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 "sbr/seq_setprev.h"
  10 #include "sbr/seq_setcur.h"
  11 #include "sbr/seq_save.h"
  12 #include "sbr/smatch.h"
  13 #include "sbr/uprf.h"
  14 #include "sbr/m_convert.h"
  15 #include "sbr/getfolder.h"
  16 #include "sbr/ext_hook.h"
  17 #include "sbr/folder_read.h"
  18 #include "sbr/folder_free.h"
  19 #include "sbr/context_save.h"
  20 #include "sbr/context_replace.h"
  21 #include "sbr/context_find.h"
  22 #include "sbr/ambigsw.h"
  23 #include "sbr/path.h"
  24 #include "sbr/print_version.h"
  25 #include "sbr/print_help.h"
  26 #include "sbr/error.h"
  27 #include "h/tws.h"
  28 #include "h/done.h"
  29 #include "h/utils.h"
  30 #include "sbr/m_maildir.h"
  31
  32 #define SORTM_SWITCHES \
  33     X("datefield field", 0, DATESW) \
  34     X("textfield field", 0, TEXTSW) \
  35     X("notextfield", 0, NSUBJSW) \
  36     X("subject", -3, SUBJSW) /* backward-compatibility */ \
  37     X("limit days", 0, LIMSW) \
  38     X("nolimit", 0, NLIMSW) \
  39     X("verbose", 0, VERBSW) \
  40     X("noverbose", 0, NVERBSW) \
  41     X("all", 0, ALLMSGS) \
  42     X("noall", 0, NALLMSGS) \
  43     X("check", 0, CHECKSW) \
  44     X("nocheck", 0, NCHECKSW) \
  45     X("version", 0, VERSIONSW) \
  46     X("help", 0, HELPSW) \
  47
  48 #define X(sw, minchars, id) id,
  49 DEFINE_SWITCH_ENUM(SORTM);
  50 #undef X
  51
  52 #define X(sw, minchars, id) { sw, minchars, id },
  53 DEFINE_SWITCH_ARRAY(SORTM, switches);
  54 #undef X
  55
  56 struct smsg {
  57     int s_msg;
  58     time_t s_clock;
  59     char *s_subj;
  60 };
  61
  62 static struct smsg *smsgs;
  63 int nmsgs;
  64
  65 char *subjsort;                 /* sort on subject if != 0 */
  66 time_t datelimit = 0;
  67 bool submajor;                  /* if true, sort on subject-major */
  68 bool verbose;
  69 int allmsgs = 1;
  70 int check_failed = 0;
  71
  72 /* This keeps compiler happy on calls to qsort */
  73 typedef int (*qsort_comp) (const void *, const void *);
  74
  75 /*
  76  * static prototypes
  77  */
  78 static int read_hdrs (struct msgs *, char *);
  79 static int get_fields (char *, int, struct smsg *);
  80 static int dsort (struct smsg **, struct smsg **);
  81 static int subsort (struct smsg **, struct smsg **);
  82 static int txtsort (struct smsg **, struct smsg **);
  83 static void rename_chain (struct msgs *, struct smsg **, int, int);
  84 static void rename_msgs (struct msgs *, struct smsg **);
  85
  86
  87 int
  88 main (int argc, char **argv)
  89 {
  90     int i, msgnum;
  91     char *cp, *maildir, *datesw = NULL;
  92     char *folder = NULL, buf[BUFSIZ], **argp;
  93     char **arguments;
  94     struct msgs_array msgs = { 0, 0, NULL };
  95     struct msgs *mp;
  96     struct smsg **dlist;
  97     bool checksw = false;
  98
  99     if (nmh_init(argv[0], true, true)) { return 1; }
 100
 101     arguments = getarguments (invo_name, argc, argv, 1);
 102     argp = arguments;
 103
 104     /*
 105      * Parse arguments
 106      */
 107     while ((cp = *argp++)) {
 108         if (*cp == '-') {
 109             switch (smatch (++cp, switches)) {
 110             case AMBIGSW:
 111                 ambigsw (cp, switches);
 112                 done (1);
 113             case UNKWNSW:
 114                 die("-%s unknown", cp);
 115
 116             case HELPSW:
 117                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 118                         invo_name);
 119                 print_help (buf, switches, 1);
 120                 done (0);
 121             case VERSIONSW:
 122                 print_version(invo_name);
 123                 done (0);
 124
 125             case DATESW:
 126                 if (datesw)
 127                     die("only one date field at a time");
 128                 if (!(datesw = *argp++) || *datesw == '-')
 129                     die("missing argument to %s", argp[-2]);
 130                 continue;
 131
 132             case TEXTSW:
 133                 if (subjsort)
 134                     die("only one text field at a time");
 135                 if (!(subjsort = *argp++) || *subjsort == '-')
 136                     die("missing argument to %s", argp[-2]);
 137                 continue;
 138
 139             case SUBJSW:
 140                 subjsort = "subject";
 141                 continue;
 142             case NSUBJSW:
 143                 subjsort = NULL;
 144                 continue;
 145
 146             case LIMSW:
 147                 if (!(cp = *argp++) || *cp == '-')
 148                         die("missing argument to %s", argp[-2]);
 149                 while (*cp == '0')
 150                     cp++;               /* skip any leading zeros */
 151                 if (!*cp) {             /* hit end of string */
 152                     submajor = true;    /* sort subject-major */
 153                     continue;
 154                 }
 155                 if (!isdigit((unsigned char) *cp) || !(datelimit = atoi(cp)))
 156                     die("impossible limit %s", cp);
 157                 datelimit *= 60*60*24;
 158                 continue;
 159             case NLIMSW:
 160                 submajor = false;       /* use date-major, but */
 161                 datelimit = 0;          /* use no limit */
 162                 continue;
 163
 164             case VERBSW:
 165                 verbose = true;
 166                 continue;
 167             case NVERBSW:
 168                 verbose = false;
 169                 continue;
 170
 171             case ALLMSGS:
 172                 allmsgs = 1;
 173                 continue;
 174             case NALLMSGS:
 175                 allmsgs = 0;
 176                 continue;
 177
 178             case CHECKSW:
 179                 checksw = true;
 180                 continue;
 181             case NCHECKSW:
 182                 checksw = false;
 183                 continue;
 184             }
 185         }
 186         if (*cp == '+' || *cp == '@') {
 187             if (folder)
 188                 die("only one folder at a time!");
 189             folder = pluspath (cp);
 190         } else
 191             app_msgarg(&msgs, cp);
 192     }
 193
 194     if (!context_find ("path"))
 195         free (path ("./", TFOLDER));
 196     if (!msgs.size) {
 197         if (allmsgs) {
 198             app_msgarg(&msgs, "all");
 199         } else {
 200             die("must specify messages to sort with -noall");
 201         }
 202     }
 203     if (!datesw)
 204         datesw = "date";
 205     if (!folder)
 206         folder = getfolder (1);
 207     maildir = m_maildir (folder);
 208
 209     if (chdir (maildir) == NOTOK)
 210         adios (maildir, "unable to change directory to");
 211
 212     /* read folder and create message structure */
 213     if (!(mp = folder_read (folder, 1)))
 214         die("unable to read folder %s", folder);
 215
 216     /* check for empty folder */
 217     if (mp->nummsg == 0)
 218         die("no messages in %s", folder);
 219
 220     /* parse all the message ranges/sequences and set SELECTED */
 221     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 222         if (!m_convert (mp, msgs.msgs[msgnum]))
 223             done (1);
 224     seq_setprev (mp);   /* set the previous sequence */
 225
 226     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 227         die("no messages to sort");
 228
 229     if (checksw  &&  check_failed) {
 230         die("errors found, no messages sorted");
 231     }
 232
 233     /*
 234      * sort a list of pointers to our "messages to be sorted".
 235      */
 236     dlist = mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 237     for (i = 0; i < nmsgs; i++)
 238         dlist[i] = &smsgs[i];
 239     dlist[nmsgs] = 0;
 240
 241     if (verbose) {      /* announce what we're doing */
 242         if (subjsort)
 243             if (submajor)
 244                 printf ("sorting by %s\n", subjsort);
 245             else
 246                 printf ("sorting by %s-major %s-minor\n", subjsort, datesw);
 247         else
 248             printf ("sorting by datefield %s\n", datesw);
 249     }
 250
 251     /* first sort by date, or by subject-major, date-minor */
 252     qsort (dlist, nmsgs, sizeof(*dlist),
 253             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 254
 255     /*
 256      * if we're sorting on subject, we need another list
 257      * in subject order, then a merge pass to collate the
 258      * two sorts.
 259      */
 260     if (!submajor && subjsort) {        /* already date sorted */
 261         struct smsg **slist, **flist;
 262         struct smsg ***il, **fp, **dp;
 263
 264         slist = mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 265         memcpy(slist, dlist, (nmsgs+1)*sizeof(*slist));
 266         qsort(slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 267
 268         /*
 269          * make an inversion list so we can quickly find
 270          * the collection of messages with the same subj
 271          * given a message number.
 272          */
 273         il = mh_xcalloc(mp->hghsel + 1, sizeof *il);
 274         for (i = 0; i < nmsgs; i++)
 275             il[slist[i]->s_msg] = &slist[i];
 276         /*
 277          * make up the final list, chronological but with
 278          * all the same subjects grouped together.
 279          */
 280         flist = mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 281         fp = flist;
 282         for (dp = dlist; *dp;) {
 283             struct smsg **s = il[(*dp++)->s_msg];
 284
 285             /* see if we already did this guy */
 286             if (! s)
 287                 continue;
 288
 289             *fp++ = *s++;
 290             /*
 291              * take the next message(s) if there is one,
 292              * its subject isn't null and its subject
 293              * is the same as this one and it's not too
 294              * far away in time.
 295              */
 296             while (*s && (*s)->s_subj[0] &&
 297                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 298                    (datelimit == 0 ||
 299                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 300                 il[(*s)->s_msg] = 0;
 301                 *fp++ = *s++;
 302             }
 303         }
 304         *fp = 0;
 305         free (il);
 306         free (slist);
 307         free (dlist);
 308         dlist = flist;
 309     }
 310
 311     /*
 312      * At this point, dlist is a sorted array of pointers to smsg structures,
 313      * each of which contains a message number.
 314      */
 315
 316     rename_msgs (mp, dlist);
 317
 318     context_replace (pfolder, folder);  /* update current folder         */
 319     seq_save (mp);                      /* synchronize message sequences */
 320     context_save ();                    /* save the context file         */
 321     folder_free (mp);                   /* free folder/message structure */
 322     done (0);
 323     return 1;
 324 }
 325
 326 static int
 327 read_hdrs (struct msgs *mp, char *datesw)
 328 {
 329     int msgnum;
 330     struct smsg *s;
 331
 332     smsgs = mh_xcalloc(mp->hghsel - mp->lowsel + 2, sizeof *smsgs);
 333     s = smsgs;
 334     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 335         if (is_selected(mp, msgnum)) {
 336             if (get_fields (datesw, msgnum, s)) {
 337                 s->s_msg = msgnum;
 338                 s++;
 339             }
 340         }
 341     }
 342     s->s_msg = 0;
 343     return s - smsgs;
 344 }
 345
 346
 347 /*
 348  * Parse the message and get the data or subject field,
 349  * if needed.
 350  */
 351
 352 static int
 353 get_fields (char *datesw, int msg, struct smsg *smsg)
 354 {
 355     int state;
 356     int compnum;
 357     char *msgnam, buf[NMH_BUFSIZ], nam[NAMESZ];
 358     struct tws *tw;
 359     char *datecomp = NULL, *subjcomp = NULL;
 360     FILE *in;
 361     m_getfld_state_t gstate;
 362
 363     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 364         admonish (msgnam, "unable to read message");
 365         return 0;
 366     }
 367     gstate = m_getfld_state_init(in);
 368     for (compnum = 1;;) {
 369         int bufsz = sizeof buf;
 370         switch (state = m_getfld2(&gstate, nam, buf, &bufsz)) {
 371         case FLD:
 372         case FLDPLUS:
 373             compnum++;
 374             if (!strcasecmp (nam, datesw)) {
 375                 datecomp = add (buf, datecomp);
 376                 while (state == FLDPLUS) {
 377                     bufsz = sizeof buf;
 378                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 379                     datecomp = add (buf, datecomp);
 380                 }
 381                 if (!subjsort || subjcomp)
 382                     break;
 383             } else if (subjsort && !strcasecmp (nam, subjsort)) {
 384                 subjcomp = add (buf, subjcomp);
 385                 while (state == FLDPLUS) {
 386                     bufsz = sizeof buf;
 387                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 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_getfld2(&gstate, nam, buf, &bufsz);
 397                 }
 398             }
 399             continue;
 400
 401         case BODY:
 402         case FILEEOF:
 403             break;
 404
 405         case LENERR:
 406         case FMTERR:
 407             if (state == LENERR || state == FMTERR) {
 408                 inform("format error in message %d (header #%d), continuing...",
 409                       msg, compnum);
 410                 check_failed = 1;
 411             }
 412             free(datecomp);
 413             free(subjcomp);
 414             fclose (in);
 415             return 0;
 416
 417         default:
 418             die("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         inform("can't parse %s field in message %d, "
 432             "will use file modification time", datesw, msg);
 433         fstat (fileno (in), &st);
 434         smsg->s_clock = st.st_mtime;
 435         check_failed = 1;
 436     } else {
 437         smsg->s_clock = dmktime (tw);
 438     }
 439
 440     if (subjsort) {
 441         if (subjcomp) {
 442             /*
 443              * try to make the subject "canonical": delete
 444              * leading "re:", everything but letters & smash
 445              * letters to lower case.
 446              */
 447             char  *cp, *cp2, c;
 448
 449             cp = subjcomp;
 450             cp2 = subjcomp;
 451             if (strcmp (subjsort, "subject") == 0) {
 452                 while ((c = *cp)) {
 453                     if (! isspace((unsigned char) c)) {
 454                         if(!uprf(cp, "re:"))
 455                             break;
 456                         cp += 2;
 457                     }
 458                     cp++;
 459                 }
 460             }
 461
 462             while ((c = *cp++)) {
 463                 if (isascii((unsigned char) c) && isalnum((unsigned char) c))
 464                     *cp2++ = tolower((unsigned char)c);
 465             }
 466
 467             *cp2 = '\0';
 468         }
 469         else
 470             subjcomp = "";
 471
 472         smsg->s_subj = subjcomp;
 473     }
 474     fclose (in);
 475     free(datecomp);
 476
 477     return 1;
 478 }
 479
 480 /*
 481  * sort on dates.
 482  */
 483 static int
 484 dsort (struct smsg **a, struct smsg **b)
 485 {
 486     if ((*a)->s_clock < (*b)->s_clock)
 487         return -1;
 488     if ((*a)->s_clock > (*b)->s_clock)
 489         return 1;
 490     if ((*a)->s_msg < (*b)->s_msg)
 491         return -1;
 492     return 1;
 493 }
 494
 495 /*
 496  * sort on subjects.
 497  */
 498 static int
 499 subsort (struct smsg **a, struct smsg **b)
 500 {
 501     int i;
 502
 503     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 504         return i;
 505
 506     return dsort(a, b);
 507 }
 508
 509 static int
 510 txtsort (struct smsg **a, struct smsg **b)
 511 {
 512     int i;
 513
 514     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 515         return i;
 516     if ((*a)->s_msg < (*b)->s_msg)
 517         return -1;
 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] = NULL;
 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 ();
 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 temporary 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 }