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 #include <unistd.h>
  13 #include <ctype.h>
  14 #include <sys/stat.h>
  15 #include <locale.h>
  16 #include <sysexits.h>
  17
  18 #ifdef HAVE_SYS_PARAM_H
  19 # include <sys/param.h>
  20 #endif
  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", 2 },
  29 #define LIMSW  3
  30          { "limit days", 0 },
  31 #define NLIMSW  4
  32          { "nolimit", 2 },
  33 #define VERBSW  5
  34          { "verbose", 0 },
  35 #define NVERBSW  6
  36          { "noverbose", 2 },
  37 #define VERSIONSW  7
  38          { "Version", 0 },
  39 #define HELPSW  8
  40          { "help", 0 },
  41          { NULL, 0 }
  42 };
  43
  44 struct smsg {
  45         int s_msg;
  46         time_t s_clock;
  47         char *s_subj;
  48 };
  49
  50 static struct smsg *smsgs;
  51 int nmsgs;
  52
  53 char *subjsort = NULL;  /* sort on subject if != 0 */
  54 time_t datelimit = 0;
  55 int submajor = 0;  /* if true, sort on subject-major */
  56 int verbose;
  57
  58 /* This keeps compiler happy on calls to qsort */
  59 typedef int (*qsort_comp) (const void *, const void *);
  60
  61 /*
  62 ** static prototypes
  63 */
  64 static int read_hdrs(struct msgs *, char *);
  65 static int get_fields(char *, int, struct smsg *);
  66 static int dsort(struct smsg **, struct smsg **);
  67 static int subsort(struct smsg **, struct smsg **);
  68 static int txtsort(struct smsg **, struct smsg **);
  69 static void rename_chain(struct msgs *, struct smsg **, int, int);
  70 static void rename_msgs(struct msgs *, struct smsg **);
  71
  72
  73 int
  74 main(int argc, char **argv)
  75 {
  76         int i, msgnum;
  77         unsigned char *cp;
  78         char *maildir, *datesw = NULL;
  79         char *folder = NULL, buf[BUFSIZ], **argp;
  80         char **arguments;
  81         struct msgs_array msgs = { 0, 0, NULL };
  82         struct msgs *mp;
  83         struct smsg **dlist;
  84
  85         setlocale(LC_ALL, "");
  86         invo_name = mhbasename(argv[0]);
  87
  88         /* read user profile/context */
  89         context_read();
  90
  91         arguments = getarguments(invo_name, argc, argv, 1);
  92         argp = arguments;
  93
  94         /*
  95         ** Parse arguments
  96         */
  97         while ((cp = *argp++)) {
  98                 if (*cp == '-') {
  99                         switch (smatch(++cp, switches)) {
 100                         case AMBIGSW:
 101                                 ambigsw(cp, switches);
 102                                 exit(EX_USAGE);
 103                         case UNKWNSW:
 104                                 adios(EX_USAGE, NULL, "-%s unknown", cp);
 105
 106                         case HELPSW:
 107                                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]", invo_name);
 108                                 print_help(buf, switches, 1);
 109                                 exit(argc == 2 ? EX_OK : EX_USAGE);
 110                         case VERSIONSW:
 111                                 print_version(invo_name);
 112                                 exit(argc == 2 ? EX_OK : EX_USAGE);
 113
 114                         case DATESW:
 115                                 if (datesw)
 116                                         adios(EX_USAGE, NULL, "only one date field at a time");
 117                                 if (!(datesw = *argp++) || *datesw == '-')
 118                                         adios(EX_USAGE, NULL, "missing argument to %s",
 119                                                         argp[-2]);
 120                                 continue;
 121
 122                         case TEXTSW:
 123                                 if (subjsort)
 124                                         adios(EX_USAGE, NULL, "only one text field at a time");
 125                                 if (!(subjsort = *argp++) || *subjsort == '-')
 126                                         adios(EX_USAGE, NULL, "missing argument to %s",
 127                                                         argp[-2]);
 128                                 continue;
 129
 130                         case NSUBJSW:
 131                                 subjsort = NULL;
 132                                 continue;
 133
 134                         case LIMSW:
 135                                 if (!(cp = *argp++) || *cp == '-')
 136                                                 adios(EX_USAGE, NULL, "missing argument to %s", argp[-2]);
 137                                 while (*cp == '0')
 138                                         cp++;  /* skip any leading zeros */
 139                                 if (!*cp) {  /* hit end of string */
 140                                         submajor++;  /* sort subject-major */
 141                                         continue;
 142                                 }
 143                                 if (!isdigit(*cp) || !(datelimit = atoi(cp)))
 144                                         adios(EX_USAGE, NULL, "impossible limit %s", cp);
 145                                 datelimit *= 60*60*24;
 146                                 continue;
 147                         case NLIMSW:
 148                                 submajor = 0;  /* use date-major, but */
 149                                 datelimit = 0;  /* use no limit */
 150                                 continue;
 151
 152                         case VERBSW:
 153                                 verbose++;
 154                                 continue;
 155                         case NVERBSW:
 156                                 verbose = 0;
 157                                 continue;
 158                         }
 159                 }
 160                 if (*cp == '+' || *cp == '@') {
 161                         if (folder)
 162                                 adios(EX_USAGE, NULL, "only one folder at a time!");
 163                         else
 164                                 folder = getcpy(expandfol(cp));
 165                 } else
 166                         app_msgarg(&msgs, cp);
 167         }
 168
 169         if (!msgs.size)
 170                 app_msgarg(&msgs, seq_all);
 171         if (!datesw)
 172                 datesw = "date";
 173         if (!folder)
 174                 folder = getcurfol();
 175         maildir = toabsdir(folder);
 176
 177         if (chdir(maildir) == NOTOK)
 178                 adios(EX_OSERR, maildir, "unable to change directory to");
 179
 180         /* read folder and create message structure */
 181         if (!(mp = folder_read(folder)))
 182                 adios(EX_IOERR, NULL, "unable to read folder %s", folder);
 183
 184         /* check for empty folder */
 185         if (mp->nummsg == 0)
 186                 adios(EX_DATAERR, NULL, "no messages in %s", folder);
 187
 188         /* parse all the message ranges/sequences and set SELECTED */
 189         for (msgnum = 0; msgnum < msgs.size; msgnum++)
 190                 if (!m_convert(mp, msgs.msgs[msgnum]))
 191                         exit(EX_USAGE);
 192         seq_setprev(mp);  /* set the previous sequence */
 193
 194         if ((nmsgs = read_hdrs(mp, datesw)) <= 0)
 195                 adios(EX_DATAERR, NULL, "no messages to sort");
 196
 197         /*
 198         ** sort a list of pointers to our "messages to be sorted".
 199         */
 200         dlist = (struct smsg **) mh_xcalloc((size_t) (nmsgs+1), sizeof(*dlist));
 201         for (i = 0; i < nmsgs; i++)
 202                 dlist[i] = &smsgs[i];
 203         dlist[nmsgs] = 0;
 204
 205         if (verbose) {  /* announce what we're doing */
 206                 if (subjsort)
 207                         printf("sorting by %s-major %s-minor\n",
 208                                 submajor ? subjsort : datesw,
 209                                 submajor ? datesw : subjsort);
 210                 else
 211                         printf("sorting by datefield %s\n", datesw);
 212         }
 213
 214         /* first sort by date, or by subject-major, date-minor */
 215         qsort((char *) dlist, nmsgs, sizeof(*dlist),
 216                         (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 217
 218         /*
 219         ** if we're sorting on subject, we need another list
 220         ** in subject order, then a merge pass to collate the
 221         ** two sorts.
 222         */
 223         if (!submajor && subjsort) {  /* already date sorted */
 224                 struct smsg **slist, **flist;
 225                 struct smsg ***il, **fp, **dp;
 226
 227                 slist = (struct smsg **)
 228                                 mh_xmalloc((nmsgs+1) * sizeof(*slist));
 229                 memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 230                 qsort((char *)slist, nmsgs, sizeof(*slist),
 231                                 (qsort_comp) subsort);
 232
 233                 /*
 234                 ** make an inversion list so we can quickly find
 235                 ** the collection of messages with the same subj
 236                 ** given a message number.
 237                 */
 238                 il = (struct smsg ***) mh_xcalloc(mp->hghsel+1, sizeof(*il));
 239                 if (! il)
 240                         adios(EX_OSERR, NULL, "couldn't allocate msg list");
 241                 for (i = 0; i < nmsgs; i++)
 242                         il[slist[i]->s_msg] = &slist[i];
 243                 /*
 244                 ** make up the final list, chronological but with
 245                 ** all the same subjects grouped together.
 246                 */
 247                 flist = (struct smsg **)
 248                                 mh_xmalloc((nmsgs+1) * sizeof(*flist));
 249                 fp = flist;
 250                 for (dp = dlist; *dp;) {
 251                         struct smsg **s = il[(*dp++)->s_msg];
 252
 253                         /* see if we already did this guy */
 254                         if (! s)
 255                                 continue;
 256
 257                         *fp++ = *s++;
 258                         /*
 259                         ** take the next message(s) if there is one,
 260                         ** its subject isn't null and its subject
 261                         ** is the same as this one and it's not too
 262                         ** far away in time.
 263                         */
 264                         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)) {
 265                                 il[(*s)->s_msg] = 0;
 266                                 *fp++ = *s++;
 267                         }
 268                 }
 269                 *fp = 0;
 270                 free(slist);
 271                 free(dlist);
 272                 dlist = flist;
 273         }
 274
 275         /*
 276         ** At this point, dlist is a sorted array of pointers to smsg
 277         ** structures, each of which contains a message number.
 278         */
 279
 280         rename_msgs(mp, dlist);
 281
 282         context_replace(curfolder, folder);  /* update current folder */
 283         seq_save(mp);  /* synchronize message sequences */
 284         context_save();  /* save the context file */
 285         folder_free(mp);  /* free folder/message structure */
 286         return 0;
 287 }
 288
 289 static int
 290 read_hdrs(struct msgs *mp, char *datesw)
 291 {
 292         int msgnum;
 293         struct tws tb;
 294         struct smsg *s;
 295
 296         twscopy(&tb, dlocaltimenow());
 297
 298         smsgs = (struct smsg *) mh_xcalloc((size_t) (mp->hghsel - mp->lowsel + 2),
 299                         sizeof(*smsgs));
 300         if (smsgs == NULL)
 301                 adios(EX_OSERR, NULL, "unable to allocate sort storage");
 302
 303         s = smsgs;
 304         for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 305                 if (is_selected(mp, msgnum)) {
 306                         if (get_fields(datesw, msgnum, s)) {
 307                                 s->s_msg = msgnum;
 308                                 s++;
 309                         }
 310                 }
 311         }
 312         s->s_msg = 0;
 313         return(s - smsgs);
 314 }
 315
 316
 317 /*
 318 ** Parse the message and get the data or subject field,
 319 ** if needed.
 320 */
 321
 322 static int
 323 get_fields(char *datesw, int msg, struct smsg *smsg)
 324 {
 325         int state;
 326         int compnum;
 327         char *msgnam, buf[BUFSIZ], nam[NAMESZ];
 328         struct tws *tw;
 329         char *datecomp = NULL, *subjcomp = NULL;
 330         FILE *in;
 331
 332         if ((in = fopen(msgnam = m_name(msg), "r")) == NULL) {
 333                 admonish(msgnam, "unable to read message");
 334                 return (0);
 335         }
 336         for (compnum = 1, state = FLD;;) {
 337                 switch (state = m_getfld(state, nam, buf, sizeof(buf), in)) {
 338                 case FLD:
 339                 case FLDEOF:
 340                 case FLDPLUS:
 341                         compnum++;
 342                         if (!mh_strcasecmp(nam, datesw)) {
 343                                 datecomp = add(buf, datecomp);
 344                                 while (state == FLDPLUS) {
 345                                         state = m_getfld(state, nam, buf,
 346                                                         sizeof(buf), in);
 347                                         datecomp = add(buf, datecomp);
 348                                 }
 349                                 if (!subjsort || subjcomp)
 350                                         break;
 351                         } else if (subjsort && !mh_strcasecmp(nam, subjsort)) {
 352                                 subjcomp = add(buf, subjcomp);
 353                                 while (state == FLDPLUS) {
 354                                         state = m_getfld(state, nam, buf,
 355                                                         sizeof(buf), in);
 356                                         subjcomp = add(buf, subjcomp);
 357                                 }
 358                                 if (datecomp)
 359                                         break;
 360                         } else {
 361                                 /* just flush this guy */
 362                                 while (state == FLDPLUS)
 363                                         state = m_getfld(state, nam, buf,
 364                                                         sizeof(buf), in);
 365                         }
 366                         continue;
 367
 368                 case BODY:
 369                 case BODYEOF:
 370                 case FILEEOF:
 371                         break;
 372
 373                 case LENERR:
 374                 case FMTERR:
 375                         if (state == LENERR || state == FMTERR)
 376                                 admonish(NULL, "format error in message %d (header #%d)", msg, compnum);
 377                         if (datecomp)
 378                                 free(datecomp);
 379                         if (subjcomp)
 380                                 free(subjcomp);
 381                         fclose(in);
 382                         return (0);
 383
 384                 default:
 385                         adios(EX_SOFTWARE, NULL, "internal error -- you lose");
 386                 }
 387                 break;
 388         }
 389
 390         /*
 391         ** If no date component, then use the modification
 392         ** time of the file as its date
 393         */
 394         if (!datecomp || (tw = dparsetime(datecomp)) == NULL) {
 395                 struct stat st;
 396
 397                 admonish(NULL, "can't parse %s field in message %d",
 398                                 datesw, msg);
 399                 fstat(fileno(in), &st);
 400                 smsg->s_clock = st.st_mtime;
 401         } else {
 402                 smsg->s_clock = dmktime(tw);
 403         }
 404
 405         if (subjsort) {
 406                 if (subjcomp) {
 407                         /*
 408                         ** try to make the subject "canonical": delete
 409                         ** leading "re:", everything but letters & smash
 410                         ** letters to lower case.
 411                         */
 412                         char  *cp, *cp2;
 413                         unsigned char c;
 414
 415                         cp = subjcomp;
 416                         cp2 = subjcomp;
 417                         if (strcmp(subjsort, "subject") == 0) {
 418                                 while ((c = *cp)) {
 419                                         if (! isspace(c)) {
 420                                                 if(uprf(cp, "re:"))
 421                                                         cp += 2;
 422                                                 else
 423                                                         break;
 424                                         }
 425                                         cp++;
 426                                 }
 427                         }
 428
 429                         while ((c = *cp++)) {
 430                                 if (isalnum(c))
 431                                         *cp2++ = isupper(c) ? tolower(c) : c;
 432                         }
 433
 434                         *cp2 = '\0';
 435                 } else
 436                         subjcomp = "";
 437
 438                 smsg->s_subj = subjcomp;
 439         }
 440         fclose(in);
 441         if (datecomp)
 442                 free(datecomp);
 443
 444         return (1);
 445 }
 446
 447 /*
 448 ** sort on dates.
 449 */
 450 static int
 451 dsort(struct smsg **a, struct smsg **b)
 452 {
 453         if ((*a)->s_clock < (*b)->s_clock)
 454                 return (-1);
 455         else if ((*a)->s_clock > (*b)->s_clock)
 456                 return (1);
 457         else if ((*a)->s_msg < (*b)->s_msg)
 458                 return (-1);
 459         else
 460                 return (1);
 461 }
 462
 463 /*
 464 ** sort on subjects.
 465 */
 466 static int
 467 subsort(struct smsg **a, struct smsg **b)
 468 {
 469         int i;
 470
 471         if ((i = strcmp((*a)->s_subj, (*b)->s_subj)))
 472                 return (i);
 473
 474         return (dsort(a, b));
 475 }
 476
 477 static int
 478 txtsort(struct smsg **a, struct smsg **b)
 479 {
 480         int i;
 481
 482         if ((i = strcmp((*a)->s_subj, (*b)->s_subj)))
 483                 return (i);
 484         else if ((*a)->s_msg < (*b)->s_msg)
 485                 return (-1);
 486         else
 487                 return (1);
 488 }
 489
 490 static void
 491 rename_chain(struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 492 {
 493         int nxt, old, new;
 494         char *newname, oldname[BUFSIZ];
 495         char newbuf[MAXPATHLEN + 1];
 496
 497         for (;;) {
 498                 nxt = mlist[msg] - smsgs;  /* mlist[msg] is a ptr into smsgs */
 499                 mlist[msg] = (struct smsg *)0;
 500                 old = smsgs[nxt].s_msg;
 501                 new = smsgs[msg].s_msg;
 502                 strncpy(oldname, m_name(old), sizeof(oldname));
 503                 newname = m_name(new);
 504                 if (verbose)
 505                         printf("message %d becomes message %d\n", old, new);
 506
 507                 snprintf(oldname, sizeof (oldname), "%s/%d",
 508                                 mp->foldpath, old);
 509                 snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 510                 ext_hook("ref-hook", oldname, newbuf);
 511
 512                 if (rename(oldname, newname) == NOTOK)
 513                         adios(EX_IOERR, newname, "unable to rename %s to", oldname);
 514
 515                 copy_msg_flags(mp, new, old);
 516                 if (mp->curmsg == old)
 517                         seq_setcur(mp, new);
 518
 519                 if (nxt == endmsg)
 520                         break;
 521
 522                 msg = nxt;
 523         }
 524 /* if (nxt != endmsg); */
 525 /* rename_chain(mp, mlist, nxt, endmsg); */
 526 }
 527
 528 static void
 529 rename_msgs(struct msgs *mp, struct smsg **mlist)
 530 {
 531         int i, j, old, new;
 532         seqset_t tmpset;
 533         char f1[BUFSIZ], tmpfil[BUFSIZ];
 534         char newbuf[MAXPATHLEN + 1];
 535         struct smsg *sp;
 536
 537         strncpy(tmpfil, m_name(mp->hghmsg + 1), sizeof(tmpfil));
 538
 539         for (i = 0; i < nmsgs; i++) {
 540                 if (! (sp = mlist[i]))
 541                         continue;   /* did this one */
 542
 543                 j = sp - smsgs;
 544                 if (j == i)
 545                         continue;   /* this one doesn't move */
 546
 547                 /*
 548                 ** the guy that was msg j is about to become msg i.
 549                 ** rename 'j' to make a hole, then recursively rename
 550                 ** guys to fill up the hole.
 551                 */
 552                 old = smsgs[j].s_msg;
 553                 new = smsgs[i].s_msg;
 554                 strncpy(f1, m_name(old), sizeof(f1));
 555
 556                 if (verbose)
 557                         printf("renaming message chain from %d to %d\n",
 558                                         old, new);
 559
 560                 /*
 561                 ** Run the external hook to refile the old message as the
 562                 ** temporary message number that is off of the end of the
 563                 ** messages in the folder.
 564                 */
 565
 566                 snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 567                 snprintf(newbuf, sizeof (newbuf), "%s/%d",
 568                                 mp->foldpath, mp->hghmsg + 1);
 569                 ext_hook("ref-hook", f1, newbuf);
 570
 571                 if (rename(f1, tmpfil) == NOTOK)
 572                         adios(EX_IOERR, tmpfil, "unable to rename %s to ", f1);
 573
 574                 get_msg_flags(mp, &tmpset, old);
 575
 576                 rename_chain(mp, mlist, j, i);
 577
 578                 /*
 579                 ** Run the external hook to refile the temorary message number
 580                 ** to the real place.
 581                 */
 582
 583                 snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 584                 ext_hook("ref-hook", newbuf, f1);
 585
 586                 if (rename(tmpfil, m_name(new)) == NOTOK)
 587                         adios(EX_IOERR, m_name(new), "unable to rename %s to", tmpfil);
 588
 589                 set_msg_flags(mp, &tmpset, new);
 590                 mp->msgflags |= SEQMOD;
 591         }
 592 }