
  28
  29 #define X(sw, minchars, id) id,
  30 DEFINE_SWITCH_ENUM(SORTM);
  31 #undef X
  32
  33 #define X(sw, minchars, id) { sw, minchars, id },
  34 DEFINE_SWITCH_ARRAY(SORTM, switches);
  35 #undef X
  36
  37 struct smsg {
  38     int s_msg;
  39     time_t s_clock;
  40     char *s_subj;
  41 };
  42
  43 static struct smsg *smsgs;
  44 int nmsgs;
  45
  46 char *subjsort = (char *) 0;    /* sort on subject if != 0 */
  47 time_t datelimit = 0;
  48 int submajor = 0;               /* if true, sort on subject-major */
  49 int verbose;
  50 int allmsgs = 1;
  51 int check_failed = 0;
  52
  53 /* This keeps compiler happy on calls to qsort */
  54 typedef int (*qsort_comp) (const void *, const void *);
  55
  56 /*
  57  * static prototypes
  58  */
  59 static int read_hdrs (struct msgs *, char *);
  60 static int get_fields (char *, int, struct smsg *);
  61 static int dsort (struct smsg **, struct smsg **);
  62 static int subsort (struct smsg **, struct smsg **);
  63 static int txtsort (struct smsg **, struct smsg **);
  64 static void rename_chain (struct msgs *, struct smsg **, int, int);
  65 static void rename_msgs (struct msgs *, struct smsg **);
  66
  67
  68 int
  69 main (int argc, char **argv)
  70 {
  71     int i, msgnum;
  72     char *cp, *maildir, *datesw = NULL;
  73     char *folder = NULL, buf[BUFSIZ], **argp;
  74     char **arguments;
  75     struct msgs_array msgs = { 0, 0, NULL };
  76     struct msgs *mp;
  77     struct smsg **dlist;
  78     int checksw = 0;
  79
  80     if (nmh_init(argv[0], 1)) { return 1; }
  81
  82     arguments = getarguments (invo_name, argc, argv, 1);
  83     argp = arguments;
  84
  85     /*
  86      * Parse arguments
  87      */
  88     while ((cp = *argp++)) {
  89         if (*cp == '-') {
  90             switch (smatch (++cp, switches)) {
  91             case AMBIGSW:
  92                 ambigsw (cp, switches);
  93                 done (1);
  94             case UNKWNSW:
  95                 adios (NULL, "-%s unknown", cp);
  96
  97             case HELPSW:
  98                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
  99                         invo_name);
 100                 print_help (buf, switches, 1);
 101                 done (0);
 102             case VERSIONSW:
 103                 print_version(invo_name);
 104                 done (0);
 105
 106             case DATESW:
 107                 if (datesw)
 108                     adios (NULL, "only one date field at a time");
 109                 if (!(datesw = *argp++) || *datesw == '-')
 110                     adios (NULL, "missing argument to %s", argp[-2]);
 111                 continue;
 112
 113             case TEXTSW:
 114                 if (subjsort)
 115                     adios (NULL, "only one text field at a time");
 116                 if (!(subjsort = *argp++) || *subjsort == '-')
 117                     adios (NULL, "missing argument to %s", argp[-2]);
 118                 continue;
 119
 120             case SUBJSW:
 121                 subjsort = "subject";
 122                 continue;
 123             case NSUBJSW:
 124                 subjsort = NULL;
 125                 continue;
 126
 127             case LIMSW:
 128                 if (!(cp = *argp++) || *cp == '-')
 129                         adios (NULL, "missing argument to %s", argp[-2]);
 130                 while (*cp == '0')
 131                     cp++;               /* skip any leading zeros */
 132                 if (!*cp) {             /* hit end of string */
 133                     submajor++;         /* sort subject-major */
 134                     continue;
 135                 }
 136                 if (!isdigit((unsigned char) *cp) || !(datelimit = atoi(cp)))
 137                     adios (NULL, "impossible limit %s", cp);
 138                 datelimit *= 60*60*24;
 139                 continue;
 140             case NLIMSW:
 141                 submajor = 0;   /* use date-major, but */
 142                 datelimit = 0;  /* use no limit */
 143                 continue;
 144
 145             case VERBSW:
 146                 verbose++;
 147                 continue;
 148             case NVERBSW:
 149                 verbose = 0;
 150                 continue;
 151
 152             case ALLMSGS:
 153                 allmsgs = 1;
 154                 continue;
 155             case NALLMSGS:
 156                 allmsgs = 0;
 157                 continue;
 158
 159             case CHECKSW:
 160                 checksw = 1;
 161                 continue;
 162             case NCHECKSW:
 163                 checksw = 0;
 164                 continue;
 165             }
 166         }
 167         if (*cp == '+' || *cp == '@') {
 168             if (folder)
 169                 adios (NULL, "only one folder at a time!");
 170             else
 171                 folder = pluspath (cp);
 172         } else
 173                 app_msgarg(&msgs, cp);
 174     }
 175
 176     if (!context_find ("path"))
 177         free (path ("./", TFOLDER));
 178     if (!msgs.size) {
 179         if (allmsgs) {
 180             app_msgarg(&msgs, "all");
 181         } else {
 182             adios (NULL, "must specify messages to sort with -noall");
 183         }
 184     }
 185     if (!datesw)
 186         datesw = "date";
 187     if (!folder)
 188         folder = getfolder (1);
 189     maildir = m_maildir (folder);
 190
 191     if (chdir (maildir) == NOTOK)
 192         adios (maildir, "unable to change directory to");
 193
 194     /* read folder and create message structure */
 195     if (!(mp = folder_read (folder, 1)))
 196         adios (NULL, "unable to read folder %s", folder);
 197
 198     /* check for empty folder */
 199     if (mp->nummsg == 0)
 200         adios (NULL, "no messages in %s", folder);
 201
 202     /* parse all the message ranges/sequences and set SELECTED */
 203     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 204         if (!m_convert (mp, msgs.msgs[msgnum]))
 205             done (1);
 206     seq_setprev (mp);   /* set the previous sequence */
 207
 208     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 209         adios (NULL, "no messages to sort");
 210
 211     if (checksw  &&  check_failed) {
 212         adios (NULL, "errors found, no messages sorted");
 213     }
 214
 215     /*
 216      * sort a list of pointers to our "messages to be sorted".
 217      */
 218     dlist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 219     for (i = 0; i < nmsgs; i++)
 220         dlist[i] = &smsgs[i];
 221     dlist[nmsgs] = 0;
 222
 223     if (verbose) {      /* announce what we're doing */
 224         if (subjsort)
 225             if (submajor)
 226                 printf ("sorting by %s\n", subjsort);
 227             else
 228                 printf ("sorting by %s-major %s-minor\n", subjsort, datesw);
 229         else
 230             printf ("sorting by datefield %s\n", datesw);
 231     }
 232
 233     /* first sort by date, or by subject-major, date-minor */
 234     qsort ((char *) dlist, nmsgs, sizeof(*dlist),
 235             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 236
 237     /*
 238      * if we're sorting on subject, we need another list
 239      * in subject order, then a merge pass to collate the
 240      * two sorts.
 241      */
 242     if (!submajor && subjsort) {        /* already date sorted */
 243         struct smsg **slist, **flist;
 244         struct smsg ***il, **fp, **dp;
 245
 246         slist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 247         memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 248         qsort((char *)slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 249
 250         /*
 251          * make an inversion list so we can quickly find
 252          * the collection of messages with the same subj
 253          * given a message number.
 254          */
 255         il = mh_xcalloc(mp->hghsel + 1, sizeof *il);
 256         for (i = 0; i < nmsgs; i++)
 257             il[slist[i]->s_msg] = &slist[i];
 258         /*
 259          * make up the final list, chronological but with
 260          * all the same subjects grouped together.
 261          */
 262         flist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 263         fp = flist;
 264         for (dp = dlist; *dp;) {
 265             struct smsg **s = il[(*dp++)->s_msg];
 266
 267             /* see if we already did this guy */
 268             if (! s)
 269                 continue;
 270
 271             *fp++ = *s++;
 272             /*
 273              * take the next message(s) if there is one,
 274              * its subject isn't null and its subject
 275              * is the same as this one and it's not too
 276              * far away in time.
 277              */
 278             while (*s && (*s)->s_subj[0] &&
 279                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 280                    (datelimit == 0 ||
 281                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 282                 il[(*s)->s_msg] = 0;
 283                 *fp++ = *s++;
 284             }
 285         }
 286         *fp = 0;
 287         free (il);
 288         free (slist);
 289         free (dlist);
 290         dlist = flist;
 291     }
 292
 293     /*
 294      * At this point, dlist is a sorted array of pointers to smsg structures,
 295      * each of which contains a message number.
 296      */
 297
 298     rename_msgs (mp, dlist);
 299
 300     context_replace (pfolder, folder);  /* update current folder         */
 301     seq_save (mp);                      /* synchronize message sequences */
 302     context_save ();                    /* save the context file         */
 303     folder_free (mp);                   /* free folder/message structure */
 304     done (0);
 305     return 1;
 306 }
 307
 308 static int
 309 read_hdrs (struct msgs *mp, char *datesw)
 310 {
 311     int msgnum;
 312     struct tws tb;
 313     struct smsg *s;
 314
 315     twscopy (&tb, dlocaltimenow ());
 316
 317     smsgs = mh_xcalloc(mp->hghsel - mp->lowsel + 2, sizeof *smsgs);
 318     s = smsgs;
 319     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 320         if (is_selected(mp, msgnum)) {
 321             if (get_fields (datesw, msgnum, s)) {
 322                 s->s_msg = msgnum;
 323                 s++;
 324             }
 325         }
 326     }
 327     s->s_msg = 0;
 328     return(s - smsgs);
 329 }
 330
 331
 332 /*
 333  * Parse the message and get the data or subject field,
 334  * if needed.
 335  */
 336
 337 static int
 338 get_fields (char *datesw, int msg, struct smsg *smsg)
 339 {
 340     int state;
 341     int compnum;
 342     char *msgnam, buf[BUFSIZ], nam[NAMESZ];
 343     struct tws *tw;
 344     char *datecomp = NULL, *subjcomp = NULL;
 345     FILE *in;
 346     m_getfld_state_t gstate = 0;
 347
 348     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 349         admonish (msgnam, "unable to read message");
 350         return (0);
 351     }
 352     for (compnum = 1;;) {
 353         int bufsz = sizeof buf;
 354         switch (state = m_getfld (&gstate, nam, buf, &bufsz, in)) {
 355         case FLD:
 356         case FLDPLUS:
 357             compnum++;
 358             if (!strcasecmp (nam, datesw)) {
 359                 datecomp = add (buf, datecomp);
 360                 while (state == FLDPLUS) {
 361                     bufsz = sizeof buf;
 362                     state = m_getfld (&gstate, nam, buf, &bufsz, in);
 363                     datecomp = add (buf, datecomp);
 364                 }
 365                 if (!subjsort || subjcomp)
 366                     break;
 367             } else if (subjsort && !strcasecmp (nam, subjsort)) {
 368                 subjcomp = add (buf, subjcomp);
 369                 while (state == FLDPLUS) {
 370                     bufsz = sizeof buf;
 371                     state = m_getfld (&gstate, nam, buf, &bufsz, in);
 372                     subjcomp = add (buf, subjcomp);
 373                 }
 374                 if (datecomp)
 375                     break;
 376             } else {
 377                 /* just flush this guy */
 378                 while (state == FLDPLUS) {
 379                     bufsz = sizeof buf;
 380                     state = m_getfld (&gstate, nam, buf, &bufsz, in);
 381                 }
 382             }
 383             continue;
 384
 385         case BODY:
 386         case FILEEOF:
 387             break;
 388
 389         case LENERR:
 390         case FMTERR:
 391             if (state == LENERR || state == FMTERR) {
 392                 inform("format error in message %d (header #%d), continuing...",
 393                       msg, compnum);
 394                 check_failed = 1;
 395             }
 396             mh_xfree(datecomp);
 397             mh_xfree(subjcomp);
 398             fclose (in);
 399             return (0);
 400
 401         default:
 402             adios (NULL, "internal error -- you lose");
 403         }
 404         break;
 405     }
 406     m_getfld_state_destroy (&gstate);
 407
 408     /*
 409      * If no date component, then use the modification
 410      * time of the file as its date
 411      */
 412     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 413         struct stat st;
 414
 415         inform("can't parse %s field in message %d, "
 416             "will use file modification time", datesw, msg);
 417         fstat (fileno (in), &st);
 418         smsg->s_clock = st.st_mtime;
 419         check_failed = 1;
 420     } else {
 421         smsg->s_clock = dmktime (tw);
 422     }
 423
 424     if (subjsort) {
 425         if (subjcomp) {
 426             /*
 427              * try to make the subject "canonical": delete
 428              * leading "re:", everything but letters & smash
 429              * letters to lower case.
 430              */
 431             char  *cp, *cp2, c;
 432
 433             cp = subjcomp;
 434             cp2 = subjcomp;
 435             if (strcmp (subjsort, "subject") == 0) {
 436                 while ((c = *cp)) {
 437                     if (! isspace((unsigned char) c)) {
 438                         if(uprf(cp, "re:"))
 439                             cp += 2;
 440                         else
 441                             break;
 442                     }
 443                     cp++;
 444                 }
 445             }
 446
 447             while ((c = *cp++)) {
 448                 if (isascii((unsigned char) c) && isalnum((unsigned char) c))
 449                     *cp2++ = tolower((unsigned char)c);
 450             }
 451
 452             *cp2 = '\0';
 453         }
 454         else
 455             subjcomp = "";
 456
 457         smsg->s_subj = subjcomp;
 458     }
 459     fclose (in);
 460     mh_xfree(datecomp);
 461
 462     return (1);
 463 }
 464
 465 /*
 466  * sort on dates.
 467  */
 468 static int
 469 dsort (struct smsg **a, struct smsg **b)
 470 {
 471     if ((*a)->s_clock < (*b)->s_clock)
 472         return (-1);
 473     if ((*a)->s_clock > (*b)->s_clock)
 474         return (1);
 475     if ((*a)->s_msg < (*b)->s_msg)
 476         return (-1);
 477     return (1);
 478 }
 479
 480 /*
 481  * sort on subjects.
 482  */
 483 static int
 484 subsort (struct smsg **a, struct smsg **b)
 485 {
 486     int i;
 487
 488     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 489         return (i);
 490
 491     return (dsort (a, b));
 492 }
 493
 494 static int
 495 txtsort (struct smsg **a, struct smsg **b)
 496 {
 497     int i;
 498
 499     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 500         return (i);
 501     if ((*a)->s_msg < (*b)->s_msg)
 502         return (-1);
 503     return (1);
 504 }
 505
 506 static void
 507 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 508 {
 509     int nxt, old, new;
 510     char *newname, oldname[BUFSIZ];
 511     char newbuf[PATH_MAX + 1];
 512
 513     for (;;) {
 514         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 515         mlist[msg] = NULL;
 516         old = smsgs[nxt].s_msg;
 517         new = smsgs[msg].s_msg;
 518         strncpy (oldname, m_name (old), sizeof(oldname));
 519         newname = m_name (new);
 520         if (verbose)
 521             printf ("message %d becomes message %d\n", old, new);
 522
 523         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 524         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 525         ext_hook("ref-hook", oldname, newbuf);
 526
 527         if (rename (oldname, newname) == NOTOK)
 528             adios (newname, "unable to rename %s to", oldname);
 529
 530         copy_msg_flags (mp, new, old);
 531         if (mp->curmsg == old)
 532             seq_setcur (mp, new);
 533
 534         if (nxt == endmsg)
 535             break;
 536
 537         msg = nxt;
 538     }
 539 /*      if (nxt != endmsg); */
 540 /*      rename_chain (mp, mlist, nxt, endmsg); */
 541 }
 542
 543 static void
 544 rename_msgs (struct msgs *mp, struct smsg **mlist)
 545 {
 546     int i, j, old, new;
 547     bvector_t tmpset = bvector_create (0);
 548     char f1[BUFSIZ], tmpfil[BUFSIZ];
 549     char newbuf[PATH_MAX + 1];
 550     struct smsg *sp;
 551
 552     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 553
 554     for (i = 0; i < nmsgs; i++) {
 555         if (! (sp = mlist[i]))
 556             continue;   /* did this one */
 557
 558         j = sp - smsgs;
 559         if (j == i)
 560             continue;   /* this one doesn't move */
 561
 562         /*
 563          * the guy that was msg j is about to become msg i.
 564          * rename 'j' to make a hole, then recursively rename
 565          * guys to fill up the hole.
 566          */
 567         old = smsgs[j].s_msg;
 568         new = smsgs[i].s_msg;
 569         strncpy (f1, m_name (old), sizeof(f1));
 570
 571         if (verbose)
 572             printf ("renaming message chain from %d to %d\n", old, new);
 573
 574         /*
 575          *      Run the external hook to refile the old message as the
 576          *      temporary message number that is off of the end of the
 577          *      messages in the folder.
 578          */
 579
 580         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 581         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 582         ext_hook("ref-hook", f1, newbuf);
 583
 584         if (rename (f1, tmpfil) == NOTOK)
 585             adios (tmpfil, "unable to rename %s to ", f1);
 586
 587         get_msg_flags (mp, tmpset, old);
 588
 589         rename_chain (mp, mlist, j, i);
 590
 591         /*
 592          *      Run the external hook to refile the temorary message number
 593          *      to the real place.
 594          */
 595
 596         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 597         ext_hook("ref-hook", newbuf, f1);
 598
 599         if (rename (tmpfil, m_name(new)) == NOTOK)
 600             adios (m_name(new), "unable to rename %s to", tmpfil);
 601
 602         set_msg_flags (mp, tmpset, new);
 603         mp->msgflags |= SEQMOD;
 604     }
 605
 606     bvector_free (tmpset);
 607 }