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/m_name.h"
  10 #include "sbr/m_getfld.h"
  11 #include "sbr/getarguments.h"
  12 #include "sbr/seq_setprev.h"
  13 #include "sbr/seq_setcur.h"
  14 #include "sbr/seq_save.h"
  15 #include "sbr/smatch.h"
  16 #include "sbr/uprf.h"
  17 #include "sbr/m_convert.h"
  18 #include "sbr/getfolder.h"
  19 #include "sbr/ext_hook.h"
  20 #include "sbr/folder_read.h"
  21 #include "sbr/folder_free.h"
  22 #include "sbr/context_save.h"
  23 #include "sbr/context_replace.h"
  24 #include "sbr/context_find.h"
  25 #include "sbr/ambigsw.h"
  26 #include "sbr/path.h"
  27 #include "sbr/print_version.h"
  28 #include "sbr/print_help.h"
  29 #include "sbr/error.h"
  30 #include "h/tws.h"
  31 #include "h/done.h"
  32 #include "h/utils.h"
  33 #include "sbr/m_maildir.h"
  34
  35 #define SORTM_SWITCHES \
  36     X("datefield field", 0, DATESW) \
  37     X("textfield field", 0, TEXTSW) \
  38     X("notextfield", 0, NSUBJSW) \
  39     X("subject", -3, SUBJSW) /* backward-compatibility */ \
  40     X("limit days", 0, LIMSW) \
  41     X("nolimit", 0, NLIMSW) \
  42     X("verbose", 0, VERBSW) \
  43     X("noverbose", 0, NVERBSW) \
  44     X("all", 0, ALLMSGS) \
  45     X("noall", 0, NALLMSGS) \
  46     X("check", 0, CHECKSW) \
  47     X("nocheck", 0, NCHECKSW) \
  48     X("version", 0, VERSIONSW) \
  49     X("help", 0, HELPSW) \
  50
  51 #define X(sw, minchars, id) id,
  52 DEFINE_SWITCH_ENUM(SORTM);
  53 #undef X
  54
  55 #define X(sw, minchars, id) { sw, minchars, id },
  56 DEFINE_SWITCH_ARRAY(SORTM, switches);
  57 #undef X
  58
  59 struct smsg {
  60     int s_msg;
  61     time_t s_clock;
  62     char *s_subj;
  63 };
  64
  65 static struct smsg *smsgs;
  66 int nmsgs;
  67
  68 char *subjsort;                 /* sort on subject if != 0 */
  69 time_t datelimit = 0;
  70 bool submajor;                  /* if true, sort on subject-major */
  71 bool verbose;
  72 int allmsgs = 1;
  73 int check_failed = 0;
  74
  75 /* This keeps compiler happy on calls to qsort */
  76 typedef int (*qsort_comp) (const void *, const void *);
  77
  78 /*
  79  * static prototypes
  80  */
  81 static int read_hdrs (struct msgs *, char *);
  82 static int get_fields (char *, int, struct smsg *);
  83 static int dsort (struct smsg **, struct smsg **);
  84 static int subsort (struct smsg **, struct smsg **);
  85 static int txtsort (struct smsg **, struct smsg **);
  86 static void rename_chain (struct msgs *, struct smsg **, int, int);
  87 static void rename_msgs (struct msgs *, struct smsg **);
  88
  89
  90 int
  91 main (int argc, char **argv)
  92 {
  93     int i, msgnum;
  94     char *cp, *maildir, *datesw = NULL;
  95     char *folder = NULL, buf[BUFSIZ], **argp;
  96     char **arguments;
  97     struct msgs_array msgs = { 0, 0, NULL };
  98     struct msgs *mp;
  99     struct smsg **dlist;
 100     bool checksw = false;
 101
 102     if (nmh_init(argv[0], true, true)) { return 1; }
 103
 104     arguments = getarguments (invo_name, argc, argv, 1);
 105     argp = arguments;
 106
 107     /*
 108      * Parse arguments
 109      */
 110     while ((cp = *argp++)) {
 111         if (*cp == '-') {
 112             switch (smatch (++cp, switches)) {
 113             case AMBIGSW:
 114                 ambigsw (cp, switches);
 115                 done (1);
 116             case UNKWNSW:
 117                 die("-%s unknown", cp);
 118
 119             case HELPSW:
 120                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 121                         invo_name);
 122                 print_help (buf, switches, 1);
 123                 done (0);
 124             case VERSIONSW:
 125                 print_version(invo_name);
 126                 done (0);
 127
 128             case DATESW:
 129                 if (datesw)
 130                     die("only one date field at a time");
 131                 if (!(datesw = *argp++) || *datesw == '-')
 132                     die("missing argument to %s", argp[-2]);
 133                 continue;
 134
 135             case TEXTSW:
 136                 if (subjsort)
 137                     die("only one text field at a time");
 138                 if (!(subjsort = *argp++) || *subjsort == '-')
 139                     die("missing argument to %s", argp[-2]);
 140                 continue;
 141
 142             case SUBJSW:
 143                 subjsort = "subject";
 144                 continue;
 145             case NSUBJSW:
 146                 subjsort = NULL;
 147                 continue;
 148
 149             case LIMSW:
 150                 if (!(cp = *argp++) || *cp == '-')
 151                         die("missing argument to %s", argp[-2]);
 152                 while (*cp == '0')
 153                     cp++;               /* skip any leading zeros */
 154                 if (!*cp) {             /* hit end of string */
 155                     submajor = true;    /* sort subject-major */
 156                     continue;
 157                 }
 158                 if (!isdigit((unsigned char) *cp) || !(datelimit = atoi(cp)))
 159                     die("impossible limit %s", cp);
 160                 datelimit *= 60*60*24;
 161                 continue;
 162             case NLIMSW:
 163                 submajor = false;       /* use date-major, but */
 164                 datelimit = 0;          /* use no limit */
 165                 continue;
 166
 167             case VERBSW:
 168                 verbose = true;
 169                 continue;
 170             case NVERBSW:
 171                 verbose = false;
 172                 continue;
 173
 174             case ALLMSGS:
 175                 allmsgs = 1;
 176                 continue;
 177             case NALLMSGS:
 178                 allmsgs = 0;
 179                 continue;
 180
 181             case CHECKSW:
 182                 checksw = true;
 183                 continue;
 184             case NCHECKSW:
 185                 checksw = false;
 186                 continue;
 187             }
 188         }
 189         if (*cp == '+' || *cp == '@') {
 190             if (folder)
 191                 die("only one folder at a time!");
 192             folder = pluspath (cp);
 193         } else
 194             app_msgarg(&msgs, cp);
 195     }
 196
 197     if (!context_find ("path"))
 198         free (path ("./", TFOLDER));
 199     if (!msgs.size) {
 200         if (allmsgs) {
 201             app_msgarg(&msgs, "all");
 202         } else {
 203             die("must specify messages to sort with -noall");
 204         }
 205     }
 206     if (!datesw)
 207         datesw = "date";
 208     if (!folder)
 209         folder = getfolder (1);
 210     maildir = m_maildir (folder);
 211
 212     if (chdir (maildir) == NOTOK)
 213         adios (maildir, "unable to change directory to");
 214
 215     /* read folder and create message structure */
 216     if (!(mp = folder_read (folder, 1)))
 217         die("unable to read folder %s", folder);
 218
 219     /* check for empty folder */
 220     if (mp->nummsg == 0)
 221         die("no messages in %s", folder);
 222
 223     /* parse all the message ranges/sequences and set SELECTED */
 224     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 225         if (!m_convert (mp, msgs.msgs[msgnum]))
 226             done (1);
 227     seq_setprev (mp);   /* set the previous sequence */
 228
 229     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 230         die("no messages to sort");
 231
 232     if (checksw  &&  check_failed) {
 233         die("errors found, no messages sorted");
 234     }
 235
 236     /*
 237      * sort a list of pointers to our "messages to be sorted".
 238      */
 239     dlist = mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 240     for (i = 0; i < nmsgs; i++)
 241         dlist[i] = &smsgs[i];
 242     dlist[nmsgs] = 0;
 243
 244     if (verbose) {      /* announce what we're doing */
 245         if (subjsort)
 246             if (submajor)
 247                 printf ("sorting by %s\n", subjsort);
 248             else
 249                 printf ("sorting by %s-major %s-minor\n", subjsort, datesw);
 250         else
 251             printf ("sorting by datefield %s\n", datesw);
 252     }
 253
 254     /* first sort by date, or by subject-major, date-minor */
 255     qsort (dlist, nmsgs, sizeof(*dlist),
 256             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 257
 258     /*
 259      * if we're sorting on subject, we need another list
 260      * in subject order, then a merge pass to collate the
 261      * two sorts.
 262      */
 263     if (!submajor && subjsort) {        /* already date sorted */
 264         struct smsg **slist, **flist;
 265         struct smsg ***il, **fp, **dp;
 266
 267         slist = mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 268         memcpy(slist, dlist, (nmsgs+1)*sizeof(*slist));
 269         qsort(slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 270
 271         /*
 272          * make an inversion list so we can quickly find
 273          * the collection of messages with the same subj
 274          * given a message number.
 275          */
 276         il = mh_xcalloc(mp->hghsel + 1, sizeof *il);
 277         for (i = 0; i < nmsgs; i++)
 278             il[slist[i]->s_msg] = &slist[i];
 279         /*
 280          * make up the final list, chronological but with
 281          * all the same subjects grouped together.
 282          */
 283         flist = mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 284         fp = flist;
 285         for (dp = dlist; *dp;) {
 286             struct smsg **s = il[(*dp++)->s_msg];
 287
 288             /* see if we already did this guy */
 289             if (! s)
 290                 continue;
 291
 292             *fp++ = *s++;
 293             /*
 294              * take the next message(s) if there is one,
 295              * its subject isn't null and its subject
 296              * is the same as this one and it's not too
 297              * far away in time.
 298              */
 299             while (*s && (*s)->s_subj[0] &&
 300                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 301                    (datelimit == 0 ||
 302                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 303                 il[(*s)->s_msg] = 0;
 304                 *fp++ = *s++;
 305             }
 306         }
 307         *fp = 0;
 308         free (il);
 309         free (slist);
 310         free (dlist);
 311         dlist = flist;
 312     }
 313
 314     /*
 315      * At this point, dlist is a sorted array of pointers to smsg structures,
 316      * each of which contains a message number.
 317      */
 318
 319     rename_msgs (mp, dlist);
 320
 321     context_replace (pfolder, folder);  /* update current folder         */
 322     seq_save (mp);                      /* synchronize message sequences */
 323     context_save ();                    /* save the context file         */
 324     folder_free (mp);                   /* free folder/message structure */
 325     done (0);
 326     return 1;
 327 }
 328
 329 static int
 330 read_hdrs (struct msgs *mp, char *datesw)
 331 {
 332     int msgnum;
 333     struct smsg *s;
 334
 335     smsgs = mh_xcalloc(mp->hghsel - mp->lowsel + 2, sizeof *smsgs);
 336     s = smsgs;
 337     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 338         if (is_selected(mp, msgnum)) {
 339             if (get_fields (datesw, msgnum, s)) {
 340                 s->s_msg = msgnum;
 341                 s++;
 342             }
 343         }
 344     }
 345     s->s_msg = 0;
 346     return s - smsgs;
 347 }
 348
 349
 350 /*
 351  * Parse the message and get the data or subject field,
 352  * if needed.
 353  */
 354
 355 static int
 356 get_fields (char *datesw, int msg, struct smsg *smsg)
 357 {
 358     int state;
 359     int compnum;
 360     char *msgnam, buf[NMH_BUFSIZ], nam[NAMESZ];
 361     struct tws *tw;
 362     char *datecomp = NULL, *subjcomp = NULL;
 363     FILE *in;
 364     m_getfld_state_t gstate;
 365
 366     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 367         admonish (msgnam, "unable to read message");
 368         return 0;
 369     }
 370     gstate = m_getfld_state_init(in);
 371     for (compnum = 1;;) {
 372         int bufsz = sizeof buf;
 373         switch (state = m_getfld2(&gstate, nam, buf, &bufsz)) {
 374         case FLD:
 375         case FLDPLUS:
 376             compnum++;
 377             if (!strcasecmp (nam, datesw)) {
 378                 datecomp = add (buf, datecomp);
 379                 while (state == FLDPLUS) {
 380                     bufsz = sizeof buf;
 381                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 382                     datecomp = add (buf, datecomp);
 383                 }
 384                 if (!subjsort || subjcomp)
 385                     break;
 386             } else if (subjsort && !strcasecmp (nam, subjsort)) {
 387                 subjcomp = add (buf, subjcomp);
 388                 while (state == FLDPLUS) {
 389                     bufsz = sizeof buf;
 390                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 391                     subjcomp = add (buf, subjcomp);
 392                 }
 393                 if (datecomp)
 394                     break;
 395             } else {
 396                 /* just flush this guy */
 397                 while (state == FLDPLUS) {
 398                     bufsz = sizeof buf;
 399                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 400                 }
 401             }
 402             continue;
 403
 404         case BODY:
 405         case FILEEOF:
 406             break;
 407
 408         case LENERR:
 409         case FMTERR:
 410             if (state == LENERR || state == FMTERR) {
 411                 inform("format error in message %d (header #%d), continuing...",
 412                       msg, compnum);
 413                 check_failed = 1;
 414             }
 415             free(datecomp);
 416             free(subjcomp);
 417             fclose (in);
 418             return 0;
 419
 420         default:
 421             die("internal error -- you lose");
 422         }
 423         break;
 424     }
 425     m_getfld_state_destroy (&gstate);
 426
 427     /*
 428      * If no date component, then use the modification
 429      * time of the file as its date
 430      */
 431     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 432         struct stat st;
 433
 434         inform("can't parse %s field in message %d, "
 435             "will use file modification time", datesw, msg);
 436         fstat (fileno (in), &st);
 437         smsg->s_clock = st.st_mtime;
 438         check_failed = 1;
 439     } else {
 440         smsg->s_clock = dmktime (tw);
 441     }
 442
 443     if (subjsort) {
 444         if (subjcomp) {
 445             /*
 446              * try to make the subject "canonical": delete
 447              * leading "re:", everything but letters & smash
 448              * letters to lower case.
 449              */
 450             char  *cp, *cp2, c;
 451
 452             cp = subjcomp;
 453             cp2 = subjcomp;
 454             if (strcmp (subjsort, "subject") == 0) {
 455                 while ((c = *cp)) {
 456                     if (! isspace((unsigned char) c)) {
 457                         if(!uprf(cp, "re:"))
 458                             break;
 459                         cp += 2;
 460                     }
 461                     cp++;
 462                 }
 463             }
 464
 465             while ((c = *cp++)) {
 466                 if (isascii((unsigned char) c) && isalnum((unsigned char) c))
 467                     *cp2++ = tolower((unsigned char)c);
 468             }
 469
 470             *cp2 = '\0';
 471         }
 472         else
 473             subjcomp = "";
 474
 475         smsg->s_subj = subjcomp;
 476     }
 477     fclose (in);
 478     free(datecomp);
 479
 480     return 1;
 481 }
 482
 483 /*
 484  * sort on dates.
 485  */
 486 static int
 487 dsort (struct smsg **a, struct smsg **b)
 488 {
 489     if ((*a)->s_clock < (*b)->s_clock)
 490         return -1;
 491     if ((*a)->s_clock > (*b)->s_clock)
 492         return 1;
 493     if ((*a)->s_msg < (*b)->s_msg)
 494         return -1;
 495     return 1;
 496 }
 497
 498 /*
 499  * sort on subjects.
 500  */
 501 static int
 502 subsort (struct smsg **a, struct smsg **b)
 503 {
 504     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     int i;
 516
 517     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 518         return i;
 519     if ((*a)->s_msg < (*b)->s_msg)
 520         return -1;
 521     return 1;
 522 }
 523
 524 static void
 525 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 526 {
 527     int nxt, old, new;
 528     char *newname, oldname[BUFSIZ];
 529     char newbuf[PATH_MAX + 1];
 530
 531     for (;;) {
 532         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 533         mlist[msg] = NULL;
 534         old = smsgs[nxt].s_msg;
 535         new = smsgs[msg].s_msg;
 536         strncpy (oldname, m_name (old), sizeof(oldname));
 537         newname = m_name (new);
 538         if (verbose)
 539             printf ("message %d becomes message %d\n", old, new);
 540
 541         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 542         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 543         ext_hook("ref-hook", oldname, newbuf);
 544
 545         if (rename (oldname, newname) == NOTOK)
 546             adios (newname, "unable to rename %s to", oldname);
 547
 548         copy_msg_flags (mp, new, old);
 549         if (mp->curmsg == old)
 550             seq_setcur (mp, new);
 551
 552         if (nxt == endmsg)
 553             break;
 554
 555         msg = nxt;
 556     }
 557 /*      if (nxt != endmsg); */
 558 /*      rename_chain (mp, mlist, nxt, endmsg); */
 559 }
 560
 561 static void
 562 rename_msgs (struct msgs *mp, struct smsg **mlist)
 563 {
 564     int i, j, old, new;
 565     bvector_t tmpset = bvector_create ();
 566     char f1[BUFSIZ], tmpfil[BUFSIZ];
 567     char newbuf[PATH_MAX + 1];
 568     struct smsg *sp;
 569
 570     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 571
 572     for (i = 0; i < nmsgs; i++) {
 573         if (! (sp = mlist[i]))
 574             continue;   /* did this one */
 575
 576         j = sp - smsgs;
 577         if (j == i)
 578             continue;   /* this one doesn't move */
 579
 580         /*
 581          * the guy that was msg j is about to become msg i.
 582          * rename 'j' to make a hole, then recursively rename
 583          * guys to fill up the hole.
 584          */
 585         old = smsgs[j].s_msg;
 586         new = smsgs[i].s_msg;
 587         strncpy (f1, m_name (old), sizeof(f1));
 588
 589         if (verbose)
 590             printf ("renaming message chain from %d to %d\n", old, new);
 591
 592         /*
 593          *      Run the external hook to refile the old message as the
 594          *      temporary message number that is off of the end of the
 595          *      messages in the folder.
 596          */
 597
 598         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 599         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 600         ext_hook("ref-hook", f1, newbuf);
 601
 602         if (rename (f1, tmpfil) == NOTOK)
 603             adios (tmpfil, "unable to rename %s to ", f1);
 604
 605         get_msg_flags (mp, tmpset, old);
 606
 607         rename_chain (mp, mlist, j, i);
 608
 609         /*
 610          *      Run the external hook to refile the temporary message number
 611          *      to the real place.
 612          */
 613
 614         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 615         ext_hook("ref-hook", newbuf, f1);
 616
 617         if (rename (tmpfil, m_name(new)) == NOTOK)
 618             adios (m_name(new), "unable to rename %s to", tmpfil);
 619
 620         set_msg_flags (mp, tmpset, new);
 621         mp->msgflags |= SEQMOD;
 622     }
 623
 624     bvector_free (tmpset);
 625 }