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/ext_hook.h"
  10 #include "sbr/folder_read.h"
  11 #include "sbr/folder_free.h"
  12 #include "sbr/context_save.h"
  13 #include "sbr/context_replace.h"
  14 #include "sbr/context_find.h"
  15 #include "sbr/ambigsw.h"
  16 #include "sbr/path.h"
  17 #include "sbr/print_version.h"
  18 #include "sbr/print_help.h"
  19 #include "sbr/error.h"
  20 #include "h/tws.h"
  21 #include "h/done.h"
  22 #include "h/utils.h"
  23 #include "sbr/m_maildir.h"
  24
  25 #define SORTM_SWITCHES \
  26     X("datefield field", 0, DATESW) \
  27     X("textfield field", 0, TEXTSW) \
  28     X("notextfield", 0, NSUBJSW) \
  29     X("subject", -3, SUBJSW) /* backward-compatibility */ \
  30     X("limit days", 0, LIMSW) \
  31     X("nolimit", 0, NLIMSW) \
  32     X("verbose", 0, VERBSW) \
  33     X("noverbose", 0, NVERBSW) \
  34     X("all", 0, ALLMSGS) \
  35     X("noall", 0, NALLMSGS) \
  36     X("check", 0, CHECKSW) \
  37     X("nocheck", 0, NCHECKSW) \
  38     X("version", 0, VERSIONSW) \
  39     X("help", 0, HELPSW) \
  40
  41 #define X(sw, minchars, id) id,
  42 DEFINE_SWITCH_ENUM(SORTM);
  43 #undef X
  44
  45 #define X(sw, minchars, id) { sw, minchars, id },
  46 DEFINE_SWITCH_ARRAY(SORTM, switches);
  47 #undef X
  48
  49 struct smsg {
  50     int s_msg;
  51     time_t s_clock;
  52     char *s_subj;
  53 };
  54
  55 static struct smsg *smsgs;
  56 int nmsgs;
  57
  58 char *subjsort;                 /* sort on subject if != 0 */
  59 time_t datelimit = 0;
  60 bool submajor;                  /* if true, sort on subject-major */
  61 bool verbose;
  62 int allmsgs = 1;
  63 int check_failed = 0;
  64
  65 /* This keeps compiler happy on calls to qsort */
  66 typedef int (*qsort_comp) (const void *, const void *);
  67
  68 /*
  69  * static prototypes
  70  */
  71 static int read_hdrs (struct msgs *, char *);
  72 static int get_fields (char *, int, struct smsg *);
  73 static int dsort (struct smsg **, struct smsg **);
  74 static int subsort (struct smsg **, struct smsg **);
  75 static int txtsort (struct smsg **, struct smsg **);
  76 static void rename_chain (struct msgs *, struct smsg **, int, int);
  77 static void rename_msgs (struct msgs *, struct smsg **);
  78
  79
  80 int
  81 main (int argc, char **argv)
  82 {
  83     int i, msgnum;
  84     char *cp, *maildir, *datesw = NULL;
  85     char *folder = NULL, buf[BUFSIZ], **argp;
  86     char **arguments;
  87     struct msgs_array msgs = { 0, 0, NULL };
  88     struct msgs *mp;
  89     struct smsg **dlist;
  90     bool checksw = false;
  91
  92     if (nmh_init(argv[0], true, true)) { return 1; }
  93
  94     arguments = getarguments (invo_name, argc, argv, 1);
  95     argp = arguments;
  96
  97     /*
  98      * Parse arguments
  99      */
 100     while ((cp = *argp++)) {
 101         if (*cp == '-') {
 102             switch (smatch (++cp, switches)) {
 103             case AMBIGSW:
 104                 ambigsw (cp, switches);
 105                 done (1);
 106             case UNKWNSW:
 107                 die("-%s unknown", cp);
 108
 109             case HELPSW:
 110                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 111                         invo_name);
 112                 print_help (buf, switches, 1);
 113                 done (0);
 114             case VERSIONSW:
 115                 print_version(invo_name);
 116                 done (0);
 117
 118             case DATESW:
 119                 if (datesw)
 120                     die("only one date field at a time");
 121                 if (!(datesw = *argp++) || *datesw == '-')
 122                     die("missing argument to %s", argp[-2]);
 123                 continue;
 124
 125             case TEXTSW:
 126                 if (subjsort)
 127                     die("only one text field at a time");
 128                 if (!(subjsort = *argp++) || *subjsort == '-')
 129                     die("missing argument to %s", argp[-2]);
 130                 continue;
 131
 132             case SUBJSW:
 133                 subjsort = "subject";
 134                 continue;
 135             case NSUBJSW:
 136                 subjsort = NULL;
 137                 continue;
 138
 139             case LIMSW:
 140                 if (!(cp = *argp++) || *cp == '-')
 141                         die("missing argument to %s", argp[-2]);
 142                 while (*cp == '0')
 143                     cp++;               /* skip any leading zeros */
 144                 if (!*cp) {             /* hit end of string */
 145                     submajor = true;    /* sort subject-major */
 146                     continue;
 147                 }
 148                 if (!isdigit((unsigned char) *cp) || !(datelimit = atoi(cp)))
 149                     die("impossible limit %s", cp);
 150                 datelimit *= 60*60*24;
 151                 continue;
 152             case NLIMSW:
 153                 submajor = false;       /* use date-major, but */
 154                 datelimit = 0;          /* use no limit */
 155                 continue;
 156
 157             case VERBSW:
 158                 verbose = true;
 159                 continue;
 160             case NVERBSW:
 161                 verbose = false;
 162                 continue;
 163
 164             case ALLMSGS:
 165                 allmsgs = 1;
 166                 continue;
 167             case NALLMSGS:
 168                 allmsgs = 0;
 169                 continue;
 170
 171             case CHECKSW:
 172                 checksw = true;
 173                 continue;
 174             case NCHECKSW:
 175                 checksw = false;
 176                 continue;
 177             }
 178         }
 179         if (*cp == '+' || *cp == '@') {
 180             if (folder)
 181                 die("only one folder at a time!");
 182             folder = pluspath (cp);
 183         } else
 184             app_msgarg(&msgs, cp);
 185     }
 186
 187     if (!context_find ("path"))
 188         free (path ("./", TFOLDER));
 189     if (!msgs.size) {
 190         if (allmsgs) {
 191             app_msgarg(&msgs, "all");
 192         } else {
 193             die("must specify messages to sort with -noall");
 194         }
 195     }
 196     if (!datesw)
 197         datesw = "date";
 198     if (!folder)
 199         folder = getfolder (1);
 200     maildir = m_maildir (folder);
 201
 202     if (chdir (maildir) == NOTOK)
 203         adios (maildir, "unable to change directory to");
 204
 205     /* read folder and create message structure */
 206     if (!(mp = folder_read (folder, 1)))
 207         die("unable to read folder %s", folder);
 208
 209     /* check for empty folder */
 210     if (mp->nummsg == 0)
 211         die("no messages in %s", folder);
 212
 213     /* parse all the message ranges/sequences and set SELECTED */
 214     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 215         if (!m_convert (mp, msgs.msgs[msgnum]))
 216             done (1);
 217     seq_setprev (mp);   /* set the previous sequence */
 218
 219     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 220         die("no messages to sort");
 221
 222     if (checksw  &&  check_failed) {
 223         die("errors found, no messages sorted");
 224     }
 225
 226     /*
 227      * sort a list of pointers to our "messages to be sorted".
 228      */
 229     dlist = mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 230     for (i = 0; i < nmsgs; i++)
 231         dlist[i] = &smsgs[i];
 232     dlist[nmsgs] = 0;
 233
 234     if (verbose) {      /* announce what we're doing */
 235         if (subjsort)
 236             if (submajor)
 237                 printf ("sorting by %s\n", subjsort);
 238             else
 239                 printf ("sorting by %s-major %s-minor\n", subjsort, datesw);
 240         else
 241             printf ("sorting by datefield %s\n", datesw);
 242     }
 243
 244     /* first sort by date, or by subject-major, date-minor */
 245     qsort (dlist, nmsgs, sizeof(*dlist),
 246             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 247
 248     /*
 249      * if we're sorting on subject, we need another list
 250      * in subject order, then a merge pass to collate the
 251      * two sorts.
 252      */
 253     if (!submajor && subjsort) {        /* already date sorted */
 254         struct smsg **slist, **flist;
 255         struct smsg ***il, **fp, **dp;
 256
 257         slist = mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 258         memcpy(slist, dlist, (nmsgs+1)*sizeof(*slist));
 259         qsort(slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 260
 261         /*
 262          * make an inversion list so we can quickly find
 263          * the collection of messages with the same subj
 264          * given a message number.
 265          */
 266         il = mh_xcalloc(mp->hghsel + 1, sizeof *il);
 267         for (i = 0; i < nmsgs; i++)
 268             il[slist[i]->s_msg] = &slist[i];
 269         /*
 270          * make up the final list, chronological but with
 271          * all the same subjects grouped together.
 272          */
 273         flist = mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 274         fp = flist;
 275         for (dp = dlist; *dp;) {
 276             struct smsg **s = il[(*dp++)->s_msg];
 277
 278             /* see if we already did this guy */
 279             if (! s)
 280                 continue;
 281
 282             *fp++ = *s++;
 283             /*
 284              * take the next message(s) if there is one,
 285              * its subject isn't null and its subject
 286              * is the same as this one and it's not too
 287              * far away in time.
 288              */
 289             while (*s && (*s)->s_subj[0] &&
 290                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 291                    (datelimit == 0 ||
 292                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 293                 il[(*s)->s_msg] = 0;
 294                 *fp++ = *s++;
 295             }
 296         }
 297         *fp = 0;
 298         free (il);
 299         free (slist);
 300         free (dlist);
 301         dlist = flist;
 302     }
 303
 304     /*
 305      * At this point, dlist is a sorted array of pointers to smsg structures,
 306      * each of which contains a message number.
 307      */
 308
 309     rename_msgs (mp, dlist);
 310
 311     context_replace (pfolder, folder);  /* update current folder         */
 312     seq_save (mp);                      /* synchronize message sequences */
 313     context_save ();                    /* save the context file         */
 314     folder_free (mp);                   /* free folder/message structure */
 315     done (0);
 316     return 1;
 317 }
 318
 319 static int
 320 read_hdrs (struct msgs *mp, char *datesw)
 321 {
 322     int msgnum;
 323     struct smsg *s;
 324
 325     smsgs = mh_xcalloc(mp->hghsel - mp->lowsel + 2, sizeof *smsgs);
 326     s = smsgs;
 327     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 328         if (is_selected(mp, msgnum)) {
 329             if (get_fields (datesw, msgnum, s)) {
 330                 s->s_msg = msgnum;
 331                 s++;
 332             }
 333         }
 334     }
 335     s->s_msg = 0;
 336     return s - smsgs;
 337 }
 338
 339
 340 /*
 341  * Parse the message and get the data or subject field,
 342  * if needed.
 343  */
 344
 345 static int
 346 get_fields (char *datesw, int msg, struct smsg *smsg)
 347 {
 348     int state;
 349     int compnum;
 350     char *msgnam, buf[NMH_BUFSIZ], nam[NAMESZ];
 351     struct tws *tw;
 352     char *datecomp = NULL, *subjcomp = NULL;
 353     FILE *in;
 354     m_getfld_state_t gstate;
 355
 356     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 357         admonish (msgnam, "unable to read message");
 358         return 0;
 359     }
 360     gstate = m_getfld_state_init(in);
 361     for (compnum = 1;;) {
 362         int bufsz = sizeof buf;
 363         switch (state = m_getfld2(&gstate, nam, buf, &bufsz)) {
 364         case FLD:
 365         case FLDPLUS:
 366             compnum++;
 367             if (!strcasecmp (nam, datesw)) {
 368                 datecomp = add (buf, datecomp);
 369                 while (state == FLDPLUS) {
 370                     bufsz = sizeof buf;
 371                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 372                     datecomp = add (buf, datecomp);
 373                 }
 374                 if (!subjsort || subjcomp)
 375                     break;
 376             } else if (subjsort && !strcasecmp (nam, subjsort)) {
 377                 subjcomp = add (buf, subjcomp);
 378                 while (state == FLDPLUS) {
 379                     bufsz = sizeof buf;
 380                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 381                     subjcomp = add (buf, subjcomp);
 382                 }
 383                 if (datecomp)
 384                     break;
 385             } else {
 386                 /* just flush this guy */
 387                 while (state == FLDPLUS) {
 388                     bufsz = sizeof buf;
 389                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 390                 }
 391             }
 392             continue;
 393
 394         case BODY:
 395         case FILEEOF:
 396             break;
 397
 398         case LENERR:
 399         case FMTERR:
 400             if (state == LENERR || state == FMTERR) {
 401                 inform("format error in message %d (header #%d), continuing...",
 402                       msg, compnum);
 403                 check_failed = 1;
 404             }
 405             free(datecomp);
 406             free(subjcomp);
 407             fclose (in);
 408             return 0;
 409
 410         default:
 411             die("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         inform("can't parse %s field in message %d, "
 425             "will use file modification time", datesw, msg);
 426         fstat (fileno (in), &st);
 427         smsg->s_clock = st.st_mtime;
 428         check_failed = 1;
 429     } else {
 430         smsg->s_clock = dmktime (tw);
 431     }
 432
 433     if (subjsort) {
 434         if (subjcomp) {
 435             /*
 436              * try to make the subject "canonical": delete
 437              * leading "re:", everything but letters & smash
 438              * letters to lower case.
 439              */
 440             char  *cp, *cp2, c;
 441
 442             cp = subjcomp;
 443             cp2 = subjcomp;
 444             if (strcmp (subjsort, "subject") == 0) {
 445                 while ((c = *cp)) {
 446                     if (! isspace((unsigned char) c)) {
 447                         if(!uprf(cp, "re:"))
 448                             break;
 449                         cp += 2;
 450                     }
 451                     cp++;
 452                 }
 453             }
 454
 455             while ((c = *cp++)) {
 456                 if (isascii((unsigned char) c) && isalnum((unsigned char) c))
 457                     *cp2++ = tolower((unsigned char)c);
 458             }
 459
 460             *cp2 = '\0';
 461         }
 462         else
 463             subjcomp = "";
 464
 465         smsg->s_subj = subjcomp;
 466     }
 467     fclose (in);
 468     free(datecomp);
 469
 470     return 1;
 471 }
 472
 473 /*
 474  * sort on dates.
 475  */
 476 static int
 477 dsort (struct smsg **a, struct smsg **b)
 478 {
 479     if ((*a)->s_clock < (*b)->s_clock)
 480         return -1;
 481     if ((*a)->s_clock > (*b)->s_clock)
 482         return 1;
 483     if ((*a)->s_msg < (*b)->s_msg)
 484         return -1;
 485     return 1;
 486 }
 487
 488 /*
 489  * sort on subjects.
 490  */
 491 static int
 492 subsort (struct smsg **a, struct smsg **b)
 493 {
 494     int i;
 495
 496     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 497         return i;
 498
 499     return dsort(a, b);
 500 }
 501
 502 static int
 503 txtsort (struct smsg **a, struct smsg **b)
 504 {
 505     int i;
 506
 507     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 508         return i;
 509     if ((*a)->s_msg < (*b)->s_msg)
 510         return -1;
 511     return 1;
 512 }
 513
 514 static void
 515 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 516 {
 517     int nxt, old, new;
 518     char *newname, oldname[BUFSIZ];
 519     char newbuf[PATH_MAX + 1];
 520
 521     for (;;) {
 522         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 523         mlist[msg] = NULL;
 524         old = smsgs[nxt].s_msg;
 525         new = smsgs[msg].s_msg;
 526         strncpy (oldname, m_name (old), sizeof(oldname));
 527         newname = m_name (new);
 528         if (verbose)
 529             printf ("message %d becomes message %d\n", old, new);
 530
 531         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 532         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 533         ext_hook("ref-hook", oldname, newbuf);
 534
 535         if (rename (oldname, newname) == NOTOK)
 536             adios (newname, "unable to rename %s to", oldname);
 537
 538         copy_msg_flags (mp, new, old);
 539         if (mp->curmsg == old)
 540             seq_setcur (mp, new);
 541
 542         if (nxt == endmsg)
 543             break;
 544
 545         msg = nxt;
 546     }
 547 /*      if (nxt != endmsg); */
 548 /*      rename_chain (mp, mlist, nxt, endmsg); */
 549 }
 550
 551 static void
 552 rename_msgs (struct msgs *mp, struct smsg **mlist)
 553 {
 554     int i, j, old, new;
 555     bvector_t tmpset = bvector_create ();
 556     char f1[BUFSIZ], tmpfil[BUFSIZ];
 557     char newbuf[PATH_MAX + 1];
 558     struct smsg *sp;
 559
 560     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 561
 562     for (i = 0; i < nmsgs; i++) {
 563         if (! (sp = mlist[i]))
 564             continue;   /* did this one */
 565
 566         j = sp - smsgs;
 567         if (j == i)
 568             continue;   /* this one doesn't move */
 569
 570         /*
 571          * the guy that was msg j is about to become msg i.
 572          * rename 'j' to make a hole, then recursively rename
 573          * guys to fill up the hole.
 574          */
 575         old = smsgs[j].s_msg;
 576         new = smsgs[i].s_msg;
 577         strncpy (f1, m_name (old), sizeof(f1));
 578
 579         if (verbose)
 580             printf ("renaming message chain from %d to %d\n", old, new);
 581
 582         /*
 583          *      Run the external hook to refile the old message as the
 584          *      temporary message number that is off of the end of the
 585          *      messages in the folder.
 586          */
 587
 588         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 589         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 590         ext_hook("ref-hook", f1, newbuf);
 591
 592         if (rename (f1, tmpfil) == NOTOK)
 593             adios (tmpfil, "unable to rename %s to ", f1);
 594
 595         get_msg_flags (mp, tmpset, old);
 596
 597         rename_chain (mp, mlist, j, i);
 598
 599         /*
 600          *      Run the external hook to refile the temporary message number
 601          *      to the real place.
 602          */
 603
 604         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 605         ext_hook("ref-hook", newbuf, f1);
 606
 607         if (rename (tmpfil, m_name(new)) == NOTOK)
 608             adios (m_name(new), "unable to rename %s to", tmpfil);
 609
 610         set_msg_flags (mp, tmpset, new);
 611         mp->msgflags |= SEQMOD;
 612     }
 613
 614     bvector_free (tmpset);
 615 }