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
   8 #include <h/mh.h>
   9 #include <zotnet/tws/tws.h>
  10
  11 /*
  12  * We allocate space for messages (msgs array)
  13  * this number of elements at a time.
  14  */
  15 #define MAXMSGS  256
  16
  17
  18 static struct swit switches[] = {
  19 #define DATESW                 0
  20      { "datefield field", 0 },
  21 #define TEXTSW                 1
  22      { "textfield field", 0 },
  23 #define NSUBJSW                2
  24      { "notextfield", 0 },
  25 #define SUBJSW                 3
  26      { "subject", -3 },            /* backward-compatibility */
  27 #define LIMSW                  4
  28      { "limit days", 0 },
  29 #define NLIMSW                 5
  30      { "nolimit", 0 },
  31 #define VERBSW                 6
  32      { "verbose", 0 },
  33 #define NVERBSW                7
  34      { "noverbose", 0 },
  35 #define VERSIONSW              8
  36      { "version", 0 },
  37 #define HELPSW                 9
  38      { "help", 0 },
  39      { NULL, 0 }
  40 };
  41
  42 struct smsg {
  43     int s_msg;
  44     time_t s_clock;
  45     char *s_subj;
  46 };
  47
  48 static struct smsg *smsgs;
  49 int nmsgs;
  50
  51 char *subjsort = (char *) 0;    /* sort on subject if != 0 */
  52 unsigned long datelimit = 0;
  53 int submajor = 0;               /* if true, sort on subject-major */
  54 int verbose;
  55
  56 /* This keeps compiler happy on calls to qsort */
  57 typedef int (*qsort_comp) (const void *, const void *);
  58
  59 /*
  60  * static prototypes
  61  */
  62 static int read_hdrs (struct msgs *, char *);
  63 static int get_fields (char *, int, struct smsg *);
  64 static int dsort (struct smsg **, struct smsg **);
  65 static int subsort (struct smsg **, struct smsg **);
  66 static int txtsort (struct smsg **, struct smsg **);
  67 static void rename_chain (struct msgs *, struct smsg **, int, int);
  68 static void rename_msgs (struct msgs *, struct smsg **);
  69
  70
  71 int
  72 main (int argc, char **argv)
  73 {
  74     int nummsgs, maxmsgs, i, msgnum;
  75     char *cp, *maildir, *datesw = NULL;
  76     char *folder = NULL, buf[BUFSIZ], **argp;
  77     char **arguments, **msgs;
  78     struct msgs *mp;
  79     struct smsg **dlist;
  80
  81 #ifdef LOCALE
  82     setlocale(LC_ALL, "");
  83 #endif
  84     invo_name = r1bindex (argv[0], '/');
  85
  86     /* read user profile/context */
  87     context_read();
  88
  89     arguments = getarguments (invo_name, argc, argv, 1);
  90     argp = arguments;
  91
  92     /*
  93      * Allocate the initial space to record message
  94      * names and ranges.
  95      */
  96     nummsgs = 0;
  97     maxmsgs = MAXMSGS;
  98     if (!(msgs = (char **) malloc ((size_t) (maxmsgs * sizeof(*msgs)))))
  99         adios (NULL, "unable to allocate storage");
 100
 101     /*
 102      * Parse arguments
 103      */
 104     while ((cp = *argp++)) {
 105         if (*cp == '-') {
 106             switch (smatch (++cp, switches)) {
 107             case AMBIGSW:
 108                 ambigsw (cp, switches);
 109                 done (1);
 110             case UNKWNSW:
 111                 adios (NULL, "-%s unknown", cp);
 112
 113             case HELPSW:
 114                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 115                         invo_name);
 116                 print_help (buf, switches, 1);
 117                 done (1);
 118             case VERSIONSW:
 119                 print_version(invo_name);
 120                 done (1);
 121
 122             case DATESW:
 123                 if (datesw)
 124                     adios (NULL, "only one date field at a time");
 125                 if (!(datesw = *argp++) || *datesw == '-')
 126                     adios (NULL, "missing argument to %s", argp[-2]);
 127                 continue;
 128
 129             case TEXTSW:
 130                 if (subjsort)
 131                     adios (NULL, "only one text field at a time");
 132                 if (!(subjsort = *argp++) || *subjsort == '-')
 133                     adios (NULL, "missing argument to %s", argp[-2]);
 134                 continue;
 135
 136             case SUBJSW:
 137                 subjsort = "subject";
 138                 continue;
 139             case NSUBJSW:
 140                 subjsort = (char *)0;
 141                 continue;
 142
 143             case LIMSW:
 144                 if (!(cp = *argp++) || *cp == '-')
 145                         adios (NULL, "missing argument to %s", argp[-2]);
 146                 while (*cp == '0')
 147                     cp++;               /* skip any leading zeros */
 148                 if (!*cp) {             /* hit end of string */
 149                     submajor++;         /* sort subject-major */
 150                     continue;
 151                 }
 152                 if (!isdigit(*cp) || !(datelimit = atoi(cp)))
 153                     adios (NULL, "impossible limit %s", cp);
 154                 datelimit *= 60*60*24;
 155                 continue;
 156             case NLIMSW:
 157                 submajor = 0;   /* use date-major, but */
 158                 datelimit = 0;  /* use no limit */
 159                 continue;
 160
 161             case VERBSW:
 162                 verbose++;
 163                 continue;
 164             case NVERBSW:
 165                 verbose = 0;
 166                 continue;
 167             }
 168         }
 169         if (*cp == '+' || *cp == '@') {
 170             if (folder)
 171                 adios (NULL, "only one folder at a time!");
 172             else
 173                 folder = path (cp + 1, *cp == '+' ? TFOLDER : TSUBCWF);
 174         } else {
 175             /*
 176              * Check if we need to allocate more space
 177              * for message names/ranges.
 178              */
 179             if (nummsgs >= maxmsgs) {
 180                 maxmsgs += MAXMSGS;
 181                 if (!(msgs = (char **) realloc (msgs,
 182                         (size_t) (maxmsgs * sizeof(*msgs)))))
 183                     adios (NULL, "unable to reallocate msgs storage");
 184             }
 185             msgs[nummsgs++] = cp;
 186         }
 187     }
 188
 189     if (!context_find ("path"))
 190         free (path ("./", TFOLDER));
 191     if (!nummsgs)
 192         msgs[nummsgs++] = "all";
 193     if (!datesw)
 194         datesw = "date";
 195     if (!folder)
 196         folder = getfolder (1);
 197     maildir = m_maildir (folder);
 198
 199     if (chdir (maildir) == NOTOK)
 200         adios (maildir, "unable to change directory to");
 201
 202     /* read folder and create message structure */
 203     if (!(mp = folder_read (folder)))
 204         adios (NULL, "unable to read folder %s", folder);
 205
 206     /* check for empty folder */
 207     if (mp->nummsg == 0)
 208         adios (NULL, "no messages in %s", folder);
 209
 210     /* parse all the message ranges/sequences and set SELECTED */
 211     for (msgnum = 0; msgnum < nummsgs; msgnum++)
 212         if (!m_convert (mp, msgs[msgnum]))
 213             done (1);
 214     seq_setprev (mp);   /* set the previous sequence */
 215
 216     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 217         adios (NULL, "no messages to sort");
 218
 219     /*
 220      * sort a list of pointers to our "messages to be sorted".
 221      */
 222     dlist = (struct smsg **) malloc ((nmsgs+1) * sizeof(*dlist));
 223     if (! dlist)
 224         adios (NULL, "couldn't allocate sort memory");
 225     for (i = 0; i < nmsgs; i++)
 226         dlist[i] = &smsgs[i];
 227     dlist[nmsgs] = 0;
 228
 229     if (verbose) {      /* announce what we're doing */
 230         if (subjsort)
 231             printf ("sorting by %s-major %s-minor\n",
 232                 submajor ? subjsort : datesw,
 233                 submajor ? datesw : subjsort);
 234         else
 235             printf ("sorting by datefield %s\n", datesw);
 236     }
 237
 238     /* first sort by date, or by subject-major, date-minor */
 239     qsort ((char *) dlist, nmsgs, sizeof(*dlist),
 240             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 241
 242     /*
 243      * if we're sorting on subject, we need another list
 244      * in subject order, then a merge pass to collate the
 245      * two sorts.
 246      */
 247     if (!submajor && subjsort) {        /* already date sorted */
 248         struct smsg **slist, **flist;
 249         register struct smsg ***il, **fp, **dp;
 250
 251         slist = (struct smsg **) malloc ((nmsgs+1) * sizeof(*slist));
 252         if (! slist)
 253             adios (NULL, "couldn't allocate sort memory");
 254         memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 255         qsort((char *)slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 256
 257         /*
 258          * make an inversion list so we can quickly find
 259          * the collection of messages with the same subj
 260          * given a message number.
 261          */
 262         il = (struct smsg ***) calloc (mp->hghsel+1, sizeof(*il));
 263         if (! il)
 264             adios (NULL, "couldn't allocate msg list");
 265         for (i = 0; i < nmsgs; i++)
 266             il[slist[i]->s_msg] = &slist[i];
 267         /*
 268          * make up the final list, chronological but with
 269          * all the same subjects grouped together.
 270          */
 271         flist = (struct smsg **) malloc ((nmsgs+1) * sizeof(*flist));
 272         if (! flist)
 273             adios (NULL, "couldn't allocate msg list");
 274         fp = flist;
 275         for (dp = dlist; *dp;) {
 276             register struct smsg **s = il[(*dp++)->s_msg];
 277
 278             /* see if we already did this guy */
 279             if (! s)
 280                 continue;
 281
 282             *fp++ = *s++;
 283             /*
 284              * take the next message(s) if there is one,
 285              * its subject isn't null and its subject
 286              * is the same as this one and it's not too
 287              * far away in time.
 288              */
 289             while (*s && (*s)->s_subj[0] &&
 290                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 291                    (datelimit == 0 ||
 292                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 293                 il[(*s)->s_msg] = 0;
 294                 *fp++ = *s++;
 295             }
 296         }
 297         *fp = 0;
 298         free (slist);
 299         free (dlist);
 300         dlist = flist;
 301     }
 302     rename_msgs (mp, dlist);
 303
 304     context_replace (pfolder, folder);  /* update current folder         */
 305     seq_save (mp);                      /* synchronize message sequences */
 306     context_save ();                    /* save the context file         */
 307     folder_free (mp);                   /* free folder/message structure */
 308     return done (0);
 309 }
 310
 311 static int
 312 read_hdrs (struct msgs *mp, char *datesw)
 313 {
 314     int msgnum;
 315     struct tws tb;
 316     register struct smsg *s;
 317
 318     twscopy (&tb, dlocaltimenow ());
 319
 320     smsgs = (struct smsg *)
 321         calloc ((size_t) (mp->hghsel - mp->lowsel + 2),
 322             sizeof(*smsgs));
 323     if (smsgs == NULL)
 324         adios (NULL, "unable to allocate sort storage");
 325
 326     s = smsgs;
 327     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 328         if (is_selected(mp, msgnum)) {
 329             if (get_fields (datesw, msgnum, s)) {
 330                 s->s_msg = msgnum;
 331                 s++;
 332             }
 333         }
 334     }
 335     s->s_msg = 0;
 336     return(s - smsgs);
 337 }
 338
 339
 340 /*
 341  * Parse the message and get the data or subject field,
 342  * if needed.
 343  */
 344
 345 static int
 346 get_fields (char *datesw, int msg, struct smsg *smsg)
 347 {
 348     register int state;
 349     int compnum;
 350     char *msgnam, buf[BUFSIZ], nam[NAMESZ];
 351     register struct tws *tw;
 352     register char *datecomp = NULL, *subjcomp = NULL;
 353     register FILE *in;
 354
 355     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 356         admonish (msgnam, "unable to read message");
 357         return (0);
 358     }
 359     for (compnum = 1, state = FLD;;) {
 360         switch (state = m_getfld (state, nam, buf, sizeof(buf), in)) {
 361         case FLD:
 362         case FLDEOF:
 363         case FLDPLUS:
 364             compnum++;
 365             if (!strcasecmp (nam, datesw)) {
 366                 datecomp = add (buf, datecomp);
 367                 while (state == FLDPLUS) {
 368                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 369                     datecomp = add (buf, datecomp);
 370                 }
 371                 if (!subjsort || subjcomp)
 372                     break;
 373             } else if (subjsort && !strcasecmp (nam, subjsort)) {
 374                 subjcomp = add (buf, subjcomp);
 375                 while (state == FLDPLUS) {
 376                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 377                     subjcomp = add (buf, subjcomp);
 378                 }
 379                 if (datecomp)
 380                     break;
 381             } else {
 382                 /* just flush this guy */
 383                 while (state == FLDPLUS)
 384                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 385             }
 386             continue;
 387
 388         case BODY:
 389         case BODYEOF:
 390         case FILEEOF:
 391             break;
 392
 393         case LENERR:
 394         case FMTERR:
 395             if (state == LENERR || state == FMTERR)
 396                 admonish (NULL, "format error in message %d (header #%d)",
 397                       msg, compnum);
 398             if (datecomp)
 399                 free (datecomp);
 400             if (subjcomp)
 401                 free (subjcomp);
 402             fclose (in);
 403             return (0);
 404
 405         default:
 406             adios (NULL, "internal error -- you lose");
 407         }
 408         break;
 409     }
 410
 411     /*
 412      * If no date component, then use the modification
 413      * time of the file as its date
 414      */
 415     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 416         struct stat st;
 417
 418         admonish (NULL, "can't parse %s field in message %d", datesw, msg);
 419         fstat (fileno (in), &st);
 420         smsg->s_clock = st.st_mtime;
 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             register char  *cp, *cp2, c;
 433
 434             cp = subjcomp;
 435             cp2 = subjcomp;
 436             if (strcmp (subjsort, "subject") == 0)
 437                 while ((c = *cp)) {
 438                     if (! isspace(c)) {
 439                         if(uprf(cp, "re:"))
 440                             cp += 2;
 441                         else {
 442                             if (isalnum(c))
 443                                 *cp2++ = isupper(c) ? tolower(c) : c;
 444                             break;
 445                         }
 446                     }
 447                     cp++;
 448                 }
 449             while ((c = *cp++)) {
 450                 if (isalnum(c))
 451                     *cp2++ = isupper(c) ? tolower(c) : c;
 452
 453             }
 454             *cp2 = '\0';
 455         }
 456         else
 457             subjcomp = "";
 458
 459         smsg->s_subj = subjcomp;
 460     }
 461     fclose (in);
 462     if (datecomp)
 463         free (datecomp);
 464
 465     return (1);
 466 }
 467
 468 /*
 469  * sort on dates.
 470  */
 471 static int
 472 dsort (struct smsg **a, struct smsg **b)
 473 {
 474     if ((*a)->s_clock < (*b)->s_clock)
 475         return (-1);
 476     else if ((*a)->s_clock > (*b)->s_clock)
 477         return (1);
 478     else if ((*a)->s_msg < (*b)->s_msg)
 479         return (-1);
 480     else
 481         return (1);
 482 }
 483
 484 /*
 485  * sort on subjects.
 486  */
 487 static int
 488 subsort (struct smsg **a, struct smsg **b)
 489 {
 490     register int i;
 491
 492     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 493         return (i);
 494
 495     return (dsort (a, b));
 496 }
 497
 498 static int
 499 txtsort (struct smsg **a, struct smsg **b)
 500 {
 501     register int i;
 502
 503     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 504         return (i);
 505     else if ((*a)->s_msg < (*b)->s_msg)
 506         return (-1);
 507     else
 508         return (1);
 509 }
 510
 511 static void
 512 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 513 {
 514     int nxt, old, new;
 515     char *newname, oldname[BUFSIZ];
 516
 517     for (;;) {
 518         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 519         mlist[msg] = (struct smsg *)0;
 520         old = smsgs[nxt].s_msg;
 521         new = smsgs[msg].s_msg;
 522         strncpy (oldname, m_name (old), sizeof(oldname));
 523         newname = m_name (new);
 524         if (verbose)
 525             printf ("message %d becomes message %d\n", old, new);
 526
 527         if (rename (oldname, newname) == NOTOK)
 528             adios (newname, "unable to rename %s to", oldname);
 529
 530         copy_msg_flags (mp, new, old);
 531         if (mp->curmsg == old)
 532             seq_setcur (mp, new);
 533
 534         if (nxt == endmsg)
 535             break;
 536
 537         msg = nxt;
 538     }
 539 /*      if (nxt != endmsg); */
 540 /*      rename_chain (mp, mlist, nxt, endmsg); */
 541 }
 542
 543 static void
 544 rename_msgs (struct msgs *mp, struct smsg **mlist)
 545 {
 546     int i, j, old, new;
 547     seqset_t tmpset;
 548     char f1[BUFSIZ], tmpfil[BUFSIZ];
 549     struct smsg *sp;
 550
 551     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 552
 553     for (i = 0; i < nmsgs; i++) {
 554         if (! (sp = mlist[i]))
 555             continue;   /* did this one */
 556
 557         j = sp - smsgs;
 558         if (j == i)
 559             continue;   /* this one doesn't move */
 560
 561         /*
 562          * the guy that was msg j is about to become msg i.
 563          * rename 'j' to make a hole, then recursively rename
 564          * guys to fill up the hole.
 565          */
 566         old = smsgs[j].s_msg;
 567         new = smsgs[i].s_msg;
 568         strncpy (f1, m_name (old), sizeof(f1));
 569
 570         if (verbose)
 571             printf ("renaming message chain from %d to %d\n", old, new);
 572
 573         if (rename (f1, tmpfil) == NOTOK)
 574             adios (tmpfil, "unable to rename %s to ", f1);
 575         get_msg_flags (mp, &tmpset, old);
 576
 577         rename_chain (mp, mlist, j, i);
 578         if (rename (tmpfil, m_name(new)) == NOTOK)
 579             adios (m_name(new), "unable to rename %s to", tmpfil);
 580
 581         set_msg_flags (mp, &tmpset, new);
 582         mp->msgflags |= SEQMOD;
 583     }
 584 }