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 /*
  17  * We allocate space for messages (msgs array)
  18  * this number of elements at a time.
  19  */
  20 #define MAXMSGS  256
  21
  22
  23 static struct swit switches[] = {
  24 #define DATESW                 0
  25      { "datefield field", 0 },
  26 #define TEXTSW                 1
  27      { "textfield field", 0 },
  28 #define NSUBJSW                2
  29      { "notextfield", 0 },
  30 #define SUBJSW                 3
  31      { "subject", -3 },            /* backward-compatibility */
  32 #define LIMSW                  4
  33      { "limit days", 0 },
  34 #define NLIMSW                 5
  35      { "nolimit", 0 },
  36 #define VERBSW                 6
  37      { "verbose", 0 },
  38 #define NVERBSW                7
  39      { "noverbose", 0 },
  40 #define VERSIONSW              8
  41      { "version", 0 },
  42 #define HELPSW                 9
  43      { "help", 0 },
  44      { NULL, 0 }
  45 };
  46
  47 struct smsg {
  48     int s_msg;
  49     time_t s_clock;
  50     char *s_subj;
  51 };
  52
  53 static struct smsg *smsgs;
  54 int nmsgs;
  55
  56 char *subjsort = (char *) 0;    /* sort on subject if != 0 */
  57 unsigned long datelimit = 0;
  58 int submajor = 0;               /* if true, sort on subject-major */
  59 int verbose;
  60
  61 /* This keeps compiler happy on calls to qsort */
  62 typedef int (*qsort_comp) (const void *, const void *);
  63
  64 /*
  65  * static prototypes
  66  */
  67 static int read_hdrs (struct msgs *, char *);
  68 static int get_fields (char *, int, struct smsg *);
  69 static int dsort (struct smsg **, struct smsg **);
  70 static int subsort (struct smsg **, struct smsg **);
  71 static int txtsort (struct smsg **, struct smsg **);
  72 static void rename_chain (struct msgs *, struct smsg **, int, int);
  73 static void rename_msgs (struct msgs *, struct smsg **);
  74
  75
  76 int
  77 main (int argc, char **argv)
  78 {
  79     int nummsgs, maxmsgs, i, msgnum;
  80     char *cp, *maildir, *datesw = NULL;
  81     char *folder = NULL, buf[BUFSIZ], **argp;
  82     char **arguments, **msgs;
  83     struct msgs *mp;
  84     struct smsg **dlist;
  85
  86 #ifdef LOCALE
  87     setlocale(LC_ALL, "");
  88 #endif
  89     invo_name = r1bindex (argv[0], '/');
  90
  91     /* read user profile/context */
  92     context_read();
  93
  94     arguments = getarguments (invo_name, argc, argv, 1);
  95     argp = arguments;
  96
  97     /*
  98      * Allocate the initial space to record message
  99      * names and ranges.
 100      */
 101     nummsgs = 0;
 102     maxmsgs = MAXMSGS;
 103     msgs = (char **) mh_xmalloc ((size_t) (maxmsgs * sizeof(*msgs)));
 104
 105     /*
 106      * Parse arguments
 107      */
 108     while ((cp = *argp++)) {
 109         if (*cp == '-') {
 110             switch (smatch (++cp, switches)) {
 111             case AMBIGSW:
 112                 ambigsw (cp, switches);
 113                 done (1);
 114             case UNKWNSW:
 115                 adios (NULL, "-%s unknown", cp);
 116
 117             case HELPSW:
 118                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 119                         invo_name);
 120                 print_help (buf, switches, 1);
 121                 done (1);
 122             case VERSIONSW:
 123                 print_version(invo_name);
 124                 done (1);
 125
 126             case DATESW:
 127                 if (datesw)
 128                     adios (NULL, "only one date field at a time");
 129                 if (!(datesw = *argp++) || *datesw == '-')
 130                     adios (NULL, "missing argument to %s", argp[-2]);
 131                 continue;
 132
 133             case TEXTSW:
 134                 if (subjsort)
 135                     adios (NULL, "only one text field at a time");
 136                 if (!(subjsort = *argp++) || *subjsort == '-')
 137                     adios (NULL, "missing argument to %s", argp[-2]);
 138                 continue;
 139
 140             case SUBJSW:
 141                 subjsort = "subject";
 142                 continue;
 143             case NSUBJSW:
 144                 subjsort = (char *)0;
 145                 continue;
 146
 147             case LIMSW:
 148                 if (!(cp = *argp++) || *cp == '-')
 149                         adios (NULL, "missing argument to %s", argp[-2]);
 150                 while (*cp == '0')
 151                     cp++;               /* skip any leading zeros */
 152                 if (!*cp) {             /* hit end of string */
 153                     submajor++;         /* sort subject-major */
 154                     continue;
 155                 }
 156                 if (!isdigit(*cp) || !(datelimit = atoi(cp)))
 157                     adios (NULL, "impossible limit %s", cp);
 158                 datelimit *= 60*60*24;
 159                 continue;
 160             case NLIMSW:
 161                 submajor = 0;   /* use date-major, but */
 162                 datelimit = 0;  /* use no limit */
 163                 continue;
 164
 165             case VERBSW:
 166                 verbose++;
 167                 continue;
 168             case NVERBSW:
 169                 verbose = 0;
 170                 continue;
 171             }
 172         }
 173         if (*cp == '+' || *cp == '@') {
 174             if (folder)
 175                 adios (NULL, "only one folder at a time!");
 176             else
 177                 folder = path (cp + 1, *cp == '+' ? TFOLDER : TSUBCWF);
 178         } else {
 179             /*
 180              * Check if we need to allocate more space
 181              * for message names/ranges.
 182              */
 183             if (nummsgs >= maxmsgs) {
 184                 maxmsgs += MAXMSGS;
 185                 msgs = (char **) mh_xrealloc (msgs,
 186                     (size_t) (maxmsgs * sizeof(*msgs)));
 187             }
 188             msgs[nummsgs++] = cp;
 189         }
 190     }
 191
 192     if (!context_find ("path"))
 193         free (path ("./", TFOLDER));
 194     if (!nummsgs)
 195         msgs[nummsgs++] = "all";
 196     if (!datesw)
 197         datesw = "date";
 198     if (!folder)
 199         folder = getfolder (1);
 200     maildir = m_maildir (folder);
 201
 202     if (chdir (maildir) == NOTOK)
 203         adios (maildir, "unable to change directory to");
 204
 205     /* read folder and create message structure */
 206     if (!(mp = folder_read (folder)))
 207         adios (NULL, "unable to read folder %s", folder);
 208
 209     /* check for empty folder */
 210     if (mp->nummsg == 0)
 211         adios (NULL, "no messages in %s", folder);
 212
 213     /* parse all the message ranges/sequences and set SELECTED */
 214     for (msgnum = 0; msgnum < nummsgs; msgnum++)
 215         if (!m_convert (mp, msgs[msgnum]))
 216             done (1);
 217     seq_setprev (mp);   /* set the previous sequence */
 218
 219     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 220         adios (NULL, "no messages to sort");
 221
 222     /*
 223      * sort a list of pointers to our "messages to be sorted".
 224      */
 225     dlist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 226     for (i = 0; i < nmsgs; i++)
 227         dlist[i] = &smsgs[i];
 228     dlist[nmsgs] = 0;
 229
 230     if (verbose) {      /* announce what we're doing */
 231         if (subjsort)
 232             printf ("sorting by %s-major %s-minor\n",
 233                 submajor ? subjsort : datesw,
 234                 submajor ? datesw : subjsort);
 235         else
 236             printf ("sorting by datefield %s\n", datesw);
 237     }
 238
 239     /* first sort by date, or by subject-major, date-minor */
 240     qsort ((char *) dlist, nmsgs, sizeof(*dlist),
 241             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 242
 243     /*
 244      * if we're sorting on subject, we need another list
 245      * in subject order, then a merge pass to collate the
 246      * two sorts.
 247      */
 248     if (!submajor && subjsort) {        /* already date sorted */
 249         struct smsg **slist, **flist;
 250         register struct smsg ***il, **fp, **dp;
 251
 252         slist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 253         memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 254         qsort((char *)slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 255
 256         /*
 257          * make an inversion list so we can quickly find
 258          * the collection of messages with the same subj
 259          * given a message number.
 260          */
 261         il = (struct smsg ***) calloc (mp->hghsel+1, sizeof(*il));
 262         if (! il)
 263             adios (NULL, "couldn't allocate msg list");
 264         for (i = 0; i < nmsgs; i++)
 265             il[slist[i]->s_msg] = &slist[i];
 266         /*
 267          * make up the final list, chronological but with
 268          * all the same subjects grouped together.
 269          */
 270         flist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 271         fp = flist;
 272         for (dp = dlist; *dp;) {
 273             register struct smsg **s = il[(*dp++)->s_msg];
 274
 275             /* see if we already did this guy */
 276             if (! s)
 277                 continue;
 278
 279             *fp++ = *s++;
 280             /*
 281              * take the next message(s) if there is one,
 282              * its subject isn't null and its subject
 283              * is the same as this one and it's not too
 284              * far away in time.
 285              */
 286             while (*s && (*s)->s_subj[0] &&
 287                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 288                    (datelimit == 0 ||
 289                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 290                 il[(*s)->s_msg] = 0;
 291                 *fp++ = *s++;
 292             }
 293         }
 294         *fp = 0;
 295         free (slist);
 296         free (dlist);
 297         dlist = flist;
 298     }
 299
 300     /*
 301      * At this point, dlist is a sorted array of pointers to smsg structures,
 302      * each of which contains a message number.
 303      */
 304
 305     rename_msgs (mp, dlist);
 306
 307     context_replace (pfolder, folder);  /* update current folder         */
 308     seq_save (mp);                      /* synchronize message sequences */
 309     context_save ();                    /* save the context file         */
 310     folder_free (mp);                   /* free folder/message structure */
 311     return done (0);
 312 }
 313
 314 static int
 315 read_hdrs (struct msgs *mp, char *datesw)
 316 {
 317     int msgnum;
 318     struct tws tb;
 319     register struct smsg *s;
 320
 321     twscopy (&tb, dlocaltimenow ());
 322
 323     smsgs = (struct smsg *)
 324         calloc ((size_t) (mp->hghsel - mp->lowsel + 2),
 325             sizeof(*smsgs));
 326     if (smsgs == NULL)
 327         adios (NULL, "unable to allocate sort storage");
 328
 329     s = smsgs;
 330     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 331         if (is_selected(mp, msgnum)) {
 332             if (get_fields (datesw, msgnum, s)) {
 333                 s->s_msg = msgnum;
 334                 s++;
 335             }
 336         }
 337     }
 338     s->s_msg = 0;
 339     return(s - smsgs);
 340 }
 341
 342
 343 /*
 344  * Parse the message and get the data or subject field,
 345  * if needed.
 346  */
 347
 348 static int
 349 get_fields (char *datesw, int msg, struct smsg *smsg)
 350 {
 351     register int state;
 352     int compnum;
 353     char *msgnam, buf[BUFSIZ], nam[NAMESZ];
 354     register struct tws *tw;
 355     register char *datecomp = NULL, *subjcomp = NULL;
 356     register FILE *in;
 357
 358     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 359         admonish (msgnam, "unable to read message");
 360         return (0);
 361     }
 362     for (compnum = 1, state = FLD;;) {
 363         switch (state = m_getfld (state, nam, buf, sizeof(buf), in)) {
 364         case FLD:
 365         case FLDEOF:
 366         case FLDPLUS:
 367             compnum++;
 368             if (!mh_strcasecmp (nam, datesw)) {
 369                 datecomp = add (buf, datecomp);
 370                 while (state == FLDPLUS) {
 371                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 372                     datecomp = add (buf, datecomp);
 373                 }
 374                 if (!subjsort || subjcomp)
 375                     break;
 376             } else if (subjsort && !mh_strcasecmp (nam, subjsort)) {
 377                 subjcomp = add (buf, subjcomp);
 378                 while (state == FLDPLUS) {
 379                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 380                     subjcomp = add (buf, subjcomp);
 381                 }
 382                 if (datecomp)
 383                     break;
 384             } else {
 385                 /* just flush this guy */
 386                 while (state == FLDPLUS)
 387                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 388             }
 389             continue;
 390
 391         case BODY:
 392         case BODYEOF:
 393         case FILEEOF:
 394             break;
 395
 396         case LENERR:
 397         case FMTERR:
 398             if (state == LENERR || state == FMTERR)
 399                 admonish (NULL, "format error in message %d (header #%d)",
 400                       msg, compnum);
 401             if (datecomp)
 402                 free (datecomp);
 403             if (subjcomp)
 404                 free (subjcomp);
 405             fclose (in);
 406             return (0);
 407
 408         default:
 409             adios (NULL, "internal error -- you lose");
 410         }
 411         break;
 412     }
 413
 414     /*
 415      * If no date component, then use the modification
 416      * time of the file as its date
 417      */
 418     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 419         struct stat st;
 420
 421         admonish (NULL, "can't parse %s field in message %d", datesw, msg);
 422         fstat (fileno (in), &st);
 423         smsg->s_clock = st.st_mtime;
 424     } else {
 425         smsg->s_clock = dmktime (tw);
 426     }
 427
 428     if (subjsort) {
 429         if (subjcomp) {
 430             /*
 431              * try to make the subject "canonical": delete
 432              * leading "re:", everything but letters & smash
 433              * letters to lower case.
 434              */
 435             register char  *cp, *cp2, c;
 436
 437             cp = subjcomp;
 438             cp2 = subjcomp;
 439             if (strcmp (subjsort, "subject") == 0) {
 440                 while ((c = *cp)) {
 441                     if (! isspace(c)) {
 442                         if(uprf(cp, "re:"))
 443                             cp += 2;
 444                         else
 445                             break;
 446                     }
 447                     cp++;
 448                 }
 449             }
 450
 451             while ((c = *cp++)) {
 452                 if (isalnum(c))
 453                     *cp2++ = isupper(c) ? tolower(c) : c;
 454             }
 455
 456             *cp2 = '\0';
 457         }
 458         else
 459             subjcomp = "";
 460
 461         smsg->s_subj = subjcomp;
 462     }
 463     fclose (in);
 464     if (datecomp)
 465         free (datecomp);
 466
 467     return (1);
 468 }
 469
 470 /*
 471  * sort on dates.
 472  */
 473 static int
 474 dsort (struct smsg **a, struct smsg **b)
 475 {
 476     if ((*a)->s_clock < (*b)->s_clock)
 477         return (-1);
 478     else if ((*a)->s_clock > (*b)->s_clock)
 479         return (1);
 480     else if ((*a)->s_msg < (*b)->s_msg)
 481         return (-1);
 482     else
 483         return (1);
 484 }
 485
 486 /*
 487  * sort on subjects.
 488  */
 489 static int
 490 subsort (struct smsg **a, struct smsg **b)
 491 {
 492     register int i;
 493
 494     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 495         return (i);
 496
 497     return (dsort (a, b));
 498 }
 499
 500 static int
 501 txtsort (struct smsg **a, struct smsg **b)
 502 {
 503     register int i;
 504
 505     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 506         return (i);
 507     else if ((*a)->s_msg < (*b)->s_msg)
 508         return (-1);
 509     else
 510         return (1);
 511 }
 512
 513 static void
 514 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 515 {
 516     int nxt, old, new;
 517     char *newname, oldname[BUFSIZ];
 518     char newbuf[MAXPATHLEN + 1];
 519
 520     for (;;) {
 521         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 522         mlist[msg] = (struct smsg *)0;
 523         old = smsgs[nxt].s_msg;
 524         new = smsgs[msg].s_msg;
 525         strncpy (oldname, m_name (old), sizeof(oldname));
 526         newname = m_name (new);
 527         if (verbose)
 528             printf ("message %d becomes message %d\n", old, new);
 529
 530         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 531         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 532         ext_hook("ref-hook", oldname, newbuf);
 533
 534         if (rename (oldname, newname) == NOTOK)
 535             adios (newname, "unable to rename %s to", oldname);
 536
 537         copy_msg_flags (mp, new, old);
 538         if (mp->curmsg == old)
 539             seq_setcur (mp, new);
 540
 541         if (nxt == endmsg)
 542             break;
 543
 544         msg = nxt;
 545     }
 546 /*      if (nxt != endmsg); */
 547 /*      rename_chain (mp, mlist, nxt, endmsg); */
 548 }
 549
 550 static void
 551 rename_msgs (struct msgs *mp, struct smsg **mlist)
 552 {
 553     int i, j, old, new;
 554     seqset_t tmpset;
 555     char f1[BUFSIZ], tmpfil[BUFSIZ];
 556     char newbuf[MAXPATHLEN + 1];
 557     struct smsg *sp;
 558
 559     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 560
 561     for (i = 0; i < nmsgs; i++) {
 562         if (! (sp = mlist[i]))
 563             continue;   /* did this one */
 564
 565         j = sp - smsgs;
 566         if (j == i)
 567             continue;   /* this one doesn't move */
 568
 569         /*
 570          * the guy that was msg j is about to become msg i.
 571          * rename 'j' to make a hole, then recursively rename
 572          * guys to fill up the hole.
 573          */
 574         old = smsgs[j].s_msg;
 575         new = smsgs[i].s_msg;
 576         strncpy (f1, m_name (old), sizeof(f1));
 577
 578         if (verbose)
 579             printf ("renaming message chain from %d to %d\n", old, new);
 580
 581         /*
 582          *      Run the external hook to refile the old message as the
 583          *      temporary message number that is off of the end of the
 584          *      messages in the folder.
 585          */
 586
 587         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 588         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 589         ext_hook("ref-hook", f1, newbuf);
 590
 591         if (rename (f1, tmpfil) == NOTOK)
 592             adios (tmpfil, "unable to rename %s to ", f1);
 593
 594         get_msg_flags (mp, &tmpset, old);
 595
 596         rename_chain (mp, mlist, j, i);
 597
 598         /*
 599          *      Run the external hook to refile the temorary message number
 600          *      to the real place.
 601          */
 602
 603         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 604         ext_hook("ref-hook", newbuf, f1);
 605
 606         if (rename (tmpfil, m_name(new)) == NOTOK)
 607             adios (m_name(new), "unable to rename %s to", tmpfil);
 608
 609         set_msg_flags (mp, &tmpset, new);
 610         mp->msgflags |= SEQMOD;
 611     }
 612 }