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 static struct swit switches[] = {
  15 #define DATESW                 0
  16      { "datefield field", 0 },
  17 #define TEXTSW                 1
  18      { "textfield field", 0 },
  19 #define NSUBJSW                2
  20      { "notextfield", 0 },
  21 #define SUBJSW                 3
  22      { "subject", -3 },            /* backward-compatibility */
  23 #define LIMSW                  4
  24      { "limit days", 0 },
  25 #define NLIMSW                 5
  26      { "nolimit", 0 },
  27 #define VERBSW                 6
  28      { "verbose", 0 },
  29 #define NVERBSW                7
  30      { "noverbose", 0 },
  31 #define VERSIONSW              8
  32      { "version", 0 },
  33 #define HELPSW                 9
  34      { "help", 0 },
  35      { NULL, 0 }
  36 };
  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
  52 /* This keeps compiler happy on calls to qsort */
  53 typedef int (*qsort_comp) (const void *, const void *);
  54
  55 /*
  56  * static prototypes
  57  */
  58 static int read_hdrs (struct msgs *, char *);
  59 static int get_fields (char *, int, struct smsg *);
  60 static int dsort (struct smsg **, struct smsg **);
  61 static int subsort (struct smsg **, struct smsg **);
  62 static int txtsort (struct smsg **, struct smsg **);
  63 static void rename_chain (struct msgs *, struct smsg **, int, int);
  64 static void rename_msgs (struct msgs *, struct smsg **);
  65
  66
  67 int
  68 main (int argc, char **argv)
  69 {
  70     int i, msgnum;
  71     unsigned char *cp;
  72     char *maildir, *datesw = NULL;
  73     char *folder = NULL, buf[BUFSIZ], **argp;
  74     char **arguments;
  75     struct msgs_array msgs = { 0, 0, NULL };
  76     struct msgs *mp;
  77     struct smsg **dlist;
  78
  79 #ifdef LOCALE
  80     setlocale(LC_ALL, "");
  81 #endif
  82     invo_name = r1bindex (argv[0], '/');
  83
  84     /* read user profile/context */
  85     context_read();
  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                 adios (NULL, "-%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 (1);
 107             case VERSIONSW:
 108                 print_version(invo_name);
 109                 done (1);
 110
 111             case DATESW:
 112                 if (datesw)
 113                     adios (NULL, "only one date field at a time");
 114                 if (!(datesw = *argp++) || *datesw == '-')
 115                     adios (NULL, "missing argument to %s", argp[-2]);
 116                 continue;
 117
 118             case TEXTSW:
 119                 if (subjsort)
 120                     adios (NULL, "only one text field at a time");
 121                 if (!(subjsort = *argp++) || *subjsort == '-')
 122                     adios (NULL, "missing argument to %s", argp[-2]);
 123                 continue;
 124
 125             case SUBJSW:
 126                 subjsort = "subject";
 127                 continue;
 128             case NSUBJSW:
 129                 subjsort = (char *)0;
 130                 continue;
 131
 132             case LIMSW:
 133                 if (!(cp = *argp++) || *cp == '-')
 134                         adios (NULL, "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++;         /* sort subject-major */
 139                     continue;
 140                 }
 141                 if (!isdigit(*cp) || !(datelimit = atoi(cp)))
 142                     adios (NULL, "impossible limit %s", cp);
 143                 datelimit *= 60*60*24;
 144                 continue;
 145             case NLIMSW:
 146                 submajor = 0;   /* use date-major, but */
 147                 datelimit = 0;  /* use no limit */
 148                 continue;
 149
 150             case VERBSW:
 151                 verbose++;
 152                 continue;
 153             case NVERBSW:
 154                 verbose = 0;
 155                 continue;
 156             }
 157         }
 158         if (*cp == '+' || *cp == '@') {
 159             if (folder)
 160                 adios (NULL, "only one folder at a time!");
 161             else
 162                 folder = pluspath (cp);
 163         } else
 164                 app_msgarg(&msgs, cp);
 165     }
 166
 167     if (!context_find ("path"))
 168         free (path ("./", TFOLDER));
 169     if (!msgs.size)
 170         app_msgarg(&msgs, "all");
 171     if (!datesw)
 172         datesw = "date";
 173     if (!folder)
 174         folder = getfolder (1);
 175     maildir = m_maildir (folder);
 176
 177     if (chdir (maildir) == NOTOK)
 178         adios (maildir, "unable to change directory to");
 179
 180     /* read folder and create message structure */
 181     if (!(mp = folder_read (folder)))
 182         adios (NULL, "unable to read folder %s", folder);
 183
 184     /* check for empty folder */
 185     if (mp->nummsg == 0)
 186         adios (NULL, "no messages in %s", folder);
 187
 188     /* parse all the message ranges/sequences and set SELECTED */
 189     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 190         if (!m_convert (mp, msgs.msgs[msgnum]))
 191             done (1);
 192     seq_setprev (mp);   /* set the previous sequence */
 193
 194     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 195         adios (NULL, "no messages to sort");
 196
 197     /*
 198      * sort a list of pointers to our "messages to be sorted".
 199      */
 200     dlist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 201     for (i = 0; i < nmsgs; i++)
 202         dlist[i] = &smsgs[i];
 203     dlist[nmsgs] = 0;
 204
 205     if (verbose) {      /* announce what we're doing */
 206         if (subjsort)
 207             printf ("sorting by %s-major %s-minor\n",
 208                 submajor ? subjsort : datesw,
 209                 submajor ? datesw : subjsort);
 210         else
 211             printf ("sorting by datefield %s\n", datesw);
 212     }
 213
 214     /* first sort by date, or by subject-major, date-minor */
 215     qsort ((char *) dlist, nmsgs, sizeof(*dlist),
 216             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 217
 218     /*
 219      * if we're sorting on subject, we need another list
 220      * in subject order, then a merge pass to collate the
 221      * two sorts.
 222      */
 223     if (!submajor && subjsort) {        /* already date sorted */
 224         struct smsg **slist, **flist;
 225         register struct smsg ***il, **fp, **dp;
 226
 227         slist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 228         memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 229         qsort((char *)slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 230
 231         /*
 232          * make an inversion list so we can quickly find
 233          * the collection of messages with the same subj
 234          * given a message number.
 235          */
 236         il = (struct smsg ***) calloc (mp->hghsel+1, sizeof(*il));
 237         if (! il)
 238             adios (NULL, "couldn't allocate msg list");
 239         for (i = 0; i < nmsgs; i++)
 240             il[slist[i]->s_msg] = &slist[i];
 241         /*
 242          * make up the final list, chronological but with
 243          * all the same subjects grouped together.
 244          */
 245         flist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 246         fp = flist;
 247         for (dp = dlist; *dp;) {
 248             register struct smsg **s = il[(*dp++)->s_msg];
 249
 250             /* see if we already did this guy */
 251             if (! s)
 252                 continue;
 253
 254             *fp++ = *s++;
 255             /*
 256              * take the next message(s) if there is one,
 257              * its subject isn't null and its subject
 258              * is the same as this one and it's not too
 259              * far away in time.
 260              */
 261             while (*s && (*s)->s_subj[0] &&
 262                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 263                    (datelimit == 0 ||
 264                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 265                 il[(*s)->s_msg] = 0;
 266                 *fp++ = *s++;
 267             }
 268         }
 269         *fp = 0;
 270         free (slist);
 271         free (dlist);
 272         dlist = flist;
 273     }
 274
 275     /*
 276      * At this point, dlist is a sorted array of pointers to smsg structures,
 277      * each of which contains a message number.
 278      */
 279
 280     rename_msgs (mp, dlist);
 281
 282     context_replace (pfolder, folder);  /* update current folder         */
 283     seq_save (mp);                      /* synchronize message sequences */
 284     context_save ();                    /* save the context file         */
 285     folder_free (mp);                   /* free folder/message structure */
 286     done (0);
 287     return 1;
 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;
 412             register unsigned char c;
 413
 414             cp = subjcomp;
 415             cp2 = subjcomp;
 416             if (strcmp (subjsort, "subject") == 0) {
 417                 while ((c = *cp)) {
 418                     if (! isspace(c)) {
 419                         if(uprf(cp, "re:"))
 420                             cp += 2;
 421                         else
 422                             break;
 423                     }
 424                     cp++;
 425                 }
 426             }
 427
 428             while ((c = *cp++)) {
 429                 if (isalnum(c))
 430                     *cp2++ = isupper(c) ? tolower(c) : c;
 431             }
 432
 433             *cp2 = '\0';
 434         }
 435         else
 436             subjcomp = "";
 437
 438         smsg->s_subj = subjcomp;
 439     }
 440     fclose (in);
 441     if (datecomp)
 442         free (datecomp);
 443
 444     return (1);
 445 }
 446
 447 /*
 448  * sort on dates.
 449  */
 450 static int
 451 dsort (struct smsg **a, struct smsg **b)
 452 {
 453     if ((*a)->s_clock < (*b)->s_clock)
 454         return (-1);
 455     else if ((*a)->s_clock > (*b)->s_clock)
 456         return (1);
 457     else if ((*a)->s_msg < (*b)->s_msg)
 458         return (-1);
 459     else
 460         return (1);
 461 }
 462
 463 /*
 464  * sort on subjects.
 465  */
 466 static int
 467 subsort (struct smsg **a, struct smsg **b)
 468 {
 469     register int i;
 470
 471     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 472         return (i);
 473
 474     return (dsort (a, b));
 475 }
 476
 477 static int
 478 txtsort (struct smsg **a, struct smsg **b)
 479 {
 480     register int i;
 481
 482     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 483         return (i);
 484     else if ((*a)->s_msg < (*b)->s_msg)
 485         return (-1);
 486     else
 487         return (1);
 488 }
 489
 490 static void
 491 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 492 {
 493     int nxt, old, new;
 494     char *newname, oldname[BUFSIZ];
 495     char newbuf[MAXPATHLEN + 1];
 496
 497     for (;;) {
 498         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 499         mlist[msg] = (struct smsg *)0;
 500         old = smsgs[nxt].s_msg;
 501         new = smsgs[msg].s_msg;
 502         strncpy (oldname, m_name (old), sizeof(oldname));
 503         newname = m_name (new);
 504         if (verbose)
 505             printf ("message %d becomes message %d\n", old, new);
 506
 507         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 508         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 509         ext_hook("ref-hook", oldname, newbuf);
 510
 511         if (rename (oldname, newname) == NOTOK)
 512             adios (newname, "unable to rename %s to", oldname);
 513
 514         copy_msg_flags (mp, new, old);
 515         if (mp->curmsg == old)
 516             seq_setcur (mp, new);
 517
 518         if (nxt == endmsg)
 519             break;
 520
 521         msg = nxt;
 522     }
 523 /*      if (nxt != endmsg); */
 524 /*      rename_chain (mp, mlist, nxt, endmsg); */
 525 }
 526
 527 static void
 528 rename_msgs (struct msgs *mp, struct smsg **mlist)
 529 {
 530     int i, j, old, new;
 531     seqset_t tmpset;
 532     char f1[BUFSIZ], tmpfil[BUFSIZ];
 533     char newbuf[MAXPATHLEN + 1];
 534     struct smsg *sp;
 535
 536     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 537
 538     for (i = 0; i < nmsgs; i++) {
 539         if (! (sp = mlist[i]))
 540             continue;   /* did this one */
 541
 542         j = sp - smsgs;
 543         if (j == i)
 544             continue;   /* this one doesn't move */
 545
 546         /*
 547          * the guy that was msg j is about to become msg i.
 548          * rename 'j' to make a hole, then recursively rename
 549          * guys to fill up the hole.
 550          */
 551         old = smsgs[j].s_msg;
 552         new = smsgs[i].s_msg;
 553         strncpy (f1, m_name (old), sizeof(f1));
 554
 555         if (verbose)
 556             printf ("renaming message chain from %d to %d\n", old, new);
 557
 558         /*
 559          *      Run the external hook to refile the old message as the
 560          *      temporary message number that is off of the end of the
 561          *      messages in the folder.
 562          */
 563
 564         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 565         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 566         ext_hook("ref-hook", f1, newbuf);
 567
 568         if (rename (f1, tmpfil) == NOTOK)
 569             adios (tmpfil, "unable to rename %s to ", f1);
 570
 571         get_msg_flags (mp, &tmpset, old);
 572
 573         rename_chain (mp, mlist, j, i);
 574
 575         /*
 576          *      Run the external hook to refile the temorary message number
 577          *      to the real place.
 578          */
 579
 580         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 581         ext_hook("ref-hook", newbuf, f1);
 582
 583         if (rename (tmpfil, m_name(new)) == NOTOK)
 584             adios (m_name(new), "unable to rename %s to", tmpfil);
 585
 586         set_msg_flags (mp, &tmpset, new);
 587         mp->msgflags |= SEQMOD;
 588     }
 589 }