functions/imap_asearch.php

   1 <?php
   2
   3 /**
   4 * imap_search.php
   5 *
   6 * Copyright (c) 1999-2005 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 global $imap_asearch_opcodes;
  35 $imap_asearch_opcodes = array(
  36 /* <sequence-set> => 'asequence', */    // Special handling, @see sqimap_asearch_build_criteria()
  37 /*'ALL' is binary operator */
  38     'ANSWERED' => '',
  39     'BCC' => 'astring',
  40     'BEFORE' => 'adate',
  41     'BODY' => 'astring',
  42     'CC' => 'astring',
  43     'DELETED' => '',
  44     'DRAFT' => '',
  45     'FLAGGED' => '',
  46     'FROM' => 'astring',
  47     'HEADER' => 'afield',    // Special syntax for this one, @see sqimap_asearch_build_criteria()
  48     'KEYWORD' => 'akeyword',
  49     'LARGER' => 'anum',
  50     'NEW' => '',
  51 /*'NOT' is unary operator */
  52     'OLD' => '',
  53     'ON' => 'adate',
  54 /*'OR' is binary operator */
  55     'RECENT' => '',
  56     'SEEN' => '',
  57     'SENTBEFORE' => 'adate',
  58     'SENTON' => 'adate',
  59     'SENTSINCE' => 'adate',
  60     'SINCE' => 'adate',
  61     'SMALLER' => 'anum',
  62     'SUBJECT' => 'astring',
  63     'TEXT' => 'astring',
  64     'TO' => 'astring',
  65     'UID' => 'asequence',
  66     'UNANSWERED' => '',
  67     'UNDELETED' => '',
  68     'UNDRAFT' => '',
  69     'UNFLAGGED' => '',
  70     'UNKEYWORD' => 'akeyword',
  71     'UNSEEN' => ''
  72 );
  73
  74 /** Imap SEARCH month names encoding
  75 * @global array $imap_asearch_months
  76 */
  77 $imap_asearch_months = array(
  78     '01' => 'jan',
  79     '02' => 'feb',
  80     '03' => 'mar',
  81     '04' => 'apr',
  82     '05' => 'may',
  83     '06' => 'jun',
  84     '07' => 'jul',
  85     '08' => 'aug',
  86     '09' => 'sep',
  87     '10' => 'oct',
  88     '11' => 'nov',
  89     '12' => 'dec'
  90 );
  91
  92 /**
  93 * Function to display an error related to an IMAP-query.
  94 * We need to do our own error management since we may receive NO responses on purpose (even BAD with SORT or THREAD)
  95 * so we call sqimap_error_box() if the function exists (sm >= 1.5) or use our own embedded code
  96 * @global array imap_error_titles
  97 * @param string $response the imap server response code
  98 * @param string $query the failed query
  99 * @param string $message an optional error message
 100 * @param string $link an optional link to try again
 101 */
 102 //@global array color sm colors array
 103 function sqimap_asearch_error_box($response, $query, $message, $link = '')
 104 {
 105     global $color;
 106     // Error message titles according to imap server returned code
 107     $imap_error_titles = array(
 108         'OK' => '',
 109         'NO' => _("ERROR : Could not complete request."),
 110         'BAD' => _("ERROR : Bad or malformed request."),
 111         'BYE' => _("ERROR : Imap server closed the connection."),
 112         '' => _("ERROR : Connection dropped by imap-server.")
 113     );
 114
 115
 116     if (!array_key_exists($response, $imap_error_titles))
 117         $title = _("ERROR : Unknown imap response.");
 118     else
 119         $title = $imap_error_titles[$response];
 120     if ($link == '')
 121         $message_title = _("Reason Given: ");
 122     else
 123         $message_title = _("Possible reason : ");
 124     if (function_exists('sqimap_error_box'))
 125         sqimap_error_box($title, $query, $message_title, $message, $link);
 126     else {    //Straight copy of 1.5 imap_general.php:sqimap_error_box(). Can be removed at a later time
 127         global $color;
 128     require_once(SM_PATH . 'functions/display_messages.php');
 129     $string = "<font color=\"$color[2]\"><b>\n" . $title . "</b><br />\n";
 130     if ($query != '')
 131         $string .= _("Query:") . ' ' . htmlspecialchars($query) . '<br />';
 132     if ($message_title != '')
 133         $string .= $message_title;
 134     if ($message != '')
 135         $string .= htmlspecialchars($message);
 136     if ($link != '')
 137         $string .= $link;
 138     $string .= "</font><br />\n";
 139     error_box($string,$color);
 140     }
 141 }
 142
 143 /**
 144 * This is a convenient way to avoid spreading if (isset(... all over the code
 145 * @param mixed $var any variable (reference)
 146 * @param mixed $def default value to return if unset (default is zls (''), pass 0 or array() when appropriate)
 147 * @return mixed $def if $var is unset, otherwise $var
 148 */
 149 function asearch_nz(&$var, $def = '')
 150 {
 151     if (isset($var))
 152         return $var;
 153     return $def;
 154 }
 155
 156 /**
 157 * This should give the same results as PHP 4 >= 4.3.0's html_entity_decode(),
 158 * except it doesn't handle hex constructs
 159 * @param string $string string to unhtmlentity()
 160 * @return string decoded string
 161 */
 162 function asearch_unhtmlentities($string) {
 163     $trans_tbl = array_flip(get_html_translation_table(HTML_ENTITIES));
 164     for ($i=127; $i<255; $i++)    /* Add &#<dec>; entities */
 165         $trans_tbl['&#' . $i . ';'] = chr($i);
 166     return strtr($string, $trans_tbl);
 167 /* I think the one above is quicker, though it should be benchmarked
 168     $string = strtr($string, array_flip(get_html_translation_table(HTML_ENTITIES)));
 169     return preg_replace("/&#([0-9]+);/E", "chr('\\1')", $string);
 170 */
 171 }
 172
 173 /**
 174 * Provide an easy way to dump the imap dialogue if $imap_asearch_debug_dump is TRUE
 175 * @global bool imap_asearch_debug_dump
 176 * @param string $var_name
 177 * @param string $var_var
 178 */
 179 function s_debug_dump($var_name, $var_var)
 180 {
 181     global $imap_asearch_debug_dump;
 182     if ($imap_asearch_debug_dump) {
 183         if (function_exists('sm_print_r'))      //Only exists since 1.4.2
 184             sm_print_r($var_name, $var_var);    //Better be the 'varargs' version ;)
 185         else {
 186             echo '<pre>';
 187             echo htmlentities($var_name);
 188             print_r($var_var);
 189             echo '</pre>';
 190         }
 191     }
 192 }
 193
 194 /** Encode a string to quoted or literal as defined in rfc 3501
 195 *
 196 * -  4.3 String:
 197 *        A quoted string is a sequence of zero or more 7-bit characters,
 198 *         excluding CR and LF, with double quote (<">) characters at each end.
 199 * -  9. Formal Syntax:
 200 *        quoted-specials = DQUOTE / "\"
 201 * @param string $what string to encode
 202 * @param string $charset search charset used
 203 * @return string encoded string
 204 */
 205 function sqimap_asearch_encode_string($what, $charset)
 206 {
 207     if (strtoupper($charset) == 'ISO-2022-JP')    // This should be now handled in imap_utf7_local?
 208         $what = mb_convert_encoding($what, 'JIS', 'auto');
 209     if (preg_match('/["\\\\\r\n\x80-\xff]/', $what))
 210         return '{' . strlen($what) . "}\r\n" . $what;    // 4.3 literal form
 211     return '"' . $what . '"';    // 4.3 quoted string form
 212 }
 213
 214 /**
 215 * Parses a user date string into an rfc 3501 date string
 216 * Handles space, slash, backslash, dot and comma as separators (and dash of course ;=)
 217 * @global array imap_asearch_months
 218 * @param string user date
 219 * @return array a preg_match-style array:
 220 *  - [0] = fully formatted rfc 3501 date string (<day number>-<US month TLA>-<4 digit year>)
 221 *  - [1] = day
 222 *  - [2] = month
 223 *  - [3] = year
 224 */
 225 function sqimap_asearch_parse_date($what)
 226 {
 227     global $imap_asearch_months;
 228
 229     $what = trim($what);
 230     $what = ereg_replace('[ /\\.,]+', '-', $what);
 231     if ($what) {
 232         preg_match('/^([0-9]+)-+([^\-]+)-+([0-9]+)$/', $what, $what_parts);
 233         if (count($what_parts) == 4) {
 234             $what_month = strtolower(asearch_unhtmlentities($what_parts[2]));
 235 /*                if (!in_array($what_month, $imap_asearch_months)) {*/
 236                 foreach ($imap_asearch_months as $month_number => $month_code) {
 237                     if (($what_month == $month_number)
 238                     || ($what_month == $month_code)
 239                     || ($what_month == strtolower(asearch_unhtmlentities(getMonthName($month_number))))
 240                     || ($what_month == strtolower(asearch_unhtmlentities(getMonthAbrv($month_number))))
 241                     ) {
 242                         $what_parts[2] = $month_number;
 243                         $what_parts[0] = $what_parts[1] . '-' . $month_code . '-' . $what_parts[3];
 244                         break;
 245                     }
 246                 }
 247 /*                }*/
 248         }
 249     }
 250     else
 251         $what_parts = array();
 252     return $what_parts;
 253 }
 254
 255 /**
 256 * Build one criteria sequence
 257 * @global array imap_asearch_opcodes
 258 * @param string $opcode search opcode
 259 * @param string $what opcode argument
 260 * @param string $charset search charset
 261 * @return string one full criteria sequence
 262 */
 263 function sqimap_asearch_build_criteria($opcode, $what, $charset)
 264 {
 265     global $imap_asearch_opcodes;
 266
 267     $criteria = '';
 268     switch ($imap_asearch_opcodes[$opcode]) {
 269         default:
 270         case 'anum':
 271             $what = str_replace(' ', '', $what);
 272             $what = ereg_replace('[^0-9]+[^KMG]$', '', strtoupper($what));
 273             if ($what != '') {
 274                 switch (substr($what, -1)) {
 275                     case 'G':
 276                         $what = substr($what, 0, -1) << 30;
 277                     break;
 278                     case 'M':
 279                         $what = substr($what, 0, -1) << 20;
 280                     break;
 281                     case 'K':
 282                         $what = substr($what, 0, -1) << 10;
 283                     break;
 284                 }
 285                 $criteria = $opcode . ' ' . $what . ' ';
 286             }
 287         break;
 288         case '':    //aflag
 289             $criteria = $opcode . ' ';
 290         break;
 291         case 'afield':    /* HEADER field-name: field-body */
 292             preg_match('/^([^:]+):(.*)$/', $what, $what_parts);
 293             if (count($what_parts) == 3)
 294                 $criteria = $opcode . ' ' .
 295                     sqimap_asearch_encode_string($what_parts[1], $charset) . ' ' .
 296                     sqimap_asearch_encode_string($what_parts[2], $charset) . ' ';
 297         break;
 298         case 'adate':
 299             $what_parts = sqimap_asearch_parse_date($what);
 300             if (isset($what_parts[0]))
 301                 $criteria = $opcode . ' ' . $what_parts[0] . ' ';
 302         break;
 303         case 'akeyword':
 304         case 'astring':
 305             $criteria = $opcode . ' ' . sqimap_asearch_encode_string($what, $charset) . ' ';
 306         break;
 307         case 'asequence':
 308             $what = ereg_replace('[^0-9:\(\)]+', '', $what);
 309             if ($what != '')
 310                 $criteria = $opcode . ' ' . $what . ' ';
 311         break;
 312     }
 313     return $criteria;
 314 }
 315
 316 /**
 317 * Another way to do array_values(array_unique(array_merge($to, $from)));
 318 * @param array $to to array (reference)
 319 * @param array $from from array
 320 * @return array uniquely merged array
 321 */
 322 function sqimap_array_merge_unique(&$to, $from)
 323 {
 324     if (empty($to))
 325         return $from;
 326     $count = count($from);
 327     for ($i = 0; $i < $count; $i++) {
 328         if (!in_array($from[$i], $to))
 329             $to[] = $from[$i];
 330     }
 331     return $to;
 332 }
 333
 334 /**
 335 * Run the imap SEARCH command as defined in rfc 3501
 336 * @link http://www.ietf.org/rfc/rfc3501.txt
 337 * @param resource $imapConnection the current imap stream
 338 * @param string $search_string the full search expression eg "ALL RECENT"
 339 * @param string $search_charset charset to use or zls ('')
 340 * @return array an IDs or UIDs array of matching messages or an empty array
 341 * @since 1.5.0
 342 */
 343 function sqimap_run_search($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_search = array();
 451     $search_string = '';
 452     $cur_mailbox = $mailbox_array[0];
 453     $cur_biop = '';    /* Start with ALL */
 454     /* We loop one more time than the real array count, so the last search gets fired */
 455     for ($cur_crit=0,$iCnt=count($where_array); $cur_crit <= $iCnt; ++$cur_crit) {
 456         if (empty($exclude_array[$cur_crit])) {
 457             $next_mailbox = (isset($mailbox_array[$cur_crit])) ? $mailbox_array[$cur_crit] : false;
 458             if ($next_mailbox != $cur_mailbox) {
 459                 $search_string = trim($search_string);    /* Trim out last space */
 460                 if ($cur_mailbox == 'All Folders')
 461                     $search_mboxes = $mboxes_array;
 462                 else if ((!empty($sub_array[$cur_crit - 1])) || (!in_array($cur_mailbox, $mboxes_array)))
 463                     $search_mboxes = sqimap_asearch_get_sub_mailboxes($cur_mailbox, $mboxes_array);
 464                 else
 465                     $search_mboxes = array($cur_mailbox);
 466                 foreach ($search_mboxes as $cur_mailbox) {
 467                     if (isset($mbox_search[$cur_mailbox])) {
 468                         $mbox_search[$cur_mailbox]['search'] .= ' ' . $search_string;
 469                     } else {
 470                         $mbox_search[$cur_mailbox]['search'] = $search_string;
 471                     }
 472                     $mbox_search[$cur_mailbox]['charset'] = $search_charset;
 473                 }
 474                 $cur_mailbox = $next_mailbox;
 475                 $search_string = '';
 476             }
 477             if (isset($where_array[$cur_crit]) && empty($exclude_array[$cur_crit])) {
 478                 for ($crit = $cur_crit; $crit < count($where_array); $crit++) {
 479                     $criteria = trim(sqimap_asearch_build_criteria($where_array[$crit], $what_array[$crit], $search_charset));
 480                     if (!empty($criteria) && empty($exclude_array[$crit])) {
 481                         if (asearch_nz($mailbox_array[$crit]) == $cur_mailbox) {
 482                             $unop = $unop_array[$crit];
 483                             if (!empty($unop)) {
 484                                 $criteria = $unop . ' ' . $criteria;
 485                             }
 486                             $aCriteria[] = array($biop_array[$crit], $criteria);
 487                         }
 488                     }
 489                     // unset something
 490                     $exclude_array[$crit] = true;
 491                 }
 492                 $aSearch = array();
 493                 for($i=0,$iCnt=count($aCriteria);$i<$iCnt;++$i) {
 494                     $cur_biop = $aCriteria[$i][0];
 495                     $next_biop = (isset($aCriteria[$i+1][0])) ? $aCriteria[$i+1][0] : false;
 496                     if ($next_biop != $cur_biop && $next_biop == 'OR') {
 497                         $aSearch[] = 'OR '.$aCriteria[$i][1];
 498                     } else if ($cur_biop != 'OR') {
 499                         $aSearch[] = 'ALL '.$aCriteria[$i][1];
 500                     } else { // OR only supports 2 search keys so we need to create a parenthesized list
 501                         $prev_biop = (isset($aCriteria[$i-1][0])) ? $aCriteria[$i-1][0] : false;
 502                         if ($prev_biop == $cur_biop) {
 503                             $last = $aSearch[$i-1];
 504                             if (!substr($last,-1) == ')') {
 505                                 $aSearch[$i-1] = "(OR $last";
 506                                 $aSearch[] = $aCriteria[$i][1].')';
 507                             } else {
 508                                 $sEnd = '';
 509                                 while ($last && substr($last,-1) == ')') {
 510                                     $last = substr($last,0,-1);
 511                                     $sEnd .= ')';
 512                                 }
 513                                 $aSearch[$i-1] = "(OR $last";
 514                                 $aSearch[] = $aCriteria[$i][1].$sEnd.')';
 515                             }
 516                         } else {
 517                             $aSearch[] = $aCriteria[$i][1];
 518                         }
 519                     }
 520                 }
 521                 $search_string .= implode(' ',$aSearch);
 522             }
 523         }
 524     }
 525     return ($mbox_search);
 526 }
 527
 528 ?>