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

   1
   2 /*
   3  * sortm.c -- sort messages in a folder by date/time
   4  *
   5  * This code is Copyright (c) 2002, by the authors of nmh.  See the
   6  * COPYRIGHT file in the root directory of the nmh distribution for
   7  * complete copyright information.
   8  */
   9
  10 #include <h/mh.h>
  11 #include <h/tws.h>
  12 #include <h/utils.h>
  13
  14 #define SORTM_SWITCHES \
  15     X("datefield field", 0, DATESW) \
  16     X("textfield field", 0, TEXTSW) \
  17     X("notextfield", 0, NSUBJSW) \
  18     X("subject", -3, SUBJSW) /* backward-compatibility */ \
  19     X("limit days", 0, LIMSW) \
  20     X("nolimit", 0, NLIMSW) \
  21     X("verbose", 0, VERBSW) \
  22     X("noverbose", 0, NVERBSW) \
  23     X("all", 0, ALLMSGS) \
  24     X("noall", 0, NALLMSGS) \
  25     X("check", 0, CHECKSW) \
  26     X("nocheck", 0, NCHECKSW) \
  27     X("version", 0, VERSIONSW) \
  28     X("help", 0, HELPSW) \
  29
  30 #define X(sw, minchars, id) id,
  31 DEFINE_SWITCH_ENUM(SORTM);
  32 #undef X
  33
  34 #define X(sw, minchars, id) { sw, minchars, id },
  35 DEFINE_SWITCH_ARRAY(SORTM, switches);
  36 #undef X
  37
  38 struct smsg {
  39     int s_msg;
  40     time_t s_clock;
  41     char *s_subj;
  42 };
  43
  44 static struct smsg *smsgs;
  45 int nmsgs;
  46
  47 char *subjsort = (char *) 0;    /* sort on subject if != 0 */
  48 time_t datelimit = 0;
  49 int submajor = 0;               /* if true, sort on subject-major */
  50 int verbose;
  51 int allmsgs = 1;
  52 int check_failed = 0;
  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     char *cp, *maildir, *datesw = NULL;
  74     char *folder = NULL, buf[BUFSIZ], **argp;
  75     char **arguments;
  76     struct msgs_array msgs = { 0, 0, NULL };
  77     struct msgs *mp;
  78     struct smsg **dlist;
  79     int checksw = 0;
  80
  81     if (nmh_init(argv[0], 1)) { return 1; }
  82
  83     arguments = getarguments (invo_name, argc, argv, 1);
  84     argp = arguments;
  85
  86     /*
  87      * Parse arguments
  88      */
  89     while ((cp = *argp++)) {
  90         if (*cp == '-') {
  91             switch (smatch (++cp, switches)) {
  92             case AMBIGSW:
  93                 ambigsw (cp, switches);
  94                 done (1);
  95             case UNKWNSW:
  96                 adios (NULL, "-%s unknown", cp);
  97
  98             case HELPSW:
  99                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 100                         invo_name);
 101                 print_help (buf, switches, 1);
 102                 done (0);
 103             case VERSIONSW:
 104                 print_version(invo_name);
 105                 done (0);
 106
 107             case DATESW:
 108                 if (datesw)
 109                     adios (NULL, "only one date field at a time");
 110                 if (!(datesw = *argp++) || *datesw == '-')
 111                     adios (NULL, "missing argument to %s", argp[-2]);
 112                 continue;
 113
 114             case TEXTSW:
 115                 if (subjsort)
 116                     adios (NULL, "only one text field at a time");
 117                 if (!(subjsort = *argp++) || *subjsort == '-')
 118                     adios (NULL, "missing argument to %s", argp[-2]);
 119                 continue;
 120
 121             case SUBJSW:
 122                 subjsort = "subject";
 123                 continue;
 124             case NSUBJSW:
 125                 subjsort = NULL;
 126                 continue;
 127
 128             case LIMSW:
 129                 if (!(cp = *argp++) || *cp == '-')
 130                         adios (NULL, "missing argument to %s", argp[-2]);
 131                 while (*cp == '0')
 132                     cp++;               /* skip any leading zeros */
 133                 if (!*cp) {             /* hit end of string */
 134                     submajor++;         /* sort subject-major */
 135                     continue;
 136                 }
 137                 if (!isdigit((unsigned char) *cp) || !(datelimit = atoi(cp)))
 138                     adios (NULL, "impossible limit %s", cp);
 139                 datelimit *= 60*60*24;
 140                 continue;
 141             case NLIMSW:
 142                 submajor = 0;   /* use date-major, but */
 143                 datelimit = 0;  /* use no limit */
 144                 continue;
 145
 146             case VERBSW:
 147                 verbose++;
 148                 continue;
 149             case NVERBSW:
 150                 verbose = 0;
 151                 continue;
 152
 153             case ALLMSGS:
 154                 allmsgs = 1;
 155                 continue;
 156             case NALLMSGS:
 157                 allmsgs = 0;
 158                 continue;
 159
 160             case CHECKSW:
 161                 checksw = 1;
 162                 continue;
 163             case NCHECKSW:
 164                 checksw = 0;
 165                 continue;
 166             }
 167         }
 168         if (*cp == '+' || *cp == '@') {
 169             if (folder)
 170                 adios (NULL, "only one folder at a time!");
 171             else
 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             adios (NULL, "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         adios (NULL, "unable to read folder %s", folder);
 198
 199     /* check for empty folder */
 200     if (mp->nummsg == 0)
 201         adios (NULL, "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         adios (NULL, "no messages to sort");
 211
 212     if (checksw  &&  check_failed) {
 213         adios (NULL, "errors found, no messages sorted");
 214     }
 215
 216     /*
 217      * sort a list of pointers to our "messages to be sorted".
 218      */
 219     dlist = (struct smsg **) 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 ((char *) 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 = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 248         memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 249         qsort((char *)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 = (struct smsg **) 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 tws tb;
 314     struct smsg *s;
 315
 316     twscopy (&tb, dlocaltimenow ());
 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[BUFSIZ], nam[NAMESZ];
 344     struct tws *tw;
 345     char *datecomp = NULL, *subjcomp = NULL;
 346     FILE *in;
 347     m_getfld_state_t gstate = 0;
 348
 349     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 350         admonish (msgnam, "unable to read message");
 351         return (0);
 352     }
 353     for (compnum = 1;;) {
 354         int bufsz = sizeof buf;
 355         switch (state = m_getfld (&gstate, nam, buf, &bufsz, in)) {
 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_getfld (&gstate, nam, buf, &bufsz, in);
 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_getfld (&gstate, nam, buf, &bufsz, in);
 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_getfld (&gstate, nam, buf, &bufsz, in);
 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             mh_xfree(datecomp);
 398             mh_xfree(subjcomp);
 399             fclose (in);
 400             return (0);
 401
 402         default:
 403             adios (NULL, "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                             cp += 2;
 441                         else
 442                             break;
 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     mh_xfree(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 (0);
 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 temorary 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 }