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", 0 },
  20 #define SUBJSW  3
  21          { "subject", -3 },  /* backward-compatibility */
  22 #define LIMSW  4
  23          { "limit days", 0 },
  24 #define NLIMSW  5
  25          { "nolimit", 0 },
  26 #define VERBSW  6
  27          { "verbose", 0 },
  28 #define NVERBSW  7
  29          { "noverbose", 0 },
  30 #define VERSIONSW  8
  31          { "version", 0 },
  32 #define HELPSW  9
  33          { "help", 0 },
  34          { NULL, 0 }
  35 };
  36
  37 struct smsg {
  38         int s_msg;
  39         time_t s_clock;
  40         char *s_subj;
  41 };
  42
  43 static struct smsg *smsgs;
  44 int nmsgs;
  45
  46 char *subjsort = NULL;  /* sort on subject if != 0 */
  47 unsigned long datelimit = 0;
  48 int submajor = 0;  /* if true, sort on subject-major */
  49 int verbose;
  50
  51 /* This keeps compiler happy on calls to qsort */
  52 typedef int (*qsort_comp) (const void *, const void *);
  53
  54 /*
  55 ** static prototypes
  56 */
  57 static int read_hdrs(struct msgs *, char *);
  58 static int get_fields(char *, int, struct smsg *);
  59 static int dsort(struct smsg **, struct smsg **);
  60 static int subsort(struct smsg **, struct smsg **);
  61 static int txtsort(struct smsg **, struct smsg **);
  62 static void rename_chain(struct msgs *, struct smsg **, int, int);
  63 static void rename_msgs(struct msgs *, struct smsg **);
  64
  65
  66 int
  67 main(int argc, char **argv)
  68 {
  69         int i, msgnum;
  70         unsigned char *cp;
  71         char *maildir, *datesw = NULL;
  72         char *folder = NULL, buf[BUFSIZ], **argp;
  73         char **arguments;
  74         struct msgs_array msgs = { 0, 0, NULL };
  75         struct msgs *mp;
  76         struct smsg **dlist;
  77
  78 #ifdef LOCALE
  79         setlocale(LC_ALL, "");
  80 #endif
  81         invo_name = mhbasename(argv[0]);
  82
  83         /* read user profile/context */
  84         context_read();
  85
  86         arguments = getarguments(invo_name, argc, argv, 1);
  87         argp = arguments;
  88
  89         /*
  90         ** Parse arguments
  91         */
  92         while ((cp = *argp++)) {
  93                 if (*cp == '-') {
  94                         switch (smatch(++cp, switches)) {
  95                         case AMBIGSW:
  96                                 ambigsw(cp, switches);
  97                                 done(1);
  98                         case UNKWNSW:
  99                                 adios(NULL, "-%s unknown", cp);
 100
 101                         case HELPSW:
 102                                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]", invo_name);
 103                                 print_help(buf, switches, 1);
 104                                 done(1);
 105                         case VERSIONSW:
 106                                 print_version(invo_name);
 107                                 done(1);
 108
 109                         case DATESW:
 110                                 if (datesw)
 111                                         adios(NULL, "only one date field at a time");
 112                                 if (!(datesw = *argp++) || *datesw == '-')
 113                                         adios(NULL, "missing argument to %s",
 114                                                         argp[-2]);
 115                                 continue;
 116
 117                         case TEXTSW:
 118                                 if (subjsort)
 119                                         adios(NULL, "only one text field at a time");
 120                                 if (!(subjsort = *argp++) || *subjsort == '-')
 121                                         adios(NULL, "missing argument to %s",
 122                                                         argp[-2]);
 123                                 continue;
 124
 125                         case SUBJSW:
 126                                 subjsort = "subject";
 127                                 continue;
 128                         case NSUBJSW:
 129                                 subjsort = NULL;
 130                                 continue;
 131
 132                         case LIMSW:
 133                                 if (!(cp = *argp++) || *cp == '-')
 134                                                 adios(NULL, "missing argument to %s", argp[-2]);
 135                                 while (*cp == '0')
 136                                         cp++;  /* skip any leading zeros */
 137                                 if (!*cp) {  /* hit end of string */
 138                                         submajor++;  /* sort subject-major */
 139                                         continue;
 140                                 }
 141                                 if (!isdigit(*cp) || !(datelimit = atoi(cp)))
 142                                         adios(NULL, "impossible limit %s", cp);
 143                                 datelimit *= 60*60*24;
 144                                 continue;
 145                         case NLIMSW:
 146                                 submajor = 0;  /* use date-major, but */
 147                                 datelimit = 0;  /* use no limit */
 148                                 continue;
 149
 150                         case VERBSW:
 151                                 verbose++;
 152                                 continue;
 153                         case NVERBSW:
 154                                 verbose = 0;
 155                                 continue;
 156                         }
 157                 }
 158                 if (*cp == '+' || *cp == '@') {
 159                         if (folder)
 160                                 adios(NULL, "only one folder at a time!");
 161                         else
 162                                 folder = getcpy(expandfol(cp));
 163                 } else
 164                         app_msgarg(&msgs, cp);
 165         }
 166
 167         if (!msgs.size)
 168                 app_msgarg(&msgs, seq_all);
 169         if (!datesw)
 170                 datesw = "date";
 171         if (!folder)
 172                 folder = getcurfol();
 173         maildir = toabsdir(folder);
 174
 175         if (chdir(maildir) == NOTOK)
 176                 adios(maildir, "unable to change directory to");
 177
 178         /* read folder and create message structure */
 179         if (!(mp = folder_read(folder)))
 180                 adios(NULL, "unable to read folder %s", folder);
 181
 182         /* check for empty folder */
 183         if (mp->nummsg == 0)
 184                 adios(NULL, "no messages in %s", folder);
 185
 186         /* parse all the message ranges/sequences and set SELECTED */
 187         for (msgnum = 0; msgnum < msgs.size; msgnum++)
 188                 if (!m_convert(mp, msgs.msgs[msgnum]))
 189                         done(1);
 190         seq_setprev(mp);  /* set the previous sequence */
 191
 192         if ((nmsgs = read_hdrs(mp, datesw)) <= 0)
 193                 adios(NULL, "no messages to sort");
 194
 195         /*
 196         ** sort a list of pointers to our "messages to be sorted".
 197         */
 198         dlist = (struct smsg **) mh_xmalloc((nmsgs+1) * sizeof(*dlist));
 199         for (i = 0; i < nmsgs; i++)
 200                 dlist[i] = &smsgs[i];
 201         dlist[nmsgs] = 0;
 202
 203         if (verbose) {  /* announce what we're doing */
 204                 if (subjsort)
 205                         printf("sorting by %s-major %s-minor\n",
 206                                 submajor ? subjsort : datesw,
 207                                 submajor ? datesw : subjsort);
 208                 else
 209                         printf("sorting by datefield %s\n", datesw);
 210         }
 211
 212         /* first sort by date, or by subject-major, date-minor */
 213         qsort((char *) dlist, nmsgs, sizeof(*dlist),
 214                         (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 215
 216         /*
 217         ** if we're sorting on subject, we need another list
 218         ** in subject order, then a merge pass to collate the
 219         ** two sorts.
 220         */
 221         if (!submajor && subjsort) {  /* already date sorted */
 222                 struct smsg **slist, **flist;
 223                 register struct smsg ***il, **fp, **dp;
 224
 225                 slist = (struct smsg **)
 226                                 mh_xmalloc((nmsgs+1) * sizeof(*slist));
 227                 memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 228                 qsort((char *)slist, nmsgs, sizeof(*slist),
 229                                 (qsort_comp) subsort);
 230
 231                 /*
 232                 ** make an inversion list so we can quickly find
 233                 ** the collection of messages with the same subj
 234                 ** given a message number.
 235                 */
 236                 il = (struct smsg ***) calloc(mp->hghsel+1, sizeof(*il));
 237                 if (! il)
 238                         adios(NULL, "couldn't allocate msg list");
 239                 for (i = 0; i < nmsgs; i++)
 240                         il[slist[i]->s_msg] = &slist[i];
 241                 /*
 242                 ** make up the final list, chronological but with
 243                 ** all the same subjects grouped together.
 244                 */
 245                 flist = (struct smsg **)
 246                                 mh_xmalloc((nmsgs+1) * sizeof(*flist));
 247                 fp = flist;
 248                 for (dp = dlist; *dp;) {
 249                         register struct smsg **s = il[(*dp++)->s_msg];
 250
 251                         /* see if we already did this guy */
 252                         if (! s)
 253                                 continue;
 254
 255                         *fp++ = *s++;
 256                         /*
 257                         ** take the next message(s) if there is one,
 258                         ** its subject isn't null and its subject
 259                         ** is the same as this one and it's not too
 260                         ** far away in time.
 261                         */
 262                         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)) {
 263                                 il[(*s)->s_msg] = 0;
 264                                 *fp++ = *s++;
 265                         }
 266                 }
 267                 *fp = 0;
 268                 free(slist);
 269                 free(dlist);
 270                 dlist = flist;
 271         }
 272
 273         /*
 274         ** At this point, dlist is a sorted array of pointers to smsg
 275         ** structures, each of which contains a message number.
 276         */
 277
 278         rename_msgs(mp, dlist);
 279
 280         context_replace(curfolder, folder);  /* update current folder */
 281         seq_save(mp);  /* synchronize message sequences */
 282         context_save();  /* save the context file */
 283         folder_free(mp);  /* free folder/message structure */
 284         done(0);
 285         return 1;
 286 }
 287
 288 static int
 289 read_hdrs(struct msgs *mp, char *datesw)
 290 {
 291         int msgnum;
 292         struct tws tb;
 293         register struct smsg *s;
 294
 295         twscopy(&tb, dlocaltimenow());
 296
 297         smsgs = (struct smsg *) calloc((size_t) (mp->hghsel - mp->lowsel + 2),
 298                         sizeof(*smsgs));
 299         if (smsgs == NULL)
 300                 adios(NULL, "unable to allocate sort storage");
 301
 302         s = smsgs;
 303         for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 304                 if (is_selected(mp, msgnum)) {
 305                         if (get_fields(datesw, msgnum, s)) {
 306                                 s->s_msg = msgnum;
 307                                 s++;
 308                         }
 309                 }
 310         }
 311         s->s_msg = 0;
 312         return(s - smsgs);
 313 }
 314
 315
 316 /*
 317 ** Parse the message and get the data or subject field,
 318 ** if needed.
 319 */
 320
 321 static int
 322 get_fields(char *datesw, int msg, struct smsg *smsg)
 323 {
 324         register int state;
 325         int compnum;
 326         char *msgnam, buf[BUFSIZ], nam[NAMESZ];
 327         register struct tws *tw;
 328         register char *datecomp = NULL, *subjcomp = NULL;
 329         register FILE *in;
 330
 331         if ((in = fopen(msgnam = m_name(msg), "r")) == NULL) {
 332                 admonish(msgnam, "unable to read message");
 333                 return (0);
 334         }
 335         for (compnum = 1, state = FLD;;) {
 336                 switch (state = m_getfld(state, nam, buf, sizeof(buf), in)) {
 337                 case FLD:
 338                 case FLDEOF:
 339                 case FLDPLUS:
 340                         compnum++;
 341                         if (!mh_strcasecmp(nam, datesw)) {
 342                                 datecomp = add(buf, datecomp);
 343                                 while (state == FLDPLUS) {
 344                                         state = m_getfld(state, nam, buf,
 345                                                         sizeof(buf), in);
 346                                         datecomp = add(buf, datecomp);
 347                                 }
 348                                 if (!subjsort || subjcomp)
 349                                         break;
 350                         } else if (subjsort && !mh_strcasecmp(nam, subjsort)) {
 351                                 subjcomp = add(buf, subjcomp);
 352                                 while (state == FLDPLUS) {
 353                                         state = m_getfld(state, nam, buf,
 354                                                         sizeof(buf), in);
 355                                         subjcomp = add(buf, subjcomp);
 356                                 }
 357                                 if (datecomp)
 358                                         break;
 359                         } else {
 360                                 /* just flush this guy */
 361                                 while (state == FLDPLUS)
 362                                         state = m_getfld(state, nam, buf,
 363                                                         sizeof(buf), in);
 364                         }
 365                         continue;
 366
 367                 case BODY:
 368                 case BODYEOF:
 369                 case FILEEOF:
 370                         break;
 371
 372                 case LENERR:
 373                 case FMTERR:
 374                         if (state == LENERR || state == FMTERR)
 375                                 admonish(NULL, "format error in message %d (header #%d)", msg, compnum);
 376                         if (datecomp)
 377                                 free(datecomp);
 378                         if (subjcomp)
 379                                 free(subjcomp);
 380                         fclose(in);
 381                         return (0);
 382
 383                 default:
 384                         adios(NULL, "internal error -- you lose");
 385                 }
 386                 break;
 387         }
 388
 389         /*
 390         ** If no date component, then use the modification
 391         ** time of the file as its date
 392         */
 393         if (!datecomp || (tw = dparsetime(datecomp)) == NULL) {
 394                 struct stat st;
 395
 396                 admonish(NULL, "can't parse %s field in message %d",
 397                                 datesw, msg);
 398                 fstat(fileno(in), &st);
 399                 smsg->s_clock = st.st_mtime;
 400         } else {
 401                 smsg->s_clock = dmktime(tw);
 402         }
 403
 404         if (subjsort) {
 405                 if (subjcomp) {
 406                         /*
 407                         ** try to make the subject "canonical": delete
 408                         ** leading "re:", everything but letters & smash
 409                         ** letters to lower case.
 410                         */
 411                         register char  *cp, *cp2;
 412                         register unsigned char c;
 413
 414                         cp = subjcomp;
 415                         cp2 = subjcomp;
 416                         if (strcmp(subjsort, "subject") == 0) {
 417                                 while ((c = *cp)) {
 418                                         if (! isspace(c)) {
 419                                                 if(uprf(cp, "re:"))
 420                                                         cp += 2;
 421                                                 else
 422                                                         break;
 423                                         }
 424                                         cp++;
 425                                 }
 426                         }
 427
 428                         while ((c = *cp++)) {
 429                                 if (isalnum(c))
 430                                         *cp2++ = isupper(c) ? tolower(c) : c;
 431                         }
 432
 433                         *cp2 = '\0';
 434                 } else
 435                         subjcomp = "";
 436
 437                 smsg->s_subj = subjcomp;
 438         }
 439         fclose(in);
 440         if (datecomp)
 441                 free(datecomp);
 442
 443         return (1);
 444 }
 445
 446 /*
 447 ** sort on dates.
 448 */
 449 static int
 450 dsort(struct smsg **a, struct smsg **b)
 451 {
 452         if ((*a)->s_clock < (*b)->s_clock)
 453                 return (-1);
 454         else if ((*a)->s_clock > (*b)->s_clock)
 455                 return (1);
 456         else if ((*a)->s_msg < (*b)->s_msg)
 457                 return (-1);
 458         else
 459                 return (1);
 460 }
 461
 462 /*
 463 ** sort on subjects.
 464 */
 465 static int
 466 subsort(struct smsg **a, struct smsg **b)
 467 {
 468         register int i;
 469
 470         if ((i = strcmp((*a)->s_subj, (*b)->s_subj)))
 471                 return (i);
 472
 473         return (dsort(a, b));
 474 }
 475
 476 static int
 477 txtsort(struct smsg **a, struct smsg **b)
 478 {
 479         register int i;
 480
 481         if ((i = strcmp((*a)->s_subj, (*b)->s_subj)))
 482                 return (i);
 483         else if ((*a)->s_msg < (*b)->s_msg)
 484                 return (-1);
 485         else
 486                 return (1);
 487 }
 488
 489 static void
 490 rename_chain(struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 491 {
 492         int nxt, old, new;
 493         char *newname, oldname[BUFSIZ];
 494         char newbuf[MAXPATHLEN + 1];
 495
 496         for (;;) {
 497                 nxt = mlist[msg] - smsgs;  /* mlist[msg] is a ptr into smsgs */
 498                 mlist[msg] = (struct smsg *)0;
 499                 old = smsgs[nxt].s_msg;
 500                 new = smsgs[msg].s_msg;
 501                 strncpy(oldname, m_name(old), sizeof(oldname));
 502                 newname = m_name(new);
 503                 if (verbose)
 504                         printf("message %d becomes message %d\n", old, new);
 505
 506                 snprintf(oldname, sizeof (oldname), "%s/%d",
 507                                 mp->foldpath, old);
 508                 snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 509                 ext_hook("ref-hook", oldname, newbuf);
 510
 511                 if (rename(oldname, newname) == NOTOK)
 512                         adios(newname, "unable to rename %s to", oldname);
 513
 514                 copy_msg_flags(mp, new, old);
 515                 if (mp->curmsg == old)
 516                         seq_setcur(mp, new);
 517
 518                 if (nxt == endmsg)
 519                         break;
 520
 521                 msg = nxt;
 522         }
 523 /* if (nxt != endmsg); */
 524 /* rename_chain(mp, mlist, nxt, endmsg); */
 525 }
 526
 527 static void
 528 rename_msgs(struct msgs *mp, struct smsg **mlist)
 529 {
 530         int i, j, old, new;
 531         seqset_t tmpset;
 532         char f1[BUFSIZ], tmpfil[BUFSIZ];
 533         char newbuf[MAXPATHLEN + 1];
 534         struct smsg *sp;
 535
 536         strncpy(tmpfil, m_name(mp->hghmsg + 1), sizeof(tmpfil));
 537
 538         for (i = 0; i < nmsgs; i++) {
 539                 if (! (sp = mlist[i]))
 540                         continue;   /* did this one */
 541
 542                 j = sp - smsgs;
 543                 if (j == i)
 544                         continue;   /* this one doesn't move */
 545
 546                 /*
 547                 ** the guy that was msg j is about to become msg i.
 548                 ** rename 'j' to make a hole, then recursively rename
 549                 ** guys to fill up the hole.
 550                 */
 551                 old = smsgs[j].s_msg;
 552                 new = smsgs[i].s_msg;
 553                 strncpy(f1, m_name(old), sizeof(f1));
 554
 555                 if (verbose)
 556                         printf("renaming message chain from %d to %d\n",
 557                                         old, new);
 558
 559                 /*
 560                 ** Run the external hook to refile the old message as the
 561                 ** temporary message number that is off of the end of the
 562                 ** messages in the folder.
 563                 */
 564
 565                 snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 566                 snprintf(newbuf, sizeof (newbuf), "%s/%d",
 567                                 mp->foldpath, mp->hghmsg + 1);
 568                 ext_hook("ref-hook", f1, newbuf);
 569
 570                 if (rename(f1, tmpfil) == NOTOK)
 571                         adios(tmpfil, "unable to rename %s to ", f1);
 572
 573                 get_msg_flags(mp, &tmpset, old);
 574
 575                 rename_chain(mp, mlist, j, i);
 576
 577                 /*
 578                 ** Run the external hook to refile the temorary message number
 579                 ** to the real place.
 580                 */
 581
 582                 snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 583                 ext_hook("ref-hook", newbuf, f1);
 584
 585                 if (rename(tmpfil, m_name(new)) == NOTOK)
 586                         adios(m_name(new), "unable to rename %s to", tmpfil);
 587
 588                 set_msg_flags(mp, &tmpset, new);
 589                 mp->msgflags |= SEQMOD;
 590         }
 591 }