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     if (nmh_init(argv[0], 1)) { return 1; }
  82
  83     arguments = getarguments (invo_name, argc, argv, 1);
  84     argp = arguments;
  85
  86     /*
  87      * Parse arguments
  88      */
  89     while ((cp = *argp++)) {
  90         if (*cp == '-') {
  91             switch (smatch (++cp, switches)) {
  92             case AMBIGSW:
  93                 ambigsw (cp, switches);
  94                 done (1);
  95             case UNKWNSW:
  96                 adios (NULL, "-%s unknown", cp);
  97
  98             case HELPSW:
  99                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 100                         invo_name);
 101                 print_help (buf, switches, 1);
 102                 done (0);
 103             case VERSIONSW:
 104                 print_version(invo_name);
 105                 done (0);
 106
 107             case DATESW:
 108                 if (datesw)
 109                     adios (NULL, "only one date field at a time");
 110                 if (!(datesw = *argp++) || *datesw == '-')
 111                     adios (NULL, "missing argument to %s", argp[-2]);
 112                 continue;
 113
 114             case TEXTSW:
 115                 if (subjsort)
 116                     adios (NULL, "only one text field at a time");
 117                 if (!(subjsort = *argp++) || *subjsort == '-')
 118                     adios (NULL, "missing argument to %s", argp[-2]);
 119                 continue;
 120
 121             case SUBJSW:
 122                 subjsort = "subject";
 123                 continue;
 124             case NSUBJSW:
 125                 subjsort = (char *)0;
 126                 continue;
 127
 128             case LIMSW:
 129                 if (!(cp = *argp++) || *cp == '-')
 130                         adios (NULL, "missing argument to %s", argp[-2]);
 131                 while (*cp == '0')
 132                     cp++;               /* skip any leading zeros */
 133                 if (!*cp) {             /* hit end of string */
 134                     submajor++;         /* sort subject-major */
 135                     continue;
 136                 }
 137                 if (!isdigit((unsigned char) *cp) || !(datelimit = atoi(cp)))
 138                     adios (NULL, "impossible limit %s", cp);
 139                 datelimit *= 60*60*24;
 140                 continue;
 141             case NLIMSW:
 142                 submajor = 0;   /* use date-major, but */
 143                 datelimit = 0;  /* use no limit */
 144                 continue;
 145
 146             case VERBSW:
 147                 verbose++;
 148                 continue;
 149             case NVERBSW:
 150                 verbose = 0;
 151                 continue;
 152
 153             case ALLMSGS:
 154                 allmsgs = 1;
 155                 continue;
 156             case NALLMSGS:
 157                 allmsgs = 0;
 158                 continue;
 159
 160             case CHECKSW:
 161                 checksw = 1;
 162                 continue;
 163             case NCHECKSW:
 164                 checksw = 0;
 165                 continue;
 166             }
 167         }
 168         if (*cp == '+' || *cp == '@') {
 169             if (folder)
 170                 adios (NULL, "only one folder at a time!");
 171             else
 172                 folder = pluspath (cp);
 173         } else
 174                 app_msgarg(&msgs, cp);
 175     }
 176
 177     if (!context_find ("path"))
 178         free (path ("./", TFOLDER));
 179     if (!msgs.size) {
 180         if (allmsgs) {
 181             app_msgarg(&msgs, "all");
 182         } else {
 183             adios (NULL, "must specify messages to sort with -noall");
 184         }
 185     }
 186     if (!datesw)
 187         datesw = "date";
 188     if (!folder)
 189         folder = getfolder (1);
 190     maildir = m_maildir (folder);
 191
 192     if (chdir (maildir) == NOTOK)
 193         adios (maildir, "unable to change directory to");
 194
 195     /* read folder and create message structure */
 196     if (!(mp = folder_read (folder, 1)))
 197         adios (NULL, "unable to read folder %s", folder);
 198
 199     /* check for empty folder */
 200     if (mp->nummsg == 0)
 201         adios (NULL, "no messages in %s", folder);
 202
 203     /* parse all the message ranges/sequences and set SELECTED */
 204     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 205         if (!m_convert (mp, msgs.msgs[msgnum]))
 206             done (1);
 207     seq_setprev (mp);   /* set the previous sequence */
 208
 209     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 210         adios (NULL, "no messages to sort");
 211
 212     if (checksw  &&  check_failed) {
 213         adios (NULL, "errors found, no messages sorted");
 214     }
 215
 216     /*
 217      * sort a list of pointers to our "messages to be sorted".
 218      */
 219     dlist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 220     for (i = 0; i < nmsgs; i++)
 221         dlist[i] = &smsgs[i];
 222     dlist[nmsgs] = 0;
 223
 224     if (verbose) {      /* announce what we're doing */
 225         if (subjsort)
 226             if (submajor)
 227                 printf ("sorting by %s\n", subjsort);
 228             else
 229                 printf ("sorting by %s-major %s-minor\n", subjsort, datesw);
 230         else
 231             printf ("sorting by datefield %s\n", datesw);
 232     }
 233
 234     /* first sort by date, or by subject-major, date-minor */
 235     qsort ((char *) dlist, nmsgs, sizeof(*dlist),
 236             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 237
 238     /*
 239      * if we're sorting on subject, we need another list
 240      * in subject order, then a merge pass to collate the
 241      * two sorts.
 242      */
 243     if (!submajor && subjsort) {        /* already date sorted */
 244         struct smsg **slist, **flist;
 245         register struct smsg ***il, **fp, **dp;
 246
 247         slist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 248         memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 249         qsort((char *)slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 250
 251         /*
 252          * make an inversion list so we can quickly find
 253          * the collection of messages with the same subj
 254          * given a message number.
 255          */
 256         il = (struct smsg ***) calloc (mp->hghsel+1, sizeof(*il));
 257         if (! il)
 258             adios (NULL, "couldn't allocate msg list");
 259         for (i = 0; i < nmsgs; i++)
 260             il[slist[i]->s_msg] = &slist[i];
 261         /*
 262          * make up the final list, chronological but with
 263          * all the same subjects grouped together.
 264          */
 265         flist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 266         fp = flist;
 267         for (dp = dlist; *dp;) {
 268             register struct smsg **s = il[(*dp++)->s_msg];
 269
 270             /* see if we already did this guy */
 271             if (! s)
 272                 continue;
 273
 274             *fp++ = *s++;
 275             /*
 276              * take the next message(s) if there is one,
 277              * its subject isn't null and its subject
 278              * is the same as this one and it's not too
 279              * far away in time.
 280              */
 281             while (*s && (*s)->s_subj[0] &&
 282                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 283                    (datelimit == 0 ||
 284                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 285                 il[(*s)->s_msg] = 0;
 286                 *fp++ = *s++;
 287             }
 288         }
 289         *fp = 0;
 290         free (slist);
 291         free (dlist);
 292         dlist = flist;
 293     }
 294
 295     /*
 296      * At this point, dlist is a sorted array of pointers to smsg structures,
 297      * each of which contains a message number.
 298      */
 299
 300     rename_msgs (mp, dlist);
 301
 302     context_replace (pfolder, folder);  /* update current folder         */
 303     seq_save (mp);                      /* synchronize message sequences */
 304     context_save ();                    /* save the context file         */
 305     folder_free (mp);                   /* free folder/message structure */
 306     done (0);
 307     return 1;
 308 }
 309
 310 static int
 311 read_hdrs (struct msgs *mp, char *datesw)
 312 {
 313     int msgnum;
 314     struct tws tb;
 315     register struct smsg *s;
 316
 317     twscopy (&tb, dlocaltimenow ());
 318
 319     smsgs = (struct smsg *)
 320         calloc ((size_t) (mp->hghsel - mp->lowsel + 2),
 321             sizeof(*smsgs));
 322     if (smsgs == NULL)
 323         adios (NULL, "unable to allocate sort storage");
 324
 325     s = smsgs;
 326     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 327         if (is_selected(mp, msgnum)) {
 328             if (get_fields (datesw, msgnum, s)) {
 329                 s->s_msg = msgnum;
 330                 s++;
 331             }
 332         }
 333     }
 334     s->s_msg = 0;
 335     return(s - smsgs);
 336 }
 337
 338
 339 /*
 340  * Parse the message and get the data or subject field,
 341  * if needed.
 342  */
 343
 344 static int
 345 get_fields (char *datesw, int msg, struct smsg *smsg)
 346 {
 347     register int state;
 348     int compnum;
 349     char *msgnam, buf[BUFSIZ], nam[NAMESZ];
 350     register struct tws *tw;
 351     register char *datecomp = NULL, *subjcomp = NULL;
 352     register FILE *in;
 353     m_getfld_state_t gstate = 0;
 354
 355     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 356         admonish (msgnam, "unable to read message");
 357         return (0);
 358     }
 359     for (compnum = 1;;) {
 360         int bufsz = sizeof buf;
 361         switch (state = m_getfld (&gstate, nam, buf, &bufsz, in)) {
 362         case FLD:
 363         case FLDPLUS:
 364             compnum++;
 365             if (!strcasecmp (nam, datesw)) {
 366                 datecomp = add (buf, datecomp);
 367                 while (state == FLDPLUS) {
 368                     bufsz = sizeof buf;
 369                     state = m_getfld (&gstate, nam, buf, &bufsz, in);
 370                     datecomp = add (buf, datecomp);
 371                 }
 372                 if (!subjsort || subjcomp)
 373                     break;
 374             } else if (subjsort && !strcasecmp (nam, subjsort)) {
 375                 subjcomp = add (buf, subjcomp);
 376                 while (state == FLDPLUS) {
 377                     bufsz = sizeof buf;
 378                     state = m_getfld (&gstate, nam, buf, &bufsz, in);
 379                     subjcomp = add (buf, subjcomp);
 380                 }
 381                 if (datecomp)
 382                     break;
 383             } else {
 384                 /* just flush this guy */
 385                 while (state == FLDPLUS) {
 386                     bufsz = sizeof buf;
 387                     state = m_getfld (&gstate, nam, buf, &bufsz, in);
 388                 }
 389             }
 390             continue;
 391
 392         case BODY:
 393         case FILEEOF:
 394             break;
 395
 396         case LENERR:
 397         case FMTERR:
 398             if (state == LENERR || state == FMTERR) {
 399                 admonish (NULL, "format error in message %d (header #%d)",
 400                       msg, compnum);
 401                 check_failed = 1;
 402             }
 403             if (datecomp)
 404                 free (datecomp);
 405             if (subjcomp)
 406                 free (subjcomp);
 407             fclose (in);
 408             return (0);
 409
 410         default:
 411             adios (NULL, "internal error -- you lose");
 412         }
 413         break;
 414     }
 415     m_getfld_state_destroy (&gstate);
 416
 417     /*
 418      * If no date component, then use the modification
 419      * time of the file as its date
 420      */
 421     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 422         struct stat st;
 423
 424         advise (NULL,
 425                 "can't parse %s field in message %d, "
 426                 "will use file modification time",
 427                 datesw, msg);
 428         fstat (fileno (in), &st);
 429         smsg->s_clock = st.st_mtime;
 430         check_failed = 1;
 431     } else {
 432         smsg->s_clock = dmktime (tw);
 433     }
 434
 435     if (subjsort) {
 436         if (subjcomp) {
 437             /*
 438              * try to make the subject "canonical": delete
 439              * leading "re:", everything but letters & smash
 440              * letters to lower case.
 441              */
 442             register char  *cp, *cp2, c;
 443
 444             cp = subjcomp;
 445             cp2 = subjcomp;
 446             if (strcmp (subjsort, "subject") == 0) {
 447                 while ((c = *cp)) {
 448                     if (! isspace((unsigned char) c)) {
 449                         if(uprf(cp, "re:"))
 450                             cp += 2;
 451                         else
 452                             break;
 453                     }
 454                     cp++;
 455                 }
 456             }
 457
 458             while ((c = *cp++)) {
 459                 if (isascii((unsigned char) c) && isalnum((unsigned char) c))
 460                     *cp2++ = isupper((unsigned char) c) ?
 461                                         tolower((unsigned char) c) : c;
 462             }
 463
 464             *cp2 = '\0';
 465         }
 466         else
 467             subjcomp = "";
 468
 469         smsg->s_subj = subjcomp;
 470     }
 471     fclose (in);
 472     if (datecomp)
 473         free (datecomp);
 474
 475     return (1);
 476 }
 477
 478 /*
 479  * sort on dates.
 480  */
 481 static int
 482 dsort (struct smsg **a, struct smsg **b)
 483 {
 484     if ((*a)->s_clock < (*b)->s_clock)
 485         return (-1);
 486     else if ((*a)->s_clock > (*b)->s_clock)
 487         return (1);
 488     else if ((*a)->s_msg < (*b)->s_msg)
 489         return (-1);
 490     else
 491         return (1);
 492 }
 493
 494 /*
 495  * sort on subjects.
 496  */
 497 static int
 498 subsort (struct smsg **a, struct smsg **b)
 499 {
 500     register int i;
 501
 502     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 503         return (i);
 504
 505     return (dsort (a, b));
 506 }
 507
 508 static int
 509 txtsort (struct smsg **a, struct smsg **b)
 510 {
 511     register int i;
 512
 513     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 514         return (i);
 515     else if ((*a)->s_msg < (*b)->s_msg)
 516         return (-1);
 517     else
 518         return (1);
 519 }
 520
 521 static void
 522 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 523 {
 524     int nxt, old, new;
 525     char *newname, oldname[BUFSIZ];
 526     char newbuf[PATH_MAX + 1];
 527
 528     for (;;) {
 529         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 530         mlist[msg] = (struct smsg *)0;
 531         old = smsgs[nxt].s_msg;
 532         new = smsgs[msg].s_msg;
 533         strncpy (oldname, m_name (old), sizeof(oldname));
 534         newname = m_name (new);
 535         if (verbose)
 536             printf ("message %d becomes message %d\n", old, new);
 537
 538         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 539         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 540         ext_hook("ref-hook", oldname, newbuf);
 541
 542         if (rename (oldname, newname) == NOTOK)
 543             adios (newname, "unable to rename %s to", oldname);
 544
 545         copy_msg_flags (mp, new, old);
 546         if (mp->curmsg == old)
 547             seq_setcur (mp, new);
 548
 549         if (nxt == endmsg)
 550             break;
 551
 552         msg = nxt;
 553     }
 554 /*      if (nxt != endmsg); */
 555 /*      rename_chain (mp, mlist, nxt, endmsg); */
 556 }
 557
 558 static void
 559 rename_msgs (struct msgs *mp, struct smsg **mlist)
 560 {
 561     int i, j, old, new;
 562     bvector_t tmpset = bvector_create (0);
 563     char f1[BUFSIZ], tmpfil[BUFSIZ];
 564     char newbuf[PATH_MAX + 1];
 565     struct smsg *sp;
 566
 567     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 568
 569     for (i = 0; i < nmsgs; i++) {
 570         if (! (sp = mlist[i]))
 571             continue;   /* did this one */
 572
 573         j = sp - smsgs;
 574         if (j == i)
 575             continue;   /* this one doesn't move */
 576
 577         /*
 578          * the guy that was msg j is about to become msg i.
 579          * rename 'j' to make a hole, then recursively rename
 580          * guys to fill up the hole.
 581          */
 582         old = smsgs[j].s_msg;
 583         new = smsgs[i].s_msg;
 584         strncpy (f1, m_name (old), sizeof(f1));
 585
 586         if (verbose)
 587             printf ("renaming message chain from %d to %d\n", old, new);
 588
 589         /*
 590          *      Run the external hook to refile the old message as the
 591          *      temporary message number that is off of the end of the
 592          *      messages in the folder.
 593          */
 594
 595         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 596         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 597         ext_hook("ref-hook", f1, newbuf);
 598
 599         if (rename (f1, tmpfil) == NOTOK)
 600             adios (tmpfil, "unable to rename %s to ", f1);
 601
 602         get_msg_flags (mp, tmpset, old);
 603
 604         rename_chain (mp, mlist, j, i);
 605
 606         /*
 607          *      Run the external hook to refile the temorary message number
 608          *      to the real place.
 609          */
 610
 611         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 612         ext_hook("ref-hook", newbuf, f1);
 613
 614         if (rename (tmpfil, m_name(new)) == NOTOK)
 615             adios (m_name(new), "unable to rename %s to", tmpfil);
 616
 617         set_msg_flags (mp, tmpset, new);
 618         mp->msgflags |= SEQMOD;
 619     }
 620
 621     bvector_free (tmpset);
 622 }