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