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

   1 /* sortm.c -- sort messages in a folder by date/time
   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/path.h"
  10 #include "sbr/print_version.h"
  11 #include "sbr/print_help.h"
  12 #include "sbr/error.h"
  13 #include "h/tws.h"
  14 #include "h/done.h"
  15 #include "h/utils.h"
  16 #include "sbr/m_maildir.h"
  17
  18 #define SORTM_SWITCHES \
  19     X("datefield field", 0, DATESW) \
  20     X("textfield field", 0, TEXTSW) \
  21     X("notextfield", 0, NSUBJSW) \
  22     X("subject", -3, SUBJSW) /* backward-compatibility */ \
  23     X("limit days", 0, LIMSW) \
  24     X("nolimit", 0, NLIMSW) \
  25     X("verbose", 0, VERBSW) \
  26     X("noverbose", 0, NVERBSW) \
  27     X("all", 0, ALLMSGS) \
  28     X("noall", 0, NALLMSGS) \
  29     X("check", 0, CHECKSW) \
  30     X("nocheck", 0, NCHECKSW) \
  31     X("version", 0, VERSIONSW) \
  32     X("help", 0, HELPSW) \
  33
  34 #define X(sw, minchars, id) id,
  35 DEFINE_SWITCH_ENUM(SORTM);
  36 #undef X
  37
  38 #define X(sw, minchars, id) { sw, minchars, id },
  39 DEFINE_SWITCH_ARRAY(SORTM, switches);
  40 #undef X
  41
  42 struct smsg {
  43     int s_msg;
  44     time_t s_clock;
  45     char *s_subj;
  46 };
  47
  48 static struct smsg *smsgs;
  49 int nmsgs;
  50
  51 char *subjsort;                 /* sort on subject if != 0 */
  52 time_t datelimit = 0;
  53 bool submajor;                  /* if true, sort on subject-major */
  54 bool verbose;
  55 int allmsgs = 1;
  56 int check_failed = 0;
  57
  58 /* This keeps compiler happy on calls to qsort */
  59 typedef int (*qsort_comp) (const void *, const void *);
  60
  61 /*
  62  * static prototypes
  63  */
  64 static int read_hdrs (struct msgs *, char *);
  65 static int get_fields (char *, int, struct smsg *);
  66 static int dsort (struct smsg **, struct smsg **);
  67 static int subsort (struct smsg **, struct smsg **);
  68 static int txtsort (struct smsg **, struct smsg **);
  69 static void rename_chain (struct msgs *, struct smsg **, int, int);
  70 static void rename_msgs (struct msgs *, struct smsg **);
  71
  72
  73 int
  74 main (int argc, char **argv)
  75 {
  76     int i, msgnum;
  77     char *cp, *maildir, *datesw = NULL;
  78     char *folder = NULL, buf[BUFSIZ], **argp;
  79     char **arguments;
  80     struct msgs_array msgs = { 0, 0, NULL };
  81     struct msgs *mp;
  82     struct smsg **dlist;
  83     bool checksw = false;
  84
  85     if (nmh_init(argv[0], true, true)) { return 1; }
  86
  87     arguments = getarguments (invo_name, argc, argv, 1);
  88     argp = arguments;
  89
  90     /*
  91      * Parse arguments
  92      */
  93     while ((cp = *argp++)) {
  94         if (*cp == '-') {
  95             switch (smatch (++cp, switches)) {
  96             case AMBIGSW:
  97                 ambigsw (cp, switches);
  98                 done (1);
  99             case UNKWNSW:
 100                 die("-%s unknown", cp);
 101
 102             case HELPSW:
 103                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 104                         invo_name);
 105                 print_help (buf, switches, 1);
 106                 done (0);
 107             case VERSIONSW:
 108                 print_version(invo_name);
 109                 done (0);
 110
 111             case DATESW:
 112                 if (datesw)
 113                     die("only one date field at a time");
 114                 if (!(datesw = *argp++) || *datesw == '-')
 115                     die("missing argument to %s", argp[-2]);
 116                 continue;
 117
 118             case TEXTSW:
 119                 if (subjsort)
 120                     die("only one text field at a time");
 121                 if (!(subjsort = *argp++) || *subjsort == '-')
 122                     die("missing argument to %s", argp[-2]);
 123                 continue;
 124
 125             case SUBJSW:
 126                 subjsort = "subject";
 127                 continue;
 128             case NSUBJSW:
 129                 subjsort = NULL;
 130                 continue;
 131
 132             case LIMSW:
 133                 if (!(cp = *argp++) || *cp == '-')
 134                         die("missing argument to %s", argp[-2]);
 135                 while (*cp == '0')
 136                     cp++;               /* skip any leading zeros */
 137                 if (!*cp) {             /* hit end of string */
 138                     submajor = true;    /* sort subject-major */
 139                     continue;
 140                 }
 141                 if (!isdigit((unsigned char) *cp) || !(datelimit = atoi(cp)))
 142                     die("impossible limit %s", cp);
 143                 datelimit *= 60*60*24;
 144                 continue;
 145             case NLIMSW:
 146                 submajor = false;       /* use date-major, but */
 147                 datelimit = 0;          /* use no limit */
 148                 continue;
 149
 150             case VERBSW:
 151                 verbose = true;
 152                 continue;
 153             case NVERBSW:
 154                 verbose = false;
 155                 continue;
 156
 157             case ALLMSGS:
 158                 allmsgs = 1;
 159                 continue;
 160             case NALLMSGS:
 161                 allmsgs = 0;
 162                 continue;
 163
 164             case CHECKSW:
 165                 checksw = true;
 166                 continue;
 167             case NCHECKSW:
 168                 checksw = false;
 169                 continue;
 170             }
 171         }
 172         if (*cp == '+' || *cp == '@') {
 173             if (folder)
 174                 die("only one folder at a time!");
 175             folder = pluspath (cp);
 176         } else
 177             app_msgarg(&msgs, cp);
 178     }
 179
 180     if (!context_find ("path"))
 181         free (path ("./", TFOLDER));
 182     if (!msgs.size) {
 183         if (allmsgs) {
 184             app_msgarg(&msgs, "all");
 185         } else {
 186             die("must specify messages to sort with -noall");
 187         }
 188     }
 189     if (!datesw)
 190         datesw = "date";
 191     if (!folder)
 192         folder = getfolder (1);
 193     maildir = m_maildir (folder);
 194
 195     if (chdir (maildir) == NOTOK)
 196         adios (maildir, "unable to change directory to");
 197
 198     /* read folder and create message structure */
 199     if (!(mp = folder_read (folder, 1)))
 200         die("unable to read folder %s", folder);
 201
 202     /* check for empty folder */
 203     if (mp->nummsg == 0)
 204         die("no messages in %s", folder);
 205
 206     /* parse all the message ranges/sequences and set SELECTED */
 207     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 208         if (!m_convert (mp, msgs.msgs[msgnum]))
 209             done (1);
 210     seq_setprev (mp);   /* set the previous sequence */
 211
 212     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 213         die("no messages to sort");
 214
 215     if (checksw  &&  check_failed) {
 216         die("errors found, no messages sorted");
 217     }
 218
 219     /*
 220      * sort a list of pointers to our "messages to be sorted".
 221      */
 222     dlist = mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 223     for (i = 0; i < nmsgs; i++)
 224         dlist[i] = &smsgs[i];
 225     dlist[nmsgs] = 0;
 226
 227     if (verbose) {      /* announce what we're doing */
 228         if (subjsort)
 229             if (submajor)
 230                 printf ("sorting by %s\n", subjsort);
 231             else
 232                 printf ("sorting by %s-major %s-minor\n", subjsort, datesw);
 233         else
 234             printf ("sorting by datefield %s\n", datesw);
 235     }
 236
 237     /* first sort by date, or by subject-major, date-minor */
 238     qsort (dlist, nmsgs, sizeof(*dlist),
 239             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 240
 241     /*
 242      * if we're sorting on subject, we need another list
 243      * in subject order, then a merge pass to collate the
 244      * two sorts.
 245      */
 246     if (!submajor && subjsort) {        /* already date sorted */
 247         struct smsg **slist, **flist;
 248         struct smsg ***il, **fp, **dp;
 249
 250         slist = mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 251         memcpy(slist, dlist, (nmsgs+1)*sizeof(*slist));
 252         qsort(slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 253
 254         /*
 255          * make an inversion list so we can quickly find
 256          * the collection of messages with the same subj
 257          * given a message number.
 258          */
 259         il = mh_xcalloc(mp->hghsel + 1, sizeof *il);
 260         for (i = 0; i < nmsgs; i++)
 261             il[slist[i]->s_msg] = &slist[i];
 262         /*
 263          * make up the final list, chronological but with
 264          * all the same subjects grouped together.
 265          */
 266         flist = mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 267         fp = flist;
 268         for (dp = dlist; *dp;) {
 269             struct smsg **s = il[(*dp++)->s_msg];
 270
 271             /* see if we already did this guy */
 272             if (! s)
 273                 continue;
 274
 275             *fp++ = *s++;
 276             /*
 277              * take the next message(s) if there is one,
 278              * its subject isn't null and its subject
 279              * is the same as this one and it's not too
 280              * far away in time.
 281              */
 282             while (*s && (*s)->s_subj[0] &&
 283                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 284                    (datelimit == 0 ||
 285                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 286                 il[(*s)->s_msg] = 0;
 287                 *fp++ = *s++;
 288             }
 289         }
 290         *fp = 0;
 291         free (il);
 292         free (slist);
 293         free (dlist);
 294         dlist = flist;
 295     }
 296
 297     /*
 298      * At this point, dlist is a sorted array of pointers to smsg structures,
 299      * each of which contains a message number.
 300      */
 301
 302     rename_msgs (mp, dlist);
 303
 304     context_replace (pfolder, folder);  /* update current folder         */
 305     seq_save (mp);                      /* synchronize message sequences */
 306     context_save ();                    /* save the context file         */
 307     folder_free (mp);                   /* free folder/message structure */
 308     done (0);
 309     return 1;
 310 }
 311
 312 static int
 313 read_hdrs (struct msgs *mp, char *datesw)
 314 {
 315     int msgnum;
 316     struct smsg *s;
 317
 318     smsgs = mh_xcalloc(mp->hghsel - mp->lowsel + 2, sizeof *smsgs);
 319     s = smsgs;
 320     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 321         if (is_selected(mp, msgnum)) {
 322             if (get_fields (datesw, msgnum, s)) {
 323                 s->s_msg = msgnum;
 324                 s++;
 325             }
 326         }
 327     }
 328     s->s_msg = 0;
 329     return s - smsgs;
 330 }
 331
 332
 333 /*
 334  * Parse the message and get the data or subject field,
 335  * if needed.
 336  */
 337
 338 static int
 339 get_fields (char *datesw, int msg, struct smsg *smsg)
 340 {
 341     int state;
 342     int compnum;
 343     char *msgnam, buf[NMH_BUFSIZ], nam[NAMESZ];
 344     struct tws *tw;
 345     char *datecomp = NULL, *subjcomp = NULL;
 346     FILE *in;
 347     m_getfld_state_t gstate;
 348
 349     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 350         admonish (msgnam, "unable to read message");
 351         return 0;
 352     }
 353     gstate = m_getfld_state_init(in);
 354     for (compnum = 1;;) {
 355         int bufsz = sizeof buf;
 356         switch (state = m_getfld2(&gstate, nam, buf, &bufsz)) {
 357         case FLD:
 358         case FLDPLUS:
 359             compnum++;
 360             if (!strcasecmp (nam, datesw)) {
 361                 datecomp = add (buf, datecomp);
 362                 while (state == FLDPLUS) {
 363                     bufsz = sizeof buf;
 364                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 365                     datecomp = add (buf, datecomp);
 366                 }
 367                 if (!subjsort || subjcomp)
 368                     break;
 369             } else if (subjsort && !strcasecmp (nam, subjsort)) {
 370                 subjcomp = add (buf, subjcomp);
 371                 while (state == FLDPLUS) {
 372                     bufsz = sizeof buf;
 373                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 374                     subjcomp = add (buf, subjcomp);
 375                 }
 376                 if (datecomp)
 377                     break;
 378             } else {
 379                 /* just flush this guy */
 380                 while (state == FLDPLUS) {
 381                     bufsz = sizeof buf;
 382                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 383                 }
 384             }
 385             continue;
 386
 387         case BODY:
 388         case FILEEOF:
 389             break;
 390
 391         case LENERR:
 392         case FMTERR:
 393             if (state == LENERR || state == FMTERR) {
 394                 inform("format error in message %d (header #%d), continuing...",
 395                       msg, compnum);
 396                 check_failed = 1;
 397             }
 398             free(datecomp);
 399             free(subjcomp);
 400             fclose (in);
 401             return 0;
 402
 403         default:
 404             die("internal error -- you lose");
 405         }
 406         break;
 407     }
 408     m_getfld_state_destroy (&gstate);
 409
 410     /*
 411      * If no date component, then use the modification
 412      * time of the file as its date
 413      */
 414     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 415         struct stat st;
 416
 417         inform("can't parse %s field in message %d, "
 418             "will use file modification time", datesw, msg);
 419         fstat (fileno (in), &st);
 420         smsg->s_clock = st.st_mtime;
 421         check_failed = 1;
 422     } else {
 423         smsg->s_clock = dmktime (tw);
 424     }
 425
 426     if (subjsort) {
 427         if (subjcomp) {
 428             /*
 429              * try to make the subject "canonical": delete
 430              * leading "re:", everything but letters & smash
 431              * letters to lower case.
 432              */
 433             char  *cp, *cp2, c;
 434
 435             cp = subjcomp;
 436             cp2 = subjcomp;
 437             if (strcmp (subjsort, "subject") == 0) {
 438                 while ((c = *cp)) {
 439                     if (! isspace((unsigned char) c)) {
 440                         if(!uprf(cp, "re:"))
 441                             break;
 442                         cp += 2;
 443                     }
 444                     cp++;
 445                 }
 446             }
 447
 448             while ((c = *cp++)) {
 449                 if (isascii((unsigned char) c) && isalnum((unsigned char) c))
 450                     *cp2++ = tolower((unsigned char)c);
 451             }
 452
 453             *cp2 = '\0';
 454         }
 455         else
 456             subjcomp = "";
 457
 458         smsg->s_subj = subjcomp;
 459     }
 460     fclose (in);
 461     free(datecomp);
 462
 463     return 1;
 464 }
 465
 466 /*
 467  * sort on dates.
 468  */
 469 static int
 470 dsort (struct smsg **a, struct smsg **b)
 471 {
 472     if ((*a)->s_clock < (*b)->s_clock)
 473         return -1;
 474     if ((*a)->s_clock > (*b)->s_clock)
 475         return 1;
 476     if ((*a)->s_msg < (*b)->s_msg)
 477         return -1;
 478     return 1;
 479 }
 480
 481 /*
 482  * sort on subjects.
 483  */
 484 static int
 485 subsort (struct smsg **a, struct smsg **b)
 486 {
 487     int i;
 488
 489     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 490         return i;
 491
 492     return dsort(a, b);
 493 }
 494
 495 static int
 496 txtsort (struct smsg **a, struct smsg **b)
 497 {
 498     int i;
 499
 500     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 501         return i;
 502     if ((*a)->s_msg < (*b)->s_msg)
 503         return -1;
 504     return 1;
 505 }
 506
 507 static void
 508 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 509 {
 510     int nxt, old, new;
 511     char *newname, oldname[BUFSIZ];
 512     char newbuf[PATH_MAX + 1];
 513
 514     for (;;) {
 515         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 516         mlist[msg] = NULL;
 517         old = smsgs[nxt].s_msg;
 518         new = smsgs[msg].s_msg;
 519         strncpy (oldname, m_name (old), sizeof(oldname));
 520         newname = m_name (new);
 521         if (verbose)
 522             printf ("message %d becomes message %d\n", old, new);
 523
 524         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 525         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 526         ext_hook("ref-hook", oldname, newbuf);
 527
 528         if (rename (oldname, newname) == NOTOK)
 529             adios (newname, "unable to rename %s to", oldname);
 530
 531         copy_msg_flags (mp, new, old);
 532         if (mp->curmsg == old)
 533             seq_setcur (mp, new);
 534
 535         if (nxt == endmsg)
 536             break;
 537
 538         msg = nxt;
 539     }
 540 /*      if (nxt != endmsg); */
 541 /*      rename_chain (mp, mlist, nxt, endmsg); */
 542 }
 543
 544 static void
 545 rename_msgs (struct msgs *mp, struct smsg **mlist)
 546 {
 547     int i, j, old, new;
 548     bvector_t tmpset = bvector_create ();
 549     char f1[BUFSIZ], tmpfil[BUFSIZ];
 550     char newbuf[PATH_MAX + 1];
 551     struct smsg *sp;
 552
 553     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 554
 555     for (i = 0; i < nmsgs; i++) {
 556         if (! (sp = mlist[i]))
 557             continue;   /* did this one */
 558
 559         j = sp - smsgs;
 560         if (j == i)
 561             continue;   /* this one doesn't move */
 562
 563         /*
 564          * the guy that was msg j is about to become msg i.
 565          * rename 'j' to make a hole, then recursively rename
 566          * guys to fill up the hole.
 567          */
 568         old = smsgs[j].s_msg;
 569         new = smsgs[i].s_msg;
 570         strncpy (f1, m_name (old), sizeof(f1));
 571
 572         if (verbose)
 573             printf ("renaming message chain from %d to %d\n", old, new);
 574
 575         /*
 576          *      Run the external hook to refile the old message as the
 577          *      temporary message number that is off of the end of the
 578          *      messages in the folder.
 579          */
 580
 581         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 582         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 583         ext_hook("ref-hook", f1, newbuf);
 584
 585         if (rename (f1, tmpfil) == NOTOK)
 586             adios (tmpfil, "unable to rename %s to ", f1);
 587
 588         get_msg_flags (mp, tmpset, old);
 589
 590         rename_chain (mp, mlist, j, i);
 591
 592         /*
 593          *      Run the external hook to refile the temporary message number
 594          *      to the real place.
 595          */
 596
 597         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 598         ext_hook("ref-hook", newbuf, f1);
 599
 600         if (rename (tmpfil, m_name(new)) == NOTOK)
 601             adios (m_name(new), "unable to rename %s to", tmpfil);
 602
 603         set_msg_flags (mp, tmpset, new);
 604         mp->msgflags |= SEQMOD;
 605     }
 606
 607     bvector_free (tmpset);
 608 }