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/getfolder.h"
  10 #include "sbr/ext_hook.h"
  11 #include "sbr/folder_read.h"
  12 #include "sbr/folder_free.h"
  13 #include "sbr/context_save.h"
  14 #include "sbr/context_replace.h"
  15 #include "sbr/context_find.h"
  16 #include "sbr/ambigsw.h"
  17 #include "sbr/path.h"
  18 #include "sbr/print_version.h"
  19 #include "sbr/print_help.h"
  20 #include "sbr/error.h"
  21 #include "h/tws.h"
  22 #include "h/done.h"
  23 #include "h/utils.h"
  24 #include "sbr/m_maildir.h"
  25
  26 #define SORTM_SWITCHES \
  27     X("datefield field", 0, DATESW) \
  28     X("textfield field", 0, TEXTSW) \
  29     X("notextfield", 0, NSUBJSW) \
  30     X("subject", -3, SUBJSW) /* backward-compatibility */ \
  31     X("limit days", 0, LIMSW) \
  32     X("nolimit", 0, NLIMSW) \
  33     X("verbose", 0, VERBSW) \
  34     X("noverbose", 0, NVERBSW) \
  35     X("all", 0, ALLMSGS) \
  36     X("noall", 0, NALLMSGS) \
  37     X("check", 0, CHECKSW) \
  38     X("nocheck", 0, NCHECKSW) \
  39     X("version", 0, VERSIONSW) \
  40     X("help", 0, HELPSW) \
  41
  42 #define X(sw, minchars, id) id,
  43 DEFINE_SWITCH_ENUM(SORTM);
  44 #undef X
  45
  46 #define X(sw, minchars, id) { sw, minchars, id },
  47 DEFINE_SWITCH_ARRAY(SORTM, switches);
  48 #undef X
  49
  50 struct smsg {
  51     int s_msg;
  52     time_t s_clock;
  53     char *s_subj;
  54 };
  55
  56 static struct smsg *smsgs;
  57 int nmsgs;
  58
  59 char *subjsort;                 /* sort on subject if != 0 */
  60 time_t datelimit = 0;
  61 bool submajor;                  /* if true, sort on subject-major */
  62 bool verbose;
  63 int allmsgs = 1;
  64 int check_failed = 0;
  65
  66 /* This keeps compiler happy on calls to qsort */
  67 typedef int (*qsort_comp) (const void *, const void *);
  68
  69 /*
  70  * static prototypes
  71  */
  72 static int read_hdrs (struct msgs *, char *);
  73 static int get_fields (char *, int, struct smsg *);
  74 static int dsort (struct smsg **, struct smsg **);
  75 static int subsort (struct smsg **, struct smsg **);
  76 static int txtsort (struct smsg **, struct smsg **);
  77 static void rename_chain (struct msgs *, struct smsg **, int, int);
  78 static void rename_msgs (struct msgs *, struct smsg **);
  79
  80
  81 int
  82 main (int argc, char **argv)
  83 {
  84     int i, msgnum;
  85     char *cp, *maildir, *datesw = NULL;
  86     char *folder = NULL, buf[BUFSIZ], **argp;
  87     char **arguments;
  88     struct msgs_array msgs = { 0, 0, NULL };
  89     struct msgs *mp;
  90     struct smsg **dlist;
  91     bool checksw = false;
  92
  93     if (nmh_init(argv[0], true, true)) { return 1; }
  94
  95     arguments = getarguments (invo_name, argc, argv, 1);
  96     argp = arguments;
  97
  98     /*
  99      * Parse arguments
 100      */
 101     while ((cp = *argp++)) {
 102         if (*cp == '-') {
 103             switch (smatch (++cp, switches)) {
 104             case AMBIGSW:
 105                 ambigsw (cp, switches);
 106                 done (1);
 107             case UNKWNSW:
 108                 die("-%s unknown", cp);
 109
 110             case HELPSW:
 111                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 112                         invo_name);
 113                 print_help (buf, switches, 1);
 114                 done (0);
 115             case VERSIONSW:
 116                 print_version(invo_name);
 117                 done (0);
 118
 119             case DATESW:
 120                 if (datesw)
 121                     die("only one date field at a time");
 122                 if (!(datesw = *argp++) || *datesw == '-')
 123                     die("missing argument to %s", argp[-2]);
 124                 continue;
 125
 126             case TEXTSW:
 127                 if (subjsort)
 128                     die("only one text field at a time");
 129                 if (!(subjsort = *argp++) || *subjsort == '-')
 130                     die("missing argument to %s", argp[-2]);
 131                 continue;
 132
 133             case SUBJSW:
 134                 subjsort = "subject";
 135                 continue;
 136             case NSUBJSW:
 137                 subjsort = NULL;
 138                 continue;
 139
 140             case LIMSW:
 141                 if (!(cp = *argp++) || *cp == '-')
 142                         die("missing argument to %s", argp[-2]);
 143                 while (*cp == '0')
 144                     cp++;               /* skip any leading zeros */
 145                 if (!*cp) {             /* hit end of string */
 146                     submajor = true;    /* sort subject-major */
 147                     continue;
 148                 }
 149                 if (!isdigit((unsigned char) *cp) || !(datelimit = atoi(cp)))
 150                     die("impossible limit %s", cp);
 151                 datelimit *= 60*60*24;
 152                 continue;
 153             case NLIMSW:
 154                 submajor = false;       /* use date-major, but */
 155                 datelimit = 0;          /* use no limit */
 156                 continue;
 157
 158             case VERBSW:
 159                 verbose = true;
 160                 continue;
 161             case NVERBSW:
 162                 verbose = false;
 163                 continue;
 164
 165             case ALLMSGS:
 166                 allmsgs = 1;
 167                 continue;
 168             case NALLMSGS:
 169                 allmsgs = 0;
 170                 continue;
 171
 172             case CHECKSW:
 173                 checksw = true;
 174                 continue;
 175             case NCHECKSW:
 176                 checksw = false;
 177                 continue;
 178             }
 179         }
 180         if (*cp == '+' || *cp == '@') {
 181             if (folder)
 182                 die("only one folder at a time!");
 183             folder = pluspath (cp);
 184         } else
 185             app_msgarg(&msgs, cp);
 186     }
 187
 188     if (!context_find ("path"))
 189         free (path ("./", TFOLDER));
 190     if (!msgs.size) {
 191         if (allmsgs) {
 192             app_msgarg(&msgs, "all");
 193         } else {
 194             die("must specify messages to sort with -noall");
 195         }
 196     }
 197     if (!datesw)
 198         datesw = "date";
 199     if (!folder)
 200         folder = getfolder (1);
 201     maildir = m_maildir (folder);
 202
 203     if (chdir (maildir) == NOTOK)
 204         adios (maildir, "unable to change directory to");
 205
 206     /* read folder and create message structure */
 207     if (!(mp = folder_read (folder, 1)))
 208         die("unable to read folder %s", folder);
 209
 210     /* check for empty folder */
 211     if (mp->nummsg == 0)
 212         die("no messages in %s", folder);
 213
 214     /* parse all the message ranges/sequences and set SELECTED */
 215     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 216         if (!m_convert (mp, msgs.msgs[msgnum]))
 217             done (1);
 218     seq_setprev (mp);   /* set the previous sequence */
 219
 220     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 221         die("no messages to sort");
 222
 223     if (checksw  &&  check_failed) {
 224         die("errors found, no messages sorted");
 225     }
 226
 227     /*
 228      * sort a list of pointers to our "messages to be sorted".
 229      */
 230     dlist = mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 231     for (i = 0; i < nmsgs; i++)
 232         dlist[i] = &smsgs[i];
 233     dlist[nmsgs] = 0;
 234
 235     if (verbose) {      /* announce what we're doing */
 236         if (subjsort)
 237             if (submajor)
 238                 printf ("sorting by %s\n", subjsort);
 239             else
 240                 printf ("sorting by %s-major %s-minor\n", subjsort, datesw);
 241         else
 242             printf ("sorting by datefield %s\n", datesw);
 243     }
 244
 245     /* first sort by date, or by subject-major, date-minor */
 246     qsort (dlist, nmsgs, sizeof(*dlist),
 247             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 248
 249     /*
 250      * if we're sorting on subject, we need another list
 251      * in subject order, then a merge pass to collate the
 252      * two sorts.
 253      */
 254     if (!submajor && subjsort) {        /* already date sorted */
 255         struct smsg **slist, **flist;
 256         struct smsg ***il, **fp, **dp;
 257
 258         slist = mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 259         memcpy(slist, dlist, (nmsgs+1)*sizeof(*slist));
 260         qsort(slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 261
 262         /*
 263          * make an inversion list so we can quickly find
 264          * the collection of messages with the same subj
 265          * given a message number.
 266          */
 267         il = mh_xcalloc(mp->hghsel + 1, sizeof *il);
 268         for (i = 0; i < nmsgs; i++)
 269             il[slist[i]->s_msg] = &slist[i];
 270         /*
 271          * make up the final list, chronological but with
 272          * all the same subjects grouped together.
 273          */
 274         flist = mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 275         fp = flist;
 276         for (dp = dlist; *dp;) {
 277             struct smsg **s = il[(*dp++)->s_msg];
 278
 279             /* see if we already did this guy */
 280             if (! s)
 281                 continue;
 282
 283             *fp++ = *s++;
 284             /*
 285              * take the next message(s) if there is one,
 286              * its subject isn't null and its subject
 287              * is the same as this one and it's not too
 288              * far away in time.
 289              */
 290             while (*s && (*s)->s_subj[0] &&
 291                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 292                    (datelimit == 0 ||
 293                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 294                 il[(*s)->s_msg] = 0;
 295                 *fp++ = *s++;
 296             }
 297         }
 298         *fp = 0;
 299         free (il);
 300         free (slist);
 301         free (dlist);
 302         dlist = flist;
 303     }
 304
 305     /*
 306      * At this point, dlist is a sorted array of pointers to smsg structures,
 307      * each of which contains a message number.
 308      */
 309
 310     rename_msgs (mp, dlist);
 311
 312     context_replace (pfolder, folder);  /* update current folder         */
 313     seq_save (mp);                      /* synchronize message sequences */
 314     context_save ();                    /* save the context file         */
 315     folder_free (mp);                   /* free folder/message structure */
 316     done (0);
 317     return 1;
 318 }
 319
 320 static int
 321 read_hdrs (struct msgs *mp, char *datesw)
 322 {
 323     int msgnum;
 324     struct smsg *s;
 325
 326     smsgs = mh_xcalloc(mp->hghsel - mp->lowsel + 2, sizeof *smsgs);
 327     s = smsgs;
 328     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 329         if (is_selected(mp, msgnum)) {
 330             if (get_fields (datesw, msgnum, s)) {
 331                 s->s_msg = msgnum;
 332                 s++;
 333             }
 334         }
 335     }
 336     s->s_msg = 0;
 337     return s - smsgs;
 338 }
 339
 340
 341 /*
 342  * Parse the message and get the data or subject field,
 343  * if needed.
 344  */
 345
 346 static int
 347 get_fields (char *datesw, int msg, struct smsg *smsg)
 348 {
 349     int state;
 350     int compnum;
 351     char *msgnam, buf[NMH_BUFSIZ], nam[NAMESZ];
 352     struct tws *tw;
 353     char *datecomp = NULL, *subjcomp = NULL;
 354     FILE *in;
 355     m_getfld_state_t gstate;
 356
 357     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 358         admonish (msgnam, "unable to read message");
 359         return 0;
 360     }
 361     gstate = m_getfld_state_init(in);
 362     for (compnum = 1;;) {
 363         int bufsz = sizeof buf;
 364         switch (state = m_getfld2(&gstate, nam, buf, &bufsz)) {
 365         case FLD:
 366         case FLDPLUS:
 367             compnum++;
 368             if (!strcasecmp (nam, datesw)) {
 369                 datecomp = add (buf, datecomp);
 370                 while (state == FLDPLUS) {
 371                     bufsz = sizeof buf;
 372                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 373                     datecomp = add (buf, datecomp);
 374                 }
 375                 if (!subjsort || subjcomp)
 376                     break;
 377             } else if (subjsort && !strcasecmp (nam, subjsort)) {
 378                 subjcomp = add (buf, subjcomp);
 379                 while (state == FLDPLUS) {
 380                     bufsz = sizeof buf;
 381                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 382                     subjcomp = add (buf, subjcomp);
 383                 }
 384                 if (datecomp)
 385                     break;
 386             } else {
 387                 /* just flush this guy */
 388                 while (state == FLDPLUS) {
 389                     bufsz = sizeof buf;
 390                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 391                 }
 392             }
 393             continue;
 394
 395         case BODY:
 396         case FILEEOF:
 397             break;
 398
 399         case LENERR:
 400         case FMTERR:
 401             if (state == LENERR || state == FMTERR) {
 402                 inform("format error in message %d (header #%d), continuing...",
 403                       msg, compnum);
 404                 check_failed = 1;
 405             }
 406             free(datecomp);
 407             free(subjcomp);
 408             fclose (in);
 409             return 0;
 410
 411         default:
 412             die("internal error -- you lose");
 413         }
 414         break;
 415     }
 416     m_getfld_state_destroy (&gstate);
 417
 418     /*
 419      * If no date component, then use the modification
 420      * time of the file as its date
 421      */
 422     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 423         struct stat st;
 424
 425         inform("can't parse %s field in message %d, "
 426             "will use file modification time", datesw, msg);
 427         fstat (fileno (in), &st);
 428         smsg->s_clock = st.st_mtime;
 429         check_failed = 1;
 430     } else {
 431         smsg->s_clock = dmktime (tw);
 432     }
 433
 434     if (subjsort) {
 435         if (subjcomp) {
 436             /*
 437              * try to make the subject "canonical": delete
 438              * leading "re:", everything but letters & smash
 439              * letters to lower case.
 440              */
 441             char  *cp, *cp2, c;
 442
 443             cp = subjcomp;
 444             cp2 = subjcomp;
 445             if (strcmp (subjsort, "subject") == 0) {
 446                 while ((c = *cp)) {
 447                     if (! isspace((unsigned char) c)) {
 448                         if(!uprf(cp, "re:"))
 449                             break;
 450                         cp += 2;
 451                     }
 452                     cp++;
 453                 }
 454             }
 455
 456             while ((c = *cp++)) {
 457                 if (isascii((unsigned char) c) && isalnum((unsigned char) c))
 458                     *cp2++ = tolower((unsigned char)c);
 459             }
 460
 461             *cp2 = '\0';
 462         }
 463         else
 464             subjcomp = "";
 465
 466         smsg->s_subj = subjcomp;
 467     }
 468     fclose (in);
 469     free(datecomp);
 470
 471     return 1;
 472 }
 473
 474 /*
 475  * sort on dates.
 476  */
 477 static int
 478 dsort (struct smsg **a, struct smsg **b)
 479 {
 480     if ((*a)->s_clock < (*b)->s_clock)
 481         return -1;
 482     if ((*a)->s_clock > (*b)->s_clock)
 483         return 1;
 484     if ((*a)->s_msg < (*b)->s_msg)
 485         return -1;
 486     return 1;
 487 }
 488
 489 /*
 490  * sort on subjects.
 491  */
 492 static int
 493 subsort (struct smsg **a, struct smsg **b)
 494 {
 495     int i;
 496
 497     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 498         return i;
 499
 500     return dsort(a, b);
 501 }
 502
 503 static int
 504 txtsort (struct smsg **a, struct smsg **b)
 505 {
 506     int i;
 507
 508     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 509         return i;
 510     if ((*a)->s_msg < (*b)->s_msg)
 511         return -1;
 512     return 1;
 513 }
 514
 515 static void
 516 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 517 {
 518     int nxt, old, new;
 519     char *newname, oldname[BUFSIZ];
 520     char newbuf[PATH_MAX + 1];
 521
 522     for (;;) {
 523         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 524         mlist[msg] = NULL;
 525         old = smsgs[nxt].s_msg;
 526         new = smsgs[msg].s_msg;
 527         strncpy (oldname, m_name (old), sizeof(oldname));
 528         newname = m_name (new);
 529         if (verbose)
 530             printf ("message %d becomes message %d\n", old, new);
 531
 532         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 533         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 534         ext_hook("ref-hook", oldname, newbuf);
 535
 536         if (rename (oldname, newname) == NOTOK)
 537             adios (newname, "unable to rename %s to", oldname);
 538
 539         copy_msg_flags (mp, new, old);
 540         if (mp->curmsg == old)
 541             seq_setcur (mp, new);
 542
 543         if (nxt == endmsg)
 544             break;
 545
 546         msg = nxt;
 547     }
 548 /*      if (nxt != endmsg); */
 549 /*      rename_chain (mp, mlist, nxt, endmsg); */
 550 }
 551
 552 static void
 553 rename_msgs (struct msgs *mp, struct smsg **mlist)
 554 {
 555     int i, j, old, new;
 556     bvector_t tmpset = bvector_create ();
 557     char f1[BUFSIZ], tmpfil[BUFSIZ];
 558     char newbuf[PATH_MAX + 1];
 559     struct smsg *sp;
 560
 561     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 562
 563     for (i = 0; i < nmsgs; i++) {
 564         if (! (sp = mlist[i]))
 565             continue;   /* did this one */
 566
 567         j = sp - smsgs;
 568         if (j == i)
 569             continue;   /* this one doesn't move */
 570
 571         /*
 572          * the guy that was msg j is about to become msg i.
 573          * rename 'j' to make a hole, then recursively rename
 574          * guys to fill up the hole.
 575          */
 576         old = smsgs[j].s_msg;
 577         new = smsgs[i].s_msg;
 578         strncpy (f1, m_name (old), sizeof(f1));
 579
 580         if (verbose)
 581             printf ("renaming message chain from %d to %d\n", old, new);
 582
 583         /*
 584          *      Run the external hook to refile the old message as the
 585          *      temporary message number that is off of the end of the
 586          *      messages in the folder.
 587          */
 588
 589         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 590         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 591         ext_hook("ref-hook", f1, newbuf);
 592
 593         if (rename (f1, tmpfil) == NOTOK)
 594             adios (tmpfil, "unable to rename %s to ", f1);
 595
 596         get_msg_flags (mp, tmpset, old);
 597
 598         rename_chain (mp, mlist, j, i);
 599
 600         /*
 601          *      Run the external hook to refile the temporary message number
 602          *      to the real place.
 603          */
 604
 605         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 606         ext_hook("ref-hook", newbuf, f1);
 607
 608         if (rename (tmpfil, m_name(new)) == NOTOK)
 609             adios (m_name(new), "unable to rename %s to", tmpfil);
 610
 611         set_msg_flags (mp, tmpset, new);
 612         mp->msgflags |= SEQMOD;
 613     }
 614
 615     bvector_free (tmpset);
 616 }