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/smatch.h"
  10 #include "sbr/uprf.h"
  11 #include "sbr/m_convert.h"
  12 #include "sbr/getfolder.h"
  13 #include "sbr/ext_hook.h"
  14 #include "sbr/folder_read.h"
  15 #include "sbr/folder_free.h"
  16 #include "sbr/context_save.h"
  17 #include "sbr/context_replace.h"
  18 #include "sbr/context_find.h"
  19 #include "sbr/ambigsw.h"
  20 #include "sbr/path.h"
  21 #include "sbr/print_version.h"
  22 #include "sbr/print_help.h"
  23 #include "sbr/error.h"
  24 #include "h/tws.h"
  25 #include "h/done.h"
  26 #include "h/utils.h"
  27 #include "sbr/m_maildir.h"
  28
  29 #define SORTM_SWITCHES \
  30     X("datefield field", 0, DATESW) \
  31     X("textfield field", 0, TEXTSW) \
  32     X("notextfield", 0, NSUBJSW) \
  33     X("subject", -3, SUBJSW) /* backward-compatibility */ \
  34     X("limit days", 0, LIMSW) \
  35     X("nolimit", 0, NLIMSW) \
  36     X("verbose", 0, VERBSW) \
  37     X("noverbose", 0, NVERBSW) \
  38     X("all", 0, ALLMSGS) \
  39     X("noall", 0, NALLMSGS) \
  40     X("check", 0, CHECKSW) \
  41     X("nocheck", 0, NCHECKSW) \
  42     X("version", 0, VERSIONSW) \
  43     X("help", 0, HELPSW) \
  44
  45 #define X(sw, minchars, id) id,
  46 DEFINE_SWITCH_ENUM(SORTM);
  47 #undef X
  48
  49 #define X(sw, minchars, id) { sw, minchars, id },
  50 DEFINE_SWITCH_ARRAY(SORTM, switches);
  51 #undef X
  52
  53 struct smsg {
  54     int s_msg;
  55     time_t s_clock;
  56     char *s_subj;
  57 };
  58
  59 static struct smsg *smsgs;
  60 int nmsgs;
  61
  62 char *subjsort;                 /* sort on subject if != 0 */
  63 time_t datelimit = 0;
  64 bool submajor;                  /* if true, sort on subject-major */
  65 bool verbose;
  66 int allmsgs = 1;
  67 int check_failed = 0;
  68
  69 /* This keeps compiler happy on calls to qsort */
  70 typedef int (*qsort_comp) (const void *, const void *);
  71
  72 /*
  73  * static prototypes
  74  */
  75 static int read_hdrs (struct msgs *, char *);
  76 static int get_fields (char *, int, struct smsg *);
  77 static int dsort (struct smsg **, struct smsg **);
  78 static int subsort (struct smsg **, struct smsg **);
  79 static int txtsort (struct smsg **, struct smsg **);
  80 static void rename_chain (struct msgs *, struct smsg **, int, int);
  81 static void rename_msgs (struct msgs *, struct smsg **);
  82
  83
  84 int
  85 main (int argc, char **argv)
  86 {
  87     int i, msgnum;
  88     char *cp, *maildir, *datesw = NULL;
  89     char *folder = NULL, buf[BUFSIZ], **argp;
  90     char **arguments;
  91     struct msgs_array msgs = { 0, 0, NULL };
  92     struct msgs *mp;
  93     struct smsg **dlist;
  94     bool checksw = false;
  95
  96     if (nmh_init(argv[0], true, true)) { return 1; }
  97
  98     arguments = getarguments (invo_name, argc, argv, 1);
  99     argp = arguments;
 100
 101     /*
 102      * Parse arguments
 103      */
 104     while ((cp = *argp++)) {
 105         if (*cp == '-') {
 106             switch (smatch (++cp, switches)) {
 107             case AMBIGSW:
 108                 ambigsw (cp, switches);
 109                 done (1);
 110             case UNKWNSW:
 111                 die("-%s unknown", cp);
 112
 113             case HELPSW:
 114                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 115                         invo_name);
 116                 print_help (buf, switches, 1);
 117                 done (0);
 118             case VERSIONSW:
 119                 print_version(invo_name);
 120                 done (0);
 121
 122             case DATESW:
 123                 if (datesw)
 124                     die("only one date field at a time");
 125                 if (!(datesw = *argp++) || *datesw == '-')
 126                     die("missing argument to %s", argp[-2]);
 127                 continue;
 128
 129             case TEXTSW:
 130                 if (subjsort)
 131                     die("only one text field at a time");
 132                 if (!(subjsort = *argp++) || *subjsort == '-')
 133                     die("missing argument to %s", argp[-2]);
 134                 continue;
 135
 136             case SUBJSW:
 137                 subjsort = "subject";
 138                 continue;
 139             case NSUBJSW:
 140                 subjsort = NULL;
 141                 continue;
 142
 143             case LIMSW:
 144                 if (!(cp = *argp++) || *cp == '-')
 145                         die("missing argument to %s", argp[-2]);
 146                 while (*cp == '0')
 147                     cp++;               /* skip any leading zeros */
 148                 if (!*cp) {             /* hit end of string */
 149                     submajor = true;    /* sort subject-major */
 150                     continue;
 151                 }
 152                 if (!isdigit((unsigned char) *cp) || !(datelimit = atoi(cp)))
 153                     die("impossible limit %s", cp);
 154                 datelimit *= 60*60*24;
 155                 continue;
 156             case NLIMSW:
 157                 submajor = false;       /* use date-major, but */
 158                 datelimit = 0;          /* use no limit */
 159                 continue;
 160
 161             case VERBSW:
 162                 verbose = true;
 163                 continue;
 164             case NVERBSW:
 165                 verbose = false;
 166                 continue;
 167
 168             case ALLMSGS:
 169                 allmsgs = 1;
 170                 continue;
 171             case NALLMSGS:
 172                 allmsgs = 0;
 173                 continue;
 174
 175             case CHECKSW:
 176                 checksw = true;
 177                 continue;
 178             case NCHECKSW:
 179                 checksw = false;
 180                 continue;
 181             }
 182         }
 183         if (*cp == '+' || *cp == '@') {
 184             if (folder)
 185                 die("only one folder at a time!");
 186             folder = pluspath (cp);
 187         } else
 188             app_msgarg(&msgs, cp);
 189     }
 190
 191     if (!context_find ("path"))
 192         free (path ("./", TFOLDER));
 193     if (!msgs.size) {
 194         if (allmsgs) {
 195             app_msgarg(&msgs, "all");
 196         } else {
 197             die("must specify messages to sort with -noall");
 198         }
 199     }
 200     if (!datesw)
 201         datesw = "date";
 202     if (!folder)
 203         folder = getfolder (1);
 204     maildir = m_maildir (folder);
 205
 206     if (chdir (maildir) == NOTOK)
 207         adios (maildir, "unable to change directory to");
 208
 209     /* read folder and create message structure */
 210     if (!(mp = folder_read (folder, 1)))
 211         die("unable to read folder %s", folder);
 212
 213     /* check for empty folder */
 214     if (mp->nummsg == 0)
 215         die("no messages in %s", folder);
 216
 217     /* parse all the message ranges/sequences and set SELECTED */
 218     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 219         if (!m_convert (mp, msgs.msgs[msgnum]))
 220             done (1);
 221     seq_setprev (mp);   /* set the previous sequence */
 222
 223     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 224         die("no messages to sort");
 225
 226     if (checksw  &&  check_failed) {
 227         die("errors found, no messages sorted");
 228     }
 229
 230     /*
 231      * sort a list of pointers to our "messages to be sorted".
 232      */
 233     dlist = mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 234     for (i = 0; i < nmsgs; i++)
 235         dlist[i] = &smsgs[i];
 236     dlist[nmsgs] = 0;
 237
 238     if (verbose) {      /* announce what we're doing */
 239         if (subjsort)
 240             if (submajor)
 241                 printf ("sorting by %s\n", subjsort);
 242             else
 243                 printf ("sorting by %s-major %s-minor\n", subjsort, datesw);
 244         else
 245             printf ("sorting by datefield %s\n", datesw);
 246     }
 247
 248     /* first sort by date, or by subject-major, date-minor */
 249     qsort (dlist, nmsgs, sizeof(*dlist),
 250             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 251
 252     /*
 253      * if we're sorting on subject, we need another list
 254      * in subject order, then a merge pass to collate the
 255      * two sorts.
 256      */
 257     if (!submajor && subjsort) {        /* already date sorted */
 258         struct smsg **slist, **flist;
 259         struct smsg ***il, **fp, **dp;
 260
 261         slist = mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 262         memcpy(slist, dlist, (nmsgs+1)*sizeof(*slist));
 263         qsort(slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 264
 265         /*
 266          * make an inversion list so we can quickly find
 267          * the collection of messages with the same subj
 268          * given a message number.
 269          */
 270         il = mh_xcalloc(mp->hghsel + 1, sizeof *il);
 271         for (i = 0; i < nmsgs; i++)
 272             il[slist[i]->s_msg] = &slist[i];
 273         /*
 274          * make up the final list, chronological but with
 275          * all the same subjects grouped together.
 276          */
 277         flist = mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 278         fp = flist;
 279         for (dp = dlist; *dp;) {
 280             struct smsg **s = il[(*dp++)->s_msg];
 281
 282             /* see if we already did this guy */
 283             if (! s)
 284                 continue;
 285
 286             *fp++ = *s++;
 287             /*
 288              * take the next message(s) if there is one,
 289              * its subject isn't null and its subject
 290              * is the same as this one and it's not too
 291              * far away in time.
 292              */
 293             while (*s && (*s)->s_subj[0] &&
 294                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 295                    (datelimit == 0 ||
 296                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 297                 il[(*s)->s_msg] = 0;
 298                 *fp++ = *s++;
 299             }
 300         }
 301         *fp = 0;
 302         free (il);
 303         free (slist);
 304         free (dlist);
 305         dlist = flist;
 306     }
 307
 308     /*
 309      * At this point, dlist is a sorted array of pointers to smsg structures,
 310      * each of which contains a message number.
 311      */
 312
 313     rename_msgs (mp, dlist);
 314
 315     context_replace (pfolder, folder);  /* update current folder         */
 316     seq_save (mp);                      /* synchronize message sequences */
 317     context_save ();                    /* save the context file         */
 318     folder_free (mp);                   /* free folder/message structure */
 319     done (0);
 320     return 1;
 321 }
 322
 323 static int
 324 read_hdrs (struct msgs *mp, char *datesw)
 325 {
 326     int msgnum;
 327     struct smsg *s;
 328
 329     smsgs = mh_xcalloc(mp->hghsel - mp->lowsel + 2, sizeof *smsgs);
 330     s = smsgs;
 331     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 332         if (is_selected(mp, msgnum)) {
 333             if (get_fields (datesw, msgnum, s)) {
 334                 s->s_msg = msgnum;
 335                 s++;
 336             }
 337         }
 338     }
 339     s->s_msg = 0;
 340     return s - smsgs;
 341 }
 342
 343
 344 /*
 345  * Parse the message and get the data or subject field,
 346  * if needed.
 347  */
 348
 349 static int
 350 get_fields (char *datesw, int msg, struct smsg *smsg)
 351 {
 352     int state;
 353     int compnum;
 354     char *msgnam, buf[NMH_BUFSIZ], nam[NAMESZ];
 355     struct tws *tw;
 356     char *datecomp = NULL, *subjcomp = NULL;
 357     FILE *in;
 358     m_getfld_state_t gstate;
 359
 360     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 361         admonish (msgnam, "unable to read message");
 362         return 0;
 363     }
 364     gstate = m_getfld_state_init(in);
 365     for (compnum = 1;;) {
 366         int bufsz = sizeof buf;
 367         switch (state = m_getfld2(&gstate, nam, buf, &bufsz)) {
 368         case FLD:
 369         case FLDPLUS:
 370             compnum++;
 371             if (!strcasecmp (nam, datesw)) {
 372                 datecomp = add (buf, datecomp);
 373                 while (state == FLDPLUS) {
 374                     bufsz = sizeof buf;
 375                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 376                     datecomp = add (buf, datecomp);
 377                 }
 378                 if (!subjsort || subjcomp)
 379                     break;
 380             } else if (subjsort && !strcasecmp (nam, subjsort)) {
 381                 subjcomp = add (buf, subjcomp);
 382                 while (state == FLDPLUS) {
 383                     bufsz = sizeof buf;
 384                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 385                     subjcomp = add (buf, subjcomp);
 386                 }
 387                 if (datecomp)
 388                     break;
 389             } else {
 390                 /* just flush this guy */
 391                 while (state == FLDPLUS) {
 392                     bufsz = sizeof buf;
 393                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 394                 }
 395             }
 396             continue;
 397
 398         case BODY:
 399         case FILEEOF:
 400             break;
 401
 402         case LENERR:
 403         case FMTERR:
 404             if (state == LENERR || state == FMTERR) {
 405                 inform("format error in message %d (header #%d), continuing...",
 406                       msg, compnum);
 407                 check_failed = 1;
 408             }
 409             free(datecomp);
 410             free(subjcomp);
 411             fclose (in);
 412             return 0;
 413
 414         default:
 415             die("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         inform("can't parse %s field in message %d, "
 429             "will use file modification time", datesw, msg);
 430         fstat (fileno (in), &st);
 431         smsg->s_clock = st.st_mtime;
 432         check_failed = 1;
 433     } else {
 434         smsg->s_clock = dmktime (tw);
 435     }
 436
 437     if (subjsort) {
 438         if (subjcomp) {
 439             /*
 440              * try to make the subject "canonical": delete
 441              * leading "re:", everything but letters & smash
 442              * letters to lower case.
 443              */
 444             char  *cp, *cp2, c;
 445
 446             cp = subjcomp;
 447             cp2 = subjcomp;
 448             if (strcmp (subjsort, "subject") == 0) {
 449                 while ((c = *cp)) {
 450                     if (! isspace((unsigned char) c)) {
 451                         if(!uprf(cp, "re:"))
 452                             break;
 453                         cp += 2;
 454                     }
 455                     cp++;
 456                 }
 457             }
 458
 459             while ((c = *cp++)) {
 460                 if (isascii((unsigned char) c) && isalnum((unsigned char) c))
 461                     *cp2++ = tolower((unsigned char)c);
 462             }
 463
 464             *cp2 = '\0';
 465         }
 466         else
 467             subjcomp = "";
 468
 469         smsg->s_subj = subjcomp;
 470     }
 471     fclose (in);
 472     free(datecomp);
 473
 474     return 1;
 475 }
 476
 477 /*
 478  * sort on dates.
 479  */
 480 static int
 481 dsort (struct smsg **a, struct smsg **b)
 482 {
 483     if ((*a)->s_clock < (*b)->s_clock)
 484         return -1;
 485     if ((*a)->s_clock > (*b)->s_clock)
 486         return 1;
 487     if ((*a)->s_msg < (*b)->s_msg)
 488         return -1;
 489     return 1;
 490 }
 491
 492 /*
 493  * sort on subjects.
 494  */
 495 static int
 496 subsort (struct smsg **a, struct smsg **b)
 497 {
 498     int i;
 499
 500     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 501         return i;
 502
 503     return dsort(a, b);
 504 }
 505
 506 static int
 507 txtsort (struct smsg **a, struct smsg **b)
 508 {
 509     int i;
 510
 511     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 512         return i;
 513     if ((*a)->s_msg < (*b)->s_msg)
 514         return -1;
 515     return 1;
 516 }
 517
 518 static void
 519 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 520 {
 521     int nxt, old, new;
 522     char *newname, oldname[BUFSIZ];
 523     char newbuf[PATH_MAX + 1];
 524
 525     for (;;) {
 526         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 527         mlist[msg] = NULL;
 528         old = smsgs[nxt].s_msg;
 529         new = smsgs[msg].s_msg;
 530         strncpy (oldname, m_name (old), sizeof(oldname));
 531         newname = m_name (new);
 532         if (verbose)
 533             printf ("message %d becomes message %d\n", old, new);
 534
 535         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 536         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 537         ext_hook("ref-hook", oldname, newbuf);
 538
 539         if (rename (oldname, newname) == NOTOK)
 540             adios (newname, "unable to rename %s to", oldname);
 541
 542         copy_msg_flags (mp, new, old);
 543         if (mp->curmsg == old)
 544             seq_setcur (mp, new);
 545
 546         if (nxt == endmsg)
 547             break;
 548
 549         msg = nxt;
 550     }
 551 /*      if (nxt != endmsg); */
 552 /*      rename_chain (mp, mlist, nxt, endmsg); */
 553 }
 554
 555 static void
 556 rename_msgs (struct msgs *mp, struct smsg **mlist)
 557 {
 558     int i, j, old, new;
 559     bvector_t tmpset = bvector_create ();
 560     char f1[BUFSIZ], tmpfil[BUFSIZ];
 561     char newbuf[PATH_MAX + 1];
 562     struct smsg *sp;
 563
 564     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 565
 566     for (i = 0; i < nmsgs; i++) {
 567         if (! (sp = mlist[i]))
 568             continue;   /* did this one */
 569
 570         j = sp - smsgs;
 571         if (j == i)
 572             continue;   /* this one doesn't move */
 573
 574         /*
 575          * the guy that was msg j is about to become msg i.
 576          * rename 'j' to make a hole, then recursively rename
 577          * guys to fill up the hole.
 578          */
 579         old = smsgs[j].s_msg;
 580         new = smsgs[i].s_msg;
 581         strncpy (f1, m_name (old), sizeof(f1));
 582
 583         if (verbose)
 584             printf ("renaming message chain from %d to %d\n", old, new);
 585
 586         /*
 587          *      Run the external hook to refile the old message as the
 588          *      temporary message number that is off of the end of the
 589          *      messages in the folder.
 590          */
 591
 592         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 593         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 594         ext_hook("ref-hook", f1, newbuf);
 595
 596         if (rename (f1, tmpfil) == NOTOK)
 597             adios (tmpfil, "unable to rename %s to ", f1);
 598
 599         get_msg_flags (mp, tmpset, old);
 600
 601         rename_chain (mp, mlist, j, i);
 602
 603         /*
 604          *      Run the external hook to refile the temporary message number
 605          *      to the real place.
 606          */
 607
 608         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 609         ext_hook("ref-hook", newbuf, f1);
 610
 611         if (rename (tmpfil, m_name(new)) == NOTOK)
 612             adios (m_name(new), "unable to rename %s to", tmpfil);
 613
 614         set_msg_flags (mp, tmpset, new);
 615         mp->msgflags |= SEQMOD;
 616     }
 617
 618     bvector_free (tmpset);
 619 }