diplodocus.org Git - nmh/blob - uip/sortm.c

   1
   2 /*
   3  * sortm.c -- sort messages in a folder by date/time
   4  *
   5  * $Id$
   6  *
   7  * This code is Copyright (c) 2002, by the authors of nmh.  See the
   8  * COPYRIGHT file in the root directory of the nmh distribution for
   9  * complete copyright information.
  10  */
  11
  12 #include <h/mh.h>
  13 #include <h/tws.h>
  14
  15 /*
  16  * We allocate space for messages (msgs array)
  17  * this number of elements at a time.
  18  */
  19 #define MAXMSGS  256
  20
  21
  22 static struct swit switches[] = {
  23 #define DATESW                 0
  24      { "datefield field", 0 },
  25 #define TEXTSW                 1
  26      { "textfield field", 0 },
  27 #define NSUBJSW                2
  28      { "notextfield", 0 },
  29 #define SUBJSW                 3
  30      { "subject", -3 },            /* backward-compatibility */
  31 #define LIMSW                  4
  32      { "limit days", 0 },
  33 #define NLIMSW                 5
  34      { "nolimit", 0 },
  35 #define VERBSW                 6
  36      { "verbose", 0 },
  37 #define NVERBSW                7
  38      { "noverbose", 0 },
  39 #define VERSIONSW              8
  40      { "version", 0 },
  41 #define HELPSW                 9
  42      { "help", 0 },
  43      { NULL, 0 }
  44 };
  45
  46 struct smsg {
  47     int s_msg;
  48     time_t s_clock;
  49     char *s_subj;
  50 };
  51
  52 static struct smsg *smsgs;
  53 int nmsgs;
  54
  55 char *subjsort = (char *) 0;    /* sort on subject if != 0 */
  56 unsigned long datelimit = 0;
  57 int submajor = 0;               /* if true, sort on subject-major */
  58 int verbose;
  59
  60 /* This keeps compiler happy on calls to qsort */
  61 typedef int (*qsort_comp) (const void *, const void *);
  62
  63 /*
  64  * static prototypes
  65  */
  66 static int read_hdrs (struct msgs *, char *);
  67 static int get_fields (char *, int, struct smsg *);
  68 static int dsort (struct smsg **, struct smsg **);
  69 static int subsort (struct smsg **, struct smsg **);
  70 static int txtsort (struct smsg **, struct smsg **);
  71 static void rename_chain (struct msgs *, struct smsg **, int, int);
  72 static void rename_msgs (struct msgs *, struct smsg **);
  73
  74
  75 int
  76 main (int argc, char **argv)
  77 {
  78     int nummsgs, maxmsgs, i, msgnum;
  79     char *cp, *maildir, *datesw = NULL;
  80     char *folder = NULL, buf[BUFSIZ], **argp;
  81     char **arguments, **msgs;
  82     struct msgs *mp;
  83     struct smsg **dlist;
  84
  85 #ifdef LOCALE
  86     setlocale(LC_ALL, "");
  87 #endif
  88     invo_name = r1bindex (argv[0], '/');
  89
  90     /* read user profile/context */
  91     context_read();
  92
  93     arguments = getarguments (invo_name, argc, argv, 1);
  94     argp = arguments;
  95
  96     /*
  97      * Allocate the initial space to record message
  98      * names and ranges.
  99      */
 100     nummsgs = 0;
 101     maxmsgs = MAXMSGS;
 102     if (!(msgs = (char **) malloc ((size_t) (maxmsgs * sizeof(*msgs)))))
 103         adios (NULL, "unable to allocate storage");
 104
 105     /*
 106      * Parse arguments
 107      */
 108     while ((cp = *argp++)) {
 109         if (*cp == '-') {
 110             switch (smatch (++cp, switches)) {
 111             case AMBIGSW:
 112                 ambigsw (cp, switches);
 113                 done (1);
 114             case UNKWNSW:
 115                 adios (NULL, "-%s unknown", cp);
 116
 117             case HELPSW:
 118                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 119                         invo_name);
 120                 print_help (buf, switches, 1);
 121                 done (1);
 122             case VERSIONSW:
 123                 print_version(invo_name);
 124                 done (1);
 125
 126             case DATESW:
 127                 if (datesw)
 128                     adios (NULL, "only one date field at a time");
 129                 if (!(datesw = *argp++) || *datesw == '-')
 130                     adios (NULL, "missing argument to %s", argp[-2]);
 131                 continue;
 132
 133             case TEXTSW:
 134                 if (subjsort)
 135                     adios (NULL, "only one text field at a time");
 136                 if (!(subjsort = *argp++) || *subjsort == '-')
 137                     adios (NULL, "missing argument to %s", argp[-2]);
 138                 continue;
 139
 140             case SUBJSW:
 141                 subjsort = "subject";
 142                 continue;
 143             case NSUBJSW:
 144                 subjsort = (char *)0;
 145                 continue;
 146
 147             case LIMSW:
 148                 if (!(cp = *argp++) || *cp == '-')
 149                         adios (NULL, "missing argument to %s", argp[-2]);
 150                 while (*cp == '0')
 151                     cp++;               /* skip any leading zeros */
 152                 if (!*cp) {             /* hit end of string */
 153                     submajor++;         /* sort subject-major */
 154                     continue;
 155                 }
 156                 if (!isdigit(*cp) || !(datelimit = atoi(cp)))
 157                     adios (NULL, "impossible limit %s", cp);
 158                 datelimit *= 60*60*24;
 159                 continue;
 160             case NLIMSW:
 161                 submajor = 0;   /* use date-major, but */
 162                 datelimit = 0;  /* use no limit */
 163                 continue;
 164
 165             case VERBSW:
 166                 verbose++;
 167                 continue;
 168             case NVERBSW:
 169                 verbose = 0;
 170                 continue;
 171             }
 172         }
 173         if (*cp == '+' || *cp == '@') {
 174             if (folder)
 175                 adios (NULL, "only one folder at a time!");
 176             else
 177                 folder = path (cp + 1, *cp == '+' ? TFOLDER : TSUBCWF);
 178         } else {
 179             /*
 180              * Check if we need to allocate more space
 181              * for message names/ranges.
 182              */
 183             if (nummsgs >= maxmsgs) {
 184                 maxmsgs += MAXMSGS;
 185                 if (!(msgs = (char **) realloc (msgs,
 186                         (size_t) (maxmsgs * sizeof(*msgs)))))
 187                     adios (NULL, "unable to reallocate msgs storage");
 188             }
 189             msgs[nummsgs++] = cp;
 190         }
 191     }
 192
 193     if (!context_find ("path"))
 194         free (path ("./", TFOLDER));
 195     if (!nummsgs)
 196         msgs[nummsgs++] = "all";
 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)))
 208         adios (NULL, "unable to read folder %s", folder);
 209
 210     /* check for empty folder */
 211     if (mp->nummsg == 0)
 212         adios (NULL, "no messages in %s", folder);
 213
 214     /* parse all the message ranges/sequences and set SELECTED */
 215     for (msgnum = 0; msgnum < nummsgs; msgnum++)
 216         if (!m_convert (mp, msgs[msgnum]))
 217             done (1);
 218     seq_setprev (mp);   /* set the previous sequence */
 219
 220     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 221         adios (NULL, "no messages to sort");
 222
 223     /*
 224      * sort a list of pointers to our "messages to be sorted".
 225      */
 226     dlist = (struct smsg **) malloc ((nmsgs+1) * sizeof(*dlist));
 227     if (! dlist)
 228         adios (NULL, "couldn't allocate sort memory");
 229     for (i = 0; i < nmsgs; i++)
 230         dlist[i] = &smsgs[i];
 231     dlist[nmsgs] = 0;
 232
 233     if (verbose) {      /* announce what we're doing */
 234         if (subjsort)
 235             printf ("sorting by %s-major %s-minor\n",
 236                 submajor ? subjsort : datesw,
 237                 submajor ? datesw : subjsort);
 238         else
 239             printf ("sorting by datefield %s\n", datesw);
 240     }
 241
 242     /* first sort by date, or by subject-major, date-minor */
 243     qsort ((char *) dlist, nmsgs, sizeof(*dlist),
 244             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 245
 246     /*
 247      * if we're sorting on subject, we need another list
 248      * in subject order, then a merge pass to collate the
 249      * two sorts.
 250      */
 251     if (!submajor && subjsort) {        /* already date sorted */
 252         struct smsg **slist, **flist;
 253         register struct smsg ***il, **fp, **dp;
 254
 255         slist = (struct smsg **) malloc ((nmsgs+1) * sizeof(*slist));
 256         if (! slist)
 257             adios (NULL, "couldn't allocate sort memory");
 258         memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 259         qsort((char *)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 = (struct smsg ***) calloc (mp->hghsel+1, sizeof(*il));
 267         if (! il)
 268             adios (NULL, "couldn't allocate msg list");
 269         for (i = 0; i < nmsgs; i++)
 270             il[slist[i]->s_msg] = &slist[i];
 271         /*
 272          * make up the final list, chronological but with
 273          * all the same subjects grouped together.
 274          */
 275         flist = (struct smsg **) malloc ((nmsgs+1) * sizeof(*flist));
 276         if (! flist)
 277             adios (NULL, "couldn't allocate msg list");
 278         fp = flist;
 279         for (dp = dlist; *dp;) {
 280             register 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 (slist);
 303         free (dlist);
 304         dlist = flist;
 305     }
 306     rename_msgs (mp, dlist);
 307
 308     context_replace (pfolder, folder);  /* update current folder         */
 309     seq_save (mp);                      /* synchronize message sequences */
 310     context_save ();                    /* save the context file         */
 311     folder_free (mp);                   /* free folder/message structure */
 312     return done (0);
 313 }
 314
 315 static int
 316 read_hdrs (struct msgs *mp, char *datesw)
 317 {
 318     int msgnum;
 319     struct tws tb;
 320     register struct smsg *s;
 321
 322     twscopy (&tb, dlocaltimenow ());
 323
 324     smsgs = (struct smsg *)
 325         calloc ((size_t) (mp->hghsel - mp->lowsel + 2),
 326             sizeof(*smsgs));
 327     if (smsgs == NULL)
 328         adios (NULL, "unable to allocate sort storage");
 329
 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     register int state;
 353     int compnum;
 354     char *msgnam, buf[BUFSIZ], nam[NAMESZ];
 355     register struct tws *tw;
 356     register char *datecomp = NULL, *subjcomp = NULL;
 357     register FILE *in;
 358
 359     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 360         admonish (msgnam, "unable to read message");
 361         return (0);
 362     }
 363     for (compnum = 1, state = FLD;;) {
 364         switch (state = m_getfld (state, nam, buf, sizeof(buf), in)) {
 365         case FLD:
 366         case FLDEOF:
 367         case FLDPLUS:
 368             compnum++;
 369             if (!strcasecmp (nam, datesw)) {
 370                 datecomp = add (buf, datecomp);
 371                 while (state == FLDPLUS) {
 372                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 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                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 381                     subjcomp = add (buf, subjcomp);
 382                 }
 383                 if (datecomp)
 384                     break;
 385             } else {
 386                 /* just flush this guy */
 387                 while (state == FLDPLUS)
 388                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 389             }
 390             continue;
 391
 392         case BODY:
 393         case BODYEOF:
 394         case FILEEOF:
 395             break;
 396
 397         case LENERR:
 398         case FMTERR:
 399             if (state == LENERR || state == FMTERR)
 400                 admonish (NULL, "format error in message %d (header #%d)",
 401                       msg, compnum);
 402             if (datecomp)
 403                 free (datecomp);
 404             if (subjcomp)
 405                 free (subjcomp);
 406             fclose (in);
 407             return (0);
 408
 409         default:
 410             adios (NULL, "internal error -- you lose");
 411         }
 412         break;
 413     }
 414
 415     /*
 416      * If no date component, then use the modification
 417      * time of the file as its date
 418      */
 419     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 420         struct stat st;
 421
 422         admonish (NULL, "can't parse %s field in message %d", datesw, msg);
 423         fstat (fileno (in), &st);
 424         smsg->s_clock = st.st_mtime;
 425     } else {
 426         smsg->s_clock = dmktime (tw);
 427     }
 428
 429     if (subjsort) {
 430         if (subjcomp) {
 431             /*
 432              * try to make the subject "canonical": delete
 433              * leading "re:", everything but letters & smash
 434              * letters to lower case.
 435              */
 436             register char  *cp, *cp2, c;
 437
 438             cp = subjcomp;
 439             cp2 = subjcomp;
 440             if (strcmp (subjsort, "subject") == 0)
 441                 while ((c = *cp)) {
 442                     if (! isspace(c)) {
 443                         if(uprf(cp, "re:"))
 444                             cp += 2;
 445                         else {
 446                             if (isalnum(c))
 447                                 *cp2++ = isupper(c) ? tolower(c) : c;
 448                             break;
 449                         }
 450                     }
 451                     cp++;
 452                 }
 453             while ((c = *cp++)) {
 454                 if (isalnum(c))
 455                     *cp2++ = isupper(c) ? tolower(c) : c;
 456
 457             }
 458             *cp2 = '\0';
 459         }
 460         else
 461             subjcomp = "";
 462
 463         smsg->s_subj = subjcomp;
 464     }
 465     fclose (in);
 466     if (datecomp)
 467         free (datecomp);
 468
 469     return (1);
 470 }
 471
 472 /*
 473  * sort on dates.
 474  */
 475 static int
 476 dsort (struct smsg **a, struct smsg **b)
 477 {
 478     if ((*a)->s_clock < (*b)->s_clock)
 479         return (-1);
 480     else if ((*a)->s_clock > (*b)->s_clock)
 481         return (1);
 482     else if ((*a)->s_msg < (*b)->s_msg)
 483         return (-1);
 484     else
 485         return (1);
 486 }
 487
 488 /*
 489  * sort on subjects.
 490  */
 491 static int
 492 subsort (struct smsg **a, struct smsg **b)
 493 {
 494     register 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     register int i;
 506
 507     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 508         return (i);
 509     else if ((*a)->s_msg < (*b)->s_msg)
 510         return (-1);
 511     else
 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
 521     for (;;) {
 522         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 523         mlist[msg] = (struct smsg *)0;
 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         if (rename (oldname, newname) == NOTOK)
 532             adios (newname, "unable to rename %s to", oldname);
 533
 534         copy_msg_flags (mp, new, old);
 535         if (mp->curmsg == old)
 536             seq_setcur (mp, new);
 537
 538         if (nxt == endmsg)
 539             break;
 540
 541         msg = nxt;
 542     }
 543 /*      if (nxt != endmsg); */
 544 /*      rename_chain (mp, mlist, nxt, endmsg); */
 545 }
 546
 547 static void
 548 rename_msgs (struct msgs *mp, struct smsg **mlist)
 549 {
 550     int i, j, old, new;
 551     seqset_t tmpset;
 552     char f1[BUFSIZ], tmpfil[BUFSIZ];
 553     struct smsg *sp;
 554
 555     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 556
 557     for (i = 0; i < nmsgs; i++) {
 558         if (! (sp = mlist[i]))
 559             continue;   /* did this one */
 560
 561         j = sp - smsgs;
 562         if (j == i)
 563             continue;   /* this one doesn't move */
 564
 565         /*
 566          * the guy that was msg j is about to become msg i.
 567          * rename 'j' to make a hole, then recursively rename
 568          * guys to fill up the hole.
 569          */
 570         old = smsgs[j].s_msg;
 571         new = smsgs[i].s_msg;
 572         strncpy (f1, m_name (old), sizeof(f1));
 573
 574         if (verbose)
 575             printf ("renaming message chain from %d to %d\n", old, new);
 576
 577         if (rename (f1, tmpfil) == NOTOK)
 578             adios (tmpfil, "unable to rename %s to ", f1);
 579         get_msg_flags (mp, &tmpset, old);
 580
 581         rename_chain (mp, mlist, j, i);
 582         if (rename (tmpfil, m_name(new)) == NOTOK)
 583             adios (m_name(new), "unable to rename %s to", tmpfil);
 584
 585         set_msg_flags (mp, &tmpset, new);
 586         mp->msgflags |= SEQMOD;
 587     }
 588 }