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 ALLMSGS                8
  32      { "all", 0 },
  33 #define NALLMSGS               9
  34      { "noall", 0 },
  35 #define CHECKSW               10
  36      { "check", 0 },
  37 #define NCHECKSW              11
  38      { "nocheck", 0 },
  39 #define VERSIONSW             12
  40      { "version", 0 },
  41 #define HELPSW                13
  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 time_t datelimit = 0;
  57 int submajor = 0;               /* if true, sort on subject-major */
  58 int verbose;
  59 int allmsgs = 1;
  60 int check_failed = 0;
  61
  62 /* This keeps compiler happy on calls to qsort */
  63 typedef int (*qsort_comp) (const void *, const void *);
  64
  65 /*
  66  * static prototypes
  67  */
  68 static int read_hdrs (struct msgs *, char *);
  69 static int get_fields (char *, int, struct smsg *);
  70 static int dsort (struct smsg **, struct smsg **);
  71 static int subsort (struct smsg **, struct smsg **);
  72 static int txtsort (struct smsg **, struct smsg **);
  73 static void rename_chain (struct msgs *, struct smsg **, int, int);
  74 static void rename_msgs (struct msgs *, struct smsg **);
  75
  76
  77 int
  78 main (int argc, char **argv)
  79 {
  80     int i, msgnum;
  81     unsigned char *cp;
  82     char *maildir, *datesw = NULL;
  83     char *folder = NULL, buf[BUFSIZ], **argp;
  84     char **arguments;
  85     struct msgs_array msgs = { 0, 0, NULL };
  86     struct msgs *mp;
  87     struct smsg **dlist;
  88     int checksw = 0;
  89
  90 #ifdef LOCALE
  91     setlocale(LC_ALL, "");
  92 #endif
  93     invo_name = r1bindex (argv[0], '/');
  94
  95     /* read user profile/context */
  96     context_read();
  97
  98     arguments = getarguments (invo_name, argc, argv, 1);
  99     argp = arguments;
 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 (0);
 118             case VERSIONSW:
 119                 print_version(invo_name);
 120                 done (0);
 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             case ALLMSGS:
 169                 allmsgs = 1;
 170                 continue;
 171             case NALLMSGS:
 172                 allmsgs = 0;
 173                 continue;
 174
 175             case CHECKSW:
 176                 checksw = 1;
 177                 continue;
 178             case NCHECKSW:
 179                 checksw = 0;
 180                 continue;
 181             }
 182         }
 183         if (*cp == '+' || *cp == '@') {
 184             if (folder)
 185                 adios (NULL, "only one folder at a time!");
 186             else
 187                 folder = pluspath (cp);
 188         } else
 189                 app_msgarg(&msgs, cp);
 190     }
 191
 192     if (!context_find ("path"))
 193         free (path ("./", TFOLDER));
 194     if (!msgs.size) {
 195         if (allmsgs) {
 196             app_msgarg(&msgs, "all");
 197         } else {
 198             adios (NULL, "must specify messages to sort with -noall");
 199         }
 200     }
 201     if (!datesw)
 202         datesw = "date";
 203     if (!folder)
 204         folder = getfolder (1);
 205     maildir = m_maildir (folder);
 206
 207     if (chdir (maildir) == NOTOK)
 208         adios (maildir, "unable to change directory to");
 209
 210     /* read folder and create message structure */
 211     if (!(mp = folder_read (folder)))
 212         adios (NULL, "unable to read folder %s", folder);
 213
 214     /* check for empty folder */
 215     if (mp->nummsg == 0)
 216         adios (NULL, "no messages in %s", folder);
 217
 218     /* parse all the message ranges/sequences and set SELECTED */
 219     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 220         if (!m_convert (mp, msgs.msgs[msgnum]))
 221             done (1);
 222     seq_setprev (mp);   /* set the previous sequence */
 223
 224     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 225         adios (NULL, "no messages to sort");
 226
 227     if (checksw  &&  check_failed) {
 228         adios (NULL, "errors found, no messages sorted");
 229     }
 230
 231     /*
 232      * sort a list of pointers to our "messages to be sorted".
 233      */
 234     dlist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 235     for (i = 0; i < nmsgs; i++)
 236         dlist[i] = &smsgs[i];
 237     dlist[nmsgs] = 0;
 238
 239     if (verbose) {      /* announce what we're doing */
 240         if (subjsort)
 241             if (submajor)
 242                 printf ("sorting by %s\n", subjsort);
 243             else
 244                 printf ("sorting by %s-major %s-minor\n", subjsort, datesw);
 245         else
 246             printf ("sorting by datefield %s\n", datesw);
 247     }
 248
 249     /* first sort by date, or by subject-major, date-minor */
 250     qsort ((char *) dlist, nmsgs, sizeof(*dlist),
 251             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 252
 253     /*
 254      * if we're sorting on subject, we need another list
 255      * in subject order, then a merge pass to collate the
 256      * two sorts.
 257      */
 258     if (!submajor && subjsort) {        /* already date sorted */
 259         struct smsg **slist, **flist;
 260         register struct smsg ***il, **fp, **dp;
 261
 262         slist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 263         memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 264         qsort((char *)slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 265
 266         /*
 267          * make an inversion list so we can quickly find
 268          * the collection of messages with the same subj
 269          * given a message number.
 270          */
 271         il = (struct smsg ***) calloc (mp->hghsel+1, sizeof(*il));
 272         if (! il)
 273             adios (NULL, "couldn't allocate msg list");
 274         for (i = 0; i < nmsgs; i++)
 275             il[slist[i]->s_msg] = &slist[i];
 276         /*
 277          * make up the final list, chronological but with
 278          * all the same subjects grouped together.
 279          */
 280         flist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 281         fp = flist;
 282         for (dp = dlist; *dp;) {
 283             register struct smsg **s = il[(*dp++)->s_msg];
 284
 285             /* see if we already did this guy */
 286             if (! s)
 287                 continue;
 288
 289             *fp++ = *s++;
 290             /*
 291              * take the next message(s) if there is one,
 292              * its subject isn't null and its subject
 293              * is the same as this one and it's not too
 294              * far away in time.
 295              */
 296             while (*s && (*s)->s_subj[0] &&
 297                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 298                    (datelimit == 0 ||
 299                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 300                 il[(*s)->s_msg] = 0;
 301                 *fp++ = *s++;
 302             }
 303         }
 304         *fp = 0;
 305         free (slist);
 306         free (dlist);
 307         dlist = flist;
 308     }
 309
 310     /*
 311      * At this point, dlist is a sorted array of pointers to smsg structures,
 312      * each of which contains a message number.
 313      */
 314
 315     rename_msgs (mp, dlist);
 316
 317     context_replace (pfolder, folder);  /* update current folder         */
 318     seq_save (mp);                      /* synchronize message sequences */
 319     context_save ();                    /* save the context file         */
 320     folder_free (mp);                   /* free folder/message structure */
 321     done (0);
 322     return 1;
 323 }
 324
 325 static int
 326 read_hdrs (struct msgs *mp, char *datesw)
 327 {
 328     int msgnum;
 329     struct tws tb;
 330     register struct smsg *s;
 331
 332     twscopy (&tb, dlocaltimenow ());
 333
 334     smsgs = (struct smsg *)
 335         calloc ((size_t) (mp->hghsel - mp->lowsel + 2),
 336             sizeof(*smsgs));
 337     if (smsgs == NULL)
 338         adios (NULL, "unable to allocate sort storage");
 339
 340     s = smsgs;
 341     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 342         if (is_selected(mp, msgnum)) {
 343             if (get_fields (datesw, msgnum, s)) {
 344                 s->s_msg = msgnum;
 345                 s++;
 346             }
 347         }
 348     }
 349     s->s_msg = 0;
 350     return(s - smsgs);
 351 }
 352
 353
 354 /*
 355  * Parse the message and get the data or subject field,
 356  * if needed.
 357  */
 358
 359 static int
 360 get_fields (char *datesw, int msg, struct smsg *smsg)
 361 {
 362     register int state;
 363     int compnum;
 364     char *msgnam, buf[BUFSIZ], nam[NAMESZ];
 365     register struct tws *tw;
 366     register char *datecomp = NULL, *subjcomp = NULL;
 367     register FILE *in;
 368
 369     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 370         admonish (msgnam, "unable to read message");
 371         return (0);
 372     }
 373     for (compnum = 1, state = FLD;;) {
 374         switch (state = m_getfld (state, nam, buf, sizeof(buf), in)) {
 375         case FLD:
 376         case FLDEOF:
 377         case FLDPLUS:
 378             compnum++;
 379             if (!mh_strcasecmp (nam, datesw)) {
 380                 datecomp = add (buf, datecomp);
 381                 while (state == FLDPLUS) {
 382                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 383                     datecomp = add (buf, datecomp);
 384                 }
 385                 if (!subjsort || subjcomp)
 386                     break;
 387             } else if (subjsort && !mh_strcasecmp (nam, subjsort)) {
 388                 subjcomp = add (buf, subjcomp);
 389                 while (state == FLDPLUS) {
 390                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 391                     subjcomp = add (buf, subjcomp);
 392                 }
 393                 if (datecomp)
 394                     break;
 395             } else {
 396                 /* just flush this guy */
 397                 while (state == FLDPLUS)
 398                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 399             }
 400             continue;
 401
 402         case BODY:
 403         case BODYEOF:
 404         case FILEEOF:
 405             break;
 406
 407         case LENERR:
 408         case FMTERR:
 409             if (state == LENERR || state == FMTERR) {
 410                 admonish (NULL, "format error in message %d (header #%d)",
 411                       msg, compnum);
 412                 check_failed = 1;
 413             }
 414             if (datecomp)
 415                 free (datecomp);
 416             if (subjcomp)
 417                 free (subjcomp);
 418             fclose (in);
 419             return (0);
 420
 421         default:
 422             adios (NULL, "internal error -- you lose");
 423         }
 424         break;
 425     }
 426
 427     /*
 428      * If no date component, then use the modification
 429      * time of the file as its date
 430      */
 431     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 432         struct stat st;
 433
 434         admonish (NULL, "can't parse %s field in message %d", datesw, msg);
 435         fstat (fileno (in), &st);
 436         smsg->s_clock = st.st_mtime;
 437         check_failed = 1;
 438     } else {
 439         smsg->s_clock = dmktime (tw);
 440     }
 441
 442     if (subjsort) {
 443         if (subjcomp) {
 444             /*
 445              * try to make the subject "canonical": delete
 446              * leading "re:", everything but letters & smash
 447              * letters to lower case.
 448              */
 449             register char  *cp, *cp2;
 450             register unsigned char c;
 451
 452             cp = subjcomp;
 453             cp2 = subjcomp;
 454             if (strcmp (subjsort, "subject") == 0) {
 455                 while ((c = *cp)) {
 456                     if (! isspace(c)) {
 457                         if(uprf(cp, "re:"))
 458                             cp += 2;
 459                         else
 460                             break;
 461                     }
 462                     cp++;
 463                 }
 464             }
 465
 466             while ((c = *cp++)) {
 467                 if (isalnum(c))
 468                     *cp2++ = isupper(c) ? tolower(c) : c;
 469             }
 470
 471             *cp2 = '\0';
 472         }
 473         else
 474             subjcomp = "";
 475
 476         smsg->s_subj = subjcomp;
 477     }
 478     fclose (in);
 479     if (datecomp)
 480         free (datecomp);
 481
 482     return (1);
 483 }
 484
 485 /*
 486  * sort on dates.
 487  */
 488 static int
 489 dsort (struct smsg **a, struct smsg **b)
 490 {
 491     if ((*a)->s_clock < (*b)->s_clock)
 492         return (-1);
 493     else if ((*a)->s_clock > (*b)->s_clock)
 494         return (1);
 495     else if ((*a)->s_msg < (*b)->s_msg)
 496         return (-1);
 497     else
 498         return (1);
 499 }
 500
 501 /*
 502  * sort on subjects.
 503  */
 504 static int
 505 subsort (struct smsg **a, struct smsg **b)
 506 {
 507     register int i;
 508
 509     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 510         return (i);
 511
 512     return (dsort (a, b));
 513 }
 514
 515 static int
 516 txtsort (struct smsg **a, struct smsg **b)
 517 {
 518     register int i;
 519
 520     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 521         return (i);
 522     else if ((*a)->s_msg < (*b)->s_msg)
 523         return (-1);
 524     else
 525         return (1);
 526 }
 527
 528 static void
 529 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 530 {
 531     int nxt, old, new;
 532     char *newname, oldname[BUFSIZ];
 533     char newbuf[PATH_MAX + 1];
 534
 535     for (;;) {
 536         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 537         mlist[msg] = (struct smsg *)0;
 538         old = smsgs[nxt].s_msg;
 539         new = smsgs[msg].s_msg;
 540         strncpy (oldname, m_name (old), sizeof(oldname));
 541         newname = m_name (new);
 542         if (verbose)
 543             printf ("message %d becomes message %d\n", old, new);
 544
 545         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 546         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 547         ext_hook("ref-hook", oldname, newbuf);
 548
 549         if (rename (oldname, newname) == NOTOK)
 550             adios (newname, "unable to rename %s to", oldname);
 551
 552         copy_msg_flags (mp, new, old);
 553         if (mp->curmsg == old)
 554             seq_setcur (mp, new);
 555
 556         if (nxt == endmsg)
 557             break;
 558
 559         msg = nxt;
 560     }
 561 /*      if (nxt != endmsg); */
 562 /*      rename_chain (mp, mlist, nxt, endmsg); */
 563 }
 564
 565 static void
 566 rename_msgs (struct msgs *mp, struct smsg **mlist)
 567 {
 568     int i, j, old, new;
 569     seqset_t tmpset;
 570     char f1[BUFSIZ], tmpfil[BUFSIZ];
 571     char newbuf[PATH_MAX + 1];
 572     struct smsg *sp;
 573
 574     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 575
 576     for (i = 0; i < nmsgs; i++) {
 577         if (! (sp = mlist[i]))
 578             continue;   /* did this one */
 579
 580         j = sp - smsgs;
 581         if (j == i)
 582             continue;   /* this one doesn't move */
 583
 584         /*
 585          * the guy that was msg j is about to become msg i.
 586          * rename 'j' to make a hole, then recursively rename
 587          * guys to fill up the hole.
 588          */
 589         old = smsgs[j].s_msg;
 590         new = smsgs[i].s_msg;
 591         strncpy (f1, m_name (old), sizeof(f1));
 592
 593         if (verbose)
 594             printf ("renaming message chain from %d to %d\n", old, new);
 595
 596         /*
 597          *      Run the external hook to refile the old message as the
 598          *      temporary message number that is off of the end of the
 599          *      messages in the folder.
 600          */
 601
 602         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 603         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 604         ext_hook("ref-hook", f1, newbuf);
 605
 606         if (rename (f1, tmpfil) == NOTOK)
 607             adios (tmpfil, "unable to rename %s to ", f1);
 608
 609         get_msg_flags (mp, &tmpset, old);
 610
 611         rename_chain (mp, mlist, j, i);
 612
 613         /*
 614          *      Run the external hook to refile the temorary message number
 615          *      to the real place.
 616          */
 617
 618         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 619         ext_hook("ref-hook", newbuf, f1);
 620
 621         if (rename (tmpfil, m_name(new)) == NOTOK)
 622             adios (m_name(new), "unable to rename %s to", tmpfil);
 623
 624         set_msg_flags (mp, &tmpset, new);
 625         mp->msgflags |= SEQMOD;
 626     }
 627 }