citadel/modules/imap/imap_search.c

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