citadel/modules/imap/imap_search.c

   1 /*
   2  * $Id$
   3  *
   4  * Implements IMAP's gratuitously complex SEARCH command.
   5  *
   6  *
   7  * Copyright (c) 2001-2009 by the citadel.org team
   8  *
   9  *  This program is free software; you can redistribute it and/or modify
  10  *  it under the terms of the GNU General Public License as published by
  11  *  the Free Software Foundation; either version 3 of the License, or
  12  *  (at your option) any later version.
  13  *
  14  *  This program is distributed in the hope that it will be useful,
  15  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17  *  GNU General Public License for more details.
  18  *
  19  *  You should have received a copy of the GNU General Public License
  20  *  along with this program; if not, write to the Free Software
  21  *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  22  */
  23
  24 #include "ctdl_module.h"
  25
  26
  27 #include "sysdep.h"
  28 #include <stdlib.h>
  29 #include <unistd.h>
  30 #include <stdio.h>
  31 #include <fcntl.h>
  32 #include <signal.h>
  33 #include <pwd.h>
  34 #include <errno.h>
  35 #include <sys/types.h>
  36
  37 #if TIME_WITH_SYS_TIME
  38 # include <sys/time.h>
  39 # include <time.h>
  40 #else
  41 # if HAVE_SYS_TIME_H
  42 #  include <sys/time.h>
  43 # else
  44 #  include <time.h>
  45 # endif
  46 #endif
  47
  48 #include <sys/wait.h>
  49 #include <ctype.h>
  50 #include <string.h>
  51 #include <limits.h>
  52 #include <libcitadel.h>
  53 #include "citadel.h"
  54 #include "server.h"
  55 #include "sysdep_decls.h"
  56 #include "citserver.h"
  57 #include "support.h"
  58 #include "config.h"
  59 #include "user_ops.h"
  60 #include "policy.h"
  61 #include "database.h"
  62 #include "msgbase.h"
  63 #include "internet_addressing.h"
  64 #include "serv_imap.h"
  65 #include "imap_tools.h"
  66 #include "imap_fetch.h"
  67 #include "imap_search.h"
  68 #include "genstamp.h"
  69
  70
  71 /*
  72  * imap_do_search() calls imap_do_search_msg() to search an individual
  73  * message after it has been fetched from the disk.  This function returns
  74  * nonzero if there is a match.
  75  *
  76  * supplied_msg MAY be used to pass a pointer to the message in memory,
  77  * if for some reason it's already been loaded.  If not, the message will
  78  * be loaded only if one or more search criteria require it.
  79  */
  80 int imap_do_search_msg(int seq, struct CtdlMessage *supplied_msg,
  81                         int num_items, char **itemlist, int is_uid) {
  82
  83         int match = 0;
  84         int is_not = 0;
  85         int is_or = 0;
  86         int pos = 0;
  87         int i;
  88         char *fieldptr;
  89         struct CtdlMessage *msg = NULL;
  90         int need_to_free_msg = 0;
  91
  92         if (num_items == 0) {
  93                 return(0);
  94         }
  95         msg = supplied_msg;
  96
  97         /* Initially we start at the beginning. */
  98         pos = 0;
  99
 100         /* Check for the dreaded NOT criterion. */
 101         if (!strcasecmp(itemlist[0], "NOT")) {
 102                 is_not = 1;
 103                 pos = 1;
 104         }
 105
 106         /* Check for the dreaded OR criterion. */
 107         if (!strcasecmp(itemlist[0], "OR")) {
 108                 is_or = 1;
 109                 pos = 1;
 110         }
 111
 112         /* Now look for criteria. */
 113         if (!strcasecmp(itemlist[pos], "ALL")) {
 114                 match = 1;
 115                 ++pos;
 116         }
 117
 118         else if (!strcasecmp(itemlist[pos], "ANSWERED")) {
 119                 if (IMAP->flags[seq-1] & IMAP_ANSWERED) {
 120                         match = 1;
 121                 }
 122                 ++pos;
 123         }
 124
 125         else if (!strcasecmp(itemlist[pos], "BCC")) {
 126                 if (msg == NULL) {
 127                         msg = CtdlFetchMessage(IMAP->msgids[seq-1], 1);
 128                         need_to_free_msg = 1;
 129                 }
 130                 if (msg != NULL) {
 131                         fieldptr = rfc822_fetch_field(msg->cm_fields['M'], "Bcc");
 132                         if (fieldptr != NULL) {
 133                                 if (bmstrcasestr(fieldptr, itemlist[pos+1])) {
 134                                         match = 1;
 135                                 }
 136                                 free(fieldptr);
 137                         }
 138                 }
 139                 pos += 2;
 140         }
 141
 142         else if (!strcasecmp(itemlist[pos], "BEFORE")) {
 143                 if (msg == NULL) {
 144                         msg = CtdlFetchMessage(IMAP->msgids[seq-1], 1);
 145                         need_to_free_msg = 1;
 146                 }
 147                 if (msg != NULL) {
 148                         if (msg->cm_fields['T'] != NULL) {
 149                                 if (imap_datecmp(itemlist[pos+1],
 150                                                 atol(msg->cm_fields['T'])) < 0) {
 151                                         match = 1;
 152                                 }
 153                         }
 154                 }
 155                 pos += 2;
 156         }
 157
 158         else if (!strcasecmp(itemlist[pos], "BODY")) {
 159
 160                 /* If fulltext indexing is active, on this server,
 161                  *  all messages have already been qualified.
 162                  */
 163                 if (config.c_enable_fulltext) {
 164                         match = 1;
 165                 }
 166
 167                 /* Otherwise, we have to do a slow search. */
 168                 else {
 169                         if (msg == NULL) {
 170                                 msg = CtdlFetchMessage(IMAP->msgids[seq-1], 1);
 171                                 need_to_free_msg = 1;
 172                         }
 173                         if (msg != NULL) {
 174                                 if (bmstrcasestr(msg->cm_fields['M'], itemlist[pos+1])) {
 175                                         match = 1;
 176                                 }
 177                         }
 178                 }
 179
 180                 pos += 2;
 181         }
 182
 183         else if (!strcasecmp(itemlist[pos], "CC")) {
 184                 if (msg == NULL) {
 185                         msg = CtdlFetchMessage(IMAP->msgids[seq-1], 1);
 186                         need_to_free_msg = 1;
 187                 }
 188                 if (msg != NULL) {
 189                         fieldptr = msg->cm_fields['Y'];
 190                         if (fieldptr != NULL) {
 191                                 if (bmstrcasestr(fieldptr, itemlist[pos+1])) {
 192                                         match = 1;
 193                                 }
 194                         }
 195                         else {
 196                                 fieldptr = rfc822_fetch_field(msg->cm_fields['M'], "Cc");
 197                                 if (fieldptr != NULL) {
 198                                         if (bmstrcasestr(fieldptr, itemlist[pos+1])) {
 199                                                 match = 1;
 200                                         }
 201                                         free(fieldptr);
 202                                 }
 203                         }
 204                 }
 205                 pos += 2;
 206         }
 207
 208         else if (!strcasecmp(itemlist[pos], "DELETED")) {
 209                 if (IMAP->flags[seq-1] & IMAP_DELETED) {
 210                         match = 1;
 211                 }
 212                 ++pos;
 213         }
 214
 215         else if (!strcasecmp(itemlist[pos], "DRAFT")) {
 216                 if (IMAP->flags[seq-1] & IMAP_DRAFT) {
 217                         match = 1;
 218                 }
 219                 ++pos;
 220         }
 221
 222         else if (!strcasecmp(itemlist[pos], "FLAGGED")) {
 223                 if (IMAP->flags[seq-1] & IMAP_FLAGGED) {
 224                         match = 1;
 225                 }
 226                 ++pos;
 227         }
 228
 229         else if (!strcasecmp(itemlist[pos], "FROM")) {
 230                 if (msg == NULL) {
 231                         msg = CtdlFetchMessage(IMAP->msgids[seq-1], 1);
 232                         need_to_free_msg = 1;
 233                 }
 234                 if (msg != NULL) {
 235                         if (bmstrcasestr(msg->cm_fields['A'], itemlist[pos+1])) {
 236                                 match = 1;
 237                         }
 238                         if (bmstrcasestr(msg->cm_fields['F'], itemlist[pos+1])) {
 239                                 match = 1;
 240                         }
 241                 }
 242                 pos += 2;
 243         }
 244
 245         else if (!strcasecmp(itemlist[pos], "HEADER")) {
 246
 247                 /* We've got to do a slow search for this because the client
 248                  * might be asking for an RFC822 header field that has not been
 249                  * converted into a Citadel header field.  That requires
 250                  * examining the message body.
 251                  */
 252                 if (msg == NULL) {
 253                         msg = CtdlFetchMessage(IMAP->msgids[seq-1], 1);
 254                         need_to_free_msg = 1;
 255                 }
 256
 257                 if (msg != NULL) {
 258
 259                         CC->redirect_buffer = malloc(SIZ);
 260                         CC->redirect_len = 0;
 261                         CC->redirect_alloc = SIZ;
 262                         CtdlOutputPreLoadedMsg(msg, MT_RFC822, HEADERS_FAST, 0, 1, 0);
 263
 264                         fieldptr = rfc822_fetch_field(CC->redirect_buffer, itemlist[pos+1]);
 265                         if (fieldptr != NULL) {
 266                                 if (bmstrcasestr(fieldptr, itemlist[pos+2])) {
 267                                         match = 1;
 268                                 }
 269                                 free(fieldptr);
 270                         }
 271
 272                         free(CC->redirect_buffer);
 273                         CC->redirect_buffer = NULL;
 274                         CC->redirect_len = 0;
 275                         CC->redirect_alloc = 0;
 276                 }
 277
 278                 pos += 3;       /* Yes, three */
 279         }
 280
 281         else if (!strcasecmp(itemlist[pos], "KEYWORD")) {
 282                 /* not implemented */
 283                 pos += 2;
 284         }
 285
 286         else if (!strcasecmp(itemlist[pos], "LARGER")) {
 287                 if (msg == NULL) {
 288                         msg = CtdlFetchMessage(IMAP->msgids[seq-1], 1);
 289                         need_to_free_msg = 1;
 290                 }
 291                 if (msg != NULL) {
 292                         if (strlen(msg->cm_fields['M']) > atoi(itemlist[pos+1])) {
 293                                 match = 1;
 294                         }
 295                 }
 296                 pos += 2;
 297         }
 298
 299         else if (!strcasecmp(itemlist[pos], "NEW")) {
 300                 if ( (IMAP->flags[seq-1] & IMAP_RECENT) && (!(IMAP->flags[seq-1] & IMAP_SEEN))) {
 301                         match = 1;
 302                 }
 303                 ++pos;
 304         }
 305
 306         else if (!strcasecmp(itemlist[pos], "OLD")) {
 307                 if (!(IMAP->flags[seq-1] & IMAP_RECENT)) {
 308                         match = 1;
 309                 }
 310                 ++pos;
 311         }
 312
 313         else if (!strcasecmp(itemlist[pos], "ON")) {
 314                 if (msg == NULL) {
 315                         msg = CtdlFetchMessage(IMAP->msgids[seq-1], 1);
 316                         need_to_free_msg = 1;
 317                 }
 318                 if (msg != NULL) {
 319                         if (msg->cm_fields['T'] != NULL) {
 320                                 if (imap_datecmp(itemlist[pos+1],
 321                                                 atol(msg->cm_fields['T'])) == 0) {
 322                                         match = 1;
 323                                 }
 324                         }
 325                 }
 326                 pos += 2;
 327         }
 328
 329         else if (!strcasecmp(itemlist[pos], "RECENT")) {
 330                 if (IMAP->flags[seq-1] & IMAP_RECENT) {
 331                         match = 1;
 332                 }
 333                 ++pos;
 334         }
 335
 336         else if (!strcasecmp(itemlist[pos], "SEEN")) {
 337                 if (IMAP->flags[seq-1] & IMAP_SEEN) {
 338                         match = 1;
 339                 }
 340                 ++pos;
 341         }
 342
 343         else if (!strcasecmp(itemlist[pos], "SENTBEFORE")) {
 344                 if (msg == NULL) {
 345                         msg = CtdlFetchMessage(IMAP->msgids[seq-1], 1);
 346                         need_to_free_msg = 1;
 347                 }
 348                 if (msg != NULL) {
 349                         if (msg->cm_fields['T'] != NULL) {
 350                                 if (imap_datecmp(itemlist[pos+1],
 351                                                 atol(msg->cm_fields['T'])) < 0) {
 352                                         match = 1;
 353                                 }
 354                         }
 355                 }
 356                 pos += 2;
 357         }
 358
 359         else if (!strcasecmp(itemlist[pos], "SENTON")) {
 360                 if (msg == NULL) {
 361                         msg = CtdlFetchMessage(IMAP->msgids[seq-1], 1);
 362                         need_to_free_msg = 1;
 363                 }
 364                 if (msg != NULL) {
 365                         if (msg->cm_fields['T'] != NULL) {
 366                                 if (imap_datecmp(itemlist[pos+1],
 367                                                 atol(msg->cm_fields['T'])) == 0) {
 368                                         match = 1;
 369                                 }
 370                         }
 371                 }
 372                 pos += 2;
 373         }
 374
 375         else if (!strcasecmp(itemlist[pos], "SENTSINCE")) {
 376                 if (msg == NULL) {
 377                         msg = CtdlFetchMessage(IMAP->msgids[seq-1], 1);
 378                         need_to_free_msg = 1;
 379                 }
 380                 if (msg != NULL) {
 381                         if (msg->cm_fields['T'] != NULL) {
 382                                 if (imap_datecmp(itemlist[pos+1],
 383                                                 atol(msg->cm_fields['T'])) >= 0) {
 384                                         match = 1;
 385                                 }
 386                         }
 387                 }
 388                 pos += 2;
 389         }
 390
 391         else if (!strcasecmp(itemlist[pos], "SINCE")) {
 392                 if (msg == NULL) {
 393                         msg = CtdlFetchMessage(IMAP->msgids[seq-1], 1);
 394                         need_to_free_msg = 1;
 395                 }
 396                 if (msg != NULL) {
 397                         if (msg->cm_fields['T'] != NULL) {
 398                                 if (imap_datecmp(itemlist[pos+1],
 399                                                 atol(msg->cm_fields['T'])) >= 0) {
 400                                         match = 1;
 401                                 }
 402                         }
 403                 }
 404                 pos += 2;
 405         }
 406
 407         else if (!strcasecmp(itemlist[pos], "SMALLER")) {
 408                 if (msg == NULL) {
 409                         msg = CtdlFetchMessage(IMAP->msgids[seq-1], 1);
 410                         need_to_free_msg = 1;
 411                 }
 412                 if (msg != NULL) {
 413                         if (strlen(msg->cm_fields['M']) < atoi(itemlist[pos+1])) {
 414                                 match = 1;
 415                         }
 416                 }
 417                 pos += 2;
 418         }
 419
 420         else if (!strcasecmp(itemlist[pos], "SUBJECT")) {
 421                 if (msg == NULL) {
 422                         msg = CtdlFetchMessage(IMAP->msgids[seq-1], 1);
 423                         need_to_free_msg = 1;
 424                 }
 425                 if (msg != NULL) {
 426                         if (bmstrcasestr(msg->cm_fields['U'], itemlist[pos+1])) {
 427                                 match = 1;
 428                         }
 429                 }
 430                 pos += 2;
 431         }
 432
 433         else if (!strcasecmp(itemlist[pos], "TEXT")) {
 434                 if (msg == NULL) {
 435                         msg = CtdlFetchMessage(IMAP->msgids[seq-1], 1);
 436                         need_to_free_msg = 1;
 437                 }
 438                 if (msg != NULL) {
 439                         for (i='A'; i<='Z'; ++i) {
 440                                 if (bmstrcasestr(msg->cm_fields[i], itemlist[pos+1])) {
 441                                         match = 1;
 442                                 }
 443                         }
 444                 }
 445                 pos += 2;
 446         }
 447
 448         else if (!strcasecmp(itemlist[pos], "TO")) {
 449                 if (msg == NULL) {
 450                         msg = CtdlFetchMessage(IMAP->msgids[seq-1], 1);
 451                         need_to_free_msg = 1;
 452                 }
 453                 if (msg != NULL) {
 454                         if (bmstrcasestr(msg->cm_fields['R'], itemlist[pos+1])) {
 455                                 match = 1;
 456                         }
 457                 }
 458                 pos += 2;
 459         }
 460
 461         /* FIXME this is b0rken.  fix it. */
 462         else if (imap_is_message_set(itemlist[pos])) {
 463                 if (is_msg_in_sequence_set(itemlist[pos], seq)) {
 464                         match = 1;
 465                 }
 466                 pos += 1;
 467         }
 468
 469         /* FIXME this is b0rken.  fix it. */
 470         else if (!strcasecmp(itemlist[pos], "UID")) {
 471                 if (is_msg_in_sequence_set(itemlist[pos+1], IMAP->msgids[seq-1])) {
 472                         match = 1;
 473                 }
 474                 pos += 2;
 475         }
 476
 477         /* Now here come the 'UN' criteria.  Why oh why do we have to
 478          * implement *both* the 'UN' criteria *and* the 'NOT' keyword?  Why
 479          * can't there be *one* way to do things?  More gratuitous complexity.
 480          */
 481
 482         else if (!strcasecmp(itemlist[pos], "UNANSWERED")) {
 483                 if ((IMAP->flags[seq-1] & IMAP_ANSWERED) == 0) {
 484                         match = 1;
 485                 }
 486                 ++pos;
 487         }
 488
 489         else if (!strcasecmp(itemlist[pos], "UNDELETED")) {
 490                 if ((IMAP->flags[seq-1] & IMAP_DELETED) == 0) {
 491                         match = 1;
 492                 }
 493                 ++pos;
 494         }
 495
 496         else if (!strcasecmp(itemlist[pos], "UNDRAFT")) {
 497                 if ((IMAP->flags[seq-1] & IMAP_DRAFT) == 0) {
 498                         match = 1;
 499                 }
 500                 ++pos;
 501         }
 502
 503         else if (!strcasecmp(itemlist[pos], "UNFLAGGED")) {
 504                 if ((IMAP->flags[seq-1] & IMAP_FLAGGED) == 0) {
 505                         match = 1;
 506                 }
 507                 ++pos;
 508         }
 509
 510         else if (!strcasecmp(itemlist[pos], "UNKEYWORD")) {
 511                 /* FIXME */
 512                 pos += 2;
 513         }
 514
 515         else if (!strcasecmp(itemlist[pos], "UNSEEN")) {
 516                 if ((IMAP->flags[seq-1] & IMAP_SEEN) == 0) {
 517                         match = 1;
 518                 }
 519                 ++pos;
 520         }
 521
 522         /* Remember to negate if we were told to */
 523         if (is_not) {
 524                 match = !match;
 525         }
 526
 527         /* Keep going if there are more criteria! */
 528         if (pos < num_items) {
 529
 530                 if (is_or) {
 531                         match = (match || imap_do_search_msg(seq, msg,
 532                                 num_items - pos, &itemlist[pos], is_uid));
 533                 }
 534                 else {
 535                         match = (match && imap_do_search_msg(seq, msg,
 536                                 num_items - pos, &itemlist[pos], is_uid));
 537                 }
 538
 539         }
 540
 541         if (need_to_free_msg) {
 542                 CtdlFreeMessage(msg);
 543         }
 544         return(match);
 545 }
 546
 547
 548 /*
 549  * imap_search() calls imap_do_search() to do its actual work, once it's
 550  * validated and boiled down the request a bit.
 551  */
 552 void imap_do_search(int num_items, char **itemlist, int is_uid) {
 553         int i, j, k;
 554         int fts_num_msgs = 0;
 555         long *fts_msgs = NULL;
 556         int is_in_list = 0;
 557         int num_results = 0;
 558
 559         /* Strip parentheses.  We realize that this method will not work
 560          * in all cases, but it seems to work with all currently available
 561          * client software.  Revisit later...
 562          */
 563         for (i=0; i<num_items; ++i) {
 564                 if (itemlist[i][0] == '(') {
 565                         strcpy(&itemlist[i][0], &itemlist[i][1]);
 566                 }
 567                 if (itemlist[i][strlen(itemlist[i])-1] == ')') {
 568                         itemlist[i][strlen(itemlist[i])-1] = 0;
 569                 }
 570         }
 571
 572         /* If there is a BODY search criterion in the query, use our full
 573          * text index to disqualify messages that don't have any chance of
 574          * matching.  (Only do this if the index is enabled!!)
 575          */
 576         if (config.c_enable_fulltext) for (i=0; i<(num_items-1); ++i) {
 577                 if (!strcasecmp(itemlist[i], "BODY")) {
 578                         CtdlModuleDoSearch(&fts_num_msgs, &fts_msgs, itemlist[i+1], "fulltext");
 579                         if (fts_num_msgs > 0) {
 580                                 for (j=0; j < IMAP->num_msgs; ++j) {
 581                                         if (IMAP->flags[j] & IMAP_SELECTED) {
 582                                                 is_in_list = 0;
 583                                                 for (k=0; k<fts_num_msgs; ++k) {
 584                                                         if (IMAP->msgids[j] == fts_msgs[k]) {
 585                                                                 ++is_in_list;
 586                                                         }
 587                                                 }
 588                                         }
 589                                         if (!is_in_list) {
 590                                                 IMAP->flags[j] = IMAP->flags[j] & ~IMAP_SELECTED;
 591                                         }
 592                                 }
 593                         }
 594                         else {          /* no hits on the index; disqualify every message */
 595                                 for (j=0; j < IMAP->num_msgs; ++j) {
 596                                         IMAP->flags[j] = IMAP->flags[j] & ~IMAP_SELECTED;
 597                                 }
 598                         }
 599                         if (fts_msgs) {
 600                                 free(fts_msgs);
 601                         }
 602                 }
 603         }
 604
 605         /* Now go through the messages and apply all search criteria. */
 606         buffer_output();
 607         cprintf("* SEARCH ");
 608         if (IMAP->num_msgs > 0)
 609          for (i = 0; i < IMAP->num_msgs; ++i)
 610           if (IMAP->flags[i] & IMAP_SELECTED) {
 611                 if (imap_do_search_msg(i+1, NULL, num_items, itemlist, is_uid)) {
 612                         if (num_results != 0) {
 613                                 cprintf(" ");
 614                         }
 615                         if (is_uid) {
 616                                 cprintf("%ld", IMAP->msgids[i]);
 617                         }
 618                         else {
 619                                 cprintf("%d", i+1);
 620                         }
 621                         ++num_results;
 622                 }
 623         }
 624         cprintf("\r\n");
 625         unbuffer_output();
 626 }
 627
 628
 629 /*
 630  * This function is called by the main command loop.
 631  */
 632 void imap_search(int num_parms, char *parms[]) {
 633         int i;
 634
 635         if (num_parms < 3) {
 636                 cprintf("%s BAD invalid parameters\r\n", parms[0]);
 637                 return;
 638         }
 639
 640         for (i = 0; i < IMAP->num_msgs; ++i) {
 641                 IMAP->flags[i] |= IMAP_SELECTED;
 642         }
 643
 644         imap_do_search(num_parms-2, &parms[2], 0);
 645         cprintf("%s OK SEARCH completed\r\n", parms[0]);
 646 }
 647
 648 /*
 649  * This function is called by the main command loop.
 650  */
 651 void imap_uidsearch(int num_parms, char *parms[]) {
 652         int i;
 653
 654         if (num_parms < 4) {
 655                 cprintf("%s BAD invalid parameters\r\n", parms[0]);
 656                 return;
 657         }
 658
 659         for (i = 0; i < IMAP->num_msgs; ++i) {
 660                 IMAP->flags[i] |= IMAP_SELECTED;
 661         }
 662
 663         imap_do_search(num_parms-3, &parms[3], 1);
 664         cprintf("%s OK UID SEARCH completed\r\n", parms[0]);
 665 }
 666
 667