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