path: root/modules/search.module

<?php
// $Id$

function search_help() {
  $output = "<b>". t("Search guidelines") ."</b>";
  $output .= "<p>". t("The search page allows you to search the website's content.  You can specify multiple words, and they will all be searched for.  You can also use wildcards, so 'walk*' will match 'walk', 'walking', 'walker', 'walkable' and so on.  Furthermore, searches are not case sensitive so searching for 'walk', 'Walk' or 'WALK' will yield exactly the same results.") ."</p>";
  $output .= "<b>". t("Words excluded from the search") ."</b>";
  $output .= "<p>". t("Words that frequently occur, typically called 'noise words', are ignored.  Example words are 'a', 'at', 'and', 'are', 'as', 'how', 'where', etc.  Words shorter than %number letters are also being filtered.", array("%number" => variable_get("minimum_word_size", 2))) ."</p>";
  return $output;
}

function search_system($field){
  $system["description"] = t("Enables site wide keyword searching.");
  return $system[$field];
}

/**
 * Return an array of valid search access permissions
 */
function search_perm() {
  return array("search content", "administer search");
}

/**
 * Return an array of links to be displayed
 *
 * @param $type  The type of page requesting the link
 *
 */
function search_link($type) {
  $links = array();

  if ($type == "page" && user_access("search content")) {
    $links[] = l(t("search"), "search", array("title" => t("Search for older content.")));
  }

  return $links;
}

function search_settings() {
  $output = form_textfield(t("Minimum word length to index"), "minimum_word_size", variable_get("minimum_word_size", 2), 10, 10, t("The number of characters a word has to be to be indexed.  Words shorter than this will not be searchable."));
  $output .= form_textfield(t("Minimum word length to search for"), "remove_short", variable_get("remove_short", 0), 10, 10, t("The number of characters a word has to be to be searched for."));
  $output .= form_textarea(t("Noise words"), "noisewords", variable_get("noisewords", ""), 70, 10, t("These words will not be indexed, enter comma separated list, linebreaks and whitespace do not matter.  Example: and, or, not, a, to, I, it, ..."));
  $output .= form_select(t("Help text position"), "help_pos", variable_get("help_pos", 1), array("1" => t("Above search output"), "2" => t("Below search output"), "3" => t("Link from above search output"), "4" => t("Link from below search output")), t("Where to show the help text for users on the search page."));

  return $output;
}

/**
 * search engine administration actions
 *
 */
function search_admin() {
  global $op;

  // Only allow people with sufficient access.
  if (user_access("administer search")) {
    if ($op == "reindex") {
      search_invalidate();
      print t("index invalidated") ."<br />\n";
      search_cron();
      print t("index recreated") ."<br /><hr />\n";
    }
  }
}

/**
 * perform a regularly run action across all modules that have the
 * <module>_update_index function in them.
 *
 */
function search_cron() {
  foreach (module_list() as $module) {
    $module_array = module_invoke($module, "update_index");
    if ($module_array) {
      update_index($module_array);
    }
    $module_array = null;
  }
  return;
}

/**
 * Perform a search on a word(s)
 *
 * Search function called by each node that supports the indexed search
 *
 * @param $search_array  an array as returned from <module>_search
 *                       of type array("keys" => ...,
 *                       "type" => ..., "select" => ...)
 *                       see node_search in node.module for an
 *                       explanation of array items
 */
function do_search($search_array) {
  global $PHP_SELF;

  $keys = strtolower($search_array["keys"]);
  $type = $search_array["type"];
  $select = $search_array["select"];

  // Replace wildcards with mysql wildcards
  $keys = str_replace("*", "%", $keys);

  // Split the words entered into an array
  $words = explode(" ", $keys);

  foreach ($words as $word) {

    // If the word is too short, and we've got it set to skip them, loop
    if (strlen($word) < variable_get("remove_short", 0)) {
      continue;
    }

    // Put the next search word into the query and do the query
    $query = preg_replace("'\%'", $word, $select);
    $result = db_query($query);

    // If we got any results
    if (db_num_rows($result) != 0) {
      $found = 1;

      // Create an in memory array of the results,
      while ($row = db_fetch_array($result)) {
        $lno       = $row["lno"];
        $nid       = $row["nid"];
        $title     = $row["title"];
        $created   = $row["created"];
        $uid       = $row["uid"];
        $name      = $row["name"];
        $count     = $row["count"];

        // If the just fetched row is not already in the table
        if ($results[$lno]["lno"] != $lno) {
          $results[$lno]["count"] = $count;

          $results[$lno]["lno"] = $lno;
          $results[$lno]["nid"] = $nid;
          $results[$lno]["title"] = $title;
          $results[$lno]["created"] = $created;
          $results[$lno]["uid"] = $uid;
          $results[$lno]["name"] = $name;
        }
        else {
          /*
          ** Different word, but existing "lno", increase the count of
          ** matches against this "lno" by the number of times this
          ** word appears in the text
          */
          $results[$lno]["count"] = $results[$lno]["count"] + $count;
        }
      }
    }
  }

  if ($found) {
    // Black magic here to sort the results
    array_multisort($results, SORT_DESC);

    // OK, time to output the results.
    foreach ($results as $key => $value) {
      $lno       = $value["lno"];
      $nid       = $value["nid"];
      $title     = $value["title"];
      $created   = $value["created"];
      $uid       = $value["uid"];
      $name      = $value["name"];
      $count     = $value["count"];
      switch ($type) {
      case "node":
        $find[$i++] = array("count" => $count, "title" => $title, "link" => (strstr($PHP_SELF, "admin") ? url("admin/node/edit/$lno") : url("node/view/$lno")), "user" => $name, "date" => $created, "keywords" => implode("|", $words));
        break;
      case "comment":
        $find[$i++] = array("count" => $count, "title" => $title, "link" => (strstr($PHP_SELF, "admin") ? url("admin/comment/edit/$lno") : url("node/view/$nid#$lno")), "user" => $name, "date" => $created, "keywords" => implode("|", $words));
        break;
      }
    }
  }

  return $find;
}

/**
 * Update the search_index table
 *
 * @param $search_array  an array as returned from <module>_update_index
 *                       of type array("last_update" => ...,
 *                       "node_type" => ..., "select" => ...)
 *                       see node_update_index in node.module for an
 *                       explanation of array items
 */
function update_index($search_array) {
  $last_update = variable_get($search_array["last_update"], 1);
  $node_type = $search_array["node_type"];
  $select = $search_array["select"];
  $minimum_word_size = variable_get("minimum_word_size", 2);

  //watchdog("user", "$last_update<br />$node_type<br />$select");

  $result = db_query($select);

  if (db_num_rows($result)) {
    // Wohoo, found some, look through the nodes we just selected
    while ($node = db_fetch_array ($result)) {

      /*
      ** Trash any existing entries in the search index for this node,
      ** in case its a modified node.
      */
      db_query("DELETE from search_index where lno = '". $node["lno"] ."' and type = '". $node_type ."'");

      /*
      ** Build the wordlist, teaser not included, as it then gives a
      ** false count of the number of hits, and doesn't show up
      ** when clicking on a node from the search interface anyway.
      */
      $wordlist = $node["text1"] ." ". $node["text2"];

      // Strip heaps of stuff out of it
      $wordlist = preg_replace("'<[\/\!]*?[^<>]*?>'si", "", $wordlist);

      // Remove punctuation and stuff
      $wordlist = preg_replace("'(»|«|!|¡|%|,|:|;|\(|\)|\&|\"|\'|\.|-|\/|\?|\\\)'", "", $wordlist);

      // Strip out (now mangled) http and tags.
      $wordlist = preg_replace("'http\w+'", "", $wordlist);
      $wordlist = preg_replace("'www\w+'", "", $wordlist);

      // Remove all newlines of any type
      $wordlist = preg_replace("'([\r\n]|[\r]|[\n])'", " ", $wordlist);

      // Lower case the whole thing.
      $wordlist = strtolower($wordlist);

      // Remove "noisewords"
      $noise = explode(",", variable_get("noisewords", ""));
      foreach ($noise as $word) {
        $wordlist = preg_replace("' $word '", " ", $wordlist);
      }

      // Remove whitespace
      $wordlist = preg_replace("'[\s]+'", " ", $wordlist);

      // Make it an array
      $eachword = explode(" ", $wordlist);

      /*
      ** walk through the array, giving a "weight" to each word, based on
      ** the number of times it appears in a page.
      */
      foreach ($eachword as $word) {
        if (strlen($word) >= $minimum_word_size) {
          if ($newwords[$word]) {
            $newwords[$word]++;
          }
          else {
            $newwords[$word] = 1;
          }
        }
      }

      /*
      ** Walk through the weighted words array, inserting them into
      ** the search index
      */
      foreach ($newwords as $key => $value) {
        db_query("INSERT INTO search_index VALUES('%s', %d, '%s', %d)", $key, $node["lno"], $node_type, $value);
      }

      // Zap the weighted words array, so we don't add multiples.
      $newwords = array ();
    }
  }

  // update the last time this process was run.
  variable_set($search_array["last_update"], time());

  return true;
}


function search_invalidate() {
  foreach (module_list() as $module) {
    $module_array = module_invoke($module, "update_index");
    if ($module_array) {
      variable_set($module_array["last_update"], 1);
    }
    $module_array = null;
  }
  return;
}

/**
 * Save the values entered by the administrator for the search module
 *
 * @param $edit  An array of fields as setup via calling form_textfield,
 *               form_textarea etc
 */
function search_save($edit) {
  variable_set("minimum_word_size", $edit["minimum_word_size"]);

  $data = strtr($edit["noisewords"], "\n\r\t", "   ");
  $data = str_replace(" ", "", $data);
  variable_set("noisewords", $data);
  variable_set("help_pos", $edit["help_pos"]);
  variable_set("remove_short", $edit["remove_short"]);
}

function search_view($keys = NULL) {
  global $edit, $type;

  if (user_access("search content")) {
    // Construct the search form:
    $form = search_form(NULL, NULL, TRUE);

    // Collect the search results:
    $output = search_data();

    // Display form and search results:
    $help_link = l(t("search help"), "search/help");
    switch (variable_get("help_pos", 1)) {
      case "1":
        $form = search_help(). $form;
        break;
      case "2":
        $form .= search_help();
        break;
      case "3":
        $form = $help_link. $form;
        break;
      case "4":
        $form .= $help_link;
    }

    theme("header");

    if ($form) {
      theme("box", t("Search"), $form);
    }

    if ($keys) {
      if ($output) {
        theme("box", t("Result"), $output);
      }
      else {
        theme("box", t("Result"), t("Your search yielded no results."));
      }
    }

    theme("footer");
  }
  else {
    theme("header");
    theme("box", t("Access denied"), message_access());
    theme("footer");
  }

}

function search_page() {
  global $keys;

  switch (arg(1)) {
    case "help":
      theme("header");
      theme("box", t("Search Help"), search_help());
      theme("footer");
      break;
    default:
      search_view($keys);
  }
}

?>