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/print_version.h"
  10 #include "sbr/print_help.h"
  11 #include "sbr/error.h"
  12 #include "h/tws.h"
  13 #include "h/done.h"
  14 #include "h/utils.h"
  15 #include "sbr/m_maildir.h"
  16
  17 #define SORTM_SWITCHES \
  18     X("datefield field", 0, DATESW) \
  19     X("textfield field", 0, TEXTSW) \
  20     X("notextfield", 0, NSUBJSW) \
  21     X("subject", -3, SUBJSW) /* backward-compatibility */ \
  22     X("limit days", 0, LIMSW) \
  23     X("nolimit", 0, NLIMSW) \
  24     X("verbose", 0, VERBSW) \
  25     X("noverbose", 0, NVERBSW) \
  26     X("all", 0, ALLMSGS) \
  27     X("noall", 0, NALLMSGS) \
  28     X("check", 0, CHECKSW) \
  29     X("nocheck", 0, NCHECKSW) \
  30     X("version", 0, VERSIONSW) \
  31     X("help", 0, HELPSW) \
  32
  33 #define X(sw, minchars, id) id,
  34 DEFINE_SWITCH_ENUM(SORTM);
  35 #undef X
  36
  37 #define X(sw, minchars, id) { sw, minchars, id },
  38 DEFINE_SWITCH_ARRAY(SORTM, switches);
  39 #undef X
  40
  41 struct smsg {
  42     int s_msg;
  43     time_t s_clock;
  44     char *s_subj;
  45 };
  46
  47 static struct smsg *smsgs;
  48 int nmsgs;
  49
  50 char *subjsort;                 /* sort on subject if != 0 */
  51 time_t datelimit = 0;
  52 bool submajor;                  /* if true, sort on subject-major */
  53 bool verbose;
  54 int allmsgs = 1;
  55 int check_failed = 0;
  56
  57 /* This keeps compiler happy on calls to qsort */
  58 typedef int (*qsort_comp) (const void *, const void *);
  59
  60 /*
  61  * static prototypes
  62  */
  63 static int read_hdrs (struct msgs *, char *);
  64 static int get_fields (char *, int, struct smsg *);
  65 static int dsort (struct smsg **, struct smsg **);
  66 static int subsort (struct smsg **, struct smsg **);
  67 static int txtsort (struct smsg **, struct smsg **);
  68 static void rename_chain (struct msgs *, struct smsg **, int, int);
  69 static void rename_msgs (struct msgs *, struct smsg **);
  70
  71
  72 int
  73 main (int argc, char **argv)
  74 {
  75     int i, msgnum;
  76     char *cp, *maildir, *datesw = NULL;
  77     char *folder = NULL, buf[BUFSIZ], **argp;
  78     char **arguments;
  79     struct msgs_array msgs = { 0, 0, NULL };
  80     struct msgs *mp;
  81     struct smsg **dlist;
  82     bool checksw = false;
  83
  84     if (nmh_init(argv[0], true, true)) { return 1; }
  85
  86     arguments = getarguments (invo_name, argc, argv, 1);
  87     argp = arguments;
  88
  89     /*
  90      * Parse arguments
  91      */
  92     while ((cp = *argp++)) {
  93         if (*cp == '-') {
  94             switch (smatch (++cp, switches)) {
  95             case AMBIGSW:
  96                 ambigsw (cp, switches);
  97                 done (1);
  98             case UNKWNSW:
  99                 die("-%s unknown", cp);
 100
 101             case HELPSW:
 102                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 103                         invo_name);
 104                 print_help (buf, switches, 1);
 105                 done (0);
 106             case VERSIONSW:
 107                 print_version(invo_name);
 108                 done (0);
 109
 110             case DATESW:
 111                 if (datesw)
 112                     die("only one date field at a time");
 113                 if (!(datesw = *argp++) || *datesw == '-')
 114                     die("missing argument to %s", argp[-2]);
 115                 continue;
 116
 117             case TEXTSW:
 118                 if (subjsort)
 119                     die("only one text field at a time");
 120                 if (!(subjsort = *argp++) || *subjsort == '-')
 121                     die("missing argument to %s", argp[-2]);
 122                 continue;
 123
 124             case SUBJSW:
 125                 subjsort = "subject";
 126                 continue;
 127             case NSUBJSW:
 128                 subjsort = NULL;
 129                 continue;
 130
 131             case LIMSW:
 132                 if (!(cp = *argp++) || *cp == '-')
 133                         die("missing argument to %s", argp[-2]);
 134                 while (*cp == '0')
 135                     cp++;               /* skip any leading zeros */
 136                 if (!*cp) {             /* hit end of string */
 137                     submajor = true;    /* sort subject-major */
 138                     continue;
 139                 }
 140                 if (!isdigit((unsigned char) *cp) || !(datelimit = atoi(cp)))
 141                     die("impossible limit %s", cp);
 142                 datelimit *= 60*60*24;
 143                 continue;
 144             case NLIMSW:
 145                 submajor = false;       /* use date-major, but */
 146                 datelimit = 0;          /* use no limit */
 147                 continue;
 148
 149             case VERBSW:
 150                 verbose = true;
 151                 continue;
 152             case NVERBSW:
 153                 verbose = false;
 154                 continue;
 155
 156             case ALLMSGS:
 157                 allmsgs = 1;
 158                 continue;
 159             case NALLMSGS:
 160                 allmsgs = 0;
 161                 continue;
 162
 163             case CHECKSW:
 164                 checksw = true;
 165                 continue;
 166             case NCHECKSW:
 167                 checksw = false;
 168                 continue;
 169             }
 170         }
 171         if (*cp == '+' || *cp == '@') {
 172             if (folder)
 173                 die("only one folder at a time!");
 174             folder = pluspath (cp);
 175         } else
 176             app_msgarg(&msgs, cp);
 177     }
 178
 179     if (!context_find ("path"))
 180         free (path ("./", TFOLDER));
 181     if (!msgs.size) {
 182         if (allmsgs) {
 183             app_msgarg(&msgs, "all");
 184         } else {
 185             die("must specify messages to sort with -noall");
 186         }
 187     }
 188     if (!datesw)
 189         datesw = "date";
 190     if (!folder)
 191         folder = getfolder (1);
 192     maildir = m_maildir (folder);
 193
 194     if (chdir (maildir) == NOTOK)
 195         adios (maildir, "unable to change directory to");
 196
 197     /* read folder and create message structure */
 198     if (!(mp = folder_read (folder, 1)))
 199         die("unable to read folder %s", folder);
 200
 201     /* check for empty folder */
 202     if (mp->nummsg == 0)
 203         die("no messages in %s", folder);
 204
 205     /* parse all the message ranges/sequences and set SELECTED */
 206     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 207         if (!m_convert (mp, msgs.msgs[msgnum]))
 208             done (1);
 209     seq_setprev (mp);   /* set the previous sequence */
 210
 211     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 212         die("no messages to sort");
 213
 214     if (checksw  &&  check_failed) {
 215         die("errors found, no messages sorted");
 216     }
 217
 218     /*
 219      * sort a list of pointers to our "messages to be sorted".
 220      */
 221     dlist = mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 222     for (i = 0; i < nmsgs; i++)
 223         dlist[i] = &smsgs[i];
 224     dlist[nmsgs] = 0;
 225
 226     if (verbose) {      /* announce what we're doing */
 227         if (subjsort)
 228             if (submajor)
 229                 printf ("sorting by %s\n", subjsort);
 230             else
 231                 printf ("sorting by %s-major %s-minor\n", subjsort, datesw);
 232         else
 233             printf ("sorting by datefield %s\n", datesw);
 234     }
 235
 236     /* first sort by date, or by subject-major, date-minor */
 237     qsort (dlist, nmsgs, sizeof(*dlist),
 238             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 239
 240     /*
 241      * if we're sorting on subject, we need another list
 242      * in subject order, then a merge pass to collate the
 243      * two sorts.
 244      */
 245     if (!submajor && subjsort) {        /* already date sorted */
 246         struct smsg **slist, **flist;
 247         struct smsg ***il, **fp, **dp;
 248
 249         slist = mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 250         memcpy(slist, dlist, (nmsgs+1)*sizeof(*slist));
 251         qsort(slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 252
 253         /*
 254          * make an inversion list so we can quickly find
 255          * the collection of messages with the same subj
 256          * given a message number.
 257          */
 258         il = mh_xcalloc(mp->hghsel + 1, sizeof *il);
 259         for (i = 0; i < nmsgs; i++)
 260             il[slist[i]->s_msg] = &slist[i];
 261         /*
 262          * make up the final list, chronological but with
 263          * all the same subjects grouped together.
 264          */
 265         flist = mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 266         fp = flist;
 267         for (dp = dlist; *dp;) {
 268             struct smsg **s = il[(*dp++)->s_msg];
 269
 270             /* see if we already did this guy */
 271             if (! s)
 272                 continue;
 273
 274             *fp++ = *s++;
 275             /*
 276              * take the next message(s) if there is one,
 277              * its subject isn't null and its subject
 278              * is the same as this one and it's not too
 279              * far away in time.
 280              */
 281             while (*s && (*s)->s_subj[0] &&
 282                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 283                    (datelimit == 0 ||
 284                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 285                 il[(*s)->s_msg] = 0;
 286                 *fp++ = *s++;
 287             }
 288         }
 289         *fp = 0;
 290         free (il);
 291         free (slist);
 292         free (dlist);
 293         dlist = flist;
 294     }
 295
 296     /*
 297      * At this point, dlist is a sorted array of pointers to smsg structures,
 298      * each of which contains a message number.
 299      */
 300
 301     rename_msgs (mp, dlist);
 302
 303     context_replace (pfolder, folder);  /* update current folder         */
 304     seq_save (mp);                      /* synchronize message sequences */
 305     context_save ();                    /* save the context file         */
 306     folder_free (mp);                   /* free folder/message structure */
 307     done (0);
 308     return 1;
 309 }
 310
 311 static int
 312 read_hdrs (struct msgs *mp, char *datesw)
 313 {
 314     int msgnum;
 315     struct smsg *s;
 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[NMH_BUFSIZ], nam[NAMESZ];
 343     struct tws *tw;
 344     char *datecomp = NULL, *subjcomp = NULL;
 345     FILE *in;
 346     m_getfld_state_t gstate;
 347
 348     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 349         admonish (msgnam, "unable to read message");
 350         return 0;
 351     }
 352     gstate = m_getfld_state_init(in);
 353     for (compnum = 1;;) {
 354         int bufsz = sizeof buf;
 355         switch (state = m_getfld2(&gstate, nam, buf, &bufsz)) {
 356         case FLD:
 357         case FLDPLUS:
 358             compnum++;
 359             if (!strcasecmp (nam, datesw)) {
 360                 datecomp = add (buf, datecomp);
 361                 while (state == FLDPLUS) {
 362                     bufsz = sizeof buf;
 363                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 364                     datecomp = add (buf, datecomp);
 365                 }
 366                 if (!subjsort || subjcomp)
 367                     break;
 368             } else if (subjsort && !strcasecmp (nam, subjsort)) {
 369                 subjcomp = add (buf, subjcomp);
 370                 while (state == FLDPLUS) {
 371                     bufsz = sizeof buf;
 372                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 373                     subjcomp = add (buf, subjcomp);
 374                 }
 375                 if (datecomp)
 376                     break;
 377             } else {
 378                 /* just flush this guy */
 379                 while (state == FLDPLUS) {
 380                     bufsz = sizeof buf;
 381                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 382                 }
 383             }
 384             continue;
 385
 386         case BODY:
 387         case FILEEOF:
 388             break;
 389
 390         case LENERR:
 391         case FMTERR:
 392             if (state == LENERR || state == FMTERR) {
 393                 inform("format error in message %d (header #%d), continuing...",
 394                       msg, compnum);
 395                 check_failed = 1;
 396             }
 397             free(datecomp);
 398             free(subjcomp);
 399             fclose (in);
 400             return 0;
 401
 402         default:
 403             die("internal error -- you lose");
 404         }
 405         break;
 406     }
 407     m_getfld_state_destroy (&gstate);
 408
 409     /*
 410      * If no date component, then use the modification
 411      * time of the file as its date
 412      */
 413     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 414         struct stat st;
 415
 416         inform("can't parse %s field in message %d, "
 417             "will use file modification time", datesw, msg);
 418         fstat (fileno (in), &st);
 419         smsg->s_clock = st.st_mtime;
 420         check_failed = 1;
 421     } else {
 422         smsg->s_clock = dmktime (tw);
 423     }
 424
 425     if (subjsort) {
 426         if (subjcomp) {
 427             /*
 428              * try to make the subject "canonical": delete
 429              * leading "re:", everything but letters & smash
 430              * letters to lower case.
 431              */
 432             char  *cp, *cp2, c;
 433
 434             cp = subjcomp;
 435             cp2 = subjcomp;
 436             if (strcmp (subjsort, "subject") == 0) {
 437                 while ((c = *cp)) {
 438                     if (! isspace((unsigned char) c)) {
 439                         if(!uprf(cp, "re:"))
 440                             break;
 441                         cp += 2;
 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     free(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 ();
 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 temporary 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 }