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

   1
   2 /*
   3  * sortm.c -- sort messages in a folder by date/time
   4  *
   5  * $Id$
   6  *
   7  * This code is Copyright (c) 2002, by the authors of nmh.  See the
   8  * COPYRIGHT file in the root directory of the nmh distribution for
   9  * complete copyright information.
  10  */
  11
  12 #include <h/mh.h>
  13 #include <h/tws.h>
  14 #include <h/utils.h>
  15
  16 static struct swit switches[] = {
  17 #define DATESW                 0
  18      { "datefield field", 0 },
  19 #define TEXTSW                 1
  20      { "textfield field", 0 },
  21 #define NSUBJSW                2
  22      { "notextfield", 0 },
  23 #define SUBJSW                 3
  24      { "subject", -3 },            /* backward-compatibility */
  25 #define LIMSW                  4
  26      { "limit days", 0 },
  27 #define NLIMSW                 5
  28      { "nolimit", 0 },
  29 #define VERBSW                 6
  30      { "verbose", 0 },
  31 #define NVERBSW                7
  32      { "noverbose", 0 },
  33 #define VERSIONSW              8
  34      { "version", 0 },
  35 #define HELPSW                 9
  36      { "help", 0 },
  37      { NULL, 0 }
  38 };
  39
  40 struct smsg {
  41     int s_msg;
  42     time_t s_clock;
  43     char *s_subj;
  44 };
  45
  46 static struct smsg *smsgs;
  47 int nmsgs;
  48
  49 char *subjsort = (char *) 0;    /* sort on subject if != 0 */
  50 unsigned long datelimit = 0;
  51 int submajor = 0;               /* if true, sort on subject-major */
  52 int verbose;
  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
  80 #ifdef LOCALE
  81     setlocale(LC_ALL, "");
  82 #endif
  83     invo_name = r1bindex (argv[0], '/');
  84
  85     /* read user profile/context */
  86     context_read();
  87
  88     arguments = getarguments (invo_name, argc, argv, 1);
  89     argp = arguments;
  90
  91     /*
  92      * Parse arguments
  93      */
  94     while ((cp = *argp++)) {
  95         if (*cp == '-') {
  96             switch (smatch (++cp, switches)) {
  97             case AMBIGSW:
  98                 ambigsw (cp, switches);
  99                 done (1);
 100             case UNKWNSW:
 101                 adios (NULL, "-%s unknown", cp);
 102
 103             case HELPSW:
 104                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 105                         invo_name);
 106                 print_help (buf, switches, 1);
 107                 done (1);
 108             case VERSIONSW:
 109                 print_version(invo_name);
 110                 done (1);
 111
 112             case DATESW:
 113                 if (datesw)
 114                     adios (NULL, "only one date field at a time");
 115                 if (!(datesw = *argp++) || *datesw == '-')
 116                     adios (NULL, "missing argument to %s", argp[-2]);
 117                 continue;
 118
 119             case TEXTSW:
 120                 if (subjsort)
 121                     adios (NULL, "only one text field at a time");
 122                 if (!(subjsort = *argp++) || *subjsort == '-')
 123                     adios (NULL, "missing argument to %s", argp[-2]);
 124                 continue;
 125
 126             case SUBJSW:
 127                 subjsort = "subject";
 128                 continue;
 129             case NSUBJSW:
 130                 subjsort = (char *)0;
 131                 continue;
 132
 133             case LIMSW:
 134                 if (!(cp = *argp++) || *cp == '-')
 135                         adios (NULL, "missing argument to %s", argp[-2]);
 136                 while (*cp == '0')
 137                     cp++;               /* skip any leading zeros */
 138                 if (!*cp) {             /* hit end of string */
 139                     submajor++;         /* sort subject-major */
 140                     continue;
 141                 }
 142                 if (!isdigit(*cp) || !(datelimit = atoi(cp)))
 143                     adios (NULL, "impossible limit %s", cp);
 144                 datelimit *= 60*60*24;
 145                 continue;
 146             case NLIMSW:
 147                 submajor = 0;   /* use date-major, but */
 148                 datelimit = 0;  /* use no limit */
 149                 continue;
 150
 151             case VERBSW:
 152                 verbose++;
 153                 continue;
 154             case NVERBSW:
 155                 verbose = 0;
 156                 continue;
 157             }
 158         }
 159         if (*cp == '+' || *cp == '@') {
 160             if (folder)
 161                 adios (NULL, "only one folder at a time!");
 162             else
 163                 folder = pluspath (cp);
 164         } else
 165                 app_msgarg(&msgs, cp);
 166     }
 167
 168     if (!context_find ("path"))
 169         free (path ("./", TFOLDER));
 170     if (!msgs.size)
 171         app_msgarg(&msgs, "all");
 172     if (!datesw)
 173         datesw = "date";
 174     if (!folder)
 175         folder = getfolder (1);
 176     maildir = m_maildir (folder);
 177
 178     if (chdir (maildir) == NOTOK)
 179         adios (maildir, "unable to change directory to");
 180
 181     /* read folder and create message structure */
 182     if (!(mp = folder_read (folder)))
 183         adios (NULL, "unable to read folder %s", folder);
 184
 185     /* check for empty folder */
 186     if (mp->nummsg == 0)
 187         adios (NULL, "no messages in %s", folder);
 188
 189     /* parse all the message ranges/sequences and set SELECTED */
 190     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 191         if (!m_convert (mp, msgs.msgs[msgnum]))
 192             done (1);
 193     seq_setprev (mp);   /* set the previous sequence */
 194
 195     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 196         adios (NULL, "no messages to sort");
 197
 198     /*
 199      * sort a list of pointers to our "messages to be sorted".
 200      */
 201     dlist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 202     for (i = 0; i < nmsgs; i++)
 203         dlist[i] = &smsgs[i];
 204     dlist[nmsgs] = 0;
 205
 206     if (verbose) {      /* announce what we're doing */
 207         if (subjsort)
 208             printf ("sorting by %s-major %s-minor\n",
 209                 submajor ? subjsort : datesw,
 210                 submajor ? datesw : subjsort);
 211         else
 212             printf ("sorting by datefield %s\n", datesw);
 213     }
 214
 215     /* first sort by date, or by subject-major, date-minor */
 216     qsort ((char *) dlist, nmsgs, sizeof(*dlist),
 217             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 218
 219     /*
 220      * if we're sorting on subject, we need another list
 221      * in subject order, then a merge pass to collate the
 222      * two sorts.
 223      */
 224     if (!submajor && subjsort) {        /* already date sorted */
 225         struct smsg **slist, **flist;
 226         register struct smsg ***il, **fp, **dp;
 227
 228         slist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 229         memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 230         qsort((char *)slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 231
 232         /*
 233          * make an inversion list so we can quickly find
 234          * the collection of messages with the same subj
 235          * given a message number.
 236          */
 237         il = (struct smsg ***) calloc (mp->hghsel+1, sizeof(*il));
 238         if (! il)
 239             adios (NULL, "couldn't allocate msg list");
 240         for (i = 0; i < nmsgs; i++)
 241             il[slist[i]->s_msg] = &slist[i];
 242         /*
 243          * make up the final list, chronological but with
 244          * all the same subjects grouped together.
 245          */
 246         flist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 247         fp = flist;
 248         for (dp = dlist; *dp;) {
 249             register struct smsg **s = il[(*dp++)->s_msg];
 250
 251             /* see if we already did this guy */
 252             if (! s)
 253                 continue;
 254
 255             *fp++ = *s++;
 256             /*
 257              * take the next message(s) if there is one,
 258              * its subject isn't null and its subject
 259              * is the same as this one and it's not too
 260              * far away in time.
 261              */
 262             while (*s && (*s)->s_subj[0] &&
 263                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 264                    (datelimit == 0 ||
 265                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 266                 il[(*s)->s_msg] = 0;
 267                 *fp++ = *s++;
 268             }
 269         }
 270         *fp = 0;
 271         free (slist);
 272         free (dlist);
 273         dlist = flist;
 274     }
 275
 276     /*
 277      * At this point, dlist is a sorted array of pointers to smsg structures,
 278      * each of which contains a message number.
 279      */
 280
 281     rename_msgs (mp, dlist);
 282
 283     context_replace (pfolder, folder);  /* update current folder         */
 284     seq_save (mp);                      /* synchronize message sequences */
 285     context_save ();                    /* save the context file         */
 286     folder_free (mp);                   /* free folder/message structure */
 287     return done (0);
 288 }
 289
 290 static int
 291 read_hdrs (struct msgs *mp, char *datesw)
 292 {
 293     int msgnum;
 294     struct tws tb;
 295     register struct smsg *s;
 296
 297     twscopy (&tb, dlocaltimenow ());
 298
 299     smsgs = (struct smsg *)
 300         calloc ((size_t) (mp->hghsel - mp->lowsel + 2),
 301             sizeof(*smsgs));
 302     if (smsgs == NULL)
 303         adios (NULL, "unable to allocate sort storage");
 304
 305     s = smsgs;
 306     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 307         if (is_selected(mp, msgnum)) {
 308             if (get_fields (datesw, msgnum, s)) {
 309                 s->s_msg = msgnum;
 310                 s++;
 311             }
 312         }
 313     }
 314     s->s_msg = 0;
 315     return(s - smsgs);
 316 }
 317
 318
 319 /*
 320  * Parse the message and get the data or subject field,
 321  * if needed.
 322  */
 323
 324 static int
 325 get_fields (char *datesw, int msg, struct smsg *smsg)
 326 {
 327     register int state;
 328     int compnum;
 329     char *msgnam, buf[BUFSIZ], nam[NAMESZ];
 330     register struct tws *tw;
 331     register char *datecomp = NULL, *subjcomp = NULL;
 332     register FILE *in;
 333
 334     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 335         admonish (msgnam, "unable to read message");
 336         return (0);
 337     }
 338     for (compnum = 1, state = FLD;;) {
 339         switch (state = m_getfld (state, nam, buf, sizeof(buf), in)) {
 340         case FLD:
 341         case FLDEOF:
 342         case FLDPLUS:
 343             compnum++;
 344             if (!mh_strcasecmp (nam, datesw)) {
 345                 datecomp = add (buf, datecomp);
 346                 while (state == FLDPLUS) {
 347                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 348                     datecomp = add (buf, datecomp);
 349                 }
 350                 if (!subjsort || subjcomp)
 351                     break;
 352             } else if (subjsort && !mh_strcasecmp (nam, subjsort)) {
 353                 subjcomp = add (buf, subjcomp);
 354                 while (state == FLDPLUS) {
 355                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 356                     subjcomp = add (buf, subjcomp);
 357                 }
 358                 if (datecomp)
 359                     break;
 360             } else {
 361                 /* just flush this guy */
 362                 while (state == FLDPLUS)
 363                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 364             }
 365             continue;
 366
 367         case BODY:
 368         case BODYEOF:
 369         case FILEEOF:
 370             break;
 371
 372         case LENERR:
 373         case FMTERR:
 374             if (state == LENERR || state == FMTERR)
 375                 admonish (NULL, "format error in message %d (header #%d)",
 376                       msg, compnum);
 377             if (datecomp)
 378                 free (datecomp);
 379             if (subjcomp)
 380                 free (subjcomp);
 381             fclose (in);
 382             return (0);
 383
 384         default:
 385             adios (NULL, "internal error -- you lose");
 386         }
 387         break;
 388     }
 389
 390     /*
 391      * If no date component, then use the modification
 392      * time of the file as its date
 393      */
 394     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 395         struct stat st;
 396
 397         admonish (NULL, "can't parse %s field in message %d", datesw, msg);
 398         fstat (fileno (in), &st);
 399         smsg->s_clock = st.st_mtime;
 400     } else {
 401         smsg->s_clock = dmktime (tw);
 402     }
 403
 404     if (subjsort) {
 405         if (subjcomp) {
 406             /*
 407              * try to make the subject "canonical": delete
 408              * leading "re:", everything but letters & smash
 409              * letters to lower case.
 410              */
 411             register char  *cp, *cp2, c;
 412
 413             cp = subjcomp;
 414             cp2 = subjcomp;
 415             if (strcmp (subjsort, "subject") == 0) {
 416                 while ((c = *cp)) {
 417                     if (! isspace(c)) {
 418                         if(uprf(cp, "re:"))
 419                             cp += 2;
 420                         else
 421                             break;
 422                     }
 423                     cp++;
 424                 }
 425             }
 426
 427             while ((c = *cp++)) {
 428                 if (isalnum(c))
 429                     *cp2++ = isupper(c) ? tolower(c) : c;
 430             }
 431
 432             *cp2 = '\0';
 433         }
 434         else
 435             subjcomp = "";
 436
 437         smsg->s_subj = subjcomp;
 438     }
 439     fclose (in);
 440     if (datecomp)
 441         free (datecomp);
 442
 443     return (1);
 444 }
 445
 446 /*
 447  * sort on dates.
 448  */
 449 static int
 450 dsort (struct smsg **a, struct smsg **b)
 451 {
 452     if ((*a)->s_clock < (*b)->s_clock)
 453         return (-1);
 454     else if ((*a)->s_clock > (*b)->s_clock)
 455         return (1);
 456     else if ((*a)->s_msg < (*b)->s_msg)
 457         return (-1);
 458     else
 459         return (1);
 460 }
 461
 462 /*
 463  * sort on subjects.
 464  */
 465 static int
 466 subsort (struct smsg **a, struct smsg **b)
 467 {
 468     register int i;
 469
 470     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 471         return (i);
 472
 473     return (dsort (a, b));
 474 }
 475
 476 static int
 477 txtsort (struct smsg **a, struct smsg **b)
 478 {
 479     register int i;
 480
 481     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 482         return (i);
 483     else if ((*a)->s_msg < (*b)->s_msg)
 484         return (-1);
 485     else
 486         return (1);
 487 }
 488
 489 static void
 490 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 491 {
 492     int nxt, old, new;
 493     char *newname, oldname[BUFSIZ];
 494     char newbuf[MAXPATHLEN + 1];
 495
 496     for (;;) {
 497         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 498         mlist[msg] = (struct smsg *)0;
 499         old = smsgs[nxt].s_msg;
 500         new = smsgs[msg].s_msg;
 501         strncpy (oldname, m_name (old), sizeof(oldname));
 502         newname = m_name (new);
 503         if (verbose)
 504             printf ("message %d becomes message %d\n", old, new);
 505
 506         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 507         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 508         ext_hook("ref-hook", oldname, newbuf);
 509
 510         if (rename (oldname, newname) == NOTOK)
 511             adios (newname, "unable to rename %s to", oldname);
 512
 513         copy_msg_flags (mp, new, old);
 514         if (mp->curmsg == old)
 515             seq_setcur (mp, new);
 516
 517         if (nxt == endmsg)
 518             break;
 519
 520         msg = nxt;
 521     }
 522 /*      if (nxt != endmsg); */
 523 /*      rename_chain (mp, mlist, nxt, endmsg); */
 524 }
 525
 526 static void
 527 rename_msgs (struct msgs *mp, struct smsg **mlist)
 528 {
 529     int i, j, old, new;
 530     seqset_t tmpset;
 531     char f1[BUFSIZ], tmpfil[BUFSIZ];
 532     char newbuf[MAXPATHLEN + 1];
 533     struct smsg *sp;
 534
 535     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 536
 537     for (i = 0; i < nmsgs; i++) {
 538         if (! (sp = mlist[i]))
 539             continue;   /* did this one */
 540
 541         j = sp - smsgs;
 542         if (j == i)
 543             continue;   /* this one doesn't move */
 544
 545         /*
 546          * the guy that was msg j is about to become msg i.
 547          * rename 'j' to make a hole, then recursively rename
 548          * guys to fill up the hole.
 549          */
 550         old = smsgs[j].s_msg;
 551         new = smsgs[i].s_msg;
 552         strncpy (f1, m_name (old), sizeof(f1));
 553
 554         if (verbose)
 555             printf ("renaming message chain from %d to %d\n", old, new);
 556
 557         /*
 558          *      Run the external hook to refile the old message as the
 559          *      temporary message number that is off of the end of the
 560          *      messages in the folder.
 561          */
 562
 563         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 564         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 565         ext_hook("ref-hook", f1, newbuf);
 566
 567         if (rename (f1, tmpfil) == NOTOK)
 568             adios (tmpfil, "unable to rename %s to ", f1);
 569
 570         get_msg_flags (mp, &tmpset, old);
 571
 572         rename_chain (mp, mlist, j, i);
 573
 574         /*
 575          *      Run the external hook to refile the temorary message number
 576          *      to the real place.
 577          */
 578
 579         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 580         ext_hook("ref-hook", newbuf, f1);
 581
 582         if (rename (tmpfil, m_name(new)) == NOTOK)
 583             adios (m_name(new), "unable to rename %s to", tmpfil);
 584
 585         set_msg_flags (mp, &tmpset, new);
 586         mp->msgflags |= SEQMOD;
 587     }
 588 }