git.marmaro.de Git - mmh/blob - uip/sortm.c

   1 /*
   2 ** sortm.c -- sort messages in a folder by date/time
   3 **
   4 ** This code is Copyright (c) 2002, by the authors of nmh.  See the
   5 ** COPYRIGHT file in the root directory of the nmh distribution for
   6 ** complete copyright information.
   7 */
   8
   9 #include <h/mh.h>
  10 #include <h/tws.h>
  11 #include <h/utils.h>
  12
  13 static struct swit switches[] = {
  14 #define DATESW  0
  15          { "datefield field", 0 },
  16 #define TEXTSW  1
  17          { "textfield field", 0 },
  18 #define NSUBJSW  2
  19          { "notextfield", 2 },
  20 #define LIMSW  3
  21          { "limit days", 0 },
  22 #define NLIMSW  4
  23          { "nolimit", 2 },
  24 #define VERBSW  5
  25          { "verbose", 0 },
  26 #define NVERBSW  6
  27          { "noverbose", 2 },
  28 #define VERSIONSW  7
  29          { "Version", 0 },
  30 #define HELPSW  8
  31          { "help", 0 },
  32          { NULL, 0 }
  33 };
  34
  35 struct smsg {
  36         int s_msg;
  37         time_t s_clock;
  38         char *s_subj;
  39 };
  40
  41 static struct smsg *smsgs;
  42 int nmsgs;
  43
  44 char *subjsort = NULL;  /* sort on subject if != 0 */
  45 time_t datelimit = 0;
  46 int submajor = 0;  /* if true, sort on subject-major */
  47 int verbose;
  48
  49 /* This keeps compiler happy on calls to qsort */
  50 typedef int (*qsort_comp) (const void *, const void *);
  51
  52 /*
  53 ** static prototypes
  54 */
  55 static int read_hdrs(struct msgs *, char *);
  56 static int get_fields(char *, int, struct smsg *);
  57 static int dsort(struct smsg **, struct smsg **);
  58 static int subsort(struct smsg **, struct smsg **);
  59 static int txtsort(struct smsg **, struct smsg **);
  60 static void rename_chain(struct msgs *, struct smsg **, int, int);
  61 static void rename_msgs(struct msgs *, struct smsg **);
  62
  63
  64 int
  65 main(int argc, char **argv)
  66 {
  67         int i, msgnum;
  68         unsigned char *cp;
  69         char *maildir, *datesw = NULL;
  70         char *folder = NULL, buf[BUFSIZ], **argp;
  71         char **arguments;
  72         struct msgs_array msgs = { 0, 0, NULL };
  73         struct msgs *mp;
  74         struct smsg **dlist;
  75
  76         setlocale(LC_ALL, "");
  77         invo_name = mhbasename(argv[0]);
  78
  79         /* read user profile/context */
  80         context_read();
  81
  82         arguments = getarguments(invo_name, argc, argv, 1);
  83         argp = arguments;
  84
  85         /*
  86         ** Parse arguments
  87         */
  88         while ((cp = *argp++)) {
  89                 if (*cp == '-') {
  90                         switch (smatch(++cp, switches)) {
  91                         case AMBIGSW:
  92                                 ambigsw(cp, switches);
  93                                 done(1);
  94                         case UNKWNSW:
  95                                 adios(NULL, "-%s unknown", cp);
  96
  97                         case HELPSW:
  98                                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]", invo_name);
  99                                 print_help(buf, switches, 1);
 100                                 done(1);
 101                         case VERSIONSW:
 102                                 print_version(invo_name);
 103                                 done(1);
 104
 105                         case DATESW:
 106                                 if (datesw)
 107                                         adios(NULL, "only one date field at a time");
 108                                 if (!(datesw = *argp++) || *datesw == '-')
 109                                         adios(NULL, "missing argument to %s",
 110                                                         argp[-2]);
 111                                 continue;
 112
 113                         case TEXTSW:
 114                                 if (subjsort)
 115                                         adios(NULL, "only one text field at a time");
 116                                 if (!(subjsort = *argp++) || *subjsort == '-')
 117                                         adios(NULL, "missing argument to %s",
 118                                                         argp[-2]);
 119                                 continue;
 120
 121                         case NSUBJSW:
 122                                 subjsort = NULL;
 123                                 continue;
 124
 125                         case LIMSW:
 126                                 if (!(cp = *argp++) || *cp == '-')
 127                                                 adios(NULL, "missing argument to %s", argp[-2]);
 128                                 while (*cp == '0')
 129                                         cp++;  /* skip any leading zeros */
 130                                 if (!*cp) {  /* hit end of string */
 131                                         submajor++;  /* sort subject-major */
 132                                         continue;
 133                                 }
 134                                 if (!isdigit(*cp) || !(datelimit = atoi(cp)))
 135                                         adios(NULL, "impossible limit %s", cp);
 136                                 datelimit *= 60*60*24;
 137                                 continue;
 138                         case NLIMSW:
 139                                 submajor = 0;  /* use date-major, but */
 140                                 datelimit = 0;  /* use no limit */
 141                                 continue;
 142
 143                         case VERBSW:
 144                                 verbose++;
 145                                 continue;
 146                         case NVERBSW:
 147                                 verbose = 0;
 148                                 continue;
 149                         }
 150                 }
 151                 if (*cp == '+' || *cp == '@') {
 152                         if (folder)
 153                                 adios(NULL, "only one folder at a time!");
 154                         else
 155                                 folder = getcpy(expandfol(cp));
 156                 } else
 157                         app_msgarg(&msgs, cp);
 158         }
 159
 160         if (!msgs.size)
 161                 app_msgarg(&msgs, seq_all);
 162         if (!datesw)
 163                 datesw = "date";
 164         if (!folder)
 165                 folder = getcurfol();
 166         maildir = toabsdir(folder);
 167
 168         if (chdir(maildir) == NOTOK)
 169                 adios(maildir, "unable to change directory to");
 170
 171         /* read folder and create message structure */
 172         if (!(mp = folder_read(folder)))
 173                 adios(NULL, "unable to read folder %s", folder);
 174
 175         /* check for empty folder */
 176         if (mp->nummsg == 0)
 177                 adios(NULL, "no messages in %s", folder);
 178
 179         /* parse all the message ranges/sequences and set SELECTED */
 180         for (msgnum = 0; msgnum < msgs.size; msgnum++)
 181                 if (!m_convert(mp, msgs.msgs[msgnum]))
 182                         done(1);
 183         seq_setprev(mp);  /* set the previous sequence */
 184
 185         if ((nmsgs = read_hdrs(mp, datesw)) <= 0)
 186                 adios(NULL, "no messages to sort");
 187
 188         /*
 189         ** sort a list of pointers to our "messages to be sorted".
 190         */
 191         dlist = (struct smsg **) mh_xmalloc((nmsgs+1) * sizeof(*dlist));
 192         for (i = 0; i < nmsgs; i++)
 193                 dlist[i] = &smsgs[i];
 194         dlist[nmsgs] = 0;
 195
 196         if (verbose) {  /* announce what we're doing */
 197                 if (subjsort)
 198                         printf("sorting by %s-major %s-minor\n",
 199                                 submajor ? subjsort : datesw,
 200                                 submajor ? datesw : subjsort);
 201                 else
 202                         printf("sorting by datefield %s\n", datesw);
 203         }
 204
 205         /* first sort by date, or by subject-major, date-minor */
 206         qsort((char *) dlist, nmsgs, sizeof(*dlist),
 207                         (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 208
 209         /*
 210         ** if we're sorting on subject, we need another list
 211         ** in subject order, then a merge pass to collate the
 212         ** two sorts.
 213         */
 214         if (!submajor && subjsort) {  /* already date sorted */
 215                 struct smsg **slist, **flist;
 216                 register struct smsg ***il, **fp, **dp;
 217
 218                 slist = (struct smsg **)
 219                                 mh_xmalloc((nmsgs+1) * sizeof(*slist));
 220                 memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 221                 qsort((char *)slist, nmsgs, sizeof(*slist),
 222                                 (qsort_comp) subsort);
 223
 224                 /*
 225                 ** make an inversion list so we can quickly find
 226                 ** the collection of messages with the same subj
 227                 ** given a message number.
 228                 */
 229                 il = (struct smsg ***) calloc(mp->hghsel+1, sizeof(*il));
 230                 if (! il)
 231                         adios(NULL, "couldn't allocate msg list");
 232                 for (i = 0; i < nmsgs; i++)
 233                         il[slist[i]->s_msg] = &slist[i];
 234                 /*
 235                 ** make up the final list, chronological but with
 236                 ** all the same subjects grouped together.
 237                 */
 238                 flist = (struct smsg **)
 239                                 mh_xmalloc((nmsgs+1) * sizeof(*flist));
 240                 fp = flist;
 241                 for (dp = dlist; *dp;) {
 242                         register struct smsg **s = il[(*dp++)->s_msg];
 243
 244                         /* see if we already did this guy */
 245                         if (! s)
 246                                 continue;
 247
 248                         *fp++ = *s++;
 249                         /*
 250                         ** take the next message(s) if there is one,
 251                         ** its subject isn't null and its subject
 252                         ** is the same as this one and it's not too
 253                         ** far away in time.
 254                         */
 255                         while (*s && (*s)->s_subj[0] && strcmp((*s)->s_subj, s[-1]->s_subj) == 0 && (datelimit == 0 || (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 256                                 il[(*s)->s_msg] = 0;
 257                                 *fp++ = *s++;
 258                         }
 259                 }
 260                 *fp = 0;
 261                 free(slist);
 262                 free(dlist);
 263                 dlist = flist;
 264         }
 265
 266         /*
 267         ** At this point, dlist is a sorted array of pointers to smsg
 268         ** structures, each of which contains a message number.
 269         */
 270
 271         rename_msgs(mp, dlist);
 272
 273         context_replace(curfolder, folder);  /* update current folder */
 274         seq_save(mp);  /* synchronize message sequences */
 275         context_save();  /* save the context file */
 276         folder_free(mp);  /* free folder/message structure */
 277         done(0);
 278         return 1;
 279 }
 280
 281 static int
 282 read_hdrs(struct msgs *mp, char *datesw)
 283 {
 284         int msgnum;
 285         struct tws tb;
 286         register struct smsg *s;
 287
 288         twscopy(&tb, dlocaltimenow());
 289
 290         smsgs = (struct smsg *) calloc((size_t) (mp->hghsel - mp->lowsel + 2),
 291                         sizeof(*smsgs));
 292         if (smsgs == NULL)
 293                 adios(NULL, "unable to allocate sort storage");
 294
 295         s = smsgs;
 296         for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 297                 if (is_selected(mp, msgnum)) {
 298                         if (get_fields(datesw, msgnum, s)) {
 299                                 s->s_msg = msgnum;
 300                                 s++;
 301                         }
 302                 }
 303         }
 304         s->s_msg = 0;
 305         return(s - smsgs);
 306 }
 307
 308
 309 /*
 310 ** Parse the message and get the data or subject field,
 311 ** if needed.
 312 */
 313
 314 static int
 315 get_fields(char *datesw, int msg, struct smsg *smsg)
 316 {
 317         register int state;
 318         int compnum;
 319         char *msgnam, buf[BUFSIZ], nam[NAMESZ];
 320         register struct tws *tw;
 321         register char *datecomp = NULL, *subjcomp = NULL;
 322         register FILE *in;
 323
 324         if ((in = fopen(msgnam = m_name(msg), "r")) == NULL) {
 325                 admonish(msgnam, "unable to read message");
 326                 return (0);
 327         }
 328         for (compnum = 1, state = FLD;;) {
 329                 switch (state = m_getfld(state, nam, buf, sizeof(buf), in)) {
 330                 case FLD:
 331                 case FLDEOF:
 332                 case FLDPLUS:
 333                         compnum++;
 334                         if (!mh_strcasecmp(nam, datesw)) {
 335                                 datecomp = add(buf, datecomp);
 336                                 while (state == FLDPLUS) {
 337                                         state = m_getfld(state, nam, buf,
 338                                                         sizeof(buf), in);
 339                                         datecomp = add(buf, datecomp);
 340                                 }
 341                                 if (!subjsort || subjcomp)
 342                                         break;
 343                         } else if (subjsort && !mh_strcasecmp(nam, subjsort)) {
 344                                 subjcomp = add(buf, subjcomp);
 345                                 while (state == FLDPLUS) {
 346                                         state = m_getfld(state, nam, buf,
 347                                                         sizeof(buf), in);
 348                                         subjcomp = add(buf, subjcomp);
 349                                 }
 350                                 if (datecomp)
 351                                         break;
 352                         } else {
 353                                 /* just flush this guy */
 354                                 while (state == FLDPLUS)
 355                                         state = m_getfld(state, nam, buf,
 356                                                         sizeof(buf), in);
 357                         }
 358                         continue;
 359
 360                 case BODY:
 361                 case BODYEOF:
 362                 case FILEEOF:
 363                         break;
 364
 365                 case LENERR:
 366                 case FMTERR:
 367                         if (state == LENERR || state == FMTERR)
 368                                 admonish(NULL, "format error in message %d (header #%d)", msg, compnum);
 369                         if (datecomp)
 370                                 free(datecomp);
 371                         if (subjcomp)
 372                                 free(subjcomp);
 373                         fclose(in);
 374                         return (0);
 375
 376                 default:
 377                         adios(NULL, "internal error -- you lose");
 378                 }
 379                 break;
 380         }
 381
 382         /*
 383         ** If no date component, then use the modification
 384         ** time of the file as its date
 385         */
 386         if (!datecomp || (tw = dparsetime(datecomp)) == NULL) {
 387                 struct stat st;
 388
 389                 admonish(NULL, "can't parse %s field in message %d",
 390                                 datesw, msg);
 391                 fstat(fileno(in), &st);
 392                 smsg->s_clock = st.st_mtime;
 393         } else {
 394                 smsg->s_clock = dmktime(tw);
 395         }
 396
 397         if (subjsort) {
 398                 if (subjcomp) {
 399                         /*
 400                         ** try to make the subject "canonical": delete
 401                         ** leading "re:", everything but letters & smash
 402                         ** letters to lower case.
 403                         */
 404                         register char  *cp, *cp2;
 405                         register unsigned char c;
 406
 407                         cp = subjcomp;
 408                         cp2 = subjcomp;
 409                         if (strcmp(subjsort, "subject") == 0) {
 410                                 while ((c = *cp)) {
 411                                         if (! isspace(c)) {
 412                                                 if(uprf(cp, "re:"))
 413                                                         cp += 2;
 414                                                 else
 415                                                         break;
 416                                         }
 417                                         cp++;
 418                                 }
 419                         }
 420
 421                         while ((c = *cp++)) {
 422                                 if (isalnum(c))
 423                                         *cp2++ = isupper(c) ? tolower(c) : c;
 424                         }
 425
 426                         *cp2 = '\0';
 427                 } else
 428                         subjcomp = "";
 429
 430                 smsg->s_subj = subjcomp;
 431         }
 432         fclose(in);
 433         if (datecomp)
 434                 free(datecomp);
 435
 436         return (1);
 437 }
 438
 439 /*
 440 ** sort on dates.
 441 */
 442 static int
 443 dsort(struct smsg **a, struct smsg **b)
 444 {
 445         if ((*a)->s_clock < (*b)->s_clock)
 446                 return (-1);
 447         else if ((*a)->s_clock > (*b)->s_clock)
 448                 return (1);
 449         else if ((*a)->s_msg < (*b)->s_msg)
 450                 return (-1);
 451         else
 452                 return (1);
 453 }
 454
 455 /*
 456 ** sort on subjects.
 457 */
 458 static int
 459 subsort(struct smsg **a, struct smsg **b)
 460 {
 461         register int i;
 462
 463         if ((i = strcmp((*a)->s_subj, (*b)->s_subj)))
 464                 return (i);
 465
 466         return (dsort(a, b));
 467 }
 468
 469 static int
 470 txtsort(struct smsg **a, struct smsg **b)
 471 {
 472         register int i;
 473
 474         if ((i = strcmp((*a)->s_subj, (*b)->s_subj)))
 475                 return (i);
 476         else if ((*a)->s_msg < (*b)->s_msg)
 477                 return (-1);
 478         else
 479                 return (1);
 480 }
 481
 482 static void
 483 rename_chain(struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 484 {
 485         int nxt, old, new;
 486         char *newname, oldname[BUFSIZ];
 487         char newbuf[MAXPATHLEN + 1];
 488
 489         for (;;) {
 490                 nxt = mlist[msg] - smsgs;  /* mlist[msg] is a ptr into smsgs */
 491                 mlist[msg] = (struct smsg *)0;
 492                 old = smsgs[nxt].s_msg;
 493                 new = smsgs[msg].s_msg;
 494                 strncpy(oldname, m_name(old), sizeof(oldname));
 495                 newname = m_name(new);
 496                 if (verbose)
 497                         printf("message %d becomes message %d\n", old, new);
 498
 499                 snprintf(oldname, sizeof (oldname), "%s/%d",
 500                                 mp->foldpath, old);
 501                 snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 502                 ext_hook("ref-hook", oldname, newbuf);
 503
 504                 if (rename(oldname, newname) == NOTOK)
 505                         adios(newname, "unable to rename %s to", oldname);
 506
 507                 copy_msg_flags(mp, new, old);
 508                 if (mp->curmsg == old)
 509                         seq_setcur(mp, new);
 510
 511                 if (nxt == endmsg)
 512                         break;
 513
 514                 msg = nxt;
 515         }
 516 /* if (nxt != endmsg); */
 517 /* rename_chain(mp, mlist, nxt, endmsg); */
 518 }
 519
 520 static void
 521 rename_msgs(struct msgs *mp, struct smsg **mlist)
 522 {
 523         int i, j, old, new;
 524         seqset_t tmpset;
 525         char f1[BUFSIZ], tmpfil[BUFSIZ];
 526         char newbuf[MAXPATHLEN + 1];
 527         struct smsg *sp;
 528
 529         strncpy(tmpfil, m_name(mp->hghmsg + 1), sizeof(tmpfil));
 530
 531         for (i = 0; i < nmsgs; i++) {
 532                 if (! (sp = mlist[i]))
 533                         continue;   /* did this one */
 534
 535                 j = sp - smsgs;
 536                 if (j == i)
 537                         continue;   /* this one doesn't move */
 538
 539                 /*
 540                 ** the guy that was msg j is about to become msg i.
 541                 ** rename 'j' to make a hole, then recursively rename
 542                 ** guys to fill up the hole.
 543                 */
 544                 old = smsgs[j].s_msg;
 545                 new = smsgs[i].s_msg;
 546                 strncpy(f1, m_name(old), sizeof(f1));
 547
 548                 if (verbose)
 549                         printf("renaming message chain from %d to %d\n",
 550                                         old, new);
 551
 552                 /*
 553                 ** Run the external hook to refile the old message as the
 554                 ** temporary message number that is off of the end of the
 555                 ** messages in the folder.
 556                 */
 557
 558                 snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 559                 snprintf(newbuf, sizeof (newbuf), "%s/%d",
 560                                 mp->foldpath, mp->hghmsg + 1);
 561                 ext_hook("ref-hook", f1, newbuf);
 562
 563                 if (rename(f1, tmpfil) == NOTOK)
 564                         adios(tmpfil, "unable to rename %s to ", f1);
 565
 566                 get_msg_flags(mp, &tmpset, old);
 567
 568                 rename_chain(mp, mlist, j, i);
 569
 570                 /*
 571                 ** Run the external hook to refile the temorary message number
 572                 ** to the real place.
 573                 */
 574
 575                 snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 576                 ext_hook("ref-hook", newbuf, f1);
 577
 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 }