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/uprf.h"
  10 #include "sbr/m_convert.h"
  11 #include "sbr/getfolder.h"
  12 #include "sbr/ext_hook.h"
  13 #include "sbr/folder_read.h"
  14 #include "sbr/folder_free.h"
  15 #include "sbr/context_save.h"
  16 #include "sbr/context_replace.h"
  17 #include "sbr/context_find.h"
  18 #include "sbr/ambigsw.h"
  19 #include "sbr/path.h"
  20 #include "sbr/print_version.h"
  21 #include "sbr/print_help.h"
  22 #include "sbr/error.h"
  23 #include "h/tws.h"
  24 #include "h/done.h"
  25 #include "h/utils.h"
  26 #include "sbr/m_maildir.h"
  27
  28 #define SORTM_SWITCHES \
  29     X("datefield field", 0, DATESW) \
  30     X("textfield field", 0, TEXTSW) \
  31     X("notextfield", 0, NSUBJSW) \
  32     X("subject", -3, SUBJSW) /* backward-compatibility */ \
  33     X("limit days", 0, LIMSW) \
  34     X("nolimit", 0, NLIMSW) \
  35     X("verbose", 0, VERBSW) \
  36     X("noverbose", 0, NVERBSW) \
  37     X("all", 0, ALLMSGS) \
  38     X("noall", 0, NALLMSGS) \
  39     X("check", 0, CHECKSW) \
  40     X("nocheck", 0, NCHECKSW) \
  41     X("version", 0, VERSIONSW) \
  42     X("help", 0, HELPSW) \
  43
  44 #define X(sw, minchars, id) id,
  45 DEFINE_SWITCH_ENUM(SORTM);
  46 #undef X
  47
  48 #define X(sw, minchars, id) { sw, minchars, id },
  49 DEFINE_SWITCH_ARRAY(SORTM, switches);
  50 #undef X
  51
  52 struct smsg {
  53     int s_msg;
  54     time_t s_clock;
  55     char *s_subj;
  56 };
  57
  58 static struct smsg *smsgs;
  59 int nmsgs;
  60
  61 char *subjsort;                 /* sort on subject if != 0 */
  62 time_t datelimit = 0;
  63 bool submajor;                  /* if true, sort on subject-major */
  64 bool verbose;
  65 int allmsgs = 1;
  66 int check_failed = 0;
  67
  68 /* This keeps compiler happy on calls to qsort */
  69 typedef int (*qsort_comp) (const void *, const void *);
  70
  71 /*
  72  * static prototypes
  73  */
  74 static int read_hdrs (struct msgs *, char *);
  75 static int get_fields (char *, int, struct smsg *);
  76 static int dsort (struct smsg **, struct smsg **);
  77 static int subsort (struct smsg **, struct smsg **);
  78 static int txtsort (struct smsg **, struct smsg **);
  79 static void rename_chain (struct msgs *, struct smsg **, int, int);
  80 static void rename_msgs (struct msgs *, struct smsg **);
  81
  82
  83 int
  84 main (int argc, char **argv)
  85 {
  86     int i, msgnum;
  87     char *cp, *maildir, *datesw = NULL;
  88     char *folder = NULL, buf[BUFSIZ], **argp;
  89     char **arguments;
  90     struct msgs_array msgs = { 0, 0, NULL };
  91     struct msgs *mp;
  92     struct smsg **dlist;
  93     bool checksw = false;
  94
  95     if (nmh_init(argv[0], true, true)) { return 1; }
  96
  97     arguments = getarguments (invo_name, argc, argv, 1);
  98     argp = arguments;
  99
 100     /*
 101      * Parse arguments
 102      */
 103     while ((cp = *argp++)) {
 104         if (*cp == '-') {
 105             switch (smatch (++cp, switches)) {
 106             case AMBIGSW:
 107                 ambigsw (cp, switches);
 108                 done (1);
 109             case UNKWNSW:
 110                 die("-%s unknown", cp);
 111
 112             case HELPSW:
 113                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 114                         invo_name);
 115                 print_help (buf, switches, 1);
 116                 done (0);
 117             case VERSIONSW:
 118                 print_version(invo_name);
 119                 done (0);
 120
 121             case DATESW:
 122                 if (datesw)
 123                     die("only one date field at a time");
 124                 if (!(datesw = *argp++) || *datesw == '-')
 125                     die("missing argument to %s", argp[-2]);
 126                 continue;
 127
 128             case TEXTSW:
 129                 if (subjsort)
 130                     die("only one text field at a time");
 131                 if (!(subjsort = *argp++) || *subjsort == '-')
 132                     die("missing argument to %s", argp[-2]);
 133                 continue;
 134
 135             case SUBJSW:
 136                 subjsort = "subject";
 137                 continue;
 138             case NSUBJSW:
 139                 subjsort = NULL;
 140                 continue;
 141
 142             case LIMSW:
 143                 if (!(cp = *argp++) || *cp == '-')
 144                         die("missing argument to %s", argp[-2]);
 145                 while (*cp == '0')
 146                     cp++;               /* skip any leading zeros */
 147                 if (!*cp) {             /* hit end of string */
 148                     submajor = true;    /* sort subject-major */
 149                     continue;
 150                 }
 151                 if (!isdigit((unsigned char) *cp) || !(datelimit = atoi(cp)))
 152                     die("impossible limit %s", cp);
 153                 datelimit *= 60*60*24;
 154                 continue;
 155             case NLIMSW:
 156                 submajor = false;       /* use date-major, but */
 157                 datelimit = 0;          /* use no limit */
 158                 continue;
 159
 160             case VERBSW:
 161                 verbose = true;
 162                 continue;
 163             case NVERBSW:
 164                 verbose = false;
 165                 continue;
 166
 167             case ALLMSGS:
 168                 allmsgs = 1;
 169                 continue;
 170             case NALLMSGS:
 171                 allmsgs = 0;
 172                 continue;
 173
 174             case CHECKSW:
 175                 checksw = true;
 176                 continue;
 177             case NCHECKSW:
 178                 checksw = false;
 179                 continue;
 180             }
 181         }
 182         if (*cp == '+' || *cp == '@') {
 183             if (folder)
 184                 die("only one folder at a time!");
 185             folder = pluspath (cp);
 186         } else
 187             app_msgarg(&msgs, cp);
 188     }
 189
 190     if (!context_find ("path"))
 191         free (path ("./", TFOLDER));
 192     if (!msgs.size) {
 193         if (allmsgs) {
 194             app_msgarg(&msgs, "all");
 195         } else {
 196             die("must specify messages to sort with -noall");
 197         }
 198     }
 199     if (!datesw)
 200         datesw = "date";
 201     if (!folder)
 202         folder = getfolder (1);
 203     maildir = m_maildir (folder);
 204
 205     if (chdir (maildir) == NOTOK)
 206         adios (maildir, "unable to change directory to");
 207
 208     /* read folder and create message structure */
 209     if (!(mp = folder_read (folder, 1)))
 210         die("unable to read folder %s", folder);
 211
 212     /* check for empty folder */
 213     if (mp->nummsg == 0)
 214         die("no messages in %s", folder);
 215
 216     /* parse all the message ranges/sequences and set SELECTED */
 217     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 218         if (!m_convert (mp, msgs.msgs[msgnum]))
 219             done (1);
 220     seq_setprev (mp);   /* set the previous sequence */
 221
 222     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 223         die("no messages to sort");
 224
 225     if (checksw  &&  check_failed) {
 226         die("errors found, no messages sorted");
 227     }
 228
 229     /*
 230      * sort a list of pointers to our "messages to be sorted".
 231      */
 232     dlist = mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 233     for (i = 0; i < nmsgs; i++)
 234         dlist[i] = &smsgs[i];
 235     dlist[nmsgs] = 0;
 236
 237     if (verbose) {      /* announce what we're doing */
 238         if (subjsort)
 239             if (submajor)
 240                 printf ("sorting by %s\n", subjsort);
 241             else
 242                 printf ("sorting by %s-major %s-minor\n", subjsort, datesw);
 243         else
 244             printf ("sorting by datefield %s\n", datesw);
 245     }
 246
 247     /* first sort by date, or by subject-major, date-minor */
 248     qsort (dlist, nmsgs, sizeof(*dlist),
 249             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 250
 251     /*
 252      * if we're sorting on subject, we need another list
 253      * in subject order, then a merge pass to collate the
 254      * two sorts.
 255      */
 256     if (!submajor && subjsort) {        /* already date sorted */
 257         struct smsg **slist, **flist;
 258         struct smsg ***il, **fp, **dp;
 259
 260         slist = mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 261         memcpy(slist, dlist, (nmsgs+1)*sizeof(*slist));
 262         qsort(slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 263
 264         /*
 265          * make an inversion list so we can quickly find
 266          * the collection of messages with the same subj
 267          * given a message number.
 268          */
 269         il = mh_xcalloc(mp->hghsel + 1, sizeof *il);
 270         for (i = 0; i < nmsgs; i++)
 271             il[slist[i]->s_msg] = &slist[i];
 272         /*
 273          * make up the final list, chronological but with
 274          * all the same subjects grouped together.
 275          */
 276         flist = mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 277         fp = flist;
 278         for (dp = dlist; *dp;) {
 279             struct smsg **s = il[(*dp++)->s_msg];
 280
 281             /* see if we already did this guy */
 282             if (! s)
 283                 continue;
 284
 285             *fp++ = *s++;
 286             /*
 287              * take the next message(s) if there is one,
 288              * its subject isn't null and its subject
 289              * is the same as this one and it's not too
 290              * far away in time.
 291              */
 292             while (*s && (*s)->s_subj[0] &&
 293                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 294                    (datelimit == 0 ||
 295                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 296                 il[(*s)->s_msg] = 0;
 297                 *fp++ = *s++;
 298             }
 299         }
 300         *fp = 0;
 301         free (il);
 302         free (slist);
 303         free (dlist);
 304         dlist = flist;
 305     }
 306
 307     /*
 308      * At this point, dlist is a sorted array of pointers to smsg structures,
 309      * each of which contains a message number.
 310      */
 311
 312     rename_msgs (mp, dlist);
 313
 314     context_replace (pfolder, folder);  /* update current folder         */
 315     seq_save (mp);                      /* synchronize message sequences */
 316     context_save ();                    /* save the context file         */
 317     folder_free (mp);                   /* free folder/message structure */
 318     done (0);
 319     return 1;
 320 }
 321
 322 static int
 323 read_hdrs (struct msgs *mp, char *datesw)
 324 {
 325     int msgnum;
 326     struct smsg *s;
 327
 328     smsgs = mh_xcalloc(mp->hghsel - mp->lowsel + 2, sizeof *smsgs);
 329     s = smsgs;
 330     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 331         if (is_selected(mp, msgnum)) {
 332             if (get_fields (datesw, msgnum, s)) {
 333                 s->s_msg = msgnum;
 334                 s++;
 335             }
 336         }
 337     }
 338     s->s_msg = 0;
 339     return s - smsgs;
 340 }
 341
 342
 343 /*
 344  * Parse the message and get the data or subject field,
 345  * if needed.
 346  */
 347
 348 static int
 349 get_fields (char *datesw, int msg, struct smsg *smsg)
 350 {
 351     int state;
 352     int compnum;
 353     char *msgnam, buf[NMH_BUFSIZ], nam[NAMESZ];
 354     struct tws *tw;
 355     char *datecomp = NULL, *subjcomp = NULL;
 356     FILE *in;
 357     m_getfld_state_t gstate;
 358
 359     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 360         admonish (msgnam, "unable to read message");
 361         return 0;
 362     }
 363     gstate = m_getfld_state_init(in);
 364     for (compnum = 1;;) {
 365         int bufsz = sizeof buf;
 366         switch (state = m_getfld2(&gstate, nam, buf, &bufsz)) {
 367         case FLD:
 368         case FLDPLUS:
 369             compnum++;
 370             if (!strcasecmp (nam, datesw)) {
 371                 datecomp = add (buf, datecomp);
 372                 while (state == FLDPLUS) {
 373                     bufsz = sizeof buf;
 374                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 375                     datecomp = add (buf, datecomp);
 376                 }
 377                 if (!subjsort || subjcomp)
 378                     break;
 379             } else if (subjsort && !strcasecmp (nam, subjsort)) {
 380                 subjcomp = add (buf, subjcomp);
 381                 while (state == FLDPLUS) {
 382                     bufsz = sizeof buf;
 383                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 384                     subjcomp = add (buf, subjcomp);
 385                 }
 386                 if (datecomp)
 387                     break;
 388             } else {
 389                 /* just flush this guy */
 390                 while (state == FLDPLUS) {
 391                     bufsz = sizeof buf;
 392                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 393                 }
 394             }
 395             continue;
 396
 397         case BODY:
 398         case FILEEOF:
 399             break;
 400
 401         case LENERR:
 402         case FMTERR:
 403             if (state == LENERR || state == FMTERR) {
 404                 inform("format error in message %d (header #%d), continuing...",
 405                       msg, compnum);
 406                 check_failed = 1;
 407             }
 408             free(datecomp);
 409             free(subjcomp);
 410             fclose (in);
 411             return 0;
 412
 413         default:
 414             die("internal error -- you lose");
 415         }
 416         break;
 417     }
 418     m_getfld_state_destroy (&gstate);
 419
 420     /*
 421      * If no date component, then use the modification
 422      * time of the file as its date
 423      */
 424     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 425         struct stat st;
 426
 427         inform("can't parse %s field in message %d, "
 428             "will use file modification time", datesw, msg);
 429         fstat (fileno (in), &st);
 430         smsg->s_clock = st.st_mtime;
 431         check_failed = 1;
 432     } else {
 433         smsg->s_clock = dmktime (tw);
 434     }
 435
 436     if (subjsort) {
 437         if (subjcomp) {
 438             /*
 439              * try to make the subject "canonical": delete
 440              * leading "re:", everything but letters & smash
 441              * letters to lower case.
 442              */
 443             char  *cp, *cp2, c;
 444
 445             cp = subjcomp;
 446             cp2 = subjcomp;
 447             if (strcmp (subjsort, "subject") == 0) {
 448                 while ((c = *cp)) {
 449                     if (! isspace((unsigned char) c)) {
 450                         if(!uprf(cp, "re:"))
 451                             break;
 452                         cp += 2;
 453                     }
 454                     cp++;
 455                 }
 456             }
 457
 458             while ((c = *cp++)) {
 459                 if (isascii((unsigned char) c) && isalnum((unsigned char) c))
 460                     *cp2++ = tolower((unsigned char)c);
 461             }
 462
 463             *cp2 = '\0';
 464         }
 465         else
 466             subjcomp = "";
 467
 468         smsg->s_subj = subjcomp;
 469     }
 470     fclose (in);
 471     free(datecomp);
 472
 473     return 1;
 474 }
 475
 476 /*
 477  * sort on dates.
 478  */
 479 static int
 480 dsort (struct smsg **a, struct smsg **b)
 481 {
 482     if ((*a)->s_clock < (*b)->s_clock)
 483         return -1;
 484     if ((*a)->s_clock > (*b)->s_clock)
 485         return 1;
 486     if ((*a)->s_msg < (*b)->s_msg)
 487         return -1;
 488     return 1;
 489 }
 490
 491 /*
 492  * sort on subjects.
 493  */
 494 static int
 495 subsort (struct smsg **a, struct smsg **b)
 496 {
 497     int i;
 498
 499     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 500         return i;
 501
 502     return dsort(a, b);
 503 }
 504
 505 static int
 506 txtsort (struct smsg **a, struct smsg **b)
 507 {
 508     int i;
 509
 510     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 511         return i;
 512     if ((*a)->s_msg < (*b)->s_msg)
 513         return -1;
 514     return 1;
 515 }
 516
 517 static void
 518 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 519 {
 520     int nxt, old, new;
 521     char *newname, oldname[BUFSIZ];
 522     char newbuf[PATH_MAX + 1];
 523
 524     for (;;) {
 525         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 526         mlist[msg] = NULL;
 527         old = smsgs[nxt].s_msg;
 528         new = smsgs[msg].s_msg;
 529         strncpy (oldname, m_name (old), sizeof(oldname));
 530         newname = m_name (new);
 531         if (verbose)
 532             printf ("message %d becomes message %d\n", old, new);
 533
 534         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 535         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 536         ext_hook("ref-hook", oldname, newbuf);
 537
 538         if (rename (oldname, newname) == NOTOK)
 539             adios (newname, "unable to rename %s to", oldname);
 540
 541         copy_msg_flags (mp, new, old);
 542         if (mp->curmsg == old)
 543             seq_setcur (mp, new);
 544
 545         if (nxt == endmsg)
 546             break;
 547
 548         msg = nxt;
 549     }
 550 /*      if (nxt != endmsg); */
 551 /*      rename_chain (mp, mlist, nxt, endmsg); */
 552 }
 553
 554 static void
 555 rename_msgs (struct msgs *mp, struct smsg **mlist)
 556 {
 557     int i, j, old, new;
 558     bvector_t tmpset = bvector_create ();
 559     char f1[BUFSIZ], tmpfil[BUFSIZ];
 560     char newbuf[PATH_MAX + 1];
 561     struct smsg *sp;
 562
 563     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 564
 565     for (i = 0; i < nmsgs; i++) {
 566         if (! (sp = mlist[i]))
 567             continue;   /* did this one */
 568
 569         j = sp - smsgs;
 570         if (j == i)
 571             continue;   /* this one doesn't move */
 572
 573         /*
 574          * the guy that was msg j is about to become msg i.
 575          * rename 'j' to make a hole, then recursively rename
 576          * guys to fill up the hole.
 577          */
 578         old = smsgs[j].s_msg;
 579         new = smsgs[i].s_msg;
 580         strncpy (f1, m_name (old), sizeof(f1));
 581
 582         if (verbose)
 583             printf ("renaming message chain from %d to %d\n", old, new);
 584
 585         /*
 586          *      Run the external hook to refile the old message as the
 587          *      temporary message number that is off of the end of the
 588          *      messages in the folder.
 589          */
 590
 591         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 592         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 593         ext_hook("ref-hook", f1, newbuf);
 594
 595         if (rename (f1, tmpfil) == NOTOK)
 596             adios (tmpfil, "unable to rename %s to ", f1);
 597
 598         get_msg_flags (mp, tmpset, old);
 599
 600         rename_chain (mp, mlist, j, i);
 601
 602         /*
 603          *      Run the external hook to refile the temporary message number
 604          *      to the real place.
 605          */
 606
 607         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 608         ext_hook("ref-hook", newbuf, f1);
 609
 610         if (rename (tmpfil, m_name(new)) == NOTOK)
 611             adios (m_name(new), "unable to rename %s to", tmpfil);
 612
 613         set_msg_flags (mp, tmpset, new);
 614         mp->msgflags |= SEQMOD;
 615     }
 616
 617     bvector_free (tmpset);
 618 }