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
   8 #include <h/mh.h>
   9 #include <zotnet/tws/tws.h>
  10
  11 /*
  12  * We allocate space for messages (msgs array)
  13  * this number of elements at a time.
  14  */
  15 #define MAXMSGS  256
  16
  17
  18 static struct swit switches[] = {
  19 #define DATESW                 0
  20      { "datefield field", 0 },
  21 #define TEXTSW                 1
  22      { "textfield field", 0 },
  23 #define NSUBJSW                2
  24      { "notextfield", 0 },
  25 #define SUBJSW                 3
  26      { "subject", -3 },            /* backward-compatibility */
  27 #define LIMSW                  4
  28      { "limit days", 0 },
  29 #define NLIMSW                 5
  30      { "nolimit", 0 },
  31 #define VERBSW                 6
  32      { "verbose", 0 },
  33 #define NVERBSW                7
  34      { "noverbose", 0 },
  35 #define VERSIONSW              8
  36      { "version", 0 },
  37 #define HELPSW                 9
  38      { "help", 4 },
  39      { NULL, 0 }
  40 };
  41
  42 struct smsg {
  43     int s_msg;
  44     time_t s_clock;
  45     char *s_subj;
  46 };
  47
  48 static struct smsg *smsgs;
  49 int nmsgs;
  50
  51 char *subjsort = (char *) 0;    /* sort on subject if != 0 */
  52 unsigned long datelimit = 0;
  53 int submajor = 0;               /* if true, sort on subject-major */
  54 int verbose;
  55
  56 /* This keeps compiler happy on calls to qsort */
  57 typedef int (*qsort_comp) (const void *, const void *);
  58
  59 /*
  60  * static prototypes
  61  */
  62 static int read_hdrs (struct msgs *, char *);
  63 static int get_fields (char *, int, struct smsg *);
  64 static int dsort (struct smsg **, struct smsg **);
  65 static int subsort (struct smsg **, struct smsg **);
  66 static int txtsort (struct smsg **, struct smsg **);
  67 static void rename_chain (struct msgs *, struct smsg **, int, int);
  68 static void rename_msgs (struct msgs *, struct smsg **);
  69
  70
  71 int
  72 main (int argc, char **argv)
  73 {
  74     int nummsgs, maxmsgs, i, msgnum;
  75     char *cp, *maildir, *datesw = NULL;
  76     char *folder = NULL, buf[BUFSIZ], **argp;
  77     char **arguments, **msgs;
  78     struct msgs *mp;
  79     struct smsg **dlist;
  80
  81 #ifdef LOCALE
  82     setlocale(LC_ALL, "");
  83 #endif
  84     invo_name = r1bindex (argv[0], '/');
  85
  86     /* read user profile/context */
  87     context_read();
  88
  89     arguments = getarguments (invo_name, argc, argv, 1);
  90     argp = arguments;
  91
  92     /*
  93      * Allocate the initial space to record message
  94      * names and ranges.
  95      */
  96     nummsgs = 0;
  97     maxmsgs = MAXMSGS;
  98     if (!(msgs = (char **) malloc ((size_t) (maxmsgs * sizeof(*msgs)))))
  99         adios (NULL, "unable to allocate storage");
 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                 adios (NULL, "-%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 (1);
 118             case VERSIONSW:
 119                 print_version(invo_name);
 120                 done (1);
 121
 122             case DATESW:
 123                 if (datesw)
 124                     adios (NULL, "only one date field at a time");
 125                 if (!(datesw = *argp++) || *datesw == '-')
 126                     adios (NULL, "missing argument to %s", argp[-2]);
 127                 continue;
 128
 129             case TEXTSW:
 130                 if (subjsort)
 131                     adios (NULL, "only one text field at a time");
 132                 if (!(subjsort = *argp++) || *subjsort == '-')
 133                     adios (NULL, "missing argument to %s", argp[-2]);
 134                 continue;
 135
 136             case SUBJSW:
 137                 subjsort = "subject";
 138                 continue;
 139             case NSUBJSW:
 140                 subjsort = (char *)0;
 141                 continue;
 142
 143             case LIMSW:
 144                 if (!(cp = *argp++) || *cp == '-')
 145                         adios (NULL, "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++;         /* sort subject-major */
 150                     continue;
 151                 }
 152                 if (!isdigit(*cp) || !(datelimit = atoi(cp)))
 153                     adios (NULL, "impossible limit %s", cp);
 154                 datelimit *= 60*60*24;
 155                 continue;
 156             case NLIMSW:
 157                 submajor = 0;   /* use date-major, but */
 158                 datelimit = 0;  /* use no limit */
 159                 continue;
 160
 161             case VERBSW:
 162                 verbose++;
 163                 continue;
 164             case NVERBSW:
 165                 verbose = 0;
 166                 continue;
 167             }
 168         }
 169         if (*cp == '+' || *cp == '@') {
 170             if (folder)
 171                 adios (NULL, "only one folder at a time!");
 172             else
 173                 folder = path (cp + 1, *cp == '+' ? TFOLDER : TSUBCWF);
 174         } else {
 175             /*
 176              * Check if we need to allocate more space
 177              * for message names/ranges.
 178              */
 179             if (nummsgs >= maxmsgs) {
 180                 maxmsgs += MAXMSGS;
 181                 if (!(msgs = (char **) realloc (msgs,
 182                         (size_t) (maxmsgs * sizeof(*msgs)))))
 183                     adios (NULL, "unable to reallocate msgs storage");
 184             }
 185             msgs[nummsgs++] = cp;
 186         }
 187     }
 188
 189     if (!context_find ("path"))
 190         free (path ("./", TFOLDER));
 191     if (!nummsgs)
 192         msgs[nummsgs++] = "all";
 193     if (!datesw)
 194         datesw = "date";
 195     if (!folder)
 196         folder = getfolder (1);
 197     maildir = m_maildir (folder);
 198
 199     if (chdir (maildir) == NOTOK)
 200         adios (maildir, "unable to change directory to");
 201
 202     /* read folder and create message structure */
 203     if (!(mp = folder_read (folder)))
 204         adios (NULL, "unable to read folder %s", folder);
 205
 206     /* check for empty folder */
 207     if (mp->nummsg == 0)
 208         adios (NULL, "no messages in %s", folder);
 209
 210     /* parse all the message ranges/sequences and set SELECTED */
 211     for (msgnum = 0; msgnum < nummsgs; msgnum++)
 212         if (!m_convert (mp, msgs[msgnum]))
 213             done (1);
 214     seq_setprev (mp);   /* set the previous sequence */
 215
 216     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 217         adios (NULL, "no messages to sort");
 218
 219     /*
 220      * sort a list of pointers to our "messages to be sorted".
 221      */
 222     dlist = (struct smsg **) malloc ((nmsgs+1) * sizeof(*dlist));
 223     if (! dlist)
 224         adios (NULL, "couldn't allocate sort memory");
 225     for (i = 0; i < nmsgs; i++)
 226         dlist[i] = &smsgs[i];
 227     dlist[nmsgs] = 0;
 228
 229     if (verbose)        /* announce what we're doing */
 230         if (subjsort)
 231             printf ("sorting by %s-major %s-minor\n",
 232                 submajor ? subjsort : datesw,
 233                 submajor ? datesw : subjsort);
 234         else
 235             printf ("sorting by datefield %s\n", datesw);
 236
 237     /* first sort by date, or by subject-major, date-minor */
 238     qsort ((char *) dlist, nmsgs, sizeof(*dlist),
 239             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 240
 241     /*
 242      * if we're sorting on subject, we need another list
 243      * in subject order, then a merge pass to collate the
 244      * two sorts.
 245      */
 246     if (!submajor && subjsort) {        /* already date sorted */
 247         struct smsg **slist, **flist;
 248         register struct smsg ***il, **fp, **dp;
 249
 250         slist = (struct smsg **) malloc ((nmsgs+1) * sizeof(*slist));
 251         if (! slist)
 252             adios (NULL, "couldn't allocate sort memory");
 253         memcpy((char *)slist, (char *)dlist, (nmsgs+1)*sizeof(*slist));
 254         qsort((char *)slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 255
 256         /*
 257          * make an inversion list so we can quickly find
 258          * the collection of messages with the same subj
 259          * given a message number.
 260          */
 261         il = (struct smsg ***) calloc (mp->hghsel+1, sizeof(*il));
 262         if (! il)
 263             adios (NULL, "couldn't allocate msg list");
 264         for (i = 0; i < nmsgs; i++)
 265             il[slist[i]->s_msg] = &slist[i];
 266         /*
 267          * make up the final list, chronological but with
 268          * all the same subjects grouped together.
 269          */
 270         flist = (struct smsg **) malloc ((nmsgs+1) * sizeof(*flist));
 271         if (! flist)
 272             adios (NULL, "couldn't allocate msg list");
 273         fp = flist;
 274         for (dp = dlist; *dp;) {
 275             register struct smsg **s = il[(*dp++)->s_msg];
 276
 277             /* see if we already did this guy */
 278             if (! s)
 279                 continue;
 280
 281             *fp++ = *s++;
 282             /*
 283              * take the next message(s) if there is one,
 284              * its subject isn't null and its subject
 285              * is the same as this one and it's not too
 286              * far away in time.
 287              */
 288             while (*s && (*s)->s_subj[0] &&
 289                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 290                    (datelimit == 0 ||
 291                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 292                 il[(*s)->s_msg] = 0;
 293                 *fp++ = *s++;
 294             }
 295         }
 296         *fp = 0;
 297         free (slist);
 298         free (dlist);
 299         dlist = flist;
 300     }
 301     rename_msgs (mp, dlist);
 302
 303     context_replace (pfolder, folder);  /* update current folder         */
 304     seq_save (mp);                      /* synchronize message sequences */
 305     context_save ();                    /* save the context file         */
 306     folder_free (mp);                   /* free folder/message structure */
 307     done (0);
 308 }
 309
 310 static int
 311 read_hdrs (struct msgs *mp, char *datesw)
 312 {
 313     int msgnum;
 314     struct tws tb;
 315     register struct smsg *s;
 316
 317     twscopy (&tb, dlocaltimenow ());
 318
 319     smsgs = (struct smsg *)
 320         calloc ((size_t) (mp->hghsel - mp->lowsel + 2),
 321             sizeof(*smsgs));
 322     if (smsgs == NULL)
 323         adios (NULL, "unable to allocate sort storage");
 324
 325     s = smsgs;
 326     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 327         if (is_selected(mp, msgnum)) {
 328             if (get_fields (datesw, msgnum, s)) {
 329                 s->s_msg = msgnum;
 330                 s++;
 331             }
 332         }
 333     }
 334     s->s_msg = 0;
 335     return(s - smsgs);
 336 }
 337
 338
 339 /*
 340  * Parse the message and get the data or subject field,
 341  * if needed.
 342  */
 343
 344 static int
 345 get_fields (char *datesw, int msg, struct smsg *smsg)
 346 {
 347     register int state;
 348     int compnum;
 349     char *msgnam, buf[BUFSIZ], nam[NAMESZ];
 350     register struct tws *tw;
 351     register char *datecomp = NULL, *subjcomp = NULL;
 352     register FILE *in;
 353
 354     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 355         admonish (msgnam, "unable to read message");
 356         return (0);
 357     }
 358     for (compnum = 1, state = FLD;;) {
 359         switch (state = m_getfld (state, nam, buf, sizeof(buf), in)) {
 360         case FLD:
 361         case FLDEOF:
 362         case FLDPLUS:
 363             compnum++;
 364             if (!strcasecmp (nam, datesw)) {
 365                 datecomp = add (buf, datecomp);
 366                 while (state == FLDPLUS) {
 367                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 368                     datecomp = add (buf, datecomp);
 369                 }
 370                 if (!subjsort || subjcomp)
 371                     break;
 372             } else if (subjsort && !strcasecmp (nam, subjsort)) {
 373                 subjcomp = add (buf, subjcomp);
 374                 while (state == FLDPLUS) {
 375                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 376                     subjcomp = add (buf, subjcomp);
 377                 }
 378                 if (datecomp)
 379                     break;
 380             } else {
 381                 /* just flush this guy */
 382                 while (state == FLDPLUS)
 383                     state = m_getfld (state, nam, buf, sizeof(buf), in);
 384             }
 385             continue;
 386
 387         case BODY:
 388         case BODYEOF:
 389         case FILEEOF:
 390             break;
 391
 392         case LENERR:
 393         case FMTERR:
 394             if (state == LENERR || state == FMTERR)
 395                 admonish (NULL, "format error in message %d (header #%d)",
 396                       msg, compnum);
 397             if (datecomp)
 398                 free (datecomp);
 399             if (subjcomp)
 400                 free (subjcomp);
 401             fclose (in);
 402             return (0);
 403
 404         default:
 405             adios (NULL, "internal error -- you lose");
 406         }
 407         break;
 408     }
 409
 410     /*
 411      * If no date component, then use the modification
 412      * time of the file as its date
 413      */
 414     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 415         struct stat st;
 416
 417         admonish (NULL, "can't parse %s field in message %d", datesw, msg);
 418         fstat (fileno (in), &st);
 419         smsg->s_clock = st.st_mtime;
 420     } else {
 421         smsg->s_clock = dmktime (tw);
 422     }
 423
 424     if (subjsort) {
 425         if (subjcomp) {
 426             /*
 427              * try to make the subject "canonical": delete
 428              * leading "re:", everything but letters & smash
 429              * letters to lower case.
 430              */
 431             register char  *cp, *cp2, c;
 432
 433             cp = subjcomp;
 434             cp2 = subjcomp;
 435             if (strcmp (subjsort, "subject") == 0)
 436                 while ((c = *cp)) {
 437                     if (! isspace(c)) {
 438                         if(uprf(cp, "re:"))
 439                             cp += 2;
 440                         else {
 441                             if (isalnum(c))
 442                                 *cp2++ = isupper(c) ? tolower(c) : c;
 443                             break;
 444                         }
 445                     }
 446                     cp++;
 447                 }
 448             while ((c = *cp++)) {
 449                 if (isalnum(c))
 450                     *cp2++ = isupper(c) ? tolower(c) : c;
 451
 452             }
 453             *cp2 = '\0';
 454         }
 455         else
 456             subjcomp = "";
 457
 458         smsg->s_subj = subjcomp;
 459     }
 460     fclose (in);
 461     if (datecomp)
 462         free (datecomp);
 463
 464     return (1);
 465 }
 466
 467 /*
 468  * sort on dates.
 469  */
 470 static int
 471 dsort (struct smsg **a, struct smsg **b)
 472 {
 473     if ((*a)->s_clock < (*b)->s_clock)
 474         return (-1);
 475     else if ((*a)->s_clock > (*b)->s_clock)
 476         return (1);
 477     else if ((*a)->s_msg < (*b)->s_msg)
 478         return (-1);
 479     else
 480         return (1);
 481 }
 482
 483 /*
 484  * sort on subjects.
 485  */
 486 static int
 487 subsort (struct smsg **a, struct smsg **b)
 488 {
 489     register int i;
 490
 491     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 492         return (i);
 493
 494     return (dsort (a, b));
 495 }
 496
 497 static int
 498 txtsort (struct smsg **a, struct smsg **b)
 499 {
 500     register int i;
 501
 502     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 503         return (i);
 504     else if ((*a)->s_msg < (*b)->s_msg)
 505         return (-1);
 506     else
 507         return (1);
 508 }
 509
 510 static void
 511 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 512 {
 513     int nxt, old, new;
 514     char *newname, oldname[BUFSIZ];
 515
 516     for (;;) {
 517         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 518         mlist[msg] = (struct smsg *)0;
 519         old = smsgs[nxt].s_msg;
 520         new = smsgs[msg].s_msg;
 521         strncpy (oldname, m_name (old), sizeof(oldname));
 522         newname = m_name (new);
 523         if (verbose)
 524             printf ("message %d becomes message %d\n", old, new);
 525
 526         if (rename (oldname, newname) == NOTOK)
 527             adios (newname, "unable to rename %s to", oldname);
 528
 529         copy_msg_flags (mp, new, old);
 530         if (mp->curmsg == old)
 531             seq_setcur (mp, new);
 532
 533         if (nxt == endmsg)
 534             break;
 535
 536         msg = nxt;
 537     }
 538 /*      if (nxt != endmsg); */
 539 /*      rename_chain (mp, mlist, nxt, endmsg); */
 540 }
 541
 542 static void
 543 rename_msgs (struct msgs *mp, struct smsg **mlist)
 544 {
 545     int i, j, old, new;
 546     seqset_t tmpset;
 547     char f1[BUFSIZ], tmpfil[BUFSIZ];
 548     struct smsg *sp;
 549
 550     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 551
 552     for (i = 0; i < nmsgs; i++) {
 553         if (! (sp = mlist[i]))
 554             continue;   /* did this one */
 555
 556         j = sp - smsgs;
 557         if (j == i)
 558             continue;   /* this one doesn't move */
 559
 560         /*
 561          * the guy that was msg j is about to become msg i.
 562          * rename 'j' to make a hole, then recursively rename
 563          * guys to fill up the hole.
 564          */
 565         old = smsgs[j].s_msg;
 566         new = smsgs[i].s_msg;
 567         strncpy (f1, m_name (old), sizeof(f1));
 568
 569         if (verbose)
 570             printf ("renaming message chain from %d to %d\n", old, new);
 571
 572         if (rename (f1, tmpfil) == NOTOK)
 573             adios (tmpfil, "unable to rename %s to ", f1);
 574         get_msg_flags (mp, &tmpset, old);
 575
 576         rename_chain (mp, mlist, j, i);
 577         if (rename (tmpfil, m_name(new)) == NOTOK)
 578             adios (m_name(new), "unable to rename %s to", tmpfil);
 579
 580         set_msg_flags (mp, &tmpset, new);
 581         mp->msgflags |= SEQMOD;
 582     }
 583 }