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     unsigned char *cp;
  74     char *maildir, *datesw = NULL;
  75     char *folder = NULL, buf[BUFSIZ], **argp;
  76     char **arguments;
  77     struct msgs_array msgs = { 0, 0, NULL };
  78     struct msgs *mp;
  79     struct smsg **dlist;
  80     int checksw = 0;
  81
  82 #ifdef LOCALE
  83     setlocale(LC_ALL, "");
  84 #endif
  85     invo_name = r1bindex (argv[0], '/');
  86
  87     /* read user profile/context */
  88     context_read();
  89
  90     arguments = getarguments (invo_name, argc, argv, 1);
  91     argp = arguments;
  92
  93     /*
  94      * Parse arguments
  95      */
  96     while ((cp = *argp++)) {
  97         if (*cp == '-') {
  98             switch (smatch (++cp, switches)) {
  99             case AMBIGSW:
 100                 ambigsw (cp, switches);
 101                 done (1);
 102             case UNKWNSW:
 103                 adios (NULL, "-%s unknown", cp);
 104
 105             case HELPSW:
 106                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 107                         invo_name);
 108                 print_help (buf, switches, 1);
 109                 done (0);
 110             case VERSIONSW:
 111                 print_version(invo_name);
 112                 done (0);
 113
 114             case DATESW:
 115                 if (datesw)
 116                     adios (NULL, "only one date field at a time");
 117                 if (!(datesw = *argp++) || *datesw == '-')
 118                     adios (NULL, "missing argument to %s", argp[-2]);
 119                 continue;
 120
 121             case TEXTSW:
 122                 if (subjsort)
 123                     adios (NULL, "only one text field at a time");
 124                 if (!(subjsort = *argp++) || *subjsort == '-')
 125                     adios (NULL, "missing argument to %s", argp[-2]);
 126                 continue;
 127
 128             case SUBJSW:
 129                 subjsort = "subject";
 130                 continue;
 131             case NSUBJSW:
 132                 subjsort = (char *)0;
 133                 continue;
 134
 135             case LIMSW:
 136                 if (!(cp = *argp++) || *cp == '-')
 137                         adios (NULL, "missing argument to %s", argp[-2]);
 138                 while (*cp == '0')
 139                     cp++;               /* skip any leading zeros */
 140                 if (!*cp) {             /* hit end of string */
 141                     submajor++;         /* sort subject-major */
 142                     continue;
 143                 }
 144                 if (!isdigit(*cp) || !(datelimit = atoi(cp)))
 145                     adios (NULL, "impossible limit %s", cp);
 146                 datelimit *= 60*60*24;
 147                 continue;
 148             case NLIMSW:
 149                 submajor = 0;   /* use date-major, but */
 150                 datelimit = 0;  /* use no limit */
 151                 continue;
 152
 153             case VERBSW:
 154                 verbose++;
 155                 continue;
 156             case NVERBSW:
 157                 verbose = 0;
 158                 continue;
 159
 160             case ALLMSGS:
 161                 allmsgs = 1;
 162                 continue;
 163             case NALLMSGS:
 164                 allmsgs = 0;
 165                 continue;
 166
 167             case CHECKSW:
 168                 checksw = 1;
 169                 continue;
 170             case NCHECKSW:
 171                 checksw = 0;
 172                 continue;
 173             }
 174         }
 175         if (*cp == '+' || *cp == '@') {
 176             if (folder)
 177                 adios (NULL, "only one folder at a time!");
 178             else
 179                 folder = pluspath (cp);
 180         } else
 181                 app_msgarg(&msgs, cp);
 182     }
 183
 184     if (!context_find ("path"))
 185         free (path ("./", TFOLDER));
 186     if (!msgs.size) {
 187         if (allmsgs) {
 188             app_msgarg(&msgs, "all");
 189         } else {
 190             adios (NULL, "must specify messages to sort with -noall");
 191         }
 192     }
 193     if (!datesw)
 194         datesw = "date";
 195     if (!folder)
 196         folder = getfolder (1);
 197     maildir = m_maildir (folder);
 198
 199     if (chdir (maildir) == NOTOK)
 200         adios (maildir, "unable to change directory to");
 201
 202     /* read folder and create message structure */
 203     if (!(mp = folder_read (folder)))
 204         adios (NULL, "unable to read folder %s", folder);
 205
 206     /* check for empty folder */
 207     if (mp->nummsg == 0)
 208         adios (NULL, "no messages in %s", folder);
 209
 210     /* parse all the message ranges/sequences and set SELECTED */
 211     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 212         if (!m_convert (mp, msgs.msgs[msgnum]))
 213             done (1);
 214     seq_setprev (mp);   /* set the previous sequence */
 215
 216     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 217         adios (NULL, "no messages to sort");
 218
 219     if (checksw  &&  check_failed) {
 220         adios (NULL, "errors found, no messages sorted");
 221     }
 222
 223     /*
 224      * sort a list of pointers to our "messages to be sorted".
 225      */
 226     dlist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 227     for (i = 0; i < nmsgs; i++)
 228         dlist[i] = &smsgs[i];
 229     dlist[nmsgs] = 0;
 230
 231     if (verbose) {      /* announce what we're doing */
 232         if (subjsort)
 233             if (submajor)
 234                 printf ("sorting by %s\n", subjsort);
 235             else
 236                 printf ("sorting by %s-major %s-minor\n", subjsort, datesw);
 237         else
 238             printf ("sorting by datefield %s\n", datesw);
 239     }
 240
 241     /* first sort by date, or by subject-major, date-minor */
 242     qsort ((char *) dlist, nmsgs, sizeof(*dlist),
 243             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 244
 245     /*
 246      * if we're sorting on subject, we need another list
 247      * in subject order, then a merge pass to collate the
 248      * two sorts.
 249      */
 250     if (!submajor && subjsort) {        /* already date sorted */
 251         struct smsg **slist, **flist;
 252         register struct smsg ***il, **fp, **dp;
 253
 254         slist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 255         memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 256         qsort((char *)slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 257
 258         /*
 259          * make an inversion list so we can quickly find
 260          * the collection of messages with the same subj
 261          * given a message number.
 262          */
 263         il = (struct smsg ***) calloc (mp->hghsel+1, sizeof(*il));
 264         if (! il)
 265             adios (NULL, "couldn't allocate msg list");
 266         for (i = 0; i < nmsgs; i++)
 267             il[slist[i]->s_msg] = &slist[i];
 268         /*
 269          * make up the final list, chronological but with
 270          * all the same subjects grouped together.
 271          */
 272         flist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 273         fp = flist;
 274         for (dp = dlist; *dp;) {
 275             register struct smsg **s = il[(*dp++)->s_msg];
 276
 277             /* see if we already did this guy */
 278             if (! s)
 279                 continue;
 280
 281             *fp++ = *s++;
 282             /*
 283              * take the next message(s) if there is one,
 284              * its subject isn't null and its subject
 285              * is the same as this one and it's not too
 286              * far away in time.
 287              */
 288             while (*s && (*s)->s_subj[0] &&
 289                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 290                    (datelimit == 0 ||
 291                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 292                 il[(*s)->s_msg] = 0;
 293                 *fp++ = *s++;
 294             }
 295         }
 296         *fp = 0;
 297         free (slist);
 298         free (dlist);
 299         dlist = flist;
 300     }
 301
 302     /*
 303      * At this point, dlist is a sorted array of pointers to smsg structures,
 304      * each of which contains a message number.
 305      */
 306
 307     rename_msgs (mp, dlist);
 308
 309     context_replace (pfolder, folder);  /* update current folder         */
 310     seq_save (mp);                      /* synchronize message sequences */
 311     context_save ();                    /* save the context file         */
 312     folder_free (mp);                   /* free folder/message structure */
 313     done (0);
 314     return 1;
 315 }
 316
 317 static int
 318 read_hdrs (struct msgs *mp, char *datesw)
 319 {
 320     int msgnum;
 321     struct tws tb;
 322     register struct smsg *s;
 323
 324     twscopy (&tb, dlocaltimenow ());
 325
 326     smsgs = (struct smsg *)
 327         calloc ((size_t) (mp->hghsel - mp->lowsel + 2),
 328             sizeof(*smsgs));
 329     if (smsgs == NULL)
 330         adios (NULL, "unable to allocate sort storage");
 331
 332     s = smsgs;
 333     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 334         if (is_selected(mp, msgnum)) {
 335             if (get_fields (datesw, msgnum, s)) {
 336                 s->s_msg = msgnum;
 337                 s++;
 338             }
 339         }
 340     }
 341     s->s_msg = 0;
 342     return(s - smsgs);
 343 }
 344
 345
 346 /*
 347  * Parse the message and get the data or subject field,
 348  * if needed.
 349  */
 350
 351 static int
 352 get_fields (char *datesw, int msg, struct smsg *smsg)
 353 {
 354     register int state;
 355     int compnum;
 356     char *msgnam, buf[BUFSIZ], nam[NAMESZ];
 357     register struct tws *tw;
 358     register char *datecomp = NULL, *subjcomp = NULL;
 359     register FILE *in;
 360     m_getfld_state_t gstate = 0;
 361
 362     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 363         admonish (msgnam, "unable to read message");
 364         return (0);
 365     }
 366     for (compnum = 1;;) {
 367         int bufsz = sizeof buf;
 368         switch (state = m_getfld (&gstate, nam, 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, nam, buf, &bufsz, in);
 377                     datecomp = add (buf, datecomp);
 378                 }
 379                 if (!subjsort || subjcomp)
 380                     break;
 381             } else if (subjsort && !mh_strcasecmp (nam, subjsort)) {
 382                 subjcomp = add (buf, subjcomp);
 383                 while (state == FLDPLUS) {
 384                     bufsz = sizeof buf;
 385                     state = m_getfld (&gstate, nam, buf, &bufsz, in);
 386                     subjcomp = add (buf, subjcomp);
 387                 }
 388                 if (datecomp)
 389                     break;
 390             } else {
 391                 /* just flush this guy */
 392                 while (state == FLDPLUS) {
 393                     bufsz = sizeof buf;
 394                     state = m_getfld (&gstate, nam, buf, &bufsz, in);
 395                 }
 396             }
 397             continue;
 398
 399         case BODY:
 400         case FILEEOF:
 401             break;
 402
 403         case LENERR:
 404         case FMTERR:
 405             if (state == LENERR || state == FMTERR) {
 406                 admonish (NULL, "format error in message %d (header #%d)",
 407                       msg, compnum);
 408                 check_failed = 1;
 409             }
 410             if (datecomp)
 411                 free (datecomp);
 412             if (subjcomp)
 413                 free (subjcomp);
 414             fclose (in);
 415             return (0);
 416
 417         default:
 418             adios (NULL, "internal error -- you lose");
 419         }
 420         break;
 421     }
 422     m_getfld_state_destroy (&gstate);
 423
 424     /*
 425      * If no date component, then use the modification
 426      * time of the file as its date
 427      */
 428     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 429         struct stat st;
 430
 431         advise (NULL,
 432                 "can't parse %s field in message %d, "
 433                 "will use file modification time",
 434                 datesw, msg);
 435         fstat (fileno (in), &st);
 436         smsg->s_clock = st.st_mtime;
 437         check_failed = 1;
 438     } else {
 439         smsg->s_clock = dmktime (tw);
 440     }
 441
 442     if (subjsort) {
 443         if (subjcomp) {
 444             /*
 445              * try to make the subject "canonical": delete
 446              * leading "re:", everything but letters & smash
 447              * letters to lower case.
 448              */
 449             register char  *cp, *cp2;
 450             register unsigned char c;
 451
 452             cp = subjcomp;
 453             cp2 = subjcomp;
 454             if (strcmp (subjsort, "subject") == 0) {
 455                 while ((c = *cp)) {
 456                     if (! isspace(c)) {
 457                         if(uprf(cp, "re:"))
 458                             cp += 2;
 459                         else
 460                             break;
 461                     }
 462                     cp++;
 463                 }
 464             }
 465
 466             while ((c = *cp++)) {
 467                 if (isalnum(c))
 468                     *cp2++ = isupper(c) ? tolower(c) : c;
 469             }
 470
 471             *cp2 = '\0';
 472         }
 473         else
 474             subjcomp = "";
 475
 476         smsg->s_subj = subjcomp;
 477     }
 478     fclose (in);
 479     if (datecomp)
 480         free (datecomp);
 481
 482     return (1);
 483 }
 484
 485 /*
 486  * sort on dates.
 487  */
 488 static int
 489 dsort (struct smsg **a, struct smsg **b)
 490 {
 491     if ((*a)->s_clock < (*b)->s_clock)
 492         return (-1);
 493     else if ((*a)->s_clock > (*b)->s_clock)
 494         return (1);
 495     else if ((*a)->s_msg < (*b)->s_msg)
 496         return (-1);
 497     else
 498         return (1);
 499 }
 500
 501 /*
 502  * sort on subjects.
 503  */
 504 static int
 505 subsort (struct smsg **a, struct smsg **b)
 506 {
 507     register int i;
 508
 509     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 510         return (i);
 511
 512     return (dsort (a, b));
 513 }
 514
 515 static int
 516 txtsort (struct smsg **a, struct smsg **b)
 517 {
 518     register int i;
 519
 520     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 521         return (i);
 522     else if ((*a)->s_msg < (*b)->s_msg)
 523         return (-1);
 524     else
 525         return (1);
 526 }
 527
 528 static void
 529 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 530 {
 531     int nxt, old, new;
 532     char *newname, oldname[BUFSIZ];
 533     char newbuf[PATH_MAX + 1];
 534
 535     for (;;) {
 536         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 537         mlist[msg] = (struct smsg *)0;
 538         old = smsgs[nxt].s_msg;
 539         new = smsgs[msg].s_msg;
 540         strncpy (oldname, m_name (old), sizeof(oldname));
 541         newname = m_name (new);
 542         if (verbose)
 543             printf ("message %d becomes message %d\n", old, new);
 544
 545         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 546         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 547         ext_hook("ref-hook", oldname, newbuf);
 548
 549         if (rename (oldname, newname) == NOTOK)
 550             adios (newname, "unable to rename %s to", oldname);
 551
 552         copy_msg_flags (mp, new, old);
 553         if (mp->curmsg == old)
 554             seq_setcur (mp, new);
 555
 556         if (nxt == endmsg)
 557             break;
 558
 559         msg = nxt;
 560     }
 561 /*      if (nxt != endmsg); */
 562 /*      rename_chain (mp, mlist, nxt, endmsg); */
 563 }
 564
 565 static void
 566 rename_msgs (struct msgs *mp, struct smsg **mlist)
 567 {
 568     int i, j, old, new;
 569     seqset_t tmpset;
 570     char f1[BUFSIZ], tmpfil[BUFSIZ];
 571     char newbuf[PATH_MAX + 1];
 572     struct smsg *sp;
 573
 574     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 575
 576     for (i = 0; i < nmsgs; i++) {
 577         if (! (sp = mlist[i]))
 578             continue;   /* did this one */
 579
 580         j = sp - smsgs;
 581         if (j == i)
 582             continue;   /* this one doesn't move */
 583
 584         /*
 585          * the guy that was msg j is about to become msg i.
 586          * rename 'j' to make a hole, then recursively rename
 587          * guys to fill up the hole.
 588          */
 589         old = smsgs[j].s_msg;
 590         new = smsgs[i].s_msg;
 591         strncpy (f1, m_name (old), sizeof(f1));
 592
 593         if (verbose)
 594             printf ("renaming message chain from %d to %d\n", old, new);
 595
 596         /*
 597          *      Run the external hook to refile the old message as the
 598          *      temporary message number that is off of the end of the
 599          *      messages in the folder.
 600          */
 601
 602         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 603         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 604         ext_hook("ref-hook", f1, newbuf);
 605
 606         if (rename (f1, tmpfil) == NOTOK)
 607             adios (tmpfil, "unable to rename %s to ", f1);
 608
 609         get_msg_flags (mp, &tmpset, old);
 610
 611         rename_chain (mp, mlist, j, i);
 612
 613         /*
 614          *      Run the external hook to refile the temorary message number
 615          *      to the real place.
 616          */
 617
 618         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 619         ext_hook("ref-hook", newbuf, f1);
 620
 621         if (rename (tmpfil, m_name(new)) == NOTOK)
 622             adios (m_name(new), "unable to rename %s to", tmpfil);
 623
 624         set_msg_flags (mp, &tmpset, new);
 625         mp->msgflags |= SEQMOD;
 626     }
 627 }