web/drupal/modules/search/search.module
branchdrupal
changeset 74 0ff3ba646492
equal deleted inserted replaced
73:fcf75e232c5b 74:0ff3ba646492
       
     1 <?php
       
     2 // $Id: search.module,v 1.250.2.6 2009/03/10 17:20:01 goba Exp $
       
     3 
       
     4 /**
       
     5  * @file
       
     6  * Enables site-wide keyword searching.
       
     7  */
       
     8 
       
     9 /**
       
    10  * Matches Unicode character classes to exclude from the search index.
       
    11  *
       
    12  * See: http://www.unicode.org/Public/UNIDATA/UCD.html#General_Category_Values
       
    13  *
       
    14  * The index only contains the following character classes:
       
    15  * Lu     Letter, Uppercase
       
    16  * Ll     Letter, Lowercase
       
    17  * Lt     Letter, Titlecase
       
    18  * Lo     Letter, Other
       
    19  * Nd     Number, Decimal Digit
       
    20  * No     Number, Other
       
    21  */
       
    22 define('PREG_CLASS_SEARCH_EXCLUDE',
       
    23 '\x{0}-\x{2f}\x{3a}-\x{40}\x{5b}-\x{60}\x{7b}-\x{bf}\x{d7}\x{f7}\x{2b0}-'.
       
    24 '\x{385}\x{387}\x{3f6}\x{482}-\x{489}\x{559}-\x{55f}\x{589}-\x{5c7}\x{5f3}-'.
       
    25 '\x{61f}\x{640}\x{64b}-\x{65e}\x{66a}-\x{66d}\x{670}\x{6d4}\x{6d6}-\x{6ed}'.
       
    26 '\x{6fd}\x{6fe}\x{700}-\x{70f}\x{711}\x{730}-\x{74a}\x{7a6}-\x{7b0}\x{901}-'.
       
    27 '\x{903}\x{93c}\x{93e}-\x{94d}\x{951}-\x{954}\x{962}-\x{965}\x{970}\x{981}-'.
       
    28 '\x{983}\x{9bc}\x{9be}-\x{9cd}\x{9d7}\x{9e2}\x{9e3}\x{9f2}-\x{a03}\x{a3c}-'.
       
    29 '\x{a4d}\x{a70}\x{a71}\x{a81}-\x{a83}\x{abc}\x{abe}-\x{acd}\x{ae2}\x{ae3}'.
       
    30 '\x{af1}-\x{b03}\x{b3c}\x{b3e}-\x{b57}\x{b70}\x{b82}\x{bbe}-\x{bd7}\x{bf0}-'.
       
    31 '\x{c03}\x{c3e}-\x{c56}\x{c82}\x{c83}\x{cbc}\x{cbe}-\x{cd6}\x{d02}\x{d03}'.
       
    32 '\x{d3e}-\x{d57}\x{d82}\x{d83}\x{dca}-\x{df4}\x{e31}\x{e34}-\x{e3f}\x{e46}-'.
       
    33 '\x{e4f}\x{e5a}\x{e5b}\x{eb1}\x{eb4}-\x{ebc}\x{ec6}-\x{ecd}\x{f01}-\x{f1f}'.
       
    34 '\x{f2a}-\x{f3f}\x{f71}-\x{f87}\x{f90}-\x{fd1}\x{102c}-\x{1039}\x{104a}-'.
       
    35 '\x{104f}\x{1056}-\x{1059}\x{10fb}\x{10fc}\x{135f}-\x{137c}\x{1390}-\x{1399}'.
       
    36 '\x{166d}\x{166e}\x{1680}\x{169b}\x{169c}\x{16eb}-\x{16f0}\x{1712}-\x{1714}'.
       
    37 '\x{1732}-\x{1736}\x{1752}\x{1753}\x{1772}\x{1773}\x{17b4}-\x{17db}\x{17dd}'.
       
    38 '\x{17f0}-\x{180e}\x{1843}\x{18a9}\x{1920}-\x{1945}\x{19b0}-\x{19c0}\x{19c8}'.
       
    39 '\x{19c9}\x{19de}-\x{19ff}\x{1a17}-\x{1a1f}\x{1d2c}-\x{1d61}\x{1d78}\x{1d9b}-'.
       
    40 '\x{1dc3}\x{1fbd}\x{1fbf}-\x{1fc1}\x{1fcd}-\x{1fcf}\x{1fdd}-\x{1fdf}\x{1fed}-'.
       
    41 '\x{1fef}\x{1ffd}-\x{2070}\x{2074}-\x{207e}\x{2080}-\x{2101}\x{2103}-\x{2106}'.
       
    42 '\x{2108}\x{2109}\x{2114}\x{2116}-\x{2118}\x{211e}-\x{2123}\x{2125}\x{2127}'.
       
    43 '\x{2129}\x{212e}\x{2132}\x{213a}\x{213b}\x{2140}-\x{2144}\x{214a}-\x{2b13}'.
       
    44 '\x{2ce5}-\x{2cff}\x{2d6f}\x{2e00}-\x{3005}\x{3007}-\x{303b}\x{303d}-\x{303f}'.
       
    45 '\x{3099}-\x{309e}\x{30a0}\x{30fb}-\x{30fe}\x{3190}-\x{319f}\x{31c0}-\x{31cf}'.
       
    46 '\x{3200}-\x{33ff}\x{4dc0}-\x{4dff}\x{a015}\x{a490}-\x{a716}\x{a802}\x{a806}'.
       
    47 '\x{a80b}\x{a823}-\x{a82b}\x{d800}-\x{f8ff}\x{fb1e}\x{fb29}\x{fd3e}\x{fd3f}'.
       
    48 '\x{fdfc}-\x{fe6b}\x{feff}-\x{ff0f}\x{ff1a}-\x{ff20}\x{ff3b}-\x{ff40}\x{ff5b}-'.
       
    49 '\x{ff65}\x{ff70}\x{ff9e}\x{ff9f}\x{ffe0}-\x{fffd}');
       
    50 
       
    51 /**
       
    52  * Matches all 'N' Unicode character classes (numbers)
       
    53  */
       
    54 define('PREG_CLASS_NUMBERS',
       
    55 '\x{30}-\x{39}\x{b2}\x{b3}\x{b9}\x{bc}-\x{be}\x{660}-\x{669}\x{6f0}-\x{6f9}'.
       
    56 '\x{966}-\x{96f}\x{9e6}-\x{9ef}\x{9f4}-\x{9f9}\x{a66}-\x{a6f}\x{ae6}-\x{aef}'.
       
    57 '\x{b66}-\x{b6f}\x{be7}-\x{bf2}\x{c66}-\x{c6f}\x{ce6}-\x{cef}\x{d66}-\x{d6f}'.
       
    58 '\x{e50}-\x{e59}\x{ed0}-\x{ed9}\x{f20}-\x{f33}\x{1040}-\x{1049}\x{1369}-'.
       
    59 '\x{137c}\x{16ee}-\x{16f0}\x{17e0}-\x{17e9}\x{17f0}-\x{17f9}\x{1810}-\x{1819}'.
       
    60 '\x{1946}-\x{194f}\x{2070}\x{2074}-\x{2079}\x{2080}-\x{2089}\x{2153}-\x{2183}'.
       
    61 '\x{2460}-\x{249b}\x{24ea}-\x{24ff}\x{2776}-\x{2793}\x{3007}\x{3021}-\x{3029}'.
       
    62 '\x{3038}-\x{303a}\x{3192}-\x{3195}\x{3220}-\x{3229}\x{3251}-\x{325f}\x{3280}-'.
       
    63 '\x{3289}\x{32b1}-\x{32bf}\x{ff10}-\x{ff19}');
       
    64 
       
    65 /**
       
    66  * Matches all 'P' Unicode character classes (punctuation)
       
    67  */
       
    68 define('PREG_CLASS_PUNCTUATION',
       
    69 '\x{21}-\x{23}\x{25}-\x{2a}\x{2c}-\x{2f}\x{3a}\x{3b}\x{3f}\x{40}\x{5b}-\x{5d}'.
       
    70 '\x{5f}\x{7b}\x{7d}\x{a1}\x{ab}\x{b7}\x{bb}\x{bf}\x{37e}\x{387}\x{55a}-\x{55f}'.
       
    71 '\x{589}\x{58a}\x{5be}\x{5c0}\x{5c3}\x{5f3}\x{5f4}\x{60c}\x{60d}\x{61b}\x{61f}'.
       
    72 '\x{66a}-\x{66d}\x{6d4}\x{700}-\x{70d}\x{964}\x{965}\x{970}\x{df4}\x{e4f}'.
       
    73 '\x{e5a}\x{e5b}\x{f04}-\x{f12}\x{f3a}-\x{f3d}\x{f85}\x{104a}-\x{104f}\x{10fb}'.
       
    74 '\x{1361}-\x{1368}\x{166d}\x{166e}\x{169b}\x{169c}\x{16eb}-\x{16ed}\x{1735}'.
       
    75 '\x{1736}\x{17d4}-\x{17d6}\x{17d8}-\x{17da}\x{1800}-\x{180a}\x{1944}\x{1945}'.
       
    76 '\x{2010}-\x{2027}\x{2030}-\x{2043}\x{2045}-\x{2051}\x{2053}\x{2054}\x{2057}'.
       
    77 '\x{207d}\x{207e}\x{208d}\x{208e}\x{2329}\x{232a}\x{23b4}-\x{23b6}\x{2768}-'.
       
    78 '\x{2775}\x{27e6}-\x{27eb}\x{2983}-\x{2998}\x{29d8}-\x{29db}\x{29fc}\x{29fd}'.
       
    79 '\x{3001}-\x{3003}\x{3008}-\x{3011}\x{3014}-\x{301f}\x{3030}\x{303d}\x{30a0}'.
       
    80 '\x{30fb}\x{fd3e}\x{fd3f}\x{fe30}-\x{fe52}\x{fe54}-\x{fe61}\x{fe63}\x{fe68}'.
       
    81 '\x{fe6a}\x{fe6b}\x{ff01}-\x{ff03}\x{ff05}-\x{ff0a}\x{ff0c}-\x{ff0f}\x{ff1a}'.
       
    82 '\x{ff1b}\x{ff1f}\x{ff20}\x{ff3b}-\x{ff3d}\x{ff3f}\x{ff5b}\x{ff5d}\x{ff5f}-'.
       
    83 '\x{ff65}');
       
    84 
       
    85 /**
       
    86  * Matches all CJK characters that are candidates for auto-splitting
       
    87  * (Chinese, Japanese, Korean).
       
    88  * Contains kana and BMP ideographs.
       
    89  */
       
    90 define('PREG_CLASS_CJK', '\x{3041}-\x{30ff}\x{31f0}-\x{31ff}\x{3400}-\x{4db5}'.
       
    91 '\x{4e00}-\x{9fbb}\x{f900}-\x{fad9}');
       
    92 
       
    93 /**
       
    94  * Implementation of hook_help().
       
    95  */
       
    96 function search_help($path, $arg) {
       
    97   switch ($path) {
       
    98     case 'admin/help#search':
       
    99       $output = '<p>'. t('The search module adds the ability to search for content by keywords. Search is often the only practical way to find content on a large site, and is useful for finding both users and posts.') .'</p>';
       
   100       $output .= '<p>'. t('To provide keyword searching, the search engine maintains an index of words found in your site\'s content. To build and maintain this index, a correctly configured <a href="@cron">cron maintenance task</a> is required. Indexing behavior can be adjusted using the <a href="@searchsettings">search settings page</a>; for example, the <em>Number of items to index per cron run</em> sets the maximum number of items indexed in each pass of a <a href="@cron">cron maintenance task</a>. If necessary, reduce this number to prevent timeouts and memory errors when indexing.', array('@cron' => url('admin/reports/status'), '@searchsettings' => url('admin/settings/search'))) .'</p>';
       
   101       $output .= '<p>'. t('For more information, see the online handbook entry for <a href="@search">Search module</a>.', array('@search' => 'http://drupal.org/handbook/modules/search/')) .'</p>';
       
   102       return $output;
       
   103     case 'admin/settings/search':
       
   104       return '<p>'. t('The search engine maintains an index of words found in your site\'s content. To build and maintain this index, a correctly configured <a href="@cron">cron maintenance task</a> is required. Indexing behavior can be adjusted using the settings below.', array('@cron' => url('admin/reports/status'))) .'</p>';
       
   105     case 'search#noresults':
       
   106       return t('<ul>
       
   107 <li>Check if your spelling is correct.</li>
       
   108 <li>Remove quotes around phrases to match each word individually: <em>"blue smurf"</em> will match less than <em>blue smurf</em>.</li>
       
   109 <li>Consider loosening your query with <em>OR</em>: <em>blue smurf</em> will match less than <em>blue OR smurf</em>.</li>
       
   110 </ul>');
       
   111   }
       
   112 }
       
   113 
       
   114 /**
       
   115  * Implementation of hook_theme()
       
   116  */
       
   117 function search_theme() {
       
   118   return array(
       
   119     'search_theme_form' => array(
       
   120       'arguments' => array('form' => NULL),
       
   121       'template' => 'search-theme-form',
       
   122     ),
       
   123     'search_block_form' => array(
       
   124       'arguments' => array('form' => NULL),
       
   125       'template' => 'search-block-form',
       
   126     ),
       
   127     'search_result' => array(
       
   128       'arguments' => array('result' => NULL, 'type' => NULL),
       
   129       'file' => 'search.pages.inc',
       
   130       'template' => 'search-result',
       
   131     ),
       
   132     'search_results' => array(
       
   133       'arguments' => array('results' => NULL, 'type' => NULL),
       
   134       'file' => 'search.pages.inc',
       
   135       'template' => 'search-results',
       
   136     ),
       
   137   );
       
   138 }
       
   139 
       
   140 /**
       
   141  * Implementation of hook_perm().
       
   142  */
       
   143 function search_perm() {
       
   144   return array('search content', 'use advanced search', 'administer search');
       
   145 }
       
   146 
       
   147 /**
       
   148  * Implementation of hook_block().
       
   149  */
       
   150 function search_block($op = 'list', $delta = 0) {
       
   151   if ($op == 'list') {
       
   152     $blocks[0]['info'] = t('Search form');
       
   153     // Not worth caching.
       
   154     $blocks[0]['cache'] = BLOCK_NO_CACHE;
       
   155     return $blocks;
       
   156   }
       
   157   else if ($op == 'view' && user_access('search content')) {
       
   158     $block['content'] = drupal_get_form('search_block_form');
       
   159     $block['subject'] = t('Search');
       
   160     return $block;
       
   161   }
       
   162 }
       
   163 
       
   164 /**
       
   165  * Implementation of hook_menu().
       
   166  */
       
   167 function search_menu() {
       
   168   $items['search'] = array(
       
   169     'title' => 'Search',
       
   170     'page callback' => 'search_view',
       
   171     'access arguments' => array('search content'),
       
   172     'type' => MENU_SUGGESTED_ITEM,
       
   173     'file' => 'search.pages.inc',
       
   174   );
       
   175   $items['admin/settings/search'] = array(
       
   176     'title' => 'Search settings',
       
   177     'description' => 'Configure relevance settings for search and other indexing options',
       
   178     'page callback' => 'drupal_get_form',
       
   179     'page arguments' => array('search_admin_settings'),
       
   180     'access arguments' => array('administer search'),
       
   181     'type' => MENU_NORMAL_ITEM,
       
   182     'file' => 'search.admin.inc',
       
   183   );
       
   184   $items['admin/settings/search/wipe'] = array(
       
   185     'title' => 'Clear index',
       
   186     'page callback' => 'drupal_get_form',
       
   187     'page arguments' => array('search_wipe_confirm'),
       
   188     'access arguments' => array('administer search'),
       
   189     'type' => MENU_CALLBACK,
       
   190     'file' => 'search.admin.inc',
       
   191   );
       
   192   $items['admin/reports/search'] = array(
       
   193     'title' => 'Top search phrases',
       
   194     'description' => 'View most popular search phrases.',
       
   195     'page callback' => 'dblog_top',
       
   196     'page arguments' => array('search'),
       
   197     'access arguments' => array('access site reports'),
       
   198     'file' => 'dblog.admin.inc',
       
   199     'file path' => drupal_get_path('module', 'dblog'),
       
   200   );
       
   201 
       
   202   foreach (module_implements('search') as $name) {
       
   203     $items['search/'. $name .'/%menu_tail'] = array(
       
   204       'title callback' => 'module_invoke',
       
   205       'title arguments' => array($name, 'search', 'name', TRUE),
       
   206       'page callback' => 'search_view',
       
   207       'page arguments' => array($name),
       
   208       'access callback' => '_search_menu',
       
   209       'access arguments' => array($name),
       
   210       'type' => MENU_LOCAL_TASK,
       
   211       'parent' => 'search',
       
   212       'file' => 'search.pages.inc',
       
   213     );
       
   214   }
       
   215   return $items;
       
   216 }
       
   217 
       
   218 function _search_menu($name) {
       
   219   return user_access('search content') && module_invoke($name, 'search', 'name');
       
   220 }
       
   221 
       
   222 /**
       
   223  * Wipes a part of or the entire search index.
       
   224  *
       
   225  * @param $sid
       
   226  *  (optional) The SID of the item to wipe. If specified, $type must be passed
       
   227  *  too.
       
   228  * @param $type
       
   229  *  (optional) The type of item to wipe.
       
   230  */
       
   231 function search_wipe($sid = NULL, $type = NULL, $reindex = FALSE) {
       
   232   if ($type == NULL && $sid == NULL) {
       
   233     module_invoke_all('search', 'reset');
       
   234   }
       
   235   else {
       
   236     db_query("DELETE FROM {search_dataset} WHERE sid = %d AND type = '%s'", $sid, $type);
       
   237     db_query("DELETE FROM {search_index} WHERE sid = %d AND type = '%s'", $sid, $type);
       
   238     // Don't remove links if re-indexing.
       
   239     if (!$reindex) {
       
   240       db_query("DELETE FROM {search_node_links} WHERE sid = %d AND type = '%s'", $sid, $type);
       
   241     }
       
   242   }
       
   243 }
       
   244 
       
   245 /**
       
   246  * Marks a word as dirty (or retrieves the list of dirty words). This is used
       
   247  * during indexing (cron). Words which are dirty have outdated total counts in
       
   248  * the search_total table, and need to be recounted.
       
   249  */
       
   250 function search_dirty($word = NULL) {
       
   251   static $dirty = array();
       
   252   if ($word !== NULL) {
       
   253     $dirty[$word] = TRUE;
       
   254   }
       
   255   else {
       
   256     return $dirty;
       
   257   }
       
   258 }
       
   259 
       
   260 /**
       
   261  * Implementation of hook_cron().
       
   262  *
       
   263  * Fires hook_update_index() in all modules and cleans up dirty words (see
       
   264  * search_dirty).
       
   265  */
       
   266 function search_cron() {
       
   267   // We register a shutdown function to ensure that search_total is always up
       
   268   // to date.
       
   269   register_shutdown_function('search_update_totals');
       
   270 
       
   271   // Update word index
       
   272   foreach (module_list() as $module) {
       
   273     module_invoke($module, 'update_index');
       
   274   }
       
   275 }
       
   276 
       
   277 /**
       
   278  * This function is called on shutdown to ensure that search_total is always
       
   279  * up to date (even if cron times out or otherwise fails).
       
   280  */
       
   281 function search_update_totals() {
       
   282   // Update word IDF (Inverse Document Frequency) counts for new/changed words
       
   283   foreach (search_dirty() as $word => $dummy) {
       
   284     // Get total count
       
   285     $total = db_result(db_query("SELECT SUM(score) FROM {search_index} WHERE word = '%s'", $word));
       
   286     // Apply Zipf's law to equalize the probability distribution
       
   287     $total = log10(1 + 1/(max(1, $total)));
       
   288     db_query("UPDATE {search_total} SET count = %f WHERE word = '%s'", $total, $word);
       
   289     if (!db_affected_rows()) {
       
   290       db_query("INSERT INTO {search_total} (word, count) VALUES ('%s', %f)", $word, $total);
       
   291     }
       
   292   }
       
   293   // Find words that were deleted from search_index, but are still in
       
   294   // search_total. We use a LEFT JOIN between the two tables and keep only the
       
   295   // rows which fail to join.
       
   296   $result = db_query("SELECT t.word AS realword, i.word FROM {search_total} t LEFT JOIN {search_index} i ON t.word = i.word WHERE i.word IS NULL");
       
   297   while ($word = db_fetch_object($result)) {
       
   298     db_query("DELETE FROM {search_total} WHERE word = '%s'", $word->realword);
       
   299   }
       
   300 }
       
   301 
       
   302 /**
       
   303  * Simplifies a string according to indexing rules.
       
   304  */
       
   305 function search_simplify($text) {
       
   306   // Decode entities to UTF-8
       
   307   $text = decode_entities($text);
       
   308 
       
   309   // Lowercase
       
   310   $text = drupal_strtolower($text);
       
   311 
       
   312   // Call an external processor for word handling.
       
   313   search_invoke_preprocess($text);
       
   314 
       
   315   // Simple CJK handling
       
   316   if (variable_get('overlap_cjk', TRUE)) {
       
   317     $text = preg_replace_callback('/['. PREG_CLASS_CJK .']+/u', 'search_expand_cjk', $text);
       
   318   }
       
   319 
       
   320   // To improve searching for numerical data such as dates, IP addresses
       
   321   // or version numbers, we consider a group of numerical characters
       
   322   // separated only by punctuation characters to be one piece.
       
   323   // This also means that searching for e.g. '20/03/1984' also returns
       
   324   // results with '20-03-1984' in them.
       
   325   // Readable regexp: ([number]+)[punctuation]+(?=[number])
       
   326   $text = preg_replace('/(['. PREG_CLASS_NUMBERS .']+)['. PREG_CLASS_PUNCTUATION .']+(?=['. PREG_CLASS_NUMBERS .'])/u', '\1', $text);
       
   327 
       
   328   // The dot, underscore and dash are simply removed. This allows meaningful
       
   329   // search behavior with acronyms and URLs.
       
   330   $text = preg_replace('/[._-]+/', '', $text);
       
   331 
       
   332   // With the exception of the rules above, we consider all punctuation,
       
   333   // marks, spacers, etc, to be a word boundary.
       
   334   $text = preg_replace('/['. PREG_CLASS_SEARCH_EXCLUDE .']+/u', ' ', $text);
       
   335 
       
   336   return $text;
       
   337 }
       
   338 
       
   339 /**
       
   340  * Basic CJK tokenizer. Simply splits a string into consecutive, overlapping
       
   341  * sequences of characters ('minimum_word_size' long).
       
   342  */
       
   343 function search_expand_cjk($matches) {
       
   344   $min = variable_get('minimum_word_size', 3);
       
   345   $str = $matches[0];
       
   346   $l = drupal_strlen($str);
       
   347   // Passthrough short words
       
   348   if ($l <= $min) {
       
   349     return ' '. $str .' ';
       
   350   }
       
   351   $tokens = ' ';
       
   352   // FIFO queue of characters
       
   353   $chars = array();
       
   354   // Begin loop
       
   355   for ($i = 0; $i < $l; ++$i) {
       
   356     // Grab next character
       
   357     $current = drupal_substr($str, 0, 1);
       
   358     $str = substr($str, strlen($current));
       
   359     $chars[] = $current;
       
   360     if ($i >= $min - 1) {
       
   361       $tokens .= implode('', $chars) .' ';
       
   362       array_shift($chars);
       
   363     }
       
   364   }
       
   365   return $tokens;
       
   366 }
       
   367 
       
   368 /**
       
   369  * Splits a string into tokens for indexing.
       
   370  */
       
   371 function search_index_split($text) {
       
   372   static $last = NULL;
       
   373   static $lastsplit = NULL;
       
   374 
       
   375   if ($last == $text) {
       
   376     return $lastsplit;
       
   377   }
       
   378   // Process words
       
   379   $text = search_simplify($text);
       
   380   $words = explode(' ', $text);
       
   381   array_walk($words, '_search_index_truncate');
       
   382 
       
   383   // Save last keyword result
       
   384   $last = $text;
       
   385   $lastsplit = $words;
       
   386 
       
   387   return $words;
       
   388 }
       
   389 
       
   390 /**
       
   391  * Helper function for array_walk in search_index_split.
       
   392  */
       
   393 function _search_index_truncate(&$text) {
       
   394   $text = truncate_utf8($text, 50);
       
   395 }
       
   396 
       
   397 /**
       
   398  * Invokes hook_search_preprocess() in modules.
       
   399  */
       
   400 function search_invoke_preprocess(&$text) {
       
   401   foreach (module_implements('search_preprocess') as $module) {
       
   402     $text = module_invoke($module, 'search_preprocess', $text);
       
   403   }
       
   404 }
       
   405 
       
   406 /**
       
   407  * Update the full-text search index for a particular item.
       
   408  *
       
   409  * @param $sid
       
   410  *   A number identifying this particular item (e.g. node id).
       
   411  *
       
   412  * @param $type
       
   413  *   A string defining this type of item (e.g. 'node')
       
   414  *
       
   415  * @param $text
       
   416  *   The content of this item. Must be a piece of HTML text.
       
   417  *
       
   418  * @ingroup search
       
   419  */
       
   420 function search_index($sid, $type, $text) {
       
   421   $minimum_word_size = variable_get('minimum_word_size', 3);
       
   422 
       
   423   // Link matching
       
   424   global $base_url;
       
   425   $node_regexp = '@href=[\'"]?(?:'. preg_quote($base_url, '@') .'/|'. preg_quote(base_path(), '@') .')(?:\?q=)?/?((?![a-z]+:)[^\'">]+)[\'">]@i';
       
   426 
       
   427   // Multipliers for scores of words inside certain HTML tags.
       
   428   // Note: 'a' must be included for link ranking to work.
       
   429   $tags = array('h1' => 25,
       
   430                 'h2' => 18,
       
   431                 'h3' => 15,
       
   432                 'h4' => 12,
       
   433                 'h5' => 9,
       
   434                 'h6' => 6,
       
   435                 'u' => 3,
       
   436                 'b' => 3,
       
   437                 'i' => 3,
       
   438                 'strong' => 3,
       
   439                 'em' => 3,
       
   440                 'a' => 10);
       
   441 
       
   442   // Strip off all ignored tags to speed up processing, but insert space before/after
       
   443   // them to keep word boundaries.
       
   444   $text = str_replace(array('<', '>'), array(' <', '> '), $text);
       
   445   $text = strip_tags($text, '<'. implode('><', array_keys($tags)) .'>');
       
   446 
       
   447   // Split HTML tags from plain text.
       
   448   $split = preg_split('/\s*<([^>]+?)>\s*/', $text, -1, PREG_SPLIT_DELIM_CAPTURE);
       
   449   // Note: PHP ensures the array consists of alternating delimiters and literals
       
   450   // and begins and ends with a literal (inserting $null as required).
       
   451 
       
   452   $tag = FALSE; // Odd/even counter. Tag or no tag.
       
   453   $link = FALSE; // State variable for link analyser
       
   454   $score = 1; // Starting score per word
       
   455   $accum = ' '; // Accumulator for cleaned up data
       
   456   $tagstack = array(); // Stack with open tags
       
   457   $tagwords = 0; // Counter for consecutive words
       
   458   $focus = 1; // Focus state
       
   459 
       
   460   $results = array(0 => array()); // Accumulator for words for index
       
   461 
       
   462   foreach ($split as $value) {
       
   463     if ($tag) {
       
   464       // Increase or decrease score per word based on tag
       
   465       list($tagname) = explode(' ', $value, 2);
       
   466       $tagname = drupal_strtolower($tagname);
       
   467       // Closing or opening tag?
       
   468       if ($tagname[0] == '/') {
       
   469         $tagname = substr($tagname, 1);
       
   470         // If we encounter unexpected tags, reset score to avoid incorrect boosting.
       
   471         if (!count($tagstack) || $tagstack[0] != $tagname) {
       
   472           $tagstack = array();
       
   473           $score = 1;
       
   474         }
       
   475         else {
       
   476           // Remove from tag stack and decrement score
       
   477           $score = max(1, $score - $tags[array_shift($tagstack)]);
       
   478         }
       
   479         if ($tagname == 'a') {
       
   480           $link = FALSE;
       
   481         }
       
   482       }
       
   483       else {
       
   484         if (isset($tagstack[0]) && $tagstack[0] == $tagname) {
       
   485           // None of the tags we look for make sense when nested identically.
       
   486           // If they are, it's probably broken HTML.
       
   487           $tagstack = array();
       
   488           $score = 1;
       
   489         }
       
   490         else {
       
   491           // Add to open tag stack and increment score
       
   492           array_unshift($tagstack, $tagname);
       
   493           $score += $tags[$tagname];
       
   494         }
       
   495         if ($tagname == 'a') {
       
   496           // Check if link points to a node on this site
       
   497           if (preg_match($node_regexp, $value, $match)) {
       
   498             $path = drupal_get_normal_path($match[1]);
       
   499             if (preg_match('!(?:node|book)/(?:view/)?([0-9]+)!i', $path, $match)) {
       
   500               $linknid = $match[1];
       
   501               if ($linknid > 0) {
       
   502                 // Note: ignore links to uncachable nodes to avoid redirect bugs.
       
   503                 $node = db_fetch_object(db_query('SELECT n.title, n.nid, n.vid, r.format FROM {node} n INNER JOIN {node_revisions} r ON n.vid = r.vid WHERE n.nid = %d', $linknid));
       
   504                 if (filter_format_allowcache($node->format)) {
       
   505                   $link = TRUE;
       
   506                   $linktitle = $node->title;
       
   507                 }
       
   508               }
       
   509             }
       
   510           }
       
   511         }
       
   512       }
       
   513       // A tag change occurred, reset counter.
       
   514       $tagwords = 0;
       
   515     }
       
   516     else {
       
   517       // Note: use of PREG_SPLIT_DELIM_CAPTURE above will introduce empty values
       
   518       if ($value != '') {
       
   519         if ($link) {
       
   520           // Check to see if the node link text is its URL. If so, we use the target node title instead.
       
   521           if (preg_match('!^https?://!i', $value)) {
       
   522             $value = $linktitle;
       
   523           }
       
   524         }
       
   525         $words = search_index_split($value);
       
   526         foreach ($words as $word) {
       
   527           // Add word to accumulator
       
   528           $accum .= $word .' ';
       
   529           $num = is_numeric($word);
       
   530           // Check wordlength
       
   531           if ($num || drupal_strlen($word) >= $minimum_word_size) {
       
   532             // Normalize numbers
       
   533             if ($num) {
       
   534               $word = (int)ltrim($word, '-0');
       
   535             }
       
   536 
       
   537             // Links score mainly for the target.
       
   538             if ($link) {
       
   539               if (!isset($results[$linknid])) {
       
   540                 $results[$linknid] = array();
       
   541               }
       
   542               $results[$linknid][] = $word;
       
   543               // Reduce score of the link caption in the source.
       
   544               $focus *= 0.2;
       
   545             }
       
   546             // Fall-through
       
   547             if (!isset($results[0][$word])) {
       
   548               $results[0][$word] = 0;
       
   549             }
       
   550             $results[0][$word] += $score * $focus;
       
   551 
       
   552             // Focus is a decaying value in terms of the amount of unique words up to this point.
       
   553             // From 100 words and more, it decays, to e.g. 0.5 at 500 words and 0.3 at 1000 words.
       
   554             $focus = min(1, .01 + 3.5 / (2 + count($results[0]) * .015));
       
   555           }
       
   556           $tagwords++;
       
   557           // Too many words inside a single tag probably mean a tag was accidentally left open.
       
   558           if (count($tagstack) && $tagwords >= 15) {
       
   559             $tagstack = array();
       
   560             $score = 1;
       
   561           }
       
   562         }
       
   563       }
       
   564     }
       
   565     $tag = !$tag;
       
   566   }
       
   567 
       
   568   search_wipe($sid, $type, TRUE);
       
   569 
       
   570   // Insert cleaned up data into dataset
       
   571   db_query("INSERT INTO {search_dataset} (sid, type, data, reindex) VALUES (%d, '%s', '%s', %d)", $sid, $type, $accum, 0);
       
   572 
       
   573   // Insert results into search index
       
   574   foreach ($results[0] as $word => $score) {
       
   575     // Try inserting first because this will succeed most times, but because
       
   576     // the database collates similar words (accented and non-accented), the
       
   577     // insert can fail, in which case we need to add the word scores together.
       
   578     @db_query("INSERT INTO {search_index} (word, sid, type, score) VALUES ('%s', %d, '%s', %f)", $word, $sid, $type, $score);
       
   579     if (!db_affected_rows()) {
       
   580       db_query("UPDATE {search_index} SET score = score + %f WHERE word = '%s' AND sid = %d AND type = '%s'", $score, $word, $sid, $type);
       
   581     }
       
   582     search_dirty($word);
       
   583   }
       
   584   unset($results[0]);
       
   585 
       
   586   // Get all previous links from this item.
       
   587   $result = db_query("SELECT nid, caption FROM {search_node_links} WHERE sid = %d AND type = '%s'", $sid, $type);
       
   588   $links = array();
       
   589   while ($link = db_fetch_object($result)) {
       
   590     $links[$link->nid] = $link->caption;
       
   591   }
       
   592 
       
   593   // Now store links to nodes.
       
   594   foreach ($results as $nid => $words) {
       
   595     $caption = implode(' ', $words);
       
   596     if (isset($links[$nid])) {
       
   597       if ($links[$nid] != $caption) {
       
   598         // Update the existing link and mark the node for reindexing.
       
   599         db_query("UPDATE {search_node_links} SET caption = '%s' WHERE sid = %d AND type = '%s' AND nid = %d", $caption, $sid, $type, $nid);
       
   600         search_touch_node($nid);
       
   601       }
       
   602       // Unset the link to mark it as processed.
       
   603       unset($links[$nid]);
       
   604     }
       
   605     else {
       
   606       // Insert the existing link and mark the node for reindexing.
       
   607       db_query("INSERT INTO {search_node_links} (caption, sid, type, nid) VALUES ('%s', %d, '%s', %d)", $caption, $sid, $type, $nid);
       
   608       search_touch_node($nid);
       
   609     }
       
   610   }
       
   611   // Any left-over links in $links no longer exist. Delete them and mark the nodes for reindexing.
       
   612   foreach ($links as $nid => $caption) {
       
   613     db_query("DELETE FROM {search_node_links} WHERE sid = %d AND type = '%s' AND nid = %d", $sid, $type, $nid);
       
   614     search_touch_node($nid);
       
   615   }
       
   616 }
       
   617 
       
   618 /**
       
   619  * Change a node's changed timestamp to 'now' to force reindexing.
       
   620  *
       
   621  * @param $nid
       
   622  *   The nid of the node that needs reindexing.
       
   623  */
       
   624 function search_touch_node($nid) {
       
   625   db_query("UPDATE {search_dataset} SET reindex = %d WHERE sid = %d AND type = 'node'", time(), $nid);
       
   626 }
       
   627 
       
   628 /**
       
   629  * Implementation of hook_nodeapi().
       
   630  */
       
   631 function search_nodeapi(&$node, $op, $teaser = NULL, $page = NULL) {
       
   632   switch ($op) {
       
   633     // Transplant links to a node into the target node.
       
   634     case 'update index':
       
   635       $result = db_query("SELECT caption FROM {search_node_links} WHERE nid = %d", $node->nid);
       
   636       $output = array();
       
   637       while ($link = db_fetch_object($result)) {
       
   638         $output[] = $link->caption;
       
   639       }
       
   640       if (count($output)) {
       
   641         return '<a>('. implode(', ', $output) .')</a>';
       
   642       }
       
   643       break;
       
   644     // Reindex the node when it is updated.  The node is automatically indexed
       
   645     // when it is added, simply by being added to the node table.
       
   646     case 'update':
       
   647       search_touch_node($node->nid);
       
   648       break;
       
   649   }
       
   650 }
       
   651 
       
   652 /**
       
   653  * Implementation of hook_comment().
       
   654  */
       
   655 function search_comment($a1, $op) {
       
   656   switch ($op) {
       
   657     // Reindex the node when comments are added or changed
       
   658     case 'insert':
       
   659     case 'update':
       
   660     case 'delete':
       
   661     case 'publish':
       
   662     case 'unpublish':
       
   663       search_touch_node(is_array($a1) ? $a1['nid'] : $a1->nid);
       
   664       break;
       
   665   }
       
   666 }
       
   667 
       
   668 /**
       
   669  * Extract a module-specific search option from a search query. e.g. 'type:book'
       
   670  */
       
   671 function search_query_extract($keys, $option) {
       
   672   if (preg_match('/(^| )'. $option .':([^ ]*)( |$)/i', $keys, $matches)) {
       
   673     return $matches[2];
       
   674   }
       
   675 }
       
   676 
       
   677 /**
       
   678  * Return a query with the given module-specific search option inserted in.
       
   679  * e.g. 'type:book'.
       
   680  */
       
   681 function search_query_insert($keys, $option, $value = '') {
       
   682   if (search_query_extract($keys, $option)) {
       
   683     $keys = trim(preg_replace('/(^| )'. $option .':[^ ]*/i', '', $keys));
       
   684   }
       
   685   if ($value != '') {
       
   686     $keys .= ' '. $option .':'. $value;
       
   687   }
       
   688   return $keys;
       
   689 }
       
   690 
       
   691 /**
       
   692  * Parse a search query into SQL conditions.
       
   693  *
       
   694  * We build two queries that matches the dataset bodies. @See do_search for
       
   695  * more about these.
       
   696  *
       
   697  * @param $text
       
   698  *   The search keys.
       
   699  * @return
       
   700  *   A list of six elements.
       
   701  *    * A series of statements AND'd together which will be used to provide all
       
   702  *      possible matches.
       
   703  *    * Arguments for this query part.
       
   704  *    * A series of exact word matches OR'd together.
       
   705  *    * Arguments for this query part.
       
   706  *    * A bool indicating whether this is a simple query or not. Negative
       
   707  *      terms, presence of both AND / OR make this FALSE.
       
   708  *    * A bool indicating the presence of a lowercase or. Maybe the user
       
   709  *      wanted to use OR.
       
   710  */
       
   711 function search_parse_query($text) {
       
   712   $keys = array('positive' => array(), 'negative' => array());
       
   713 
       
   714   // Tokenize query string
       
   715   preg_match_all('/ (-?)("[^"]+"|[^" ]+)/i', ' '. $text, $matches, PREG_SET_ORDER);
       
   716 
       
   717   if (count($matches) < 1) {
       
   718     return NULL;
       
   719   }
       
   720 
       
   721   // Classify tokens
       
   722   $or = FALSE;
       
   723   $warning = '';
       
   724   $simple = TRUE;
       
   725   foreach ($matches as $match) {
       
   726     $phrase = FALSE;
       
   727     // Strip off phrase quotes
       
   728     if ($match[2]{0} == '"') {
       
   729       $match[2] = substr($match[2], 1, -1);
       
   730       $phrase = TRUE;
       
   731       $simple = FALSE;
       
   732     }
       
   733     // Simplify keyword according to indexing rules and external preprocessors
       
   734     $words = search_simplify($match[2]);
       
   735     // Re-explode in case simplification added more words, except when matching a phrase
       
   736     $words = $phrase ? array($words) : preg_split('/ /', $words, -1, PREG_SPLIT_NO_EMPTY);
       
   737     // Negative matches
       
   738     if ($match[1] == '-') {
       
   739       $keys['negative'] = array_merge($keys['negative'], $words);
       
   740     }
       
   741     // OR operator: instead of a single keyword, we store an array of all
       
   742     // OR'd keywords.
       
   743     elseif ($match[2] == 'OR' && count($keys['positive'])) {
       
   744       $last = array_pop($keys['positive']);
       
   745       // Starting a new OR?
       
   746       if (!is_array($last)) {
       
   747         $last = array($last);
       
   748       }
       
   749       $keys['positive'][] = $last;
       
   750       $or = TRUE;
       
   751       continue;
       
   752     }
       
   753     // AND operator: implied, so just ignore it
       
   754     elseif ($match[2] == 'AND' || $match[2] == 'and') {
       
   755       $warning = $match[2];
       
   756       continue;
       
   757     }
       
   758 
       
   759     // Plain keyword
       
   760     else {
       
   761       if ($match[2] == 'or') {
       
   762         $warning = $match[2];
       
   763       }
       
   764       if ($or) {
       
   765         // Add to last element (which is an array)
       
   766         $keys['positive'][count($keys['positive']) - 1] = array_merge($keys['positive'][count($keys['positive']) - 1], $words);
       
   767       }
       
   768       else {
       
   769         $keys['positive'] = array_merge($keys['positive'], $words);
       
   770       }
       
   771     }
       
   772     $or = FALSE;
       
   773   }
       
   774 
       
   775   // Convert keywords into SQL statements.
       
   776   $query = array();
       
   777   $query2 = array();
       
   778   $arguments = array();
       
   779   $arguments2 = array();
       
   780   $matches = 0;
       
   781   $simple_and = FALSE;
       
   782   $simple_or = FALSE;
       
   783   // Positive matches
       
   784   foreach ($keys['positive'] as $key) {
       
   785     // Group of ORed terms
       
   786     if (is_array($key) && count($key)) {
       
   787       $simple_or = TRUE;
       
   788       $queryor = array();
       
   789       $any = FALSE;
       
   790       foreach ($key as $or) {
       
   791         list($q, $num_new_scores) = _search_parse_query($or, $arguments2);
       
   792         $any |= $num_new_scores;
       
   793         if ($q) {
       
   794           $queryor[] = $q;
       
   795           $arguments[] = $or;
       
   796         }
       
   797       }
       
   798       if (count($queryor)) {
       
   799         $query[] = '('. implode(' OR ', $queryor) .')';
       
   800         // A group of OR keywords only needs to match once
       
   801         $matches += ($any > 0);
       
   802       }
       
   803     }
       
   804     // Single ANDed term
       
   805     else {
       
   806       $simple_and = TRUE;
       
   807       list($q, $num_new_scores, $num_valid_words) = _search_parse_query($key, $arguments2);
       
   808       if ($q) {
       
   809         $query[] = $q;
       
   810         $arguments[] = $key;
       
   811         if (!$num_valid_words) {
       
   812           $simple = FALSE;
       
   813         }
       
   814         // Each AND keyword needs to match at least once
       
   815         $matches += $num_new_scores;
       
   816       }
       
   817     }
       
   818   }
       
   819   if ($simple_and && $simple_or) {
       
   820     $simple = FALSE;
       
   821   }
       
   822   // Negative matches
       
   823   foreach ($keys['negative'] as $key) {
       
   824     list($q) = _search_parse_query($key, $arguments2, TRUE);
       
   825     if ($q) {
       
   826       $query[] = $q;
       
   827       $arguments[] = $key;
       
   828       $simple = FALSE;
       
   829     }
       
   830   }
       
   831   $query = implode(' AND ', $query);
       
   832 
       
   833   // Build word-index conditions for the first pass
       
   834   $query2 = substr(str_repeat("i.word = '%s' OR ", count($arguments2)), 0, -4);
       
   835 
       
   836   return array($query, $arguments, $query2, $arguments2, $matches, $simple, $warning);
       
   837 }
       
   838 
       
   839 /**
       
   840  * Helper function for search_parse_query();
       
   841  */
       
   842 function _search_parse_query(&$word, &$scores, $not = FALSE) {
       
   843   $num_new_scores = 0;
       
   844   $num_valid_words = 0;
       
   845   // Determine the scorewords of this word/phrase
       
   846   if (!$not) {
       
   847     $split = explode(' ', $word);
       
   848     foreach ($split as $s) {
       
   849       $num = is_numeric($s);
       
   850       if ($num || drupal_strlen($s) >= variable_get('minimum_word_size', 3)) {
       
   851         $s = $num ? ((int)ltrim($s, '-0')) : $s;
       
   852         if (!isset($scores[$s])) {
       
   853           $scores[$s] = $s;
       
   854           $num_new_scores++;
       
   855         }
       
   856         $num_valid_words++;
       
   857       }
       
   858     }
       
   859   }
       
   860   // Return matching snippet and number of added words
       
   861   return array("d.data ". ($not ? 'NOT ' : '') ."LIKE '%% %s %%'", $num_new_scores, $num_valid_words);
       
   862 }
       
   863 
       
   864 /**
       
   865  * Do a query on the full-text search index for a word or words.
       
   866  *
       
   867  * This function is normally only called by each module that support the
       
   868  * indexed search (and thus, implements hook_update_index()).
       
   869  *
       
   870  * Results are retrieved in two logical passes. However, the two passes are
       
   871  * joined together into a single query.  And in the case of most simple
       
   872  * queries the second pass is not even used.
       
   873  *
       
   874  * The first pass selects a set of all possible matches, which has the benefit
       
   875  * of also providing the exact result set for simple "AND" or "OR" searches.
       
   876  *
       
   877  * The second portion of the query further refines this set by verifying
       
   878  * advanced text conditions (such negative or phrase matches)
       
   879  *
       
   880  * @param $keywords
       
   881  *   A search string as entered by the user.
       
   882  *
       
   883  * @param $type
       
   884  *   A string identifying the calling module.
       
   885  *
       
   886  * @param $join1
       
   887  *   (optional) Inserted into the JOIN part of the first SQL query.
       
   888  *   For example "INNER JOIN {node} n ON n.nid = i.sid".
       
   889  *
       
   890  * @param $where1
       
   891  *   (optional) Inserted into the WHERE part of the first SQL query.
       
   892  *   For example "(n.status > %d)".
       
   893  *
       
   894  * @param $arguments1
       
   895  *   (optional) Extra SQL arguments belonging to the first query.
       
   896  *
       
   897  * @param $columns2
       
   898  *   (optional) Inserted into the SELECT pat of the second query. Must contain
       
   899  *   a column selected as 'score'.
       
   900  *   defaults to 'i.relevance AS score'
       
   901  *
       
   902  * @param $join2
       
   903  *   (optional) Inserted into the JOIN par of the second SQL query.
       
   904  *   For example "INNER JOIN {node_comment_statistics} n ON n.nid = i.sid"
       
   905  *
       
   906  * @param $arguments2
       
   907  *   (optional) Extra SQL arguments belonging to the second query parameter.
       
   908  *
       
   909  * @param $sort_parameters
       
   910  *   (optional) SQL arguments for sorting the final results.
       
   911  *              Default: 'ORDER BY score DESC'
       
   912  *
       
   913  * @return
       
   914  *   An array of SIDs for the search results.
       
   915  *
       
   916  * @ingroup search
       
   917  */
       
   918 function do_search($keywords, $type, $join1 = '', $where1 = '1 = 1', $arguments1 = array(), $columns2 = 'i.relevance AS score', $join2 = '', $arguments2 = array(), $sort_parameters = 'ORDER BY score DESC') {
       
   919   $query = search_parse_query($keywords);
       
   920 
       
   921   if ($query[2] == '') {
       
   922     form_set_error('keys', t('You must include at least one positive keyword with @count characters or more.', array('@count' => variable_get('minimum_word_size', 3))));
       
   923   }
       
   924   if ($query[6]) {
       
   925     if ($query[6] == 'or') {
       
   926       drupal_set_message(t('Search for either of the two terms with uppercase <strong>OR</strong>. For example, <strong>cats OR dogs</strong>.'));
       
   927     }
       
   928   }
       
   929   if ($query === NULL || $query[0] == '' || $query[2] == '') {
       
   930     return array();
       
   931   }
       
   932 
       
   933   // Build query for keyword normalization.
       
   934   $conditions = "$where1 AND ($query[2]) AND i.type = '%s'";
       
   935   $arguments1 = array_merge($arguments1, $query[3], array($type));
       
   936   $join = "INNER JOIN {search_total} t ON i.word = t.word $join1";
       
   937   if (!$query[5]) {
       
   938     $conditions .= " AND ($query[0])";
       
   939     $arguments1 = array_merge($arguments1, $query[1]);
       
   940     $join .= " INNER JOIN {search_dataset} d ON i.sid = d.sid AND i.type = d.type";
       
   941   }
       
   942 
       
   943   // Calculate maximum keyword relevance, to normalize it.
       
   944   $select = "SELECT SUM(i.score * t.count) AS score FROM {search_index} i $join WHERE $conditions GROUP BY i.type, i.sid HAVING COUNT(*) >= %d ORDER BY score DESC";
       
   945   $arguments = array_merge($arguments1, array($query[4]));
       
   946   $normalize = db_result(db_query_range($select, $arguments, 0, 1));
       
   947   if (!$normalize) {
       
   948     return array();
       
   949   }
       
   950   $columns2 = str_replace('i.relevance', '('. (1.0 / $normalize) .' * SUM(i.score * t.count))', $columns2);
       
   951 
       
   952   // Build query to retrieve results.
       
   953   $select = "SELECT i.type, i.sid, $columns2 FROM {search_index} i $join $join2 WHERE $conditions GROUP BY i.type, i.sid HAVING COUNT(*) >= %d";
       
   954   $count_select =  "SELECT COUNT(*) FROM ($select) n1";
       
   955   $arguments = array_merge($arguments2, $arguments1, array($query[4]));
       
   956 
       
   957   // Do actual search query
       
   958   $result = pager_query("$select $sort_parameters", 10, 0, $count_select, $arguments);
       
   959   $results = array();
       
   960   while ($item = db_fetch_object($result)) {
       
   961     $results[] = $item;
       
   962   }
       
   963   return $results;
       
   964 }
       
   965 
       
   966 /**
       
   967  * Helper function for grabbing search keys.
       
   968  */
       
   969 function search_get_keys() {
       
   970   static $return;
       
   971   if (!isset($return)) {
       
   972     // Extract keys as remainder of path
       
   973     // Note: support old GET format of searches for existing links.
       
   974     $path = explode('/', $_GET['q'], 3);
       
   975     $keys = empty($_REQUEST['keys']) ? '' : $_REQUEST['keys'];
       
   976     $return = count($path) == 3 ? $path[2] : $keys;
       
   977   }
       
   978   return $return;
       
   979 }
       
   980 
       
   981 /**
       
   982  * @defgroup search Search interface
       
   983  * @{
       
   984  * The Drupal search interface manages a global search mechanism.
       
   985  *
       
   986  * Modules may plug into this system to provide searches of different types of
       
   987  * data. Most of the system is handled by search.module, so this must be enabled
       
   988  * for all of the search features to work.
       
   989  *
       
   990  * There are three ways to interact with the search system:
       
   991  * - Specifically for searching nodes, you can implement nodeapi('update index')
       
   992  *   and nodeapi('search result'). However, note that the search system already
       
   993  *   indexes all visible output of a node, i.e. everything displayed normally
       
   994  *   by hook_view() and hook_nodeapi('view'). This is usually sufficient.
       
   995  *   You should only use this mechanism if you want additional, non-visible data
       
   996  *   to be indexed.
       
   997  * - Implement hook_search(). This will create a search tab for your module on
       
   998  *   the /search page with a simple keyword search form. You may optionally
       
   999  *   implement hook_search_item() to customize the display of your results.
       
  1000  * - Implement hook_update_index(). This allows your module to use Drupal's
       
  1001  *   HTML indexing mechanism for searching full text efficiently.
       
  1002  *
       
  1003  * If your module needs to provide a more complicated search form, then you need
       
  1004  * to implement it yourself without hook_search(). In that case, you should
       
  1005  * define it as a local task (tab) under the /search page (e.g. /search/mymodule)
       
  1006  * so that users can easily find it.
       
  1007  */
       
  1008 
       
  1009 /**
       
  1010  * Render a search form.
       
  1011  *
       
  1012  * @param $action
       
  1013  *   Form action. Defaults to "search".
       
  1014  * @param $keys
       
  1015  *   The search string entered by the user, containing keywords for the search.
       
  1016  * @param $type
       
  1017  *   The type of search to render the node for. Must be the name of module
       
  1018  *   which implements hook_search(). Defaults to 'node'.
       
  1019  * @param $prompt
       
  1020  *   A piece of text to put before the form (e.g. "Enter your keywords")
       
  1021  * @return
       
  1022  *   An HTML string containing the search form.
       
  1023  */
       
  1024 function search_form(&$form_state, $action = '', $keys = '', $type = NULL, $prompt = NULL) {
       
  1025 
       
  1026   // Add CSS
       
  1027   drupal_add_css(drupal_get_path('module', 'search') .'/search.css', 'module', 'all', FALSE);
       
  1028 
       
  1029   if (!$action) {
       
  1030     $action = url('search/'. $type);
       
  1031   }
       
  1032   if (is_null($prompt)) {
       
  1033     $prompt = t('Enter your keywords');
       
  1034   }
       
  1035 
       
  1036   $form = array(
       
  1037     '#action' => $action,
       
  1038     '#attributes' => array('class' => 'search-form'),
       
  1039   );
       
  1040   $form['module'] = array('#type' => 'value', '#value' => $type);
       
  1041   $form['basic'] = array('#type' => 'item', '#title' => $prompt);
       
  1042   $form['basic']['inline'] = array('#prefix' => '<div class="container-inline">', '#suffix' => '</div>');
       
  1043   $form['basic']['inline']['keys'] = array(
       
  1044     '#type' => 'textfield',
       
  1045     '#title' => '',
       
  1046     '#default_value' => $keys,
       
  1047     '#size' => $prompt ? 40 : 20,
       
  1048     '#maxlength' => 255,
       
  1049   );
       
  1050   // processed_keys is used to coordinate keyword passing between other forms
       
  1051   // that hook into the basic search form.
       
  1052   $form['basic']['inline']['processed_keys'] = array('#type' => 'value', '#value' => array());
       
  1053   $form['basic']['inline']['submit'] = array('#type' => 'submit', '#value' => t('Search'));
       
  1054 
       
  1055   return $form;
       
  1056 }
       
  1057 
       
  1058 /**
       
  1059  * Form builder; Output a search form for the search block and the theme's search box.
       
  1060  *
       
  1061  * @ingroup forms
       
  1062  * @see search_box_form_submit()
       
  1063  * @see theme_search_box_form()
       
  1064  */
       
  1065 function search_box(&$form_state, $form_id) {
       
  1066   $form[$form_id] = array(
       
  1067     '#title' => t('Search this site'),
       
  1068     '#type' => 'textfield',
       
  1069     '#size' => 15,
       
  1070     '#default_value' => '',
       
  1071     '#attributes' => array('title' => t('Enter the terms you wish to search for.')),
       
  1072   );
       
  1073   $form['submit'] = array('#type' => 'submit', '#value' => t('Search'));
       
  1074   $form['#submit'][] = 'search_box_form_submit';
       
  1075   $form['#validate'][] = 'search_box_form_validate';
       
  1076 
       
  1077   return $form;
       
  1078 }
       
  1079 
       
  1080 /**
       
  1081  * Process a block search form submission.
       
  1082  */
       
  1083 function search_box_form_submit($form, &$form_state) {
       
  1084   $form_id = $form['form_id']['#value'];
       
  1085   $form_state['redirect'] = 'search/node/'. trim($form_state['values'][$form_id]);
       
  1086 }
       
  1087 
       
  1088 /**
       
  1089  * Process variables for search-theme-form.tpl.php.
       
  1090  *
       
  1091  * The $variables array contains the following arguments:
       
  1092  * - $form
       
  1093  *
       
  1094  * @see search-theme-form.tpl.php
       
  1095  */
       
  1096 function template_preprocess_search_theme_form(&$variables) {
       
  1097   $variables['search'] = array();
       
  1098   $hidden = array();
       
  1099   // Provide variables named after form keys so themers can print each element independently.
       
  1100   foreach (element_children($variables['form']) as $key) {
       
  1101     $type = $variables['form'][$key]['#type'];
       
  1102     if ($type == 'hidden' || $type == 'token') {
       
  1103       $hidden[] = drupal_render($variables['form'][$key]);
       
  1104     }
       
  1105     else {
       
  1106       $variables['search'][$key] = drupal_render($variables['form'][$key]);
       
  1107     }
       
  1108   }
       
  1109   // Hidden form elements have no value to themers. No need for separation.
       
  1110   $variables['search']['hidden'] = implode($hidden);
       
  1111   // Collect all form elements to make it easier to print the whole form.
       
  1112   $variables['search_form'] = implode($variables['search']);
       
  1113 }
       
  1114 
       
  1115 /**
       
  1116  * Process variables for search-block-form.tpl.php.
       
  1117  *
       
  1118  * The $variables array contains the following arguments:
       
  1119  * - $form
       
  1120  *
       
  1121  * @see search-block-form.tpl.php
       
  1122  */
       
  1123 function template_preprocess_search_block_form(&$variables) {
       
  1124   $variables['search'] = array();
       
  1125   $hidden = array();
       
  1126   // Provide variables named after form keys so themers can print each element independently.
       
  1127   foreach (element_children($variables['form']) as $key) {
       
  1128     $type = $variables['form'][$key]['#type'];
       
  1129     if ($type == 'hidden' || $type == 'token') {
       
  1130       $hidden[] = drupal_render($variables['form'][$key]);
       
  1131     }
       
  1132     else {
       
  1133       $variables['search'][$key] = drupal_render($variables['form'][$key]);
       
  1134     }
       
  1135   }
       
  1136   // Hidden form elements have no value to themers. No need for separation.
       
  1137   $variables['search']['hidden'] = implode($hidden);
       
  1138   // Collect all form elements to make it easier to print the whole form.
       
  1139   $variables['search_form'] = implode($variables['search']);
       
  1140 }
       
  1141 
       
  1142 /**
       
  1143  * Perform a standard search on the given keys, and return the formatted results.
       
  1144  */
       
  1145 function search_data($keys = NULL, $type = 'node') {
       
  1146 
       
  1147   if (isset($keys)) {
       
  1148     if (module_hook($type, 'search')) {
       
  1149       $results = module_invoke($type, 'search', 'search', $keys);
       
  1150       if (isset($results) && is_array($results) && count($results)) {
       
  1151         if (module_hook($type, 'search_page')) {
       
  1152           return module_invoke($type, 'search_page', $results);
       
  1153         }
       
  1154         else {
       
  1155           return theme('search_results', $results, $type);
       
  1156         }
       
  1157       }
       
  1158     }
       
  1159   }
       
  1160 }
       
  1161 
       
  1162 /**
       
  1163  * Returns snippets from a piece of text, with certain keywords highlighted.
       
  1164  * Used for formatting search results.
       
  1165  *
       
  1166  * @param $keys
       
  1167  *   A string containing a search query.
       
  1168  *
       
  1169  * @param $text
       
  1170  *   The text to extract fragments from.
       
  1171  *
       
  1172  * @return
       
  1173  *   A string containing HTML for the excerpt.
       
  1174  */
       
  1175 function search_excerpt($keys, $text) {
       
  1176   // We highlight around non-indexable or CJK characters.
       
  1177   $boundary = '(?:(?<=['. PREG_CLASS_SEARCH_EXCLUDE . PREG_CLASS_CJK .'])|(?=['. PREG_CLASS_SEARCH_EXCLUDE . PREG_CLASS_CJK .']))';
       
  1178 
       
  1179   // Extract positive keywords and phrases
       
  1180   preg_match_all('/ ("([^"]+)"|(?!OR)([^" ]+))/', ' '. $keys, $matches);
       
  1181   $keys = array_merge($matches[2], $matches[3]);
       
  1182 
       
  1183   // Prepare text
       
  1184   $text = ' '. strip_tags(str_replace(array('<', '>'), array(' <', '> '), $text)) .' ';
       
  1185   array_walk($keys, '_search_excerpt_replace');
       
  1186   $workkeys = $keys;
       
  1187 
       
  1188   // Extract a fragment per keyword for at most 4 keywords.
       
  1189   // First we collect ranges of text around each keyword, starting/ending
       
  1190   // at spaces.
       
  1191   // If the sum of all fragments is too short, we look for second occurrences.
       
  1192   $ranges = array();
       
  1193   $included = array();
       
  1194   $length = 0;
       
  1195   while ($length < 256 && count($workkeys)) {
       
  1196     foreach ($workkeys as $k => $key) {
       
  1197       if (strlen($key) == 0) {
       
  1198         unset($workkeys[$k]);
       
  1199         unset($keys[$k]);
       
  1200         continue;
       
  1201       }
       
  1202       if ($length >= 256) {
       
  1203         break;
       
  1204       }
       
  1205       // Remember occurrence of key so we can skip over it if more occurrences
       
  1206       // are desired.
       
  1207       if (!isset($included[$key])) {
       
  1208         $included[$key] = 0;
       
  1209       }
       
  1210       // Locate a keyword (position $p), then locate a space in front (position
       
  1211       // $q) and behind it (position $s)
       
  1212       if (preg_match('/'. $boundary . $key . $boundary .'/iu', $text, $match, PREG_OFFSET_CAPTURE, $included[$key])) {
       
  1213         $p = $match[0][1];
       
  1214         if (($q = strpos($text, ' ', max(0, $p - 60))) !== FALSE) {
       
  1215           $end = substr($text, $p, 80);
       
  1216           if (($s = strrpos($end, ' ')) !== FALSE) {
       
  1217             $ranges[$q] = $p + $s;
       
  1218             $length += $p + $s - $q;
       
  1219             $included[$key] = $p + 1;
       
  1220           }
       
  1221           else {
       
  1222             unset($workkeys[$k]);
       
  1223           }
       
  1224         }
       
  1225         else {
       
  1226           unset($workkeys[$k]);
       
  1227         }
       
  1228       }
       
  1229       else {
       
  1230         unset($workkeys[$k]);
       
  1231       }
       
  1232     }
       
  1233   }
       
  1234 
       
  1235   // If we didn't find anything, return the beginning.
       
  1236   if (count($ranges) == 0) {
       
  1237     return truncate_utf8($text, 256) .' ...';
       
  1238   }
       
  1239 
       
  1240   // Sort the text ranges by starting position.
       
  1241   ksort($ranges);
       
  1242 
       
  1243   // Now we collapse overlapping text ranges into one. The sorting makes it O(n).
       
  1244   $newranges = array();
       
  1245   foreach ($ranges as $from2 => $to2) {
       
  1246     if (!isset($from1)) {
       
  1247       $from1 = $from2;
       
  1248       $to1 = $to2;
       
  1249       continue;
       
  1250     }
       
  1251     if ($from2 <= $to1) {
       
  1252       $to1 = max($to1, $to2);
       
  1253     }
       
  1254     else {
       
  1255       $newranges[$from1] = $to1;
       
  1256       $from1 = $from2;
       
  1257       $to1 = $to2;
       
  1258     }
       
  1259   }
       
  1260   $newranges[$from1] = $to1;
       
  1261 
       
  1262   // Fetch text
       
  1263   $out = array();
       
  1264   foreach ($newranges as $from => $to) {
       
  1265     $out[] = substr($text, $from, $to - $from);
       
  1266   }
       
  1267   $text = (isset($newranges[0]) ? '' : '... ') . implode(' ... ', $out) .' ...';
       
  1268 
       
  1269   // Highlight keywords. Must be done at once to prevent conflicts ('strong' and '<strong>').
       
  1270   $text = preg_replace('/'. $boundary .'('. implode('|', $keys) .')'. $boundary .'/iu', '<strong>\0</strong>', $text);
       
  1271   return $text;
       
  1272 }
       
  1273 
       
  1274 /**
       
  1275  * @} End of "defgroup search".
       
  1276  */
       
  1277 
       
  1278 /**
       
  1279  * Helper function for array_walk in search_except.
       
  1280  */
       
  1281 function _search_excerpt_replace(&$text) {
       
  1282   $text = preg_quote($text, '/');
       
  1283 }
       
  1284 
       
  1285 function search_forms() {
       
  1286   $forms['search_theme_form']= array(
       
  1287     'callback' => 'search_box',
       
  1288     'callback arguments' => array('search_theme_form'),
       
  1289   );
       
  1290   $forms['search_block_form']= array(
       
  1291     'callback' => 'search_box',
       
  1292     'callback arguments' => array('search_block_form'),
       
  1293   );
       
  1294   return $forms;
       
  1295 }