diplodocus.org Git - nmh/blob - uip/sortm.c

   1 /* sortm.c -- sort messages in a folder by date/time
   2  *
   3  * This code is Copyright (c) 2002, by the authors of nmh.  See the
   4  * COPYRIGHT file in the root directory of the nmh distribution for
   5  * complete copyright information.
   6  */
   7
   8 #include <h/mh.h>
   9 #include <h/tws.h>
  10 #include <h/utils.h>
  11 #include "sbr/m_maildir.h"
  12
  13 #define SORTM_SWITCHES \
  14     X("datefield field", 0, DATESW) \
  15     X("textfield field", 0, TEXTSW) \
  16     X("notextfield", 0, NSUBJSW) \
  17     X("subject", -3, SUBJSW) /* backward-compatibility */ \
  18     X("limit days", 0, LIMSW) \
  19     X("nolimit", 0, NLIMSW) \
  20     X("verbose", 0, VERBSW) \
  21     X("noverbose", 0, NVERBSW) \
  22     X("all", 0, ALLMSGS) \
  23     X("noall", 0, NALLMSGS) \
  24     X("check", 0, CHECKSW) \
  25     X("nocheck", 0, NCHECKSW) \
  26     X("version", 0, VERSIONSW) \
  27     X("help", 0, HELPSW) \
  28
  29 #define X(sw, minchars, id) id,
  30 DEFINE_SWITCH_ENUM(SORTM);
  31 #undef X
  32
  33 #define X(sw, minchars, id) { sw, minchars, id },
  34 DEFINE_SWITCH_ARRAY(SORTM, switches);
  35 #undef X
  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;                 /* sort on subject if != 0 */
  47 time_t datelimit = 0;
  48 int submajor = 0;               /* if true, sort on subject-major */
  49 int verbose;
  50 int allmsgs = 1;
  51 int check_failed = 0;
  52
  53 /* This keeps compiler happy on calls to qsort */
  54 typedef int (*qsort_comp) (const void *, const void *);
  55
  56 /*
  57  * static prototypes
  58  */
  59 static int read_hdrs (struct msgs *, char *);
  60 static int get_fields (char *, int, struct smsg *);
  61 static int dsort (struct smsg **, struct smsg **);
  62 static int subsort (struct smsg **, struct smsg **);
  63 static int txtsort (struct smsg **, struct smsg **);
  64 static void rename_chain (struct msgs *, struct smsg **, int, int);
  65 static void rename_msgs (struct msgs *, struct smsg **);
  66
  67
  68 int
  69 main (int argc, char **argv)
  70 {
  71     int i, msgnum;
  72     char *cp, *maildir, *datesw = NULL;
  73     char *folder = NULL, buf[BUFSIZ], **argp;
  74     char **arguments;
  75     struct msgs_array msgs = { 0, 0, NULL };
  76     struct msgs *mp;
  77     struct smsg **dlist;
  78     int checksw = 0;
  79
  80     if (nmh_init(argv[0], 1)) { return 1; }
  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]",
  99                         invo_name);
 100                 print_help (buf, switches, 1);
 101                 done (0);
 102             case VERSIONSW:
 103                 print_version(invo_name);
 104                 done (0);
 105
 106             case DATESW:
 107                 if (datesw)
 108                     adios (NULL, "only one date field at a time");
 109                 if (!(datesw = *argp++) || *datesw == '-')
 110                     adios (NULL, "missing argument to %s", 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", argp[-2]);
 118                 continue;
 119
 120             case SUBJSW:
 121                 subjsort = "subject";
 122                 continue;
 123             case NSUBJSW:
 124                 subjsort = NULL;
 125                 continue;
 126
 127             case LIMSW:
 128                 if (!(cp = *argp++) || *cp == '-')
 129                         adios (NULL, "missing argument to %s", argp[-2]);
 130                 while (*cp == '0')
 131                     cp++;               /* skip any leading zeros */
 132                 if (!*cp) {             /* hit end of string */
 133                     submajor++;         /* sort subject-major */
 134                     continue;
 135                 }
 136                 if (!isdigit((unsigned char) *cp) || !(datelimit = atoi(cp)))
 137                     adios (NULL, "impossible limit %s", cp);
 138                 datelimit *= 60*60*24;
 139                 continue;
 140             case NLIMSW:
 141                 submajor = 0;   /* use date-major, but */
 142                 datelimit = 0;  /* use no limit */
 143                 continue;
 144
 145             case VERBSW:
 146                 verbose++;
 147                 continue;
 148             case NVERBSW:
 149                 verbose = 0;
 150                 continue;
 151
 152             case ALLMSGS:
 153                 allmsgs = 1;
 154                 continue;
 155             case NALLMSGS:
 156                 allmsgs = 0;
 157                 continue;
 158
 159             case CHECKSW:
 160                 checksw = 1;
 161                 continue;
 162             case NCHECKSW:
 163                 checksw = 0;
 164                 continue;
 165             }
 166         }
 167         if (*cp == '+' || *cp == '@') {
 168             if (folder)
 169                 adios (NULL, "only one folder at a time!");
 170             folder = pluspath (cp);
 171         } else
 172             app_msgarg(&msgs, cp);
 173     }
 174
 175     if (!context_find ("path"))
 176         free (path ("./", TFOLDER));
 177     if (!msgs.size) {
 178         if (allmsgs) {
 179             app_msgarg(&msgs, "all");
 180         } else {
 181             adios (NULL, "must specify messages to sort with -noall");
 182         }
 183     }
 184     if (!datesw)
 185         datesw = "date";
 186     if (!folder)
 187         folder = getfolder (1);
 188     maildir = m_maildir (folder);
 189
 190     if (chdir (maildir) == NOTOK)
 191         adios (maildir, "unable to change directory to");
 192
 193     /* read folder and create message structure */
 194     if (!(mp = folder_read (folder, 1)))
 195         adios (NULL, "unable to read folder %s", folder);
 196
 197     /* check for empty folder */
 198     if (mp->nummsg == 0)
 199         adios (NULL, "no messages in %s", folder);
 200
 201     /* parse all the message ranges/sequences and set SELECTED */
 202     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 203         if (!m_convert (mp, msgs.msgs[msgnum]))
 204             done (1);
 205     seq_setprev (mp);   /* set the previous sequence */
 206
 207     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 208         adios (NULL, "no messages to sort");
 209
 210     if (checksw  &&  check_failed) {
 211         adios (NULL, "errors found, no messages sorted");
 212     }
 213
 214     /*
 215      * sort a list of pointers to our "messages to be sorted".
 216      */
 217     dlist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 218     for (i = 0; i < nmsgs; i++)
 219         dlist[i] = &smsgs[i];
 220     dlist[nmsgs] = 0;
 221
 222     if (verbose) {      /* announce what we're doing */
 223         if (subjsort)
 224             if (submajor)
 225                 printf ("sorting by %s\n", subjsort);
 226             else
 227                 printf ("sorting by %s-major %s-minor\n", subjsort, datesw);
 228         else
 229             printf ("sorting by datefield %s\n", datesw);
 230     }
 231
 232     /* first sort by date, or by subject-major, date-minor */
 233     qsort ((char *) dlist, nmsgs, sizeof(*dlist),
 234             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 235
 236     /*
 237      * if we're sorting on subject, we need another list
 238      * in subject order, then a merge pass to collate the
 239      * two sorts.
 240      */
 241     if (!submajor && subjsort) {        /* already date sorted */
 242         struct smsg **slist, **flist;
 243         struct smsg ***il, **fp, **dp;
 244
 245         slist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 246         memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 247         qsort((char *)slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 248
 249         /*
 250          * make an inversion list so we can quickly find
 251          * the collection of messages with the same subj
 252          * given a message number.
 253          */
 254         il = mh_xcalloc(mp->hghsel + 1, sizeof *il);
 255         for (i = 0; i < nmsgs; i++)
 256             il[slist[i]->s_msg] = &slist[i];
 257         /*
 258          * make up the final list, chronological but with
 259          * all the same subjects grouped together.
 260          */
 261         flist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 262         fp = flist;
 263         for (dp = dlist; *dp;) {
 264             struct smsg **s = il[(*dp++)->s_msg];
 265
 266             /* see if we already did this guy */
 267             if (! s)
 268                 continue;
 269
 270             *fp++ = *s++;
 271             /*
 272              * take the next message(s) if there is one,
 273              * its subject isn't null and its subject
 274              * is the same as this one and it's not too
 275              * far away in time.
 276              */
 277             while (*s && (*s)->s_subj[0] &&
 278                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 279                    (datelimit == 0 ||
 280                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 281                 il[(*s)->s_msg] = 0;
 282                 *fp++ = *s++;
 283             }
 284         }
 285         *fp = 0;
 286         free (il);
 287         free (slist);
 288         free (dlist);
 289         dlist = flist;
 290     }
 291
 292     /*
 293      * At this point, dlist is a sorted array of pointers to smsg structures,
 294      * each of which contains a message number.
 295      */
 296
 297     rename_msgs (mp, dlist);
 298
 299     context_replace (pfolder, folder);  /* update current folder         */
 300     seq_save (mp);                      /* synchronize message sequences */
 301     context_save ();                    /* save the context file         */
 302     folder_free (mp);                   /* free folder/message structure */
 303     done (0);
 304     return 1;
 305 }
 306
 307 static int
 308 read_hdrs (struct msgs *mp, char *datesw)
 309 {
 310     int msgnum;
 311     struct smsg *s;
 312
 313     smsgs = mh_xcalloc(mp->hghsel - mp->lowsel + 2, sizeof *smsgs);
 314     s = smsgs;
 315     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 316         if (is_selected(mp, msgnum)) {
 317             if (get_fields (datesw, msgnum, s)) {
 318                 s->s_msg = msgnum;
 319                 s++;
 320             }
 321         }
 322     }
 323     s->s_msg = 0;
 324     return(s - smsgs);
 325 }
 326
 327
 328 /*
 329  * Parse the message and get the data or subject field,
 330  * if needed.
 331  */
 332
 333 static int
 334 get_fields (char *datesw, int msg, struct smsg *smsg)
 335 {
 336     int state;
 337     int compnum;
 338     char *msgnam, buf[NMH_BUFSIZ], nam[NAMESZ];
 339     struct tws *tw;
 340     char *datecomp = NULL, *subjcomp = NULL;
 341     FILE *in;
 342     m_getfld_state_t gstate;
 343
 344     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 345         admonish (msgnam, "unable to read message");
 346         return (0);
 347     }
 348     gstate = m_getfld_state_init(in);
 349     for (compnum = 1;;) {
 350         int bufsz = sizeof buf;
 351         switch (state = m_getfld2(&gstate, nam, buf, &bufsz)) {
 352         case FLD:
 353         case FLDPLUS:
 354             compnum++;
 355             if (!strcasecmp (nam, datesw)) {
 356                 datecomp = add (buf, datecomp);
 357                 while (state == FLDPLUS) {
 358                     bufsz = sizeof buf;
 359                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 360                     datecomp = add (buf, datecomp);
 361                 }
 362                 if (!subjsort || subjcomp)
 363                     break;
 364             } else if (subjsort && !strcasecmp (nam, subjsort)) {
 365                 subjcomp = add (buf, subjcomp);
 366                 while (state == FLDPLUS) {
 367                     bufsz = sizeof buf;
 368                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 369                     subjcomp = add (buf, subjcomp);
 370                 }
 371                 if (datecomp)
 372                     break;
 373             } else {
 374                 /* just flush this guy */
 375                 while (state == FLDPLUS) {
 376                     bufsz = sizeof buf;
 377                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 378                 }
 379             }
 380             continue;
 381
 382         case BODY:
 383         case FILEEOF:
 384             break;
 385
 386         case LENERR:
 387         case FMTERR:
 388             if (state == LENERR || state == FMTERR) {
 389                 inform("format error in message %d (header #%d), continuing...",
 390                       msg, compnum);
 391                 check_failed = 1;
 392             }
 393             free(datecomp);
 394             free(subjcomp);
 395             fclose (in);
 396             return (0);
 397
 398         default:
 399             adios (NULL, "internal error -- you lose");
 400         }
 401         break;
 402     }
 403     m_getfld_state_destroy (&gstate);
 404
 405     /*
 406      * If no date component, then use the modification
 407      * time of the file as its date
 408      */
 409     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 410         struct stat st;
 411
 412         inform("can't parse %s field in message %d, "
 413             "will use file modification time", datesw, msg);
 414         fstat (fileno (in), &st);
 415         smsg->s_clock = st.st_mtime;
 416         check_failed = 1;
 417     } else {
 418         smsg->s_clock = dmktime (tw);
 419     }
 420
 421     if (subjsort) {
 422         if (subjcomp) {
 423             /*
 424              * try to make the subject "canonical": delete
 425              * leading "re:", everything but letters & smash
 426              * letters to lower case.
 427              */
 428             char  *cp, *cp2, c;
 429
 430             cp = subjcomp;
 431             cp2 = subjcomp;
 432             if (strcmp (subjsort, "subject") == 0) {
 433                 while ((c = *cp)) {
 434                     if (! isspace((unsigned char) c)) {
 435                         if(!uprf(cp, "re:"))
 436                             break;
 437                         cp += 2;
 438                     }
 439                     cp++;
 440                 }
 441             }
 442
 443             while ((c = *cp++)) {
 444                 if (isascii((unsigned char) c) && isalnum((unsigned char) c))
 445                     *cp2++ = tolower((unsigned char)c);
 446             }
 447
 448             *cp2 = '\0';
 449         }
 450         else
 451             subjcomp = "";
 452
 453         smsg->s_subj = subjcomp;
 454     }
 455     fclose (in);
 456     free(datecomp);
 457
 458     return (1);
 459 }
 460
 461 /*
 462  * sort on dates.
 463  */
 464 static int
 465 dsort (struct smsg **a, struct smsg **b)
 466 {
 467     if ((*a)->s_clock < (*b)->s_clock)
 468         return (-1);
 469     if ((*a)->s_clock > (*b)->s_clock)
 470         return (1);
 471     if ((*a)->s_msg < (*b)->s_msg)
 472         return (-1);
 473     return (1);
 474 }
 475
 476 /*
 477  * sort on subjects.
 478  */
 479 static int
 480 subsort (struct smsg **a, struct smsg **b)
 481 {
 482     int i;
 483
 484     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 485         return (i);
 486
 487     return (dsort (a, b));
 488 }
 489
 490 static int
 491 txtsort (struct smsg **a, struct smsg **b)
 492 {
 493     int i;
 494
 495     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 496         return (i);
 497     if ((*a)->s_msg < (*b)->s_msg)
 498         return (-1);
 499     return (1);
 500 }
 501
 502 static void
 503 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 504 {
 505     int nxt, old, new;
 506     char *newname, oldname[BUFSIZ];
 507     char newbuf[PATH_MAX + 1];
 508
 509     for (;;) {
 510         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 511         mlist[msg] = NULL;
 512         old = smsgs[nxt].s_msg;
 513         new = smsgs[msg].s_msg;
 514         strncpy (oldname, m_name (old), sizeof(oldname));
 515         newname = m_name (new);
 516         if (verbose)
 517             printf ("message %d becomes message %d\n", old, new);
 518
 519         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 520         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 521         ext_hook("ref-hook", oldname, newbuf);
 522
 523         if (rename (oldname, newname) == NOTOK)
 524             adios (newname, "unable to rename %s to", oldname);
 525
 526         copy_msg_flags (mp, new, old);
 527         if (mp->curmsg == old)
 528             seq_setcur (mp, new);
 529
 530         if (nxt == endmsg)
 531             break;
 532
 533         msg = nxt;
 534     }
 535 /*      if (nxt != endmsg); */
 536 /*      rename_chain (mp, mlist, nxt, endmsg); */
 537 }
 538
 539 static void
 540 rename_msgs (struct msgs *mp, struct smsg **mlist)
 541 {
 542     int i, j, old, new;
 543     bvector_t tmpset = bvector_create ();
 544     char f1[BUFSIZ], tmpfil[BUFSIZ];
 545     char newbuf[PATH_MAX + 1];
 546     struct smsg *sp;
 547
 548     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 549
 550     for (i = 0; i < nmsgs; i++) {
 551         if (! (sp = mlist[i]))
 552             continue;   /* did this one */
 553
 554         j = sp - smsgs;
 555         if (j == i)
 556             continue;   /* this one doesn't move */
 557
 558         /*
 559          * the guy that was msg j is about to become msg i.
 560          * rename 'j' to make a hole, then recursively rename
 561          * guys to fill up the hole.
 562          */
 563         old = smsgs[j].s_msg;
 564         new = smsgs[i].s_msg;
 565         strncpy (f1, m_name (old), sizeof(f1));
 566
 567         if (verbose)
 568             printf ("renaming message chain from %d to %d\n", old, new);
 569
 570         /*
 571          *      Run the external hook to refile the old message as the
 572          *      temporary message number that is off of the end of the
 573          *      messages in the folder.
 574          */
 575
 576         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 577         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 578         ext_hook("ref-hook", f1, newbuf);
 579
 580         if (rename (f1, tmpfil) == NOTOK)
 581             adios (tmpfil, "unable to rename %s to ", f1);
 582
 583         get_msg_flags (mp, tmpset, old);
 584
 585         rename_chain (mp, mlist, j, i);
 586
 587         /*
 588          *      Run the external hook to refile the temporary message number
 589          *      to the real place.
 590          */
 591
 592         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 593         ext_hook("ref-hook", newbuf, f1);
 594
 595         if (rename (tmpfil, m_name(new)) == NOTOK)
 596             adios (m_name(new), "unable to rename %s to", tmpfil);
 597
 598         set_msg_flags (mp, tmpset, new);
 599         mp->msgflags |= SEQMOD;
 600     }
 601
 602     bvector_free (tmpset);
 603 }