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
  15 /*
  16  * We allocate space for messages (msgs array)
  17  * this number of elements at a time.
  18  */
  19 #define MAXMSGS  256
  20
  21
  22 static struct swit switches[] = {
  23 #define DATESW                 0
  24      { "datefield field", 0 },
  25 #define TEXTSW                 1
  26      { "textfield field", 0 },
  27 #define NSUBJSW                2
  28      { "notextfield", 0 },
  29 #define SUBJSW                 3
  30      { "subject", -3 },            /* backward-compatibility */
  31 #define LIMSW                  4
  32      { "limit days", 0 },
  33 #define NLIMSW                 5
  34      { "nolimit", 0 },
  35 #define VERBSW                 6
  36      { "verbose", 0 },
  37 #define NVERBSW                7
  38      { "noverbose", 0 },
  39 #define VERSIONSW              8
  40      { "version", 0 },
  41 #define HELPSW                 9
  42      { "help", 0 },
  43      { NULL, 0 }
  44 };
  45
  46 struct smsg {
  47     int s_msg;
  48     time_t s_clock;
  49     char *s_subj;
  50 };
  51
  52 static struct smsg *smsgs;
  53 int nmsgs;
  54
  55 char *subjsort = (char *) 0;    /* sort on subject if != 0 */
  56 unsigned long datelimit = 0;
  57 int submajor = 0;               /* if true, sort on subject-major */
  58 int verbose;
  59
  60 /* This keeps compiler happy on calls to qsort */
  61 typedef int (*qsort_comp) (const void *, const void *);
  62
  63 /*
  64  * static prototypes
  65  */
  66 static int read_hdrs (struct msgs *, char *);
  67 static int get_fields (char *, int, struct smsg *);
  68 static int dsort (struct smsg **, struct smsg **);
  69 static int subsort (struct smsg **, struct smsg **);
  70 static int txtsort (struct smsg **, struct smsg **);
  71 static void rename_chain (struct msgs *, struct smsg **, int, int);
  72 static void rename_msgs (struct msgs *, struct smsg **);
  73
  74
  75 int
  76 main (int argc, char **argv)
  77 {
  78     int nummsgs, maxmsgs, i, msgnum;
  79     char *cp, *maildir, *datesw = NULL;
  80     char *folder = NULL, buf[BUFSIZ], **argp;
  81     char **arguments, **msgs;
  82     struct msgs *mp;
  83     struct smsg **dlist;
  84
  85 #ifdef LOCALE
  86     setlocale(LC_ALL, "");
  87 #endif
  88     invo_name = r1bindex (argv[0], '/');
  89
  90     /* read user profile/context */
  91     context_read();
  92
  93     arguments = getarguments (invo_name, argc, argv, 1);
  94     argp = arguments;
  95
  96     /*
  97      * Allocate the initial space to record message
  98      * names and ranges.
  99      */
 100     nummsgs = 0;
 101     maxmsgs = MAXMSGS;
 102     if (!(msgs = (char **) malloc ((size_t) (maxmsgs * sizeof(*msgs)))))
 103         adios (NULL, "unable to allocate storage");
 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                 if (!(msgs = (char **) realloc (msgs,
 186                         (size_t) (maxmsgs * sizeof(*msgs)))))
 187                     adios (NULL, "unable to reallocate msgs storage");
 188             }
 189             msgs[nummsgs++] = cp;
 190         }
 191     }
 192
 193     if (!context_find ("path"))
 194         free (path ("./", TFOLDER));
 195     if (!nummsgs)
 196         msgs[nummsgs++] = "all";
 197     if (!datesw)
 198         datesw = "date";
 199     if (!folder)
 200         folder = getfolder (1);
 201     maildir = m_maildir (folder);
 202
 203     if (chdir (maildir) == NOTOK)
 204         adios (maildir, "unable to change directory to");
 205
 206     /* read folder and create message structure */
 207     if (!(mp = folder_read (folder)))
 208         adios (NULL, "unable to read folder %s", folder);
 209
 210     /* check for empty folder */
 211     if (mp->nummsg == 0)
 212         adios (NULL, "no messages in %s", folder);
 213
 214     /* parse all the message ranges/sequences and set SELECTED */
 215     for (msgnum = 0; msgnum < nummsgs; msgnum++)
 216         if (!m_convert (mp, msgs[msgnum]))
 217             done (1);
 218     seq_setprev (mp);   /* set the previous sequence */
 219
 220     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 221         adios (NULL, "no messages to sort");
 222
 223     /*
 224      * sort a list of pointers to our "messages to be sorted".
 225      */
 226     dlist = (struct smsg **) malloc ((nmsgs+1) * sizeof(*dlist));
 227     if (! dlist)
 228         adios (NULL, "couldn't allocate sort memory");
 229     for (i = 0; i < nmsgs; i++)
 230         dlist[i] = &smsgs[i];
 231     dlist[nmsgs] = 0;
 232
 233     if (verbose) {      /* announce what we're doing */
 234         if (subjsort)
 235             printf ("sorting by %s-major %s-minor\n",
 236                 submajor ? subjsort : datesw,
 237                 submajor ? datesw : subjsort);
 238         else
 239             printf ("sorting by datefield %s\n", datesw);
 240     }
 241
 242     /* first sort by date, or by subject-major, date-minor */
 243     qsort ((char *) dlist, nmsgs, sizeof(*dlist),
 244             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 245
 246     /*
 247      * if we're sorting on subject, we need another list
 248      * in subject order, then a merge pass to collate the
 249      * two sorts.
 250      */
 251     if (!submajor && subjsort) {        /* already date sorted */
 252         struct smsg **slist, **flist;
 253         register struct smsg ***il, **fp, **dp;
 254
 255         slist = (struct smsg **) malloc ((nmsgs+1) * sizeof(*slist));
 256         if (! slist)
 257             adios (NULL, "couldn't allocate sort memory");
 258         memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 259         qsort((char *)slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 260
 261         /*
 262          * make an inversion list so we can quickly find
 263          * the collection of messages with the same subj
 264          * given a message number.
 265          */
 266         il = (struct smsg ***) calloc (mp->hghsel+1, sizeof(*il));
 267         if (! il)
 268             adios (NULL, "couldn't allocate msg list");
 269         for (i = 0; i < nmsgs; i++)
 270             il[slist[i]->s_msg] = &slist[i];
 271         /*
 272          * make up the final list, chronological but with
 273          * all the same subjects grouped together.
 274          */
 275         flist = (struct smsg **) malloc ((nmsgs+1) * sizeof(*flist));
 276         if (! flist)
 277             adios (NULL, "couldn't allocate msg list");
 278         fp = flist;
 279         for (dp = dlist; *dp;) {
 280             register struct smsg **s = il[(*dp++)->s_msg];
 281
 282             /* see if we already did this guy */
 283             if (! s)
 284                 continue;
 285
 286             *fp++ = *s++;
 287             /*
 288              * take the next message(s) if there is one,
 289              * its subject isn't null and its subject
 290              * is the same as this one and it's not too
 291              * far away in time.
 292              */
 293             while (*s && (*s)->s_subj[0] &&
 294                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 295                    (datelimit == 0 ||
 296                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 297                 il[(*s)->s_msg] = 0;
 298                 *fp++ = *s++;
 299             }
 300         }
 301         *fp = 0;
 302         free (slist);
 303         free (dlist);
 304         dlist = flist;
 305     }
 306
 307     /*
 308      * At this point, dlist is a sorted array of pointers to smsg structures,
 309      * each of which contains a message number.
 310      */
 311
 312     rename_msgs (mp, dlist);
 313
 314     context_replace (pfolder, folder);  /* update current folder         */
 315     seq_save (mp);                      /* synchronize message sequences */
 316     context_save ();                    /* save the context file         */
 317     folder_free (mp);                   /* free folder/message structure */
 318     return done (0);
 319 }
 320
 321 static int
 322 read_hdrs (struct msgs *mp, char *datesw)
 323 {
 324     int msgnum;
 325     struct tws tb;
 326     register struct smsg *s;
 327
 328     twscopy (&tb, dlocaltimenow ());
 329
 330     smsgs = (struct smsg *)
 331         calloc ((size_t) (mp->hghsel - mp->lowsel + 2),
 332             sizeof(*smsgs));
 333     if (smsgs == NULL)
 334         adios (NULL, "unable to allocate sort storage");
 335
 336     s = smsgs;
 337     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 338         if (is_selected(mp, msgnum)) {
 339             if (get_fields (datesw, msgnum, s)) {
 340                 s->s_msg = msgnum;
 341                 s++;
 342             }
 343         }
 344     }
 345     s->s_msg = 0;
 346     return(s - smsgs);
 347 }
 348
 349
 350 /*
 351  * Parse the message and get the data or subject field,
 352  * if needed.
 353  */
 354
 355 static int
 356 get_fields (char *datesw, int msg, struct smsg *smsg)
 357 {
 358     register int state;
 359     int compnum;
 360     char *msgnam, buf[BUFSIZ], nam[NAMESZ];
 361     register struct tws *tw;
 362     register char *datecomp = NULL, *subjcomp = NULL;
 363     register FILE *in;
 364
 365     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 366         admonish (msgnam, "unable to read message");
 367         return (0);
 368     }
 369     for (compnum = 1, state = FLD;;) {
 370         switch (state = m_getfld (state, nam, buf, sizeof(buf), in)) {
 371         case FLD:
 372         case FLDEOF:
 373         case FLDPLUS:
 374             compnum++;
 375             if (!strcasecmp (nam, datesw)) {
 376                 datecomp = add (buf, datecomp);
 377                 while (state == FLDPLUS) {
 378                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 379                     datecomp = add (buf, datecomp);
 380                 }
 381                 if (!subjsort || subjcomp)
 382                     break;
 383             } else if (subjsort && !strcasecmp (nam, subjsort)) {
 384                 subjcomp = add (buf, subjcomp);
 385                 while (state == FLDPLUS) {
 386                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 387                     subjcomp = add (buf, subjcomp);
 388                 }
 389                 if (datecomp)
 390                     break;
 391             } else {
 392                 /* just flush this guy */
 393                 while (state == FLDPLUS)
 394                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 395             }
 396             continue;
 397
 398         case BODY:
 399         case BODYEOF:
 400         case FILEEOF:
 401             break;
 402
 403         case LENERR:
 404         case FMTERR:
 405             if (state == LENERR || state == FMTERR)
 406                 admonish (NULL, "format error in message %d (header #%d)",
 407                       msg, compnum);
 408             if (datecomp)
 409                 free (datecomp);
 410             if (subjcomp)
 411                 free (subjcomp);
 412             fclose (in);
 413             return (0);
 414
 415         default:
 416             adios (NULL, "internal error -- you lose");
 417         }
 418         break;
 419     }
 420
 421     /*
 422      * If no date component, then use the modification
 423      * time of the file as its date
 424      */
 425     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 426         struct stat st;
 427
 428         admonish (NULL, "can't parse %s field in message %d", datesw, msg);
 429         fstat (fileno (in), &st);
 430         smsg->s_clock = st.st_mtime;
 431     } else {
 432         smsg->s_clock = dmktime (tw);
 433     }
 434
 435     if (subjsort) {
 436         if (subjcomp) {
 437             /*
 438              * try to make the subject "canonical": delete
 439              * leading "re:", everything but letters & smash
 440              * letters to lower case.
 441              */
 442             register char  *cp, *cp2, c;
 443
 444             cp = subjcomp;
 445             cp2 = subjcomp;
 446             if (strcmp (subjsort, "subject") == 0) {
 447                 while ((c = *cp)) {
 448                     if (! isspace(c)) {
 449                         if(uprf(cp, "re:"))
 450                             cp += 2;
 451                         else
 452                             break;
 453                     }
 454                     cp++;
 455                 }
 456             }
 457
 458             while ((c = *cp++)) {
 459                 if (isalnum(c))
 460                     *cp2++ = isupper(c) ? tolower(c) : c;
 461             }
 462
 463             *cp2 = '\0';
 464         }
 465         else
 466             subjcomp = "";
 467
 468         smsg->s_subj = subjcomp;
 469     }
 470     fclose (in);
 471     if (datecomp)
 472         free (datecomp);
 473
 474     return (1);
 475 }
 476
 477 /*
 478  * sort on dates.
 479  */
 480 static int
 481 dsort (struct smsg **a, struct smsg **b)
 482 {
 483     if ((*a)->s_clock < (*b)->s_clock)
 484         return (-1);
 485     else if ((*a)->s_clock > (*b)->s_clock)
 486         return (1);
 487     else if ((*a)->s_msg < (*b)->s_msg)
 488         return (-1);
 489     else
 490         return (1);
 491 }
 492
 493 /*
 494  * sort on subjects.
 495  */
 496 static int
 497 subsort (struct smsg **a, struct smsg **b)
 498 {
 499     register int i;
 500
 501     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 502         return (i);
 503
 504     return (dsort (a, b));
 505 }
 506
 507 static int
 508 txtsort (struct smsg **a, struct smsg **b)
 509 {
 510     register int i;
 511
 512     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 513         return (i);
 514     else if ((*a)->s_msg < (*b)->s_msg)
 515         return (-1);
 516     else
 517         return (1);
 518 }
 519
 520 static void
 521 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 522 {
 523     int nxt, old, new;
 524     char *newname, oldname[BUFSIZ];
 525     char newbuf[MAXPATHLEN + 1];
 526
 527     for (;;) {
 528         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 529         mlist[msg] = (struct smsg *)0;
 530         old = smsgs[nxt].s_msg;
 531         new = smsgs[msg].s_msg;
 532         strncpy (oldname, m_name (old), sizeof(oldname));
 533         newname = m_name (new);
 534         if (verbose)
 535             printf ("message %d becomes message %d\n", old, new);
 536
 537         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 538         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 539         ext_hook("ref-hook", oldname, newbuf);
 540
 541         if (rename (oldname, newname) == NOTOK)
 542             adios (newname, "unable to rename %s to", oldname);
 543
 544         copy_msg_flags (mp, new, old);
 545         if (mp->curmsg == old)
 546             seq_setcur (mp, new);
 547
 548         if (nxt == endmsg)
 549             break;
 550
 551         msg = nxt;
 552     }
 553 /*      if (nxt != endmsg); */
 554 /*      rename_chain (mp, mlist, nxt, endmsg); */
 555 }
 556
 557 static void
 558 rename_msgs (struct msgs *mp, struct smsg **mlist)
 559 {
 560     int i, j, old, new;
 561     seqset_t tmpset;
 562     char f1[BUFSIZ], tmpfil[BUFSIZ];
 563     char newbuf[MAXPATHLEN + 1];
 564     struct smsg *sp;
 565
 566     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 567
 568     for (i = 0; i < nmsgs; i++) {
 569         if (! (sp = mlist[i]))
 570             continue;   /* did this one */
 571
 572         j = sp - smsgs;
 573         if (j == i)
 574             continue;   /* this one doesn't move */
 575
 576         /*
 577          * the guy that was msg j is about to become msg i.
 578          * rename 'j' to make a hole, then recursively rename
 579          * guys to fill up the hole.
 580          */
 581         old = smsgs[j].s_msg;
 582         new = smsgs[i].s_msg;
 583         strncpy (f1, m_name (old), sizeof(f1));
 584
 585         if (verbose)
 586             printf ("renaming message chain from %d to %d\n", old, new);
 587
 588         /*
 589          *      Run the external hook to refile the old message as the
 590          *      temporary message number that is off of the end of the
 591          *      messages in the folder.
 592          */
 593
 594         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 595         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 596         ext_hook("ref-hook", f1, newbuf);
 597
 598         if (rename (f1, tmpfil) == NOTOK)
 599             adios (tmpfil, "unable to rename %s to ", f1);
 600
 601         get_msg_flags (mp, &tmpset, old);
 602
 603         rename_chain (mp, mlist, j, i);
 604
 605         /*
 606          *      Run the external hook to refile the temorary message number
 607          *      to the real place.
 608          */
 609
 610         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 611         ext_hook("ref-hook", newbuf, f1);
 612
 613         if (rename (tmpfil, m_name(new)) == NOTOK)
 614             adios (m_name(new), "unable to rename %s to", tmpfil);
 615
 616         set_msg_flags (mp, &tmpset, new);
 617         mp->msgflags |= SEQMOD;
 618     }
 619 }