functions/imap_asearch.php

   1 <?php
   2
   3 /**
   4 * imap_search.php
   5 *
   6 * Copyright (c) 1999-2004 The SquirrelMail Project Team
   7 * Licensed under the GNU GPL. For full terms see the file COPYING.
   8 *
   9 * IMAP asearch routines
  10 *
  11 * Subfolder search idea from Patch #806075 by Thomas Pohl xraven at users.sourceforge.net. Thanks Thomas!
  12 *
  13 * @version $Id$
  14 * @package squirrelmail
  15 * @subpackage imap
  16 * @see search.php
  17 * @link http://www.ietf.org/rfc/rfc3501.txt
  18 * @author Alex Lemaresquier - Brainstorm - alex at brainstorm.fr
  19 */
  20
  21 /** This functionality requires the IMAP and date functions
  22 */
  23 require_once(SM_PATH . 'functions/imap_general.php');
  24 require_once(SM_PATH . 'functions/date.php');
  25
  26 /** Set to TRUE to dump the imap dialogue
  27 * @global bool $imap_asearch_debug_dump
  28 */
  29 $imap_asearch_debug_dump = FALSE;
  30
  31 /** Imap SEARCH keys
  32 * @global array $imap_asearch_opcodes
  33 */
  34 $imap_asearch_opcodes = array(
  35 /* <sequence-set> => 'asequence', */    // Special handling, @see sqimap_asearch_build_criteria()
  36 /*'ALL' is binary operator */
  37     'ANSWERED' => '',
  38     'BCC' => 'astring',
  39     'BEFORE' => 'adate',
  40     'BODY' => 'astring',
  41     'CC' => 'astring',
  42     'DELETED' => '',
  43     'DRAFT' => '',
  44     'FLAGGED' => '',
  45     'FROM' => 'astring',
  46     'HEADER' => 'afield',       // Special syntax for this one, @see sqimap_asearch_build_criteria()
  47     'KEYWORD' => 'akeyword',
  48     'LARGER' => 'anum',
  49     'NEW' => '',
  50 /*'NOT' is unary operator */
  51     'OLD' => '',
  52     'ON' => 'adate',
  53 /*'OR' is binary operator */
  54     'RECENT' => '',
  55     'SEEN' => '',
  56     'SENTBEFORE' => 'adate',
  57     'SENTON' => 'adate',
  58     'SENTSINCE' => 'adate',
  59     'SINCE' => 'adate',
  60     'SMALLER' => 'anum',
  61     'SUBJECT' => 'astring',
  62     'TEXT' => 'astring',
  63     'TO' => 'astring',
  64     'UID' => 'asequence',
  65     'UNANSWERED' => '',
  66     'UNDELETED' => '',
  67     'UNDRAFT' => '',
  68     'UNFLAGGED' => '',
  69     'UNKEYWORD' => 'akeyword',
  70     'UNSEEN' => ''
  71 );
  72
  73 /** Imap SEARCH month names encoding
  74 * @global array $imap_asearch_months
  75 */
  76 $imap_asearch_months = array(
  77     '01' => 'jan',
  78     '02' => 'feb',
  79     '03' => 'mar',
  80     '04' => 'apr',
  81     '05' => 'may',
  82     '06' => 'jun',
  83     '07' => 'jul',
  84     '08' => 'aug',
  85     '09' => 'sep',
  86     '10' => 'oct',
  87     '11' => 'nov',
  88     '12' => 'dec'
  89 );
  90
  91 /** Error message titles according to imap server returned code
  92 * @global array $imap_error_titles
  93 */
  94 $imap_error_titles = array(
  95     'OK' => '',
  96     'NO' => _("ERROR : Could not complete request."),
  97     'BAD' => _("ERROR : Bad or malformed request."),
  98     'BYE' => _("ERROR : Imap server closed the connection."),
  99     '' => _("ERROR : Connection dropped by imap-server.")
 100 );
 101
 102 /**
 103 * Function to display an error related to an IMAP-query.
 104 * We need to do our own error management since we may receive NO responses on purpose (even BAD with SORT or THREAD)
 105 * so we call sqimap_error_box() if the function exists (sm >= 1.5) or use our own embedded code
 106 * @global array imap_error_titles
 107 * @param string $response the imap server response code
 108 * @param string $query the failed query
 109 * @param string $message an optional error message
 110 * @param string $link an optional link to try again
 111 */
 112 //@global array color sm colors array
 113 function sqimap_asearch_error_box($response, $query, $message, $link = '')
 114 {
 115     global $imap_error_titles;
 116
 117     if (!array_key_exists($response, $imap_error_titles))
 118         $title = _("ERROR : Unknown imap response.");
 119     else
 120         $title = $imap_error_titles[$response];
 121     if ($link == '')
 122         $message_title = _("Reason Given: ");
 123     else
 124         $message_title = _("Possible reason : ");
 125     if (function_exists('sqimap_error_box'))
 126         sqimap_error_box($title, $query, $message_title, $message, $link);
 127     else {      //Straight copy of 1.5 imap_general.php:sqimap_error_box(). Can be removed at a later time
 128         global $color;
 129     require_once(SM_PATH . 'functions/display_messages.php');
 130     $string = "<font color=$color[2]><b>\n" . $title . "</b><br>\n";
 131     if ($query != '')
 132         $string .= _("Query:") . ' ' . htmlspecialchars($query) . '<br>';
 133     if ($message_title != '')
 134         $string .= $message_title;
 135     if ($message != '')
 136         $string .= htmlspecialchars($message);
 137     if ($link != '')
 138         $string .= $link;
 139     $string .= "</font><br>\n";
 140     error_box($string,$color);
 141     }
 142 }
 143
 144 /**
 145 * This is a convenient way to avoid spreading if (isset(... all over the code
 146 * @param mixed $var any variable (reference)
 147 * @param mixed $def default value to return if unset (default is zls (''), pass 0 or array() when appropriate)
 148 * @return mixed $def if $var is unset, otherwise $var
 149 */
 150 function asearch_nz(&$var, $def = '')
 151 {
 152     if (isset($var))
 153         return $var;
 154     return $def;
 155 }
 156
 157 /**
 158 * This should give the same results as PHP 4 >= 4.3.0's html_entity_decode(),
 159 * except it doesn't handle hex constructs
 160 * @param string $string string to unhtmlentity()
 161 * @return string decoded string
 162 */
 163 function asearch_unhtmlentities($string) {
 164     $trans_tbl = array_flip(get_html_translation_table(HTML_ENTITIES));
 165     for ($i=127; $i<255; $i++)  /* Add &#<dec>; entities */
 166         $trans_tbl['&#' . $i . ';'] = chr($i);
 167     return strtr($string, $trans_tbl);
 168 /* I think the one above is quicker, though it should be benchmarked
 169     $string = strtr($string, array_flip(get_html_translation_table(HTML_ENTITIES)));
 170     return preg_replace("/&#([0-9]+);/E", "chr('\\1')", $string);
 171 */
 172 }
 173
 174 /**
 175 * Provide an easy way to dump the imap dialogue if $imap_asearch_debug_dump is TRUE
 176 * @global bool imap_asearch_debug_dump
 177 * @param string $var_name
 178 * @param string $var_var
 179 */
 180 function s_debug_dump($var_name, $var_var)
 181 {
 182     global $imap_asearch_debug_dump;
 183     if ($imap_asearch_debug_dump) {
 184         if (function_exists('sm_print_r'))      //Only exists since 1.4.2
 185             sm_print_r($var_name, $var_var);    //Better be the 'varargs' version ;)
 186         else {
 187             echo '<pre>';
 188             echo htmlentities($var_name);
 189             print_r($var_var);
 190             echo '</pre>';
 191         }
 192     }
 193 }
 194
 195 /** Encode a string to quoted or literal as defined in rfc 3501
 196 *
 197 * -  4.3 String:
 198 *       A quoted string is a sequence of zero or more 7-bit characters,
 199 *        excluding CR and LF, with double quote (<">) characters at each end.
 200 * -  9. Formal Syntax:
 201 *       quoted-specials = DQUOTE / "\"
 202 * @param string $what string to encode
 203 * @param string $charset search charset used
 204 * @return string encoded string
 205 */
 206 function sqimap_asearch_encode_string($what, $charset)
 207 {
 208     if (strtoupper($charset) == 'ISO-2022-JP')  // This should be now handled in imap_utf7_local?
 209         $what = mb_convert_encoding($what, 'JIS', 'auto');
 210     if (preg_match('/["\\\\\r\n\x80-\xff]/', $what))
 211         return '{' . strlen($what) . "}\r\n" . $what;   // 4.3 literal form
 212     return '"' . $what . '"';   // 4.3 quoted string form
 213 }
 214
 215 /**
 216 * Parses a user date string into an rfc 3501 date string
 217 * Handles space, slash, backslash, dot and comma as separators (and dash of course ;=)
 218 * @global array imap_asearch_months
 219 * @param string user date
 220 * @return array a preg_match-style array:
 221 *  - [0] = fully formatted rfc 3501 date string (<day number>-<US month TLA>-<4 digit year>)
 222 *  - [1] = day
 223 *  - [2] = month
 224 *  - [3] = year
 225 */
 226 function sqimap_asearch_parse_date($what)
 227 {
 228     global $imap_asearch_months;
 229
 230     $what = trim($what);
 231     $what = ereg_replace('[ /\\.,]+', '-', $what);
 232     if ($what) {
 233         preg_match('/^([0-9]+)-+([^\-]+)-+([0-9]+)$/', $what, $what_parts);
 234         if (count($what_parts) == 4) {
 235             $what_month = strtolower(asearch_unhtmlentities($what_parts[2]));
 236 /*              if (!in_array($what_month, $imap_asearch_months)) {*/
 237                 foreach ($imap_asearch_months as $month_number => $month_code) {
 238                     if (($what_month == $month_number)
 239                     || ($what_month == $month_code)
 240                     || ($what_month == strtolower(asearch_unhtmlentities(getMonthName($month_number))))
 241                     || ($what_month == strtolower(asearch_unhtmlentities(getMonthAbrv($month_number))))
 242                     ) {
 243                         $what_parts[2] = $month_number;
 244                         $what_parts[0] = $what_parts[1] . '-' . $month_code . '-' . $what_parts[3];
 245                         break;
 246                     }
 247                 }
 248 /*              }*/
 249         }
 250     }
 251     else
 252         $what_parts = array();
 253     return $what_parts;
 254 }
 255
 256 /**
 257 * Build one criteria sequence
 258 * @global array imap_asearch_opcodes
 259 * @param string $opcode search opcode
 260 * @param string $what opcode argument
 261 * @param string $charset search charset
 262 * @return string one full criteria sequence
 263 */
 264 function sqimap_asearch_build_criteria($opcode, $what, $charset)
 265 {
 266     global $imap_asearch_opcodes;
 267
 268     $criteria = '';
 269     switch ($imap_asearch_opcodes[$opcode]) {
 270         default:
 271         case 'anum':
 272             $what = str_replace(' ', '', $what);
 273             $what = ereg_replace('[^0-9]+[^KMG]$', '', strtoupper($what));
 274             if ($what != '') {
 275                 switch (substr($what, -1)) {
 276                     case 'G':
 277                         $what = substr($what, 0, -1) << 30;
 278                     break;
 279                     case 'M':
 280                         $what = substr($what, 0, -1) << 20;
 281                     break;
 282                     case 'K':
 283                         $what = substr($what, 0, -1) << 10;
 284                     break;
 285                 }
 286                 $criteria = $opcode . ' ' . $what . ' ';
 287             }
 288         break;
 289         case '':        //aflag
 290             $criteria = $opcode . ' ';
 291         break;
 292         case 'afield':  /* HEADER field-name: field-body */
 293             preg_match('/^([^:]+):(.*)$/', $what, $what_parts);
 294             if (count($what_parts) == 3)
 295                 $criteria = $opcode . ' ' .
 296                     sqimap_asearch_encode_string($what_parts[1], $charset) . ' ' .
 297                     sqimap_asearch_encode_string($what_parts[2], $charset) . ' ';
 298         break;
 299         case 'adate':
 300             $what_parts = sqimap_asearch_parse_date($what);
 301             if (isset($what_parts[0]))
 302                 $criteria = $opcode . ' ' . $what_parts[0] . ' ';
 303         break;
 304         case 'akeyword':
 305         case 'astring':
 306             $criteria = $opcode . ' ' . sqimap_asearch_encode_string($what, $charset) . ' ';
 307         break;
 308         case 'asequence':
 309             $what = ereg_replace('[^0-9:\(\)]+', '', $what);
 310             if ($what != '')
 311                 $criteria = $opcode . ' ' . $what . ' ';
 312         break;
 313     }
 314     return $criteria;
 315 }
 316
 317 /**
 318 * Another way to do array_values(array_unique(array_merge($to, $from)));
 319 * @param array $to to array (reference)
 320 * @param array $from from array
 321 * @return array uniquely merged array
 322 */
 323 function sqimap_array_merge_unique(&$to, $from)
 324 {
 325     if (empty($to))
 326         return $from;
 327     $count = count($from);
 328     for ($i = 0; $i < $count; $i++) {
 329         if (!in_array($from[$i], $to))
 330             $to[] = $from[$i];
 331     }
 332     return $to;
 333 }
 334
 335 /**
 336 * Run the imap SEARCH command as defined in rfc 3501
 337 * @link http://www.ietf.org/rfc/rfc3501.txt
 338 * @param resource $imapConnection the current imap stream
 339 * @param string $search_string the full search expression eg "ALL RECENT"
 340 * @param string $search_charset charset to use or zls ('')
 341 * @return array an IDs or UIDs array of matching messages or an empty array
 342 */
 343 function sqimap_run_search($imapConnection, $search_string, $search_charset)
 344 {
 345     //For some reason, this seems to happen and forbids searching servers not allowing OPTIONAL [CHARSET]
 346     if (strtoupper($search_charset) == 'US-ASCII')
 347         $search_charset = '';
 348     /* 6.4.4 try OPTIONAL [CHARSET] specification first */
 349     if ($search_charset != '')
 350         $query = 'SEARCH CHARSET "' . strtoupper($search_charset) . '" ' . $search_string;
 351     else
 352         $query = 'SEARCH ' . $search_string;
 353     s_debug_dump('C:', $query);
 354     $readin = sqimap_run_command($imapConnection, $query, false, $response, $message, TRUE);
 355
 356     /* 6.4.4 try US-ASCII charset if we tried an OPTIONAL [CHARSET] and received a tagged NO response (SHOULD be [BADCHARSET]) */
 357     if (($search_charset != '')  && (strtoupper($response) == 'NO')) {
 358         $query = 'SEARCH CHARSET US-ASCII ' . $search_string;
 359         s_debug_dump('C:', $query);
 360         $readin = sqimap_run_command($imapConnection, $query, false, $response, $message, TRUE);
 361     }
 362     if (strtoupper($response) != 'OK') {
 363         sqimap_asearch_error_box($response, $query, $message);
 364         return array();
 365     }
 366     $messagelist = parseUidList($readin,'SEARCH');
 367
 368     if (empty($messagelist))    //Empty search response, ie '* SEARCH'
 369         return array();
 370
 371     $cnt = count($messagelist);
 372     for ($q = 0; $q < $cnt; $q++)
 373         $id[$q] = trim($messagelist[$q]);
 374     return $id;
 375 }
 376
 377 /**
 378 * @global bool allow_charset_search user setting
 379 * @global array languages sm languages array
 380 * @global string squirrelmail_language user language setting
 381 * @return string the user defined charset if $allow_charset_search is TRUE else zls ('')
 382 */
 383 function sqimap_asearch_get_charset()
 384 {
 385     global $allow_charset_search, $languages, $squirrelmail_language;
 386
 387     if ($allow_charset_search)
 388         return $languages[$squirrelmail_language]['CHARSET'];
 389     return '';
 390 }
 391
 392 /**
 393 * Convert sm internal sort to imap sort taking care of:
 394 * - user defined date sorting (ARRIVAL vs DATE)
 395 * - if the searched mailbox is the sent folder then TO is being used instead of FROM
 396 * - reverse order by using REVERSE
 397 * @param string $mailbox mailbox name to sort
 398 * @param integer $sort_by sm sort criteria index
 399 * @global bool internal_date_sort sort by arrival date instead of message date
 400 * @global string sent_folder sent folder name
 401 * @return string imap sort criteria
 402 */
 403 function sqimap_asearch_get_sort_criteria($mailbox, $sort_by)
 404 {
 405     global $internal_date_sort, $sent_folder;
 406
 407     $sort_opcodes = array ('DATE', 'FROM', 'SUBJECT', 'SIZE');
 408     if ($internal_date_sort == true)
 409         $sort_opcodes[0] = 'ARRIVAL';
 410 //      if (handleAsSent($mailbox))
 411 //      if (isSentFolder($mailbox))
 412     if ($mailbox == $sent_folder)
 413         $sort_opcodes[1] = 'TO';
 414     return (($sort_by % 2) ? '' : 'REVERSE ') . $sort_opcodes[($sort_by >> 1) & 3];
 415 }
 416
 417 /**
 418 * @param string $cur_mailbox unformatted mailbox name
 419 * @param array $boxes_unformatted selectable mailbox unformatted names array (reference)
 420 * @return array sub mailboxes unformatted names
 421 */
 422 function sqimap_asearch_get_sub_mailboxes($cur_mailbox, &$mboxes_array)
 423 {
 424     $sub_mboxes_array = array();
 425     $boxcount = count($mboxes_array);
 426     for ($boxnum=0; $boxnum < $boxcount; $boxnum++) {
 427         if (isBoxBelow($mboxes_array[$boxnum], $cur_mailbox))
 428             $sub_mboxes_array[] = $mboxes_array[$boxnum];
 429     }
 430     return $sub_mboxes_array;
 431 }
 432
 433 /**
 434 * Create the search query strings for all given criteria and merge results for every mailbox
 435 * @param resource $imapConnection
 436 * @param array $mailbox_array (reference)
 437 * @param array $biop_array (reference)
 438 * @param array $unop_array (reference)
 439 * @param array $where_array (reference)
 440 * @param array $what_array (reference)
 441 * @param array $exclude_array (reference)
 442 * @param array $sub_array (reference)
 443 * @param array $mboxes_array selectable unformatted mailboxes names (reference)
 444 * @return array array(mailbox => array(UIDs))
 445 */
 446 function sqimap_asearch($imapConnection, &$mailbox_array, &$biop_array, &$unop_array, &$where_array, &$what_array, &$exclude_array, &$sub_array, &$mboxes_array)
 447 {
 448
 449     $search_charset = sqimap_asearch_get_charset();
 450     $mbox_msgs = array();
 451     $mbox_search = array();
 452     $search_string = '';
 453     $cur_mailbox = $mailbox_array[0];
 454     $cur_biop = '';     /* Start with ALL */
 455     /* We loop one more time than the real array count, so the last search gets fired */
 456     for ($cur_crit=0,$iCnt=count($where_array); $cur_crit <= $iCnt; ++$cur_crit) {
 457         if (empty($exclude_array[$cur_crit])) {
 458             $next_mailbox = (isset($mailbox_array[$cur_crit])) ? $mailbox_array[$cur_crit] : false;
 459             if ($next_mailbox != $cur_mailbox) {
 460                 $search_string = trim($search_string);  /* Trim out last space */
 461                 if ($cur_mailbox == 'All Folders')
 462                         $search_mboxes = $mboxes_array;
 463                 else if ((!empty($sub_array[$cur_crit - 1])) || (!in_array($cur_mailbox, $mboxes_array)))
 464                     $search_mboxes = sqimap_asearch_get_sub_mailboxes($cur_mailbox, $mboxes_array);
 465                 else
 466                     $search_mboxes = array($cur_mailbox);
 467                 foreach ($search_mboxes as $cur_mailbox) {
 468                     if (isset($mbox_search[$cur_mailbox])) {
 469                         $mbox_search[$cur_mailbox]['search'] .= ' ' . $search_string;
 470                     } else {
 471                         $mbox_search[$cur_mailbox]['search'] = $search_string;
 472                     }
 473                     $mbox_search[$cur_mailbox]['charset'] = $search_charset;
 474                 }
 475                 $cur_mailbox = $next_mailbox;
 476                 $search_string = '';
 477
 478             }
 479             if (isset($where_array[$cur_crit]) && empty($exclude_array[$cur_crit])) {
 480                 for ($crit = $cur_crit; $crit < count($where_array); $crit++) {
 481                     $criteria = trim(sqimap_asearch_build_criteria($where_array[$crit], $what_array[$crit], $search_charset));
 482                     if (!empty($criteria) && empty($exclude_array[$crit])) {
 483                         if (asearch_nz($mailbox_array[$crit]) == $cur_mailbox) {
 484                             $unop = $unop_array[$crit];
 485                             if (!empty($unop)) {
 486                                 $criteria = $unop . ' ' . $criteria;
 487                             }
 488                             $aCriteria[] = array($biop_array[$crit], $criteria);
 489                         }
 490                     }
 491                     // unset something
 492                     $exclude_array[$crit] = true;
 493                 }
 494                 $aSearch = array();
 495                 for($i=0,$iCnt=count($aCriteria);$i<$iCnt;++$i) {
 496                     $cur_biop = $aCriteria[$i][0];
 497                     $next_biop = (isset($aCriteria[$i+1][0])) ? $aCriteria[$i+1][0] : false;
 498                     if ($next_biop != $cur_biop && $next_biop == 'OR') {
 499                         $aSearch[] = 'OR '.$aCriteria[$i][1];
 500                     } else if ($cur_biop != 'OR') {
 501                         $aSearch[] = 'ALL '.$aCriteria[$i][1];
 502                     } else { // or only supports 2 search keys so we need to create a parenthized list
 503                         $prev_biop = (isset($aCriteria[$i-1][0])) ? $aCriteria[$i-1][0] : false;
 504                         if ($prev_biop == $cur_biop) {
 505                             $last = $aSearch[$i-1];
 506                             if (!substr($last,-1) == ')') {
 507                                 $aSearch[$i-1] = "(OR $last";
 508                                 $aSearch[] = $aCriteria[$i][1].')';
 509                             } else {
 510                                 $sEnd = '';
 511                                 while ($last && substr($last,-1) == ')') {
 512                                 $last = substr($last,0,-1);
 513                                 $sEnd .= ')';
 514                                 }
 515                                 $aSearch[$i-1] = "(OR $last";
 516                                 $aSearch[] = $aCriteria[$i][1].$sEnd.')';
 517                             }
 518                         } else {
 519                             $aSearch[] = $aCriteria[$i][1];
 520                         }
 521                     }
 522
 523                 }
 524                 $search_string .= implode(' ',$aSearch);
 525             }
 526
 527         }
 528     }
 529     return ($mbox_search);
 530 }
 531
 532 ?>