function search($searchstr, $category, $start, $per_page, $type, $domain)
{
global $length_of_link_desc, $show_meta_description, $merge_site_results, $stem_words;
global $did_you_mean_enabled, $did_you_mean_always;
global $matchless, $equivalent, $language;
global $db;
$possible_to_find = 1;
$stat = $db->prepare("SELECT domain_id FROM " . TABLE_PREFIX . "domains WHERE domain = :domain");
$stat->execute(array(':domain' => $domain));
if ($row = $stat->fetch()) {
$domain_qry = "and domain = " . $row[0];
} else {
$domain_qry = "";
}
$stat->closeCursor();
/* if there are no words to search for, quit */
if (!isset($searchstr['+']) || count($searchstr['+']) == 0) {
return null;
}
/* find all words that should _not_ be included in the result */
if (isset($searchstr['-'])) {
$wordarray = $searchstr['-'];
} else {
$wordarray = array();
}
$notlist = array();
$not_words = 0;
while ($not_words < count($wordarray)) {
if ($stem_words == 1) {
$searchword = stem($wordarray[$not_words]);
} else {
$searchword = $wordarray[$not_words];
}
$wordmd5 = substr(md5($searchword), 0, 1);
$stat = $db->prepare("SELECT link_id from " . TABLE_PREFIX . "link_keyword{$wordmd5}, " . TABLE_PREFIX . "keywords where " . TABLE_PREFIX . "link_keyword{$wordmd5}.keyword_id= " . TABLE_PREFIX . "keywords.keyword_id and keyword = :keyword");
$stat->execute(array(':keyword' => $searchword));
while ($row = $stat->fetch()) {
$notlist[$not_words]['id'][$row[0]] = 1;
}
$not_words++;
}
/* find all phrases */
if (isset($searchstr['+s'])) {
$wordarray = $searchstr['+s'];
} else {
$wordarray = array();
}
$phrase_words = 0;
while ($phrase_words < count($wordarray)) {
$searchword = $wordarray[$phrase_words];
$searchword = str_replace("|", "", $searchword);
$searchword = str_replace("%", "|%", $searchword);
$searchword = str_replace("_", "|_", $searchword);
$stat = $db->prepare("SELECT link_id from " . TABLE_PREFIX . "links where fulltxt like :keyword escape '|'");
$stat->execute(array(':keyword' => "%" . $searchword . "%"));
echo sql_errorstring(__FILE__, __LINE__);
$row = $stat->fetch();
if (!$row) {
$possible_to_find = 0;
$stat->closeCursor();
break;
}
$phraselist[$phrase_words]['id'][$row[0]] = 1;
while ($row = $stat->fetch()) {
$phraselist[$phrase_words]['id'][$row[0]] = 1;
}
$phrase_words++;
}
if ($category > 0 && $possible_to_find == 1) {
$allcats = get_cats($category);
$catlist = implode(",", $allcats);
$result = $db->query("SELECT link_id FROM " . TABLE_PREFIX . "links, " . TABLE_PREFIX . "sites, " . TABLE_PREFIX . "categories, " . TABLE_PREFIX . "site_category where " . TABLE_PREFIX . "links.site_id = " . TABLE_PREFIX . "sites.site_id and " . TABLE_PREFIX . "sites.site_id = " . TABLE_PREFIX . "site_category.site_id and " . TABLE_PREFIX . "site_category.category_id in ({$catlist})");
echo sql_errorstring(__FILE__, __LINE__);
$row = $result->fetch();
if (!$row) {
$possible_to_find = 0;
} else {
$category_list[$row[0]] = 1;
while ($row = $result->fetch()) {
$category_list[$row[0]] = 1;
}
}
$result->closeCursor();
}
/* find individual words */
$word_not_found = array();
$wordarray = $searchstr['+'];
$words = 0;
while ($words < count($wordarray) && $possible_to_find == 1) {
if ($stem_words == 1) {
$searchword = stem($wordarray[$words]);
} else {
$searchword = $wordarray[$words];
}
$wordmd5 = substr(md5($searchword), 0, 1);
$stat = $db->prepare("SELECT distinct link_id, weight, domain FROM " . TABLE_PREFIX . "link_keyword{$wordmd5}, " . TABLE_PREFIX . "keywords WHERE " . TABLE_PREFIX . "link_keyword{$wordmd5}.keyword_id= " . TABLE_PREFIX . "keywords.keyword_id AND keyword=:keyword {$domain_qry}\tORDER\tBY\tweight\tDESC");
$stat->execute(array(':keyword' => $searchword));
echo sql_errorstring(__FILE__, __LINE__);
$row = $stat->fetch();
if (!$row) {
$word_not_found[$wordarray[$words]] = 1;
if ($type != "or") {
$possible_to_find = 0;
$stat->closeCursor();
break;
}
}
if ($type == "or") {
$indx = 0;
} else {
$indx = $words;
}
do {
$linklist[$indx]['id'][] = $row[0];
$domains[$row[0]] = $row[2];
$linklist[$indx]['weight'][$row[0]] = $row[1];
} while ($row = $stat->fetch());
$words++;
}
if ($type == "or") {
$words = 1;
}
$result_array_full = array();
if ($possible_to_find != 0) {
if ($words == 1 && $not_words == 0 && $category < 1) {
//if there is only one search word, we already have the result
$result_array_full = $linklist[0]['weight'];
} else {
//otherwise build an intersection of all the results
$j = 1;
$min = 0;
while ($j < $words) {
if (count($linklist[$min]['id']) > count($linklist[$j]['id'])) {
$min = $j;
}
$j++;
}
$j = 0;
$temp_array = $linklist[$min]['id'];
$count = 0;
while ($j < count($temp_array)) {
$k = 0;
//and word counter
$n = 0;
//not word counter
$o = 0;
//phrase word counter
$weight = 1;
$break = 0;
while ($k < $words && $break == 0) {
if (isset($linklist[$k]['weight'][$temp_array[$j]]) && $linklist[$k]['weight'][$temp_array[$j]] > 0) {
$weight = $weight + $linklist[$k]['weight'][$temp_array[$j]];
} else {
$break = 1;
}
$k++;
}
while ($n < $not_words && $break == 0) {
if ($notlist[$n]['id'][$temp_array[$j]] > 0) {
$break = 1;
}
$n++;
}
while ($o < $phrase_words && $break == 0) {
if (!isset($phraselist[$n]['id'][$temp_array[$j]]) || $phraselist[$n]['id'][$temp_array[$j]] != 1) {
$break = 1;
}
$o++;
}
if ($break == 0 && $category > 0 && $category_list[$temp_array[$j]] != 1) {
$break = 1;
}
if ($break == 0) {
$result_array_full[$temp_array[$j]] = $weight;
$count++;
}
$j++;
}
}
}
if ((count($result_array_full) == 0 || $possible_to_find == 0 || $did_you_mean_always == 1) && $did_you_mean_enabled == 1) {
/* search for word pairs written as two words where a single words
for example: when the user typed "full colour", also search for
fullcolour and full-colour */
for ($idx = 0; $idx < count($searchstr['+']) - 1; $idx++) {
$word = $searchstr['+'][$idx] . " " . $searchstr['+'][$idx + 1];
$near_word = $searchstr['+'][$idx] . $searchstr['+'][$idx + 1];
/* words that are in the "nonpareil" list are excluded in searching
for alternatives */
if (!isset($matchless[$near_word])) {
$stat = $db->prepare("SELECT keyword FROM " . TABLE_PREFIX . "keywords WHERE keyword=:keyword");
if ($stat->execute(array(':keyword' => $near_word)) && ($row = $stat->fetch())) {
$near_words[$word] = latin1_to_html($near_word);
$stat->closeCursor();
}
}
$near_word = $searchstr['+'][$idx] . "-" . $searchstr['+'][$idx + 1];
if (!isset($matchless[$near_word])) {
$stat = $db->prepare("SELECT keyword FROM " . TABLE_PREFIX . "keywords WHERE keyword=:keyword");
if ($stat->execute(array(':keyword' => $near_word)) && ($row = $stat->fetch())) {
$near_words[$word] = latin1_to_html($near_word);
$stat->closeCursor();
}
}
}
/* then search for "near words" for the individual words */
reset($searchstr['+']);
foreach ($searchstr['+'] as $word) {
/* words that are in the "nonpareil" list are excluded in searching
for alternatives */
if (isset($matchless[$word]) && $matchless[$word] == 1) {
continue;
}
/* search for alternatives in the explicit equivalents word list first */
if (isset($equivalent[$word]) && strlen($equivalent[$word]) > 0) {
$near_words[$word] = latin1_to_html($equivalent[$word]);
continue;
}
/* if there are misspelled words, show only alternatives for the
misspelled words, (so, if the current word is not in the list
of misspelled words, exclude it from the search for alternatives */
if (count($word_not_found) > 0 && !(isset($word_not_found[$word]) && $word_not_found[$word] == 1)) {
continue;
}
$word = sanitize($word);
/* use the double-metaphone to find close words */
$meta = double_metaphone($word);
if (!isset($meta["primary"]) || strlen($meta["primary"]) == 0) {
continue;
}
/* no metaphone, don't match anything */
$where = "metaphone1='" . $meta["primary"] . "' OR metaphone2='" . $meta["primary"] . "'";
if (isset($meta["secondary"]) && strlen($meta["secondary"]) > 0) {
$where .= " OR metaphone1='" . $meta["secondary"] . "' OR metaphone2='" . $meta["secondary"] . "'";
}
$result = $db->query("SELECT keyword FROM " . TABLE_PREFIX . "keywords WHERE {$where}");
/* adapted from http://www.mdj.us/web-development/php-programming/creating-better-search-suggestions-with-sphider/
but using a double-metaphone filter (instead of SOUNDEX) and
adding a filter for accented characters */
$max_distance = 3;
$max_similar = 0;
$near_word = "";
while ($result && ($row = $result->fetch())) {
$item = $row[0];
if (strcasecmp($item, $word) != 0) {
$distance = levenshtein($item, $word);
$distance_na = levenshtein(remove_accents($item), $word);
if ($distance_na < $distance) {
$distance = $distance_na;
}
if ($distance < $max_distance) {
$max_distance = $distance;
$near_word = $item;
}
if ($distance == $max_distance) {
$similar = similar_text($item, $word);
if ($similar >= $max_similar) {
$max_distance = $distance;
$max_similar = $similar;
$near_word = $item;
}
}
}
}
if ($near_word != "") {
$near_words[$word] = latin1_to_html($near_word);
} else {
if (isset($word_not_found[$word]) && $word_not_found[$word] == 1 && count($wordarray) > 1) {
$near_words[$word] = "/{$word}";
}
}
}
if (!isset($near_words)) {
$near_words = "";
}
$res['did_you_mean'] = $near_words;
if (count($result_array_full) == 0 || $possible_to_find == 0) {
return $res;
}
}
if (count($result_array_full) == 0) {
return null;
}
arsort($result_array_full);
if ($merge_site_results == 1 && $domain_qry == "") {
while (list($key, $value) = each($result_array_full)) {
if (!isset($domains_to_show[$domains[$key]])) {
$result_array_temp[$key] = $value;
$domains_to_show[$domains[$key]] = 1;
} else {
if ($domains_to_show[$domains[$key]] == 1) {
$domains_to_show[$domains[$key]] = array($key => $value);
}
}
}
} else {
$result_array_temp = $result_array_full;
}
while (list($key, $value) = each($result_array_temp)) {
$result_array[$key] = $value;
if (isset($domains_to_show[$domains[$key]]) && $domains_to_show[$domains[$key]] != 1) {
list($k, $v) = each($domains_to_show[$domains[$key]]);
$result_array[$k] = $v;
}
}
$results = count($result_array);
$keys = array_keys($result_array);
$maxweight = $result_array[$keys[0]];
for ($i = ($start - 1) * $per_page; $i < min($results, ($start - 1) * $per_page + $per_page); $i++) {
$in[] = $keys[$i];
}
if (!is_array($in)) {
$res['results'] = $results;
return $res;
}
$inlist = implode(",", $in);
if ($length_of_link_desc == 0) {
$fulltxt = "fulltxt";
} else {
$fulltxt = "substring(fulltxt, 1, {$length_of_link_desc})";
}
$query = "SELECT distinct link_id, url, title, description, language, {$fulltxt}, size FROM " . TABLE_PREFIX . "links WHERE link_id in ({$inlist})";
$result = $db->query($query);
echo sql_errorstring(__FILE__, __LINE__);
$i = 0;
while ($row = $result->fetch()) {
$res[$i]['title'] = $row[2];
$res[$i]['url'] = $row[1];
if (isset($row[3]) && $row[3] != null && $show_meta_description == 1) {
$res[$i]['summary'] = $row[3];
} else {
$res[$i]['summary'] = "";
}
$res[$i]['lang'] = $row[4];
$res[$i]['fulltxt'] = $row[5];
$res[$i]['size'] = $row[6];
$res[$i]['weight'] = $result_array[$row[0]];
/* if a language has been set for this page, and it is _not_ the
* same language as the user language, decrease the weight
*/
if (isset($row[4]) && $row[4] != null && strlen($row[4]) > 0 && strcasecmp($row[4], $language) != 0) {
$res[$i]['weight'] *= 0.5;
}
$dom_result = $db->query("select domain from " . TABLE_PREFIX . "domains where domain_id='" . $domains[$row[0]] . "'");
$dom_row = $dom_result->fetch();
$res[$i]['domain'] = $dom_row[0];
$i++;
}
if ($merge_site_results && $domain_qry == "") {
sort_with_domains($res);
} else {
usort($res, "cmp");
}
echo sql_errorstring(__FILE__, __LINE__);
/* sorting destroys the other columns in the array, restore */
if (isset($near_words)) {
$res['did_you_mean'] = $near_words;
}
$res['maxweight'] = $maxweight;
$res['results'] = $results;
return $res;
/**/
}
function save_keywords($wordarray, $link_id, $domain)
{
global $all_keywords;
global $db;
reset($wordarray);
$inserts = array();
while ($thisword = each($wordarray)) {
$word = $thisword[1][1];
$wordmd5 = substr(md5($word), 0, 1);
$weight = $thisword[1][2];
if (strlen($word) <= 30) {
if (isset($all_keywords[$word])) {
$keyword_id = $all_keywords[$word];
} else {
/* create the double metaphone */
$meta = double_metaphone($word);
$meta1 = $meta["primary"];
$meta2 = $meta["secondary"];
$res = $db->exec("insert into " . TABLE_PREFIX . "keywords (keyword,metaphone1,metaphone2) values ('{$word}','{$meta1}','{$meta2}')");
if (!$res) {
//error
$query = "select keyword_ID from " . TABLE_PREFIX . "keywords where keyword='{$word}'";
$result = $db->query($query);
$error = sql_errorstring(__FILE__, __LINE__);
if ($error) {
echo $error;
}
echo sql_errorstring(__FILE__, __LINE__);
$row = $result->fetch();
$keyword_id = $row[0];
$result->closeCursor();
} else {
$keyword_id = $db->lastInsertId();
$all_keywords[$word] = $keyword_id;
echo sql_errorstring(__FILE__, __LINE__);
}
}
if (!isset($inserts[$wordmd5])) {
$inserts[$wordmd5] = "";
}
$inserts[$wordmd5] .= ";({$link_id}, {$keyword_id}, {$weight}, {$domain})";
}
}
for ($i = 0; $i <= 15; $i++) {
$char = dechex($i);
if (isset($inserts[$char])) {
$values = substr($inserts[$char], 1);
} else {
$values = "";
}
if ($values != "") {
$arr = explode(';', $values);
foreach ($arr as $insert_sql) {
$query = "insert into " . TABLE_PREFIX . "link_keyword{$char} (link_id, keyword_id, weight, domain) values {$insert_sql}";
$db->exec($query);
$error = sql_errorstring(__FILE__, __LINE__);
if ($error) {
echo $error;
}
}
}
}
}
