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 VERSIONSW             10
  36      { "version", 0 },
  37 #define HELPSW                11
  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 time_t datelimit = 0;
  53 int submajor = 0;               /* if true, sort on subject-major */
  54 int verbose;
  55 int allmsgs = 1;
  56
  57 /* This keeps compiler happy on calls to qsort */
  58 typedef int (*qsort_comp) (const void *, const void *);
  59
  60 /*
  61  * static prototypes
  62  */
  63 static int read_hdrs (struct msgs *, char *);
  64 static int get_fields (char *, int, struct smsg *);
  65 static int dsort (struct smsg **, struct smsg **);
  66 static int subsort (struct smsg **, struct smsg **);
  67 static int txtsort (struct smsg **, struct smsg **);
  68 static void rename_chain (struct msgs *, struct smsg **, int, int);
  69 static void rename_msgs (struct msgs *, struct smsg **);
  70
  71
  72 int
  73 main (int argc, char **argv)
  74 {
  75     int i, msgnum;
  76     unsigned char *cp;
  77     char *maildir, *datesw = NULL;
  78     char *folder = NULL, buf[BUFSIZ], **argp;
  79     char **arguments;
  80     struct msgs_array msgs = { 0, 0, NULL };
  81     struct msgs *mp;
  82     struct smsg **dlist;
  83
  84 #ifdef LOCALE
  85     setlocale(LC_ALL, "");
  86 #endif
  87     invo_name = r1bindex (argv[0], '/');
  88
  89     /* read user profile/context */
  90     context_read();
  91
  92     arguments = getarguments (invo_name, argc, argv, 1);
  93     argp = arguments;
  94
  95     /*
  96      * Parse arguments
  97      */
  98     while ((cp = *argp++)) {
  99         if (*cp == '-') {
 100             switch (smatch (++cp, switches)) {
 101             case AMBIGSW:
 102                 ambigsw (cp, switches);
 103                 done (1);
 104             case UNKWNSW:
 105                 adios (NULL, "-%s unknown", cp);
 106
 107             case HELPSW:
 108                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 109                         invo_name);
 110                 print_help (buf, switches, 1);
 111                 done (0);
 112             case VERSIONSW:
 113                 print_version(invo_name);
 114                 done (0);
 115
 116             case DATESW:
 117                 if (datesw)
 118                     adios (NULL, "only one date field at a time");
 119                 if (!(datesw = *argp++) || *datesw == '-')
 120                     adios (NULL, "missing argument to %s", argp[-2]);
 121                 continue;
 122
 123             case TEXTSW:
 124                 if (subjsort)
 125                     adios (NULL, "only one text field at a time");
 126                 if (!(subjsort = *argp++) || *subjsort == '-')
 127                     adios (NULL, "missing argument to %s", argp[-2]);
 128                 continue;
 129
 130             case SUBJSW:
 131                 subjsort = "subject";
 132                 continue;
 133             case NSUBJSW:
 134                 subjsort = (char *)0;
 135                 continue;
 136
 137             case LIMSW:
 138                 if (!(cp = *argp++) || *cp == '-')
 139                         adios (NULL, "missing argument to %s", argp[-2]);
 140                 while (*cp == '0')
 141                     cp++;               /* skip any leading zeros */
 142                 if (!*cp) {             /* hit end of string */
 143                     submajor++;         /* sort subject-major */
 144                     continue;
 145                 }
 146                 if (!isdigit(*cp) || !(datelimit = atoi(cp)))
 147                     adios (NULL, "impossible limit %s", cp);
 148                 datelimit *= 60*60*24;
 149                 continue;
 150             case NLIMSW:
 151                 submajor = 0;   /* use date-major, but */
 152                 datelimit = 0;  /* use no limit */
 153                 continue;
 154
 155             case VERBSW:
 156                 verbose++;
 157                 continue;
 158             case NVERBSW:
 159                 verbose = 0;
 160                 continue;
 161
 162             case ALLMSGS:
 163                 allmsgs = 1;
 164                 continue;
 165             case NALLMSGS:
 166                 allmsgs = 0;
 167                 continue;
 168             }
 169         }
 170         if (*cp == '+' || *cp == '@') {
 171             if (folder)
 172                 adios (NULL, "only one folder at a time!");
 173             else
 174                 folder = pluspath (cp);
 175         } else
 176                 app_msgarg(&msgs, cp);
 177     }
 178
 179     if (!context_find ("path"))
 180         free (path ("./", TFOLDER));
 181     if (!msgs.size) {
 182         if (allmsgs) {
 183             app_msgarg(&msgs, "all");
 184         } else {
 185             adios (NULL, "must specify messages to sort with -noall");
 186         }
 187     }
 188     if (!datesw)
 189         datesw = "date";
 190     if (!folder)
 191         folder = getfolder (1);
 192     maildir = m_maildir (folder);
 193
 194     if (chdir (maildir) == NOTOK)
 195         adios (maildir, "unable to change directory to");
 196
 197     /* read folder and create message structure */
 198     if (!(mp = folder_read (folder)))
 199         adios (NULL, "unable to read folder %s", folder);
 200
 201     /* check for empty folder */
 202     if (mp->nummsg == 0)
 203         adios (NULL, "no messages in %s", folder);
 204
 205     /* parse all the message ranges/sequences and set SELECTED */
 206     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 207         if (!m_convert (mp, msgs.msgs[msgnum]))
 208             done (1);
 209     seq_setprev (mp);   /* set the previous sequence */
 210
 211     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 212         adios (NULL, "no messages to sort");
 213
 214     /*
 215      * sort a list of pointers to our "messages to be sorted".
 216      */
 217     dlist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 218     for (i = 0; i < nmsgs; i++)
 219         dlist[i] = &smsgs[i];
 220     dlist[nmsgs] = 0;
 221
 222     if (verbose) {      /* announce what we're doing */
 223         if (subjsort)
 224             if (submajor)
 225                 printf ("sorting by %s\n", subjsort);
 226             else
 227                 printf ("sorting by %s-major %s-minor\n", subjsort, datesw);
 228         else
 229             printf ("sorting by datefield %s\n", datesw);
 230     }
 231
 232     /* first sort by date, or by subject-major, date-minor */
 233     qsort ((char *) dlist, nmsgs, sizeof(*dlist),
 234             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 235
 236     /*
 237      * if we're sorting on subject, we need another list
 238      * in subject order, then a merge pass to collate the
 239      * two sorts.
 240      */
 241     if (!submajor && subjsort) {        /* already date sorted */
 242         struct smsg **slist, **flist;
 243         register struct smsg ***il, **fp, **dp;
 244
 245         slist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 246         memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 247         qsort((char *)slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 248
 249         /*
 250          * make an inversion list so we can quickly find
 251          * the collection of messages with the same subj
 252          * given a message number.
 253          */
 254         il = (struct smsg ***) calloc (mp->hghsel+1, sizeof(*il));
 255         if (! il)
 256             adios (NULL, "couldn't allocate msg list");
 257         for (i = 0; i < nmsgs; i++)
 258             il[slist[i]->s_msg] = &slist[i];
 259         /*
 260          * make up the final list, chronological but with
 261          * all the same subjects grouped together.
 262          */
 263         flist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 264         fp = flist;
 265         for (dp = dlist; *dp;) {
 266             register struct smsg **s = il[(*dp++)->s_msg];
 267
 268             /* see if we already did this guy */
 269             if (! s)
 270                 continue;
 271
 272             *fp++ = *s++;
 273             /*
 274              * take the next message(s) if there is one,
 275              * its subject isn't null and its subject
 276              * is the same as this one and it's not too
 277              * far away in time.
 278              */
 279             while (*s && (*s)->s_subj[0] &&
 280                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 281                    (datelimit == 0 ||
 282                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 283                 il[(*s)->s_msg] = 0;
 284                 *fp++ = *s++;
 285             }
 286         }
 287         *fp = 0;
 288         free (slist);
 289         free (dlist);
 290         dlist = flist;
 291     }
 292
 293     /*
 294      * At this point, dlist is a sorted array of pointers to smsg structures,
 295      * each of which contains a message number.
 296      */
 297
 298     rename_msgs (mp, dlist);
 299
 300     context_replace (pfolder, folder);  /* update current folder         */
 301     seq_save (mp);                      /* synchronize message sequences */
 302     context_save ();                    /* save the context file         */
 303     folder_free (mp);                   /* free folder/message structure */
 304     done (0);
 305     return 1;
 306 }
 307
 308 static int
 309 read_hdrs (struct msgs *mp, char *datesw)
 310 {
 311     int msgnum;
 312     struct tws tb;
 313     register struct smsg *s;
 314
 315     twscopy (&tb, dlocaltimenow ());
 316
 317     smsgs = (struct smsg *)
 318         calloc ((size_t) (mp->hghsel - mp->lowsel + 2),
 319             sizeof(*smsgs));
 320     if (smsgs == NULL)
 321         adios (NULL, "unable to allocate sort storage");
 322
 323     s = smsgs;
 324     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 325         if (is_selected(mp, msgnum)) {
 326             if (get_fields (datesw, msgnum, s)) {
 327                 s->s_msg = msgnum;
 328                 s++;
 329             }
 330         }
 331     }
 332     s->s_msg = 0;
 333     return(s - smsgs);
 334 }
 335
 336
 337 /*
 338  * Parse the message and get the data or subject field,
 339  * if needed.
 340  */
 341
 342 static int
 343 get_fields (char *datesw, int msg, struct smsg *smsg)
 344 {
 345     register int state;
 346     int compnum;
 347     char *msgnam, buf[BUFSIZ], nam[NAMESZ];
 348     register struct tws *tw;
 349     register char *datecomp = NULL, *subjcomp = NULL;
 350     register FILE *in;
 351
 352     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 353         admonish (msgnam, "unable to read message");
 354         return (0);
 355     }
 356     for (compnum = 1, state = FLD;;) {
 357         switch (state = m_getfld (state, nam, buf, sizeof(buf), in)) {
 358         case FLD:
 359         case FLDEOF:
 360         case FLDPLUS:
 361             compnum++;
 362             if (!mh_strcasecmp (nam, datesw)) {
 363                 datecomp = add (buf, datecomp);
 364                 while (state == FLDPLUS) {
 365                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 366                     datecomp = add (buf, datecomp);
 367                 }
 368                 if (!subjsort || subjcomp)
 369                     break;
 370             } else if (subjsort && !mh_strcasecmp (nam, subjsort)) {
 371                 subjcomp = add (buf, subjcomp);
 372                 while (state == FLDPLUS) {
 373                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 374                     subjcomp = add (buf, subjcomp);
 375                 }
 376                 if (datecomp)
 377                     break;
 378             } else {
 379                 /* just flush this guy */
 380                 while (state == FLDPLUS)
 381                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 382             }
 383             continue;
 384
 385         case BODY:
 386         case BODYEOF:
 387         case FILEEOF:
 388             break;
 389
 390         case LENERR:
 391         case FMTERR:
 392             if (state == LENERR || state == FMTERR)
 393                 admonish (NULL, "format error in message %d (header #%d)",
 394                       msg, compnum);
 395             if (datecomp)
 396                 free (datecomp);
 397             if (subjcomp)
 398                 free (subjcomp);
 399             fclose (in);
 400             return (0);
 401
 402         default:
 403             adios (NULL, "internal error -- you lose");
 404         }
 405         break;
 406     }
 407
 408     /*
 409      * If no date component, then use the modification
 410      * time of the file as its date
 411      */
 412     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 413         struct stat st;
 414
 415         admonish (NULL, "can't parse %s field in message %d", datesw, msg);
 416         fstat (fileno (in), &st);
 417         smsg->s_clock = st.st_mtime;
 418     } else {
 419         smsg->s_clock = dmktime (tw);
 420     }
 421
 422     if (subjsort) {
 423         if (subjcomp) {
 424             /*
 425              * try to make the subject "canonical": delete
 426              * leading "re:", everything but letters & smash
 427              * letters to lower case.
 428              */
 429             register char  *cp, *cp2;
 430             register unsigned char c;
 431
 432             cp = subjcomp;
 433             cp2 = subjcomp;
 434             if (strcmp (subjsort, "subject") == 0) {
 435                 while ((c = *cp)) {
 436                     if (! isspace(c)) {
 437                         if(uprf(cp, "re:"))
 438                             cp += 2;
 439                         else
 440                             break;
 441                     }
 442                     cp++;
 443                 }
 444             }
 445
 446             while ((c = *cp++)) {
 447                 if (isalnum(c))
 448                     *cp2++ = isupper(c) ? tolower(c) : c;
 449             }
 450
 451             *cp2 = '\0';
 452         }
 453         else
 454             subjcomp = "";
 455
 456         smsg->s_subj = subjcomp;
 457     }
 458     fclose (in);
 459     if (datecomp)
 460         free (datecomp);
 461
 462     return (1);
 463 }
 464
 465 /*
 466  * sort on dates.
 467  */
 468 static int
 469 dsort (struct smsg **a, struct smsg **b)
 470 {
 471     if ((*a)->s_clock < (*b)->s_clock)
 472         return (-1);
 473     else if ((*a)->s_clock > (*b)->s_clock)
 474         return (1);
 475     else if ((*a)->s_msg < (*b)->s_msg)
 476         return (-1);
 477     else
 478         return (1);
 479 }
 480
 481 /*
 482  * sort on subjects.
 483  */
 484 static int
 485 subsort (struct smsg **a, struct smsg **b)
 486 {
 487     register int i;
 488
 489     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 490         return (i);
 491
 492     return (dsort (a, b));
 493 }
 494
 495 static int
 496 txtsort (struct smsg **a, struct smsg **b)
 497 {
 498     register int i;
 499
 500     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 501         return (i);
 502     else if ((*a)->s_msg < (*b)->s_msg)
 503         return (-1);
 504     else
 505         return (1);
 506 }
 507
 508 static void
 509 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 510 {
 511     int nxt, old, new;
 512     char *newname, oldname[BUFSIZ];
 513     char newbuf[PATH_MAX + 1];
 514
 515     for (;;) {
 516         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 517         mlist[msg] = (struct smsg *)0;
 518         old = smsgs[nxt].s_msg;
 519         new = smsgs[msg].s_msg;
 520         strncpy (oldname, m_name (old), sizeof(oldname));
 521         newname = m_name (new);
 522         if (verbose)
 523             printf ("message %d becomes message %d\n", old, new);
 524
 525         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 526         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 527         ext_hook("ref-hook", oldname, newbuf);
 528
 529         if (rename (oldname, newname) == NOTOK)
 530             adios (newname, "unable to rename %s to", oldname);
 531
 532         copy_msg_flags (mp, new, old);
 533         if (mp->curmsg == old)
 534             seq_setcur (mp, new);
 535
 536         if (nxt == endmsg)
 537             break;
 538
 539         msg = nxt;
 540     }
 541 /*      if (nxt != endmsg); */
 542 /*      rename_chain (mp, mlist, nxt, endmsg); */
 543 }
 544
 545 static void
 546 rename_msgs (struct msgs *mp, struct smsg **mlist)
 547 {
 548     int i, j, old, new;
 549     seqset_t tmpset;
 550     char f1[BUFSIZ], tmpfil[BUFSIZ];
 551     char newbuf[PATH_MAX + 1];
 552     struct smsg *sp;
 553
 554     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 555
 556     for (i = 0; i < nmsgs; i++) {
 557         if (! (sp = mlist[i]))
 558             continue;   /* did this one */
 559
 560         j = sp - smsgs;
 561         if (j == i)
 562             continue;   /* this one doesn't move */
 563
 564         /*
 565          * the guy that was msg j is about to become msg i.
 566          * rename 'j' to make a hole, then recursively rename
 567          * guys to fill up the hole.
 568          */
 569         old = smsgs[j].s_msg;
 570         new = smsgs[i].s_msg;
 571         strncpy (f1, m_name (old), sizeof(f1));
 572
 573         if (verbose)
 574             printf ("renaming message chain from %d to %d\n", old, new);
 575
 576         /*
 577          *      Run the external hook to refile the old message as the
 578          *      temporary message number that is off of the end of the
 579          *      messages in the folder.
 580          */
 581
 582         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 583         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 584         ext_hook("ref-hook", f1, newbuf);
 585
 586         if (rename (f1, tmpfil) == NOTOK)
 587             adios (tmpfil, "unable to rename %s to ", f1);
 588
 589         get_msg_flags (mp, &tmpset, old);
 590
 591         rename_chain (mp, mlist, j, i);
 592
 593         /*
 594          *      Run the external hook to refile the temorary message number
 595          *      to the real place.
 596          */
 597
 598         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 599         ext_hook("ref-hook", newbuf, f1);
 600
 601         if (rename (tmpfil, m_name(new)) == NOTOK)
 602             adios (m_name(new), "unable to rename %s to", tmpfil);
 603
 604         set_msg_flags (mp, &tmpset, new);
 605         mp->msgflags |= SEQMOD;
 606     }
 607 }