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     setlocale(LC_ALL, "");
  82     invo_name = r1bindex (argv[0], '/');
  83
  84     /* read user profile/context */
  85     context_read();
  86
  87     arguments = getarguments (invo_name, argc, argv, 1);
  88     argp = arguments;
  89
  90     /*
  91      * Parse arguments
  92      */
  93     while ((cp = *argp++)) {
  94         if (*cp == '-') {
  95             switch (smatch (++cp, switches)) {
  96             case AMBIGSW:
  97                 ambigsw (cp, switches);
  98                 done (1);
  99             case UNKWNSW:
 100                 adios (NULL, "-%s unknown", cp);
 101
 102             case HELPSW:
 103                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 104                         invo_name);
 105                 print_help (buf, switches, 1);
 106                 done (0);
 107             case VERSIONSW:
 108                 print_version(invo_name);
 109                 done (0);
 110
 111             case DATESW:
 112                 if (datesw)
 113                     adios (NULL, "only one date field at a time");
 114                 if (!(datesw = *argp++) || *datesw == '-')
 115                     adios (NULL, "missing argument to %s", argp[-2]);
 116                 continue;
 117
 118             case TEXTSW:
 119                 if (subjsort)
 120                     adios (NULL, "only one text field at a time");
 121                 if (!(subjsort = *argp++) || *subjsort == '-')
 122                     adios (NULL, "missing argument to %s", argp[-2]);
 123                 continue;
 124
 125             case SUBJSW:
 126                 subjsort = "subject";
 127                 continue;
 128             case NSUBJSW:
 129                 subjsort = (char *)0;
 130                 continue;
 131
 132             case LIMSW:
 133                 if (!(cp = *argp++) || *cp == '-')
 134                         adios (NULL, "missing argument to %s", argp[-2]);
 135                 while (*cp == '0')
 136                     cp++;               /* skip any leading zeros */
 137                 if (!*cp) {             /* hit end of string */
 138                     submajor++;         /* sort subject-major */
 139                     continue;
 140                 }
 141                 if (!isdigit((unsigned char) *cp) || !(datelimit = atoi(cp)))
 142                     adios (NULL, "impossible limit %s", cp);
 143                 datelimit *= 60*60*24;
 144                 continue;
 145             case NLIMSW:
 146                 submajor = 0;   /* use date-major, but */
 147                 datelimit = 0;  /* use no limit */
 148                 continue;
 149
 150             case VERBSW:
 151                 verbose++;
 152                 continue;
 153             case NVERBSW:
 154                 verbose = 0;
 155                 continue;
 156
 157             case ALLMSGS:
 158                 allmsgs = 1;
 159                 continue;
 160             case NALLMSGS:
 161                 allmsgs = 0;
 162                 continue;
 163
 164             case CHECKSW:
 165                 checksw = 1;
 166                 continue;
 167             case NCHECKSW:
 168                 checksw = 0;
 169                 continue;
 170             }
 171         }
 172         if (*cp == '+' || *cp == '@') {
 173             if (folder)
 174                 adios (NULL, "only one folder at a time!");
 175             else
 176                 folder = pluspath (cp);
 177         } else
 178                 app_msgarg(&msgs, cp);
 179     }
 180
 181     if (!context_find ("path"))
 182         free (path ("./", TFOLDER));
 183     if (!msgs.size) {
 184         if (allmsgs) {
 185             app_msgarg(&msgs, "all");
 186         } else {
 187             adios (NULL, "must specify messages to sort with -noall");
 188         }
 189     }
 190     if (!datesw)
 191         datesw = "date";
 192     if (!folder)
 193         folder = getfolder (1);
 194     maildir = m_maildir (folder);
 195
 196     if (chdir (maildir) == NOTOK)
 197         adios (maildir, "unable to change directory to");
 198
 199     /* read folder and create message structure */
 200     if (!(mp = folder_read (folder, 1)))
 201         adios (NULL, "unable to read folder %s", folder);
 202
 203     /* check for empty folder */
 204     if (mp->nummsg == 0)
 205         adios (NULL, "no messages in %s", folder);
 206
 207     /* parse all the message ranges/sequences and set SELECTED */
 208     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 209         if (!m_convert (mp, msgs.msgs[msgnum]))
 210             done (1);
 211     seq_setprev (mp);   /* set the previous sequence */
 212
 213     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 214         adios (NULL, "no messages to sort");
 215
 216     if (checksw  &&  check_failed) {
 217         adios (NULL, "errors found, no messages sorted");
 218     }
 219
 220     /*
 221      * sort a list of pointers to our "messages to be sorted".
 222      */
 223     dlist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 224     for (i = 0; i < nmsgs; i++)
 225         dlist[i] = &smsgs[i];
 226     dlist[nmsgs] = 0;
 227
 228     if (verbose) {      /* announce what we're doing */
 229         if (subjsort)
 230             if (submajor)
 231                 printf ("sorting by %s\n", subjsort);
 232             else
 233                 printf ("sorting by %s-major %s-minor\n", subjsort, datesw);
 234         else
 235             printf ("sorting by datefield %s\n", datesw);
 236     }
 237
 238     /* first sort by date, or by subject-major, date-minor */
 239     qsort ((char *) dlist, nmsgs, sizeof(*dlist),
 240             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 241
 242     /*
 243      * if we're sorting on subject, we need another list
 244      * in subject order, then a merge pass to collate the
 245      * two sorts.
 246      */
 247     if (!submajor && subjsort) {        /* already date sorted */
 248         struct smsg **slist, **flist;
 249         register struct smsg ***il, **fp, **dp;
 250
 251         slist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 252         memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 253         qsort((char *)slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 254
 255         /*
 256          * make an inversion list so we can quickly find
 257          * the collection of messages with the same subj
 258          * given a message number.
 259          */
 260         il = (struct smsg ***) calloc (mp->hghsel+1, sizeof(*il));
 261         if (! il)
 262             adios (NULL, "couldn't allocate msg list");
 263         for (i = 0; i < nmsgs; i++)
 264             il[slist[i]->s_msg] = &slist[i];
 265         /*
 266          * make up the final list, chronological but with
 267          * all the same subjects grouped together.
 268          */
 269         flist = (struct smsg **) mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 270         fp = flist;
 271         for (dp = dlist; *dp;) {
 272             register struct smsg **s = il[(*dp++)->s_msg];
 273
 274             /* see if we already did this guy */
 275             if (! s)
 276                 continue;
 277
 278             *fp++ = *s++;
 279             /*
 280              * take the next message(s) if there is one,
 281              * its subject isn't null and its subject
 282              * is the same as this one and it's not too
 283              * far away in time.
 284              */
 285             while (*s && (*s)->s_subj[0] &&
 286                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 287                    (datelimit == 0 ||
 288                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 289                 il[(*s)->s_msg] = 0;
 290                 *fp++ = *s++;
 291             }
 292         }
 293         *fp = 0;
 294         free (slist);
 295         free (dlist);
 296         dlist = flist;
 297     }
 298
 299     /*
 300      * At this point, dlist is a sorted array of pointers to smsg structures,
 301      * each of which contains a message number.
 302      */
 303
 304     rename_msgs (mp, dlist);
 305
 306     context_replace (pfolder, folder);  /* update current folder         */
 307     seq_save (mp);                      /* synchronize message sequences */
 308     context_save ();                    /* save the context file         */
 309     folder_free (mp);                   /* free folder/message structure */
 310     done (0);
 311     return 1;
 312 }
 313
 314 static int
 315 read_hdrs (struct msgs *mp, char *datesw)
 316 {
 317     int msgnum;
 318     struct tws tb;
 319     register struct smsg *s;
 320
 321     twscopy (&tb, dlocaltimenow ());
 322
 323     smsgs = (struct smsg *)
 324         calloc ((size_t) (mp->hghsel - mp->lowsel + 2),
 325             sizeof(*smsgs));
 326     if (smsgs == NULL)
 327         adios (NULL, "unable to allocate sort storage");
 328
 329     s = smsgs;
 330     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 331         if (is_selected(mp, msgnum)) {
 332             if (get_fields (datesw, msgnum, s)) {
 333                 s->s_msg = msgnum;
 334                 s++;
 335             }
 336         }
 337     }
 338     s->s_msg = 0;
 339     return(s - smsgs);
 340 }
 341
 342
 343 /*
 344  * Parse the message and get the data or subject field,
 345  * if needed.
 346  */
 347
 348 static int
 349 get_fields (char *datesw, int msg, struct smsg *smsg)
 350 {
 351     register int state;
 352     int compnum;
 353     char *msgnam, buf[BUFSIZ], nam[NAMESZ];
 354     register struct tws *tw;
 355     register char *datecomp = NULL, *subjcomp = NULL;
 356     register FILE *in;
 357     m_getfld_state_t gstate = 0;
 358
 359     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 360         admonish (msgnam, "unable to read message");
 361         return (0);
 362     }
 363     for (compnum = 1;;) {
 364         int bufsz = sizeof buf;
 365         switch (state = m_getfld (&gstate, nam, buf, &bufsz, in)) {
 366         case FLD:
 367         case FLDPLUS:
 368             compnum++;
 369             if (!strcasecmp (nam, datesw)) {
 370                 datecomp = add (buf, datecomp);
 371                 while (state == FLDPLUS) {
 372                     bufsz = sizeof buf;
 373                     state = m_getfld (&gstate, nam, buf, &bufsz, in);
 374                     datecomp = add (buf, datecomp);
 375                 }
 376                 if (!subjsort || subjcomp)
 377                     break;
 378             } else if (subjsort && !strcasecmp (nam, subjsort)) {
 379                 subjcomp = add (buf, subjcomp);
 380                 while (state == FLDPLUS) {
 381                     bufsz = sizeof buf;
 382                     state = m_getfld (&gstate, nam, buf, &bufsz, in);
 383                     subjcomp = add (buf, subjcomp);
 384                 }
 385                 if (datecomp)
 386                     break;
 387             } else {
 388                 /* just flush this guy */
 389                 while (state == FLDPLUS) {
 390                     bufsz = sizeof buf;
 391                     state = m_getfld (&gstate, nam, buf, &bufsz, in);
 392                 }
 393             }
 394             continue;
 395
 396         case BODY:
 397         case FILEEOF:
 398             break;
 399
 400         case LENERR:
 401         case FMTERR:
 402             if (state == LENERR || state == FMTERR) {
 403                 admonish (NULL, "format error in message %d (header #%d)",
 404                       msg, compnum);
 405                 check_failed = 1;
 406             }
 407             if (datecomp)
 408                 free (datecomp);
 409             if (subjcomp)
 410                 free (subjcomp);
 411             fclose (in);
 412             return (0);
 413
 414         default:
 415             adios (NULL, "internal error -- you lose");
 416         }
 417         break;
 418     }
 419     m_getfld_state_destroy (&gstate);
 420
 421     /*
 422      * If no date component, then use the modification
 423      * time of the file as its date
 424      */
 425     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 426         struct stat st;
 427
 428         advise (NULL,
 429                 "can't parse %s field in message %d, "
 430                 "will use file modification time",
 431                 datesw, msg);
 432         fstat (fileno (in), &st);
 433         smsg->s_clock = st.st_mtime;
 434         check_failed = 1;
 435     } else {
 436         smsg->s_clock = dmktime (tw);
 437     }
 438
 439     if (subjsort) {
 440         if (subjcomp) {
 441             /*
 442              * try to make the subject "canonical": delete
 443              * leading "re:", everything but letters & smash
 444              * letters to lower case.
 445              */
 446             register char  *cp, *cp2, c;
 447
 448             cp = subjcomp;
 449             cp2 = subjcomp;
 450             if (strcmp (subjsort, "subject") == 0) {
 451                 while ((c = *cp)) {
 452                     if (! isspace((unsigned char) c)) {
 453                         if(uprf(cp, "re:"))
 454                             cp += 2;
 455                         else
 456                             break;
 457                     }
 458                     cp++;
 459                 }
 460             }
 461
 462             while ((c = *cp++)) {
 463                 if (isascii((unsigned char) c) && isalnum((unsigned char) c))
 464                     *cp2++ = isupper((unsigned char) c) ?
 465                                         tolower((unsigned char) c) : c;
 466             }
 467
 468             *cp2 = '\0';
 469         }
 470         else
 471             subjcomp = "";
 472
 473         smsg->s_subj = subjcomp;
 474     }
 475     fclose (in);
 476     if (datecomp)
 477         free (datecomp);
 478
 479     return (1);
 480 }
 481
 482 /*
 483  * sort on dates.
 484  */
 485 static int
 486 dsort (struct smsg **a, struct smsg **b)
 487 {
 488     if ((*a)->s_clock < (*b)->s_clock)
 489         return (-1);
 490     else if ((*a)->s_clock > (*b)->s_clock)
 491         return (1);
 492     else if ((*a)->s_msg < (*b)->s_msg)
 493         return (-1);
 494     else
 495         return (1);
 496 }
 497
 498 /*
 499  * sort on subjects.
 500  */
 501 static int
 502 subsort (struct smsg **a, struct smsg **b)
 503 {
 504     register int i;
 505
 506     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 507         return (i);
 508
 509     return (dsort (a, b));
 510 }
 511
 512 static int
 513 txtsort (struct smsg **a, struct smsg **b)
 514 {
 515     register int i;
 516
 517     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 518         return (i);
 519     else if ((*a)->s_msg < (*b)->s_msg)
 520         return (-1);
 521     else
 522         return (1);
 523 }
 524
 525 static void
 526 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 527 {
 528     int nxt, old, new;
 529     char *newname, oldname[BUFSIZ];
 530     char newbuf[PATH_MAX + 1];
 531
 532     for (;;) {
 533         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 534         mlist[msg] = (struct smsg *)0;
 535         old = smsgs[nxt].s_msg;
 536         new = smsgs[msg].s_msg;
 537         strncpy (oldname, m_name (old), sizeof(oldname));
 538         newname = m_name (new);
 539         if (verbose)
 540             printf ("message %d becomes message %d\n", old, new);
 541
 542         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 543         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 544         ext_hook("ref-hook", oldname, newbuf);
 545
 546         if (rename (oldname, newname) == NOTOK)
 547             adios (newname, "unable to rename %s to", oldname);
 548
 549         copy_msg_flags (mp, new, old);
 550         if (mp->curmsg == old)
 551             seq_setcur (mp, new);
 552
 553         if (nxt == endmsg)
 554             break;
 555
 556         msg = nxt;
 557     }
 558 /*      if (nxt != endmsg); */
 559 /*      rename_chain (mp, mlist, nxt, endmsg); */
 560 }
 561
 562 static void
 563 rename_msgs (struct msgs *mp, struct smsg **mlist)
 564 {
 565     int i, j, old, new;
 566     bvector_t tmpset = bvector_create (0);
 567     char f1[BUFSIZ], tmpfil[BUFSIZ];
 568     char newbuf[PATH_MAX + 1];
 569     struct smsg *sp;
 570
 571     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 572
 573     for (i = 0; i < nmsgs; i++) {
 574         if (! (sp = mlist[i]))
 575             continue;   /* did this one */
 576
 577         j = sp - smsgs;
 578         if (j == i)
 579             continue;   /* this one doesn't move */
 580
 581         /*
 582          * the guy that was msg j is about to become msg i.
 583          * rename 'j' to make a hole, then recursively rename
 584          * guys to fill up the hole.
 585          */
 586         old = smsgs[j].s_msg;
 587         new = smsgs[i].s_msg;
 588         strncpy (f1, m_name (old), sizeof(f1));
 589
 590         if (verbose)
 591             printf ("renaming message chain from %d to %d\n", old, new);
 592
 593         /*
 594          *      Run the external hook to refile the old message as the
 595          *      temporary message number that is off of the end of the
 596          *      messages in the folder.
 597          */
 598
 599         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 600         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 601         ext_hook("ref-hook", f1, newbuf);
 602
 603         if (rename (f1, tmpfil) == NOTOK)
 604             adios (tmpfil, "unable to rename %s to ", f1);
 605
 606         get_msg_flags (mp, tmpset, old);
 607
 608         rename_chain (mp, mlist, j, i);
 609
 610         /*
 611          *      Run the external hook to refile the temorary message number
 612          *      to the real place.
 613          */
 614
 615         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 616         ext_hook("ref-hook", newbuf, f1);
 617
 618         if (rename (tmpfil, m_name(new)) == NOTOK)
 619             adios (m_name(new), "unable to rename %s to", tmpfil);
 620
 621         set_msg_flags (mp, tmpset, new);
 622         mp->msgflags |= SEQMOD;
 623     }
 624
 625     bvector_free (tmpset);
 626 }