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     $message_title .= ' ';
 125     if (function_exists('sqimap_error_box'))
 126         sqimap_error_box($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 $var_var
 179  * @deprecated contains workarounds for 1.4.0 and older code.
 180  *   code without workarounds uses regular sm 1.4.2+ functions.
 181  *   it is not compatible with 1.4.1
 182  * @todo remove debugging function
 183  */
 184 function s_debug_dump($var_name, $var_var)
 185 {
 186     global $imap_asearch_debug_dump;
 187     if ($imap_asearch_debug_dump) {
 188         if (function_exists('sm_print_r'))      //Only exists since 1.4.2
 189             sm_print_r($var_name, $var_var);    //Better be the 'varargs' version ;)
 190         else {
 191             echo '<pre>';
 192             echo htmlentities($var_name);
 193             print_r($var_var);
 194             echo '</pre>';
 195         }
 196     }
 197 }
 198
 199 /** Encode a string to quoted or literal as defined in rfc 3501
 200  *
 201  * -  4.3 String:
 202  *        A quoted string is a sequence of zero or more 7-bit characters,
 203  *         excluding CR and LF, with double quote (<">) characters at each end.
 204  * -  9. Formal Syntax:
 205  *        quoted-specials = DQUOTE / "\"
 206  * @param string $what string to encode
 207  * @param string $charset search charset used
 208  * @return string encoded string
 209  */
 210 function sqimap_asearch_encode_string($what, $charset)
 211 {
 212     if (strtoupper($charset) == 'ISO-2022-JP')    // This should be now handled in imap_utf7_local?
 213         $what = mb_convert_encoding($what, 'JIS', 'auto');
 214     if (preg_match('/["\\\\\r\n\x80-\xff]/', $what))
 215         return '{' . strlen($what) . "}\r\n" . $what;    // 4.3 literal form
 216     return '"' . $what . '"';    // 4.3 quoted string form
 217 }
 218
 219 /**
 220  * Parses a user date string into an rfc 3501 date string
 221  * Handles space, slash, backslash, dot and comma as separators (and dash of course ;=)
 222  * @global array imap_asearch_months
 223  * @param string user date
 224  * @return array a preg_match-style array:
 225  *  - [0] = fully formatted rfc 3501 date string (<day number>-<US month TLA>-<4 digit year>)
 226  *  - [1] = day
 227  *  - [2] = month
 228  *  - [3] = year
 229  */
 230 function sqimap_asearch_parse_date($what)
 231 {
 232     global $imap_asearch_months;
 233
 234     $what = trim($what);
 235     $what = ereg_replace('[ /\\.,]+', '-', $what);
 236     if ($what) {
 237         preg_match('/^([0-9]+)-+([^\-]+)-+([0-9]+)$/', $what, $what_parts);
 238         if (count($what_parts) == 4) {
 239             $what_month = strtolower(asearch_unhtmlentities($what_parts[2]));
 240 /*                if (!in_array($what_month, $imap_asearch_months)) {*/
 241                 foreach ($imap_asearch_months as $month_number => $month_code) {
 242                     if (($what_month == $month_number)
 243                     || ($what_month == $month_code)
 244                     || ($what_month == strtolower(asearch_unhtmlentities(getMonthName($month_number))))
 245                     || ($what_month == strtolower(asearch_unhtmlentities(getMonthAbrv($month_number))))
 246                     ) {
 247                         $what_parts[2] = $month_number;
 248                         $what_parts[0] = $what_parts[1] . '-' . $month_code . '-' . $what_parts[3];
 249                         break;
 250                     }
 251                 }
 252 /*                }*/
 253         }
 254     }
 255     else
 256         $what_parts = array();
 257     return $what_parts;
 258 }
 259
 260 /**
 261  * Build one criteria sequence
 262  * @global array imap_asearch_opcodes
 263  * @param string $opcode search opcode
 264  * @param string $what opcode argument
 265  * @param string $charset search charset
 266  * @return string one full criteria sequence
 267  */
 268 function sqimap_asearch_build_criteria($opcode, $what, $charset)
 269 {
 270     global $imap_asearch_opcodes;
 271
 272     $criteria = '';
 273     switch ($imap_asearch_opcodes[$opcode]) {
 274         default:
 275         case 'anum':
 276             $what = str_replace(' ', '', $what);
 277             $what = ereg_replace('[^0-9]+[^KMG]$', '', strtoupper($what));
 278             if ($what != '') {
 279                 switch (substr($what, -1)) {
 280                     case 'G':
 281                         $what = substr($what, 0, -1) << 30;
 282                     break;
 283                     case 'M':
 284                         $what = substr($what, 0, -1) << 20;
 285                     break;
 286                     case 'K':
 287                         $what = substr($what, 0, -1) << 10;
 288                     break;
 289                 }
 290                 $criteria = $opcode . ' ' . $what . ' ';
 291             }
 292         break;
 293         case '':    //aflag
 294             $criteria = $opcode . ' ';
 295         break;
 296         case 'afield':    /* HEADER field-name: field-body */
 297             preg_match('/^([^:]+):(.*)$/', $what, $what_parts);
 298             if (count($what_parts) == 3)
 299                 $criteria = $opcode . ' ' .
 300                     sqimap_asearch_encode_string($what_parts[1], $charset) . ' ' .
 301                     sqimap_asearch_encode_string($what_parts[2], $charset) . ' ';
 302         break;
 303         case 'adate':
 304             $what_parts = sqimap_asearch_parse_date($what);
 305             if (isset($what_parts[0]))
 306                 $criteria = $opcode . ' ' . $what_parts[0] . ' ';
 307         break;
 308         case 'akeyword':
 309         case 'astring':
 310             $criteria = $opcode . ' ' . sqimap_asearch_encode_string($what, $charset) . ' ';
 311         break;
 312         case 'asequence':
 313             $what = ereg_replace('[^0-9:\(\)]+', '', $what);
 314             if ($what != '')
 315                 $criteria = $opcode . ' ' . $what . ' ';
 316         break;
 317     }
 318     return $criteria;
 319 }
 320
 321 /**
 322  * Another way to do array_values(array_unique(array_merge($to, $from)));
 323  * @param array $to to array (reference)
 324  * @param array $from from array
 325  * @return array uniquely merged array
 326  */
 327 function sqimap_array_merge_unique(&$to, $from)
 328 {
 329     if (empty($to))
 330         return $from;
 331     $count = count($from);
 332     for ($i = 0; $i < $count; $i++) {
 333         if (!in_array($from[$i], $to))
 334             $to[] = $from[$i];
 335     }
 336     return $to;
 337 }
 338
 339 /**
 340  * Run the IMAP SEARCH command as defined in rfc 3501
 341  * @link http://www.ietf.org/rfc/rfc3501.txt
 342  * @param resource $imapConnection the current imap stream
 343  * @param string $search_string the full search expression eg "ALL RECENT"
 344  * @param string $search_charset charset to use or zls ('')
 345  * @return array an IDs or UIDs array of matching messages or an empty array
 346  * @since 1.5.0
 347  */
 348 function sqimap_run_search($imapConnection, $search_string, $search_charset)
 349 {
 350     //For some reason, this seems to happen and forbids searching servers not allowing OPTIONAL [CHARSET]
 351     if (strtoupper($search_charset) == 'US-ASCII')
 352         $search_charset = '';
 353     /* 6.4.4 try OPTIONAL [CHARSET] specification first */
 354     if ($search_charset != '')
 355         $query = 'SEARCH CHARSET "' . strtoupper($search_charset) . '" ' . $search_string;
 356     else
 357         $query = 'SEARCH ' . $search_string;
 358     s_debug_dump('C:', $query);
 359     $readin = sqimap_run_command($imapConnection, $query, false, $response, $message, TRUE);
 360
 361     /* 6.4.4 try US-ASCII charset if we tried an OPTIONAL [CHARSET] and received a tagged NO response (SHOULD be [BADCHARSET]) */
 362     if (($search_charset != '')  && (strtoupper($response) == 'NO')) {
 363         $query = 'SEARCH CHARSET US-ASCII ' . $search_string;
 364         s_debug_dump('C:', $query);
 365         $readin = sqimap_run_command($imapConnection, $query, false, $response, $message, TRUE);
 366     }
 367     if (strtoupper($response) != 'OK') {
 368         sqimap_asearch_error_box($response, $query, $message);
 369         return array();
 370     }
 371     $messagelist = parseUidList($readin,'SEARCH');
 372
 373     if (empty($messagelist))    //Empty search response, ie '* SEARCH'
 374         return array();
 375
 376     $cnt = count($messagelist);
 377     for ($q = 0; $q < $cnt; $q++)
 378         $id[$q] = trim($messagelist[$q]);
 379     return $id;
 380 }
 381
 382 /**
 383  * @global bool allow_charset_search user setting
 384  * @global array languages sm languages array
 385  * @global string squirrelmail_language user language setting
 386  * @return string the user defined charset if $allow_charset_search is TRUE else zls ('')
 387  */
 388 function sqimap_asearch_get_charset()
 389 {
 390     global $allow_charset_search, $languages, $squirrelmail_language;
 391
 392     if ($allow_charset_search)
 393         return $languages[$squirrelmail_language]['CHARSET'];
 394     return '';
 395 }
 396
 397 /**
 398  * Convert SquirrelMail internal sort to IMAP sort taking care of:
 399  * - user defined date sorting (ARRIVAL vs DATE)
 400  * - if the searched mailbox is the sent folder then TO is being used instead of FROM
 401  * - reverse order by using REVERSE
 402  * @param string $mailbox mailbox name to sort
 403  * @param integer $sort_by sm sort criteria index
 404  * @global bool internal_date_sort sort by arrival date instead of message date
 405  * @global string sent_folder sent folder name
 406  * @return string imap sort criteria
 407  */
 408 function sqimap_asearch_get_sort_criteria($mailbox, $sort_by)
 409 {
 410     global $internal_date_sort, $sent_folder;
 411
 412     $sort_opcodes = array ('DATE', 'FROM', 'SUBJECT', 'SIZE');
 413     if ($internal_date_sort == true)
 414         $sort_opcodes[0] = 'ARRIVAL';
 415 //        if (handleAsSent($mailbox))
 416 //        if (isSentFolder($mailbox))
 417     if ($mailbox == $sent_folder)
 418         $sort_opcodes[1] = 'TO';
 419     return (($sort_by % 2) ? '' : 'REVERSE ') . $sort_opcodes[($sort_by >> 1) & 3];
 420 }
 421
 422 /**
 423  * @param string $cur_mailbox unformatted mailbox name
 424  * @param array $boxes_unformatted selectable mailbox unformatted names array (reference)
 425  * @return array sub mailboxes unformatted names
 426  */
 427 function sqimap_asearch_get_sub_mailboxes($cur_mailbox, &$mboxes_array)
 428 {
 429     $sub_mboxes_array = array();
 430     $boxcount = count($mboxes_array);
 431     for ($boxnum=0; $boxnum < $boxcount; $boxnum++) {
 432         if (isBoxBelow($mboxes_array[$boxnum], $cur_mailbox))
 433             $sub_mboxes_array[] = $mboxes_array[$boxnum];
 434     }
 435     return $sub_mboxes_array;
 436 }
 437
 438 /**
 439  * Create the search query strings for all given criteria and merge results for every mailbox
 440  * @param resource $imapConnection
 441  * @param array $mailbox_array (reference)
 442  * @param array $biop_array (reference)
 443  * @param array $unop_array (reference)
 444  * @param array $where_array (reference)
 445  * @param array $what_array (reference)
 446  * @param array $exclude_array (reference)
 447  * @param array $sub_array (reference)
 448  * @param array $mboxes_array selectable unformatted mailboxes names (reference)
 449  * @return array array(mailbox => array(UIDs))
 450  */
 451 function sqimap_asearch($imapConnection, &$mailbox_array, &$biop_array, &$unop_array, &$where_array, &$what_array, &$exclude_array, &$sub_array, &$mboxes_array)
 452 {
 453
 454     $search_charset = sqimap_asearch_get_charset();
 455     $mbox_search = array();
 456     $search_string = '';
 457     $cur_mailbox = $mailbox_array[0];
 458     $cur_biop = '';    /* Start with ALL */
 459     /* We loop one more time than the real array count, so the last search gets fired */
 460     for ($cur_crit=0,$iCnt=count($where_array); $cur_crit <= $iCnt; ++$cur_crit) {
 461         if (empty($exclude_array[$cur_crit])) {
 462             $next_mailbox = (isset($mailbox_array[$cur_crit])) ? $mailbox_array[$cur_crit] : false;
 463             if ($next_mailbox != $cur_mailbox) {
 464                 $search_string = trim($search_string);    /* Trim out last space */
 465                 if ($cur_mailbox == 'All Folders')
 466                     $search_mboxes = $mboxes_array;
 467                 else if ((!empty($sub_array[$cur_crit - 1])) || (!in_array($cur_mailbox, $mboxes_array)))
 468                     $search_mboxes = sqimap_asearch_get_sub_mailboxes($cur_mailbox, $mboxes_array);
 469                 else
 470                     $search_mboxes = array($cur_mailbox);
 471                 foreach ($search_mboxes as $cur_mailbox) {
 472                     if (isset($mbox_search[$cur_mailbox])) {
 473                         $mbox_search[$cur_mailbox]['search'] .= ' ' . $search_string;
 474                     } else {
 475                         $mbox_search[$cur_mailbox]['search'] = $search_string;
 476                     }
 477                     $mbox_search[$cur_mailbox]['charset'] = $search_charset;
 478                 }
 479                 $cur_mailbox = $next_mailbox;
 480                 $search_string = '';
 481             }
 482             if (isset($where_array[$cur_crit]) && empty($exclude_array[$cur_crit])) {
 483                 for ($crit = $cur_crit; $crit < count($where_array); $crit++) {
 484                     $criteria = trim(sqimap_asearch_build_criteria($where_array[$crit], $what_array[$crit], $search_charset));
 485                     if (!empty($criteria) && empty($exclude_array[$crit])) {
 486                         if (asearch_nz($mailbox_array[$crit]) == $cur_mailbox) {
 487                             $unop = $unop_array[$crit];
 488                             if (!empty($unop)) {
 489                                 $criteria = $unop . ' ' . $criteria;
 490                             }
 491                             $aCriteria[] = array($biop_array[$crit], $criteria);
 492                         }
 493                     }
 494                     // unset something
 495                     $exclude_array[$crit] = true;
 496                 }
 497                 $aSearch = array();
 498                 for($i=0,$iCnt=count($aCriteria);$i<$iCnt;++$i) {
 499                     $cur_biop = $aCriteria[$i][0];
 500                     $next_biop = (isset($aCriteria[$i+1][0])) ? $aCriteria[$i+1][0] : false;
 501                     if ($next_biop != $cur_biop && $next_biop == 'OR') {
 502                         $aSearch[] = 'OR '.$aCriteria[$i][1];
 503                     } else if ($cur_biop != 'OR') {
 504                         $aSearch[] = 'ALL '.$aCriteria[$i][1];
 505                     } else { // OR only supports 2 search keys so we need to create a parenthesized list
 506                         $prev_biop = (isset($aCriteria[$i-1][0])) ? $aCriteria[$i-1][0] : false;
 507                         if ($prev_biop == $cur_biop) {
 508                             $last = $aSearch[$i-1];
 509                             if (!substr($last,-1) == ')') {
 510                                 $aSearch[$i-1] = "(OR $last";
 511                                 $aSearch[] = $aCriteria[$i][1].')';
 512                             } else {
 513                                 $sEnd = '';
 514                                 while ($last && substr($last,-1) == ')') {
 515                                     $last = substr($last,0,-1);
 516                                     $sEnd .= ')';
 517                                 }
 518                                 $aSearch[$i-1] = "(OR $last";
 519                                 $aSearch[] = $aCriteria[$i][1].$sEnd.')';
 520                             }
 521                         } else {
 522                             $aSearch[] = $aCriteria[$i][1];
 523                         }
 524                     }
 525                 }
 526                 $search_string .= implode(' ',$aSearch);
 527             }
 528         }
 529     }
 530     return ($mbox_search);
 531 }
 532
 533 ?>