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 #define SORTM_SWITCHES \
  15     X("datefield field", 0, DATESW) \
  16     X("textfield field", 0, TEXTSW) \
  17     X("notextfield", 0, NSUBJSW) \
  18     X("subject", -3, SUBJSW) /* backward-compatibility */ \
  19     X("limit days", 0, LIMSW) \
  20     X("nolimit", 0, NLIMSW) \
  21     X("verbose", 0, VERBSW) \
  22     X("noverbose", 0, NVERBSW) \
  23     X("all", 0, ALLMSGS) \
  24     X("noall", 0, NALLMSGS) \
  25     X("check", 0, CHECKSW) \
  26     X("nocheck", 0, NCHECKSW) \
  27     X("version", 0, VERSIONSW) \
  28     X("help", 0, HELPSW) \
  29
  30 #define X(sw, minchars, id) id,
  31 DEFINE_SWITCH_ENUM(SORTM);
  32 #undef X
  33
  34 #define X(sw, minchars, id) { sw, minchars, id },
  35 DEFINE_SWITCH_ARRAY(SORTM, switches);
  36 #undef X
  37
  38 struct smsg {
  39     int s_msg;
  40     time_t s_clock;
  41     char *s_subj;
  42 };
  43
  44 static struct smsg *smsgs;
  45 int nmsgs;
  46
  47 char *subjsort = (char *) 0;    /* sort on subject if != 0 */
  48 time_t datelimit = 0;
  49 int submajor = 0;               /* if true, sort on subject-major */
  50 int verbose;
  51 int allmsgs = 1;
  52 int check_failed = 0;
  53
  54 /* This keeps compiler happy on calls to qsort */
  55 typedef int (*qsort_comp) (const void *, const void *);
  56
  57 /*
  58  * static prototypes
  59  */
  60 static int read_hdrs (struct msgs *, char *);
  61 static int get_fields (char *, int, struct smsg *);
  62 static int dsort (struct smsg **, struct smsg **);
  63 static int subsort (struct smsg **, struct smsg **);
  64 static int txtsort (struct smsg **, struct smsg **);
  65 static void rename_chain (struct msgs *, struct smsg **, int, int);
  66 static void rename_msgs (struct msgs *, struct smsg **);
  67
  68
  69 int
  70 main (int argc, char **argv)
  71 {
  72     int i, msgnum;
  73     char *cp, *maildir, *datesw = NULL;
  74     char *folder = NULL, buf[BUFSIZ], **argp;
  75     char **arguments;
  76     struct msgs_array msgs = { 0, 0, NULL };
  77     struct msgs *mp;
  78     struct smsg **dlist;
  79     int checksw = 0;
  80
  81 #ifdef LOCALE
  82     setlocale(LC_ALL, "");
  83 #endif
  84     invo_name = r1bindex (argv[0], '/');
  85
  86     /* read user profile/context */
  87     context_read();
  88
  89     arguments = getarguments (invo_name, argc, argv, 1);
  90     argp = arguments;
  91
  92     /*
  93      * Parse arguments
  94      */
  95     while ((cp = *argp++)) {
  96         if (*cp == '-') {
  97             switch (smatch (++cp, switches)) {
  98             case AMBIGSW:
  99                 ambigsw (cp, switches);
 100                 done (1);
 101             case UNKWNSW:
 102                 adios (NULL, "-%s unknown", cp);
 103
 104             case HELPSW:
 105                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 106                         invo_name);
 107                 print_help (buf, switches, 1);
 108                 done (0);
 109             case VERSIONSW:
 110                 print_version(invo_name);
 111                 done (0);
 112
 113             case DATESW:
 114                 if (datesw)
 115                     adios (NULL, "only one date field at a time");
 116                 if (!(datesw = *argp++) || *datesw == '-')
 117                     adios (NULL, "missing argument to %s", argp[-2]);
 118                 continue;
 119
 120             case TEXTSW:
 121                 if (subjsort)
 122                     adios (NULL, "only one text field at a time");
 123                 if (!(subjsort = *argp++) || *subjsort == '-')
 124                     adios (NULL, "missing argument to %s", argp[-2]);
 125                 continue;
 126
 127             case SUBJSW:
 128                 subjsort = "subject";
 129                 continue;
 130             case NSUBJSW:
 131                 subjsort = (char *)0;
 132                 continue;
 133
 134             case LIMSW:
 135                 if (!(cp = *argp++) || *cp == '-')
 136                         adios (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((unsigned char) *cp) || !(datelimit = atoi(cp)))
 144                     adios (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             case ALLMSGS:
 160                 allmsgs = 1;
 161                 continue;
 162             case NALLMSGS:
 163                 allmsgs = 0;
 164                 continue;
 165
 166             case CHECKSW:
 167                 checksw = 1;
 168                 continue;
 169             case NCHECKSW:
 170                 checksw = 0;
 171                 continue;
 172             }
 173         }
 174         if (*cp == '+' || *cp == '@') {
 175             if (folder)
 176                 adios (NULL, "only one folder at a time!");
 177             else
 178                 folder = pluspath (cp);
 179         } else
 180                 app_msgarg(&msgs, cp);
 181     }
 182
 183     if (!context_find ("path"))
 184         free (path ("./", TFOLDER));
 185     if (!msgs.size) {
 186         if (allmsgs) {
 187             app_msgarg(&msgs, "all");
 188         } else {
 189             adios (NULL, "must specify messages to sort with -noall");
 190         }
 191     }
 192     if (!datesw)
 193         datesw = "date";
 194     if (!folder)
 195         folder = getfolder (1);
 196     maildir = m_maildir (folder);
 197
 198     if (chdir (maildir) == NOTOK)
 199         adios (maildir, "unable to change directory to");
 200
 201     /* read folder and create message structure */
 202     if (!(mp = folder_read (folder)))
 203         adios (NULL, "unable to read folder %s", folder);
 204
 205     /* check for empty folder */
 206     if (mp->nummsg == 0)
 207         adios (NULL, "no messages in %s", folder);
 208
 209     /* parse all the message ranges/sequences and set SELECTED */
 210     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 211         if (!m_convert (mp, msgs.msgs[msgnum]))
 212             done (1);
 213     seq_setprev (mp);   /* set the previous sequence */
 214
 215     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 216         adios (NULL, "no messages to sort");
 217
 218     if (checksw  &&  check_failed) {
 219         adios (NULL, "errors found, no messages sorted");
 220     }
 221
 222     /*
 223      * sort a list of pointers to our "messages to be sorted".
 224      */
 225     dlist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 226     for (i = 0; i < nmsgs; i++)
 227         dlist[i] = &smsgs[i];
 228     dlist[nmsgs] = 0;
 229
 230     if (verbose) {      /* announce what we're doing */
 231         if (subjsort)
 232             if (submajor)
 233                 printf ("sorting by %s\n", subjsort);
 234             else
 235                 printf ("sorting by %s-major %s-minor\n", subjsort, datesw);
 236         else
 237             printf ("sorting by datefield %s\n", datesw);
 238     }
 239
 240     /* first sort by date, or by subject-major, date-minor */
 241     qsort ((char *) dlist, nmsgs, sizeof(*dlist),
 242             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 243
 244     /*
 245      * if we're sorting on subject, we need another list
 246      * in subject order, then a merge pass to collate the
 247      * two sorts.
 248      */
 249     if (!submajor && subjsort) {        /* already date sorted */
 250         struct smsg **slist, **flist;
 251         register struct smsg ***il, **fp, **dp;
 252
 253         slist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 254         memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 255         qsort((char *)slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 256
 257         /*
 258          * make an inversion list so we can quickly find
 259          * the collection of messages with the same subj
 260          * given a message number.
 261          */
 262         il = (struct smsg ***) calloc (mp->hghsel+1, sizeof(*il));
 263         if (! il)
 264             adios (NULL, "couldn't allocate msg list");
 265         for (i = 0; i < nmsgs; i++)
 266             il[slist[i]->s_msg] = &slist[i];
 267         /*
 268          * make up the final list, chronological but with
 269          * all the same subjects grouped together.
 270          */
 271         flist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 272         fp = flist;
 273         for (dp = dlist; *dp;) {
 274             register struct smsg **s = il[(*dp++)->s_msg];
 275
 276             /* see if we already did this guy */
 277             if (! s)
 278                 continue;
 279
 280             *fp++ = *s++;
 281             /*
 282              * take the next message(s) if there is one,
 283              * its subject isn't null and its subject
 284              * is the same as this one and it's not too
 285              * far away in time.
 286              */
 287             while (*s && (*s)->s_subj[0] &&
 288                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 289                    (datelimit == 0 ||
 290                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 291                 il[(*s)->s_msg] = 0;
 292                 *fp++ = *s++;
 293             }
 294         }
 295         *fp = 0;
 296         free (slist);
 297         free (dlist);
 298         dlist = flist;
 299     }
 300
 301     /*
 302      * At this point, dlist is a sorted array of pointers to smsg structures,
 303      * each of which contains a message number.
 304      */
 305
 306     rename_msgs (mp, dlist);
 307
 308     context_replace (pfolder, folder);  /* update current folder         */
 309     seq_save (mp);                      /* synchronize message sequences */
 310     context_save ();                    /* save the context file         */
 311     folder_free (mp);                   /* free folder/message structure */
 312     done (0);
 313     return 1;
 314 }
 315
 316 static int
 317 read_hdrs (struct msgs *mp, char *datesw)
 318 {
 319     int msgnum;
 320     struct tws tb;
 321     register struct smsg *s;
 322
 323     twscopy (&tb, dlocaltimenow ());
 324
 325     smsgs = (struct smsg *)
 326         calloc ((size_t) (mp->hghsel - mp->lowsel + 2),
 327             sizeof(*smsgs));
 328     if (smsgs == NULL)
 329         adios (NULL, "unable to allocate sort storage");
 330
 331     s = smsgs;
 332     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 333         if (is_selected(mp, msgnum)) {
 334             if (get_fields (datesw, msgnum, s)) {
 335                 s->s_msg = msgnum;
 336                 s++;
 337             }
 338         }
 339     }
 340     s->s_msg = 0;
 341     return(s - smsgs);
 342 }
 343
 344
 345 /*
 346  * Parse the message and get the data or subject field,
 347  * if needed.
 348  */
 349
 350 static int
 351 get_fields (char *datesw, int msg, struct smsg *smsg)
 352 {
 353     register int state;
 354     int compnum;
 355     char *msgnam, buf[BUFSIZ], nam[NAMESZ];
 356     register struct tws *tw;
 357     register char *datecomp = NULL, *subjcomp = NULL;
 358     register FILE *in;
 359     m_getfld_state_t gstate = 0;
 360
 361     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 362         admonish (msgnam, "unable to read message");
 363         return (0);
 364     }
 365     for (compnum = 1;;) {
 366         int bufsz = sizeof buf;
 367         switch (state = m_getfld (&gstate, (unsigned char *) nam,
 368                                   (unsigned char *) buf, &bufsz, in)) {
 369         case FLD:
 370         case FLDPLUS:
 371             compnum++;
 372             if (!mh_strcasecmp (nam, datesw)) {
 373                 datecomp = add (buf, datecomp);
 374                 while (state == FLDPLUS) {
 375                     bufsz = sizeof buf;
 376                     state = m_getfld (&gstate, (unsigned char *) nam,
 377                                       (unsigned char *) buf, &bufsz, in);
 378                     datecomp = add (buf, datecomp);
 379                 }
 380                 if (!subjsort || subjcomp)
 381                     break;
 382             } else if (subjsort && !mh_strcasecmp (nam, subjsort)) {
 383                 subjcomp = add (buf, subjcomp);
 384                 while (state == FLDPLUS) {
 385                     bufsz = sizeof buf;
 386                     state = m_getfld (&gstate, (unsigned char *) nam,
 387                                       (unsigned char *) buf, &bufsz, in);
 388                     subjcomp = add (buf, subjcomp);
 389                 }
 390                 if (datecomp)
 391                     break;
 392             } else {
 393                 /* just flush this guy */
 394                 while (state == FLDPLUS) {
 395                     bufsz = sizeof buf;
 396                     state = m_getfld (&gstate, (unsigned char *) nam,
 397                                       (unsigned char *) buf, &bufsz, in);
 398                 }
 399             }
 400             continue;
 401
 402         case BODY:
 403         case FILEEOF:
 404             break;
 405
 406         case LENERR:
 407         case FMTERR:
 408             if (state == LENERR || state == FMTERR) {
 409                 admonish (NULL, "format error in message %d (header #%d)",
 410                       msg, compnum);
 411                 check_failed = 1;
 412             }
 413             if (datecomp)
 414                 free (datecomp);
 415             if (subjcomp)
 416                 free (subjcomp);
 417             fclose (in);
 418             return (0);
 419
 420         default:
 421             adios (NULL, "internal error -- you lose");
 422         }
 423         break;
 424     }
 425     m_getfld_state_destroy (&gstate);
 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         advise (NULL,
 435                 "can't parse %s field in message %d, "
 436                 "will use file modification time",
 437                 datesw, msg);
 438         fstat (fileno (in), &st);
 439         smsg->s_clock = st.st_mtime;
 440         check_failed = 1;
 441     } else {
 442         smsg->s_clock = dmktime (tw);
 443     }
 444
 445     if (subjsort) {
 446         if (subjcomp) {
 447             /*
 448              * try to make the subject "canonical": delete
 449              * leading "re:", everything but letters & smash
 450              * letters to lower case.
 451              */
 452             register char  *cp, *cp2, c;
 453
 454             cp = subjcomp;
 455             cp2 = subjcomp;
 456             if (strcmp (subjsort, "subject") == 0) {
 457                 while ((c = *cp)) {
 458                     if (! isspace((unsigned char) c)) {
 459                         if(uprf(cp, "re:"))
 460                             cp += 2;
 461                         else
 462                             break;
 463                     }
 464                     cp++;
 465                 }
 466             }
 467
 468             while ((c = *cp++)) {
 469                 if (isascii((unsigned char) c) && isalnum((unsigned char) c))
 470                     *cp2++ = isupper((unsigned char) c) ?
 471                                         tolower((unsigned char) c) : c;
 472             }
 473
 474             *cp2 = '\0';
 475         }
 476         else
 477             subjcomp = "";
 478
 479         smsg->s_subj = subjcomp;
 480     }
 481     fclose (in);
 482     if (datecomp)
 483         free (datecomp);
 484
 485     return (1);
 486 }
 487
 488 /*
 489  * sort on dates.
 490  */
 491 static int
 492 dsort (struct smsg **a, struct smsg **b)
 493 {
 494     if ((*a)->s_clock < (*b)->s_clock)
 495         return (-1);
 496     else if ((*a)->s_clock > (*b)->s_clock)
 497         return (1);
 498     else if ((*a)->s_msg < (*b)->s_msg)
 499         return (-1);
 500     else
 501         return (1);
 502 }
 503
 504 /*
 505  * sort on subjects.
 506  */
 507 static int
 508 subsort (struct smsg **a, struct smsg **b)
 509 {
 510     register int i;
 511
 512     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 513         return (i);
 514
 515     return (dsort (a, b));
 516 }
 517
 518 static int
 519 txtsort (struct smsg **a, struct smsg **b)
 520 {
 521     register int i;
 522
 523     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 524         return (i);
 525     else if ((*a)->s_msg < (*b)->s_msg)
 526         return (-1);
 527     else
 528         return (1);
 529 }
 530
 531 static void
 532 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 533 {
 534     int nxt, old, new;
 535     char *newname, oldname[BUFSIZ];
 536     char newbuf[PATH_MAX + 1];
 537
 538     for (;;) {
 539         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 540         mlist[msg] = (struct smsg *)0;
 541         old = smsgs[nxt].s_msg;
 542         new = smsgs[msg].s_msg;
 543         strncpy (oldname, m_name (old), sizeof(oldname));
 544         newname = m_name (new);
 545         if (verbose)
 546             printf ("message %d becomes message %d\n", old, new);
 547
 548         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 549         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 550         ext_hook("ref-hook", oldname, newbuf);
 551
 552         if (rename (oldname, newname) == NOTOK)
 553             adios (newname, "unable to rename %s to", oldname);
 554
 555         copy_msg_flags (mp, new, old);
 556         if (mp->curmsg == old)
 557             seq_setcur (mp, new);
 558
 559         if (nxt == endmsg)
 560             break;
 561
 562         msg = nxt;
 563     }
 564 /*      if (nxt != endmsg); */
 565 /*      rename_chain (mp, mlist, nxt, endmsg); */
 566 }
 567
 568 static void
 569 rename_msgs (struct msgs *mp, struct smsg **mlist)
 570 {
 571     int i, j, old, new;
 572     seqset_t tmpset;
 573     char f1[BUFSIZ], tmpfil[BUFSIZ];
 574     char newbuf[PATH_MAX + 1];
 575     struct smsg *sp;
 576
 577     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 578
 579     for (i = 0; i < nmsgs; i++) {
 580         if (! (sp = mlist[i]))
 581             continue;   /* did this one */
 582
 583         j = sp - smsgs;
 584         if (j == i)
 585             continue;   /* this one doesn't move */
 586
 587         /*
 588          * the guy that was msg j is about to become msg i.
 589          * rename 'j' to make a hole, then recursively rename
 590          * guys to fill up the hole.
 591          */
 592         old = smsgs[j].s_msg;
 593         new = smsgs[i].s_msg;
 594         strncpy (f1, m_name (old), sizeof(f1));
 595
 596         if (verbose)
 597             printf ("renaming message chain from %d to %d\n", old, new);
 598
 599         /*
 600          *      Run the external hook to refile the old message as the
 601          *      temporary message number that is off of the end of the
 602          *      messages in the folder.
 603          */
 604
 605         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 606         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 607         ext_hook("ref-hook", f1, newbuf);
 608
 609         if (rename (f1, tmpfil) == NOTOK)
 610             adios (tmpfil, "unable to rename %s to ", f1);
 611
 612         get_msg_flags (mp, &tmpset, old);
 613
 614         rename_chain (mp, mlist, j, i);
 615
 616         /*
 617          *      Run the external hook to refile the temorary message number
 618          *      to the real place.
 619          */
 620
 621         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 622         ext_hook("ref-hook", newbuf, f1);
 623
 624         if (rename (tmpfil, m_name(new)) == NOTOK)
 625             adios (m_name(new), "unable to rename %s to", tmpfil);
 626
 627         set_msg_flags (mp, &tmpset, new);
 628         mp->msgflags |= SEQMOD;
 629     }
 630 }