function relevanssi_recognize_phrases($q)
{
global $wpdb;
$phrases = relevanssi_extract_phrases($q);
if (count($phrases) > 0) {
$phrase_matches = array();
foreach ($phrases as $phrase) {
$phrase = esc_sql($phrase);
"on" == get_option("relevanssi_index_excerpt") ? $excerpt = " OR post_excerpt LIKE '%{$phrase}%'" : ($excerpt = "");
$query = "SELECT ID FROM {$wpdb->posts} \n\t\t\t\tWHERE (post_content LIKE '%{$phrase}%' OR post_title LIKE '%{$phrase}%' {$excerpt})\n\t\t\t\tAND post_status IN ('publish', 'draft', 'private', 'pending', 'future', 'inherit')";
$docs = $wpdb->get_results($query);
if (is_array($docs)) {
foreach ($docs as $doc) {
if (!isset($phrase_matches[$phrase])) {
$phrase_matches[$phrase] = array();
}
$phrase_matches[$phrase][] = $doc->ID;
}
}
$query = "SELECT ID FROM {$wpdb->posts} as p, {$wpdb->term_relationships} as r, {$wpdb->term_taxonomy} as s, {$wpdb->terms} as t\n\t\t\t\tWHERE r.term_taxonomy_id = s.term_taxonomy_id AND s.term_id = t.term_id AND p.ID = r.object_id\n\t\t\t\tAND t.name LIKE '%{$phrase}%' AND p.post_status IN ('publish', 'draft', 'private', 'pending', 'future', 'inherit')";
$docs = $wpdb->get_results($query);
if (is_array($docs)) {
foreach ($docs as $doc) {
if (!isset($phrase_matches[$phrase])) {
$phrase_matches[$phrase] = array();
}
$phrase_matches[$phrase][] = $doc->ID;
}
}
$query = "SELECT ID\n FROM {$wpdb->posts} AS p, {$wpdb->postmeta} AS m\n WHERE p.ID = m.post_id\n AND m.meta_value LIKE '%{$phrase}%'\n AND p.post_status IN ('publish', 'draft', 'private', 'pending', 'future', 'inherit')";
$docs = $wpdb->get_results($query);
if (is_array($docs)) {
foreach ($docs as $doc) {
if (!isset($phrase_matches[$phrase])) {
$phrase_matches[$phrase] = array();
}
$phrase_matches[$phrase][] = $doc->ID;
}
}
}
if (count($phrase_matches) < 1) {
$phrases = "-1";
} else {
// Complicated mess, but necessary...
$i = 0;
$phms = array();
foreach ($phrase_matches as $phm) {
$phms[$i++] = $phm;
}
$phrases = $phms[0];
if ($i > 1) {
for ($i = 1; $i < count($phms); $i++) {
$phrases = array_intersect($phrases, $phms[$i]);
}
}
if (count($phrases) < 1) {
$phrases = "-1";
} else {
$phrases = implode(",", $phrases);
}
}
} else {
$phrases = false;
}
return $phrases;
}
function relevanssi_recognize_phrases($q)
{
global $wpdb;
$phrases = relevanssi_extract_phrases($q);
$all_queries = array();
if (count($phrases) > 0) {
foreach ($phrases as $phrase) {
$queries = array();
$phrase = esc_sql($phrase);
"on" == get_option("relevanssi_index_excerpt") ? $excerpt = " OR post_excerpt LIKE '%{$phrase}%'" : ($excerpt = "");
$query = "(SELECT ID FROM {$wpdb->posts} \n\t\t\t\tWHERE (post_content LIKE '%{$phrase}%' OR post_title LIKE '%{$phrase}%' {$excerpt})\n\t\t\t\tAND post_status IN ('publish', 'draft', 'private', 'pending', 'future', 'inherit'))";
$queries[] = $query;
$query = "(SELECT ID FROM {$wpdb->posts} as p, {$wpdb->term_relationships} as r, {$wpdb->term_taxonomy} as s, {$wpdb->terms} as t\n\t\t\t\tWHERE r.term_taxonomy_id = s.term_taxonomy_id AND s.term_id = t.term_id AND p.ID = r.object_id\n\t\t\t\tAND t.name LIKE '%{$phrase}%' AND p.post_status IN ('publish', 'draft', 'private', 'pending', 'future', 'inherit'))";
$queries[] = $query;
$query = "(SELECT ID\n FROM {$wpdb->posts} AS p, {$wpdb->postmeta} AS m\n WHERE p.ID = m.post_id\n AND m.meta_value LIKE '%{$phrase}%'\n AND p.post_status IN ('publish', 'draft', 'private', 'pending', 'future', 'inherit'))";
$queries[] = $query;
$queries = implode(' OR relevanssi.doc IN ', $queries);
$queries = "AND (relevanssi.doc IN {$queries})";
$all_queries[] = $queries;
}
} else {
$phrases = "";
}
$all_queries = implode(" ", $all_queries);
return $all_queries;
}
function relevanssi_highlight_terms($excerpt, $query)
{
$type = get_option("relevanssi_highlight");
if ("none" == $type) {
return $excerpt;
}
switch ($type) {
case "mark":
// thanks to Jeff Byrnes
$start_emp = "<mark>";
$end_emp = "</mark>";
break;
case "strong":
$start_emp = "<strong>";
$end_emp = "</strong>";
break;
case "em":
$start_emp = "<em>";
$end_emp = "</em>";
break;
case "col":
$col = get_option("relevanssi_txt_col");
if (!$col) {
$col = "#ff0000";
}
$start_emp = "<span style='color: {$col}'>";
$end_emp = "</span>";
break;
case "bgcol":
$col = get_option("relevanssi_bg_col");
if (!$col) {
$col = "#ff0000";
}
$start_emp = "<span style='background-color: {$col}'>";
$end_emp = "</span>";
break;
case "css":
$css = get_option("relevanssi_css");
if (!$css) {
$css = "color: #ff0000";
}
$start_emp = "<span style='{$css}'>";
$end_emp = "</span>";
break;
case "class":
$css = get_option("relevanssi_class");
if (!$css) {
$css = "relevanssi-query-term";
}
$start_emp = "<span class='{$css}'>";
$end_emp = "</span>";
break;
default:
return $excerpt;
}
$start_emp_token = "**[";
$end_emp_token = "]**";
if (function_exists('mb_internal_encoding')) {
mb_internal_encoding("UTF-8");
}
$terms = array_keys(relevanssi_tokenize($query, $remove_stopwords = true, $min_word_length = -1));
if (is_array($query)) {
$query = implode(' ', $query);
}
// just in case
$phrases = relevanssi_extract_phrases(stripslashes($query));
$non_phrase_terms = array();
foreach ($phrases as $phrase) {
$phrase_terms = array_keys(relevanssi_tokenize($phrase, $remove_stopwords = false));
foreach ($terms as $term) {
if (!in_array($term, $phrase_terms)) {
$non_phrase_terms[] = $term;
}
}
$terms = $non_phrase_terms;
$terms[] = $phrase;
}
uksort($terms, 'relevanssi_strlen_sort');
get_option('relevanssi_word_boundaries', 'on') == 'on' ? $word_boundaries = true : ($word_boundaries = false);
foreach ($terms as $term) {
$pr_term = relevanssi_replace_punctuation(preg_quote($term, '/'));
if ($word_boundaries) {
$excerpt = preg_replace("/(\\b{$pr_term}|{$pr_term}\\b)(?!(^&+)?(;))/iu", $start_emp_token . '\\1' . $end_emp_token, $excerpt);
} else {
$excerpt = preg_replace("/({$pr_term})(?!(^&+)?(;))/iu", $start_emp_token . '\\1' . $end_emp_token, $excerpt);
}
$preg_start = preg_quote($start_emp_token);
$preg_end = preg_quote($end_emp_token);
if (preg_match_all('/<.*>/U', $excerpt, $matches) > 0) {
// Remove highlights from inside HTML tags
foreach ($matches as $match) {
$new_match = str_replace($start_emp_token, '', $match);
$new_match = str_replace($end_emp_token, '', $new_match);
$excerpt = str_replace($match, $new_match, $excerpt);
}
}
if (preg_match_all('/<(style|script|object|embed)>.*<\\/(style|script|object|embed)>/U', $excerpt, $matches) > 0) {
// Remove highlights in style, object, embed and script tags
foreach ($matches as $match) {
$new_match = str_replace($start_emp_token, '', $match);
$new_match = str_replace($end_emp_token, '', $new_match);
$excerpt = str_replace($match, $new_match, $excerpt);
}
}
}
$excerpt = relevanssi_remove_nested_highlights($excerpt, $start_emp_token, $end_emp_token);
/*
$excerpt = htmlentities($excerpt, ENT_QUOTES, 'UTF-8');
// return the HTML entities that were stripped before
*/
$excerpt = str_replace($start_emp_token, $start_emp, $excerpt);
$excerpt = str_replace($end_emp_token, $end_emp, $excerpt);
$excerpt = str_replace($end_emp . $start_emp, "", $excerpt);
if (function_exists('mb_ereg_replace')) {
$pattern = $end_emp . '\\s*' . $start_emp;
$excerpt = mb_ereg_replace($pattern, " ", $excerpt);
}
return $excerpt;
}
function relevanssi_highlight_terms($excerpt, $query) {
$type = get_option("relevanssi_highlight");
if ("none" == $type) {
return $excerpt;
}
switch ($type) {
case "mark": // thanks to Jeff Byrnes
$start_emp = "<mark>";
$end_emp = "</mark>";
break;
case "strong":
$start_emp = "<strong>";
$end_emp = "</strong>";
break;
case "em":
$start_emp = "<em>";
$end_emp = "</em>";
break;
case "col":
$col = get_option("relevanssi_txt_col");
if (!$col) $col = "#ff0000";
$start_emp = "<span style='color: $col'>";
$end_emp = "</span>";
break;
case "bgcol":
$col = get_option("relevanssi_bg_col");
if (!$col) $col = "#ff0000";
$start_emp = "<span style='background-color: $col'>";
$end_emp = "</span>";
break;
case "css":
$css = get_option("relevanssi_css");
if (!$css) $css = "color: #ff0000";
$start_emp = "<span style='$css'>";
$end_emp = "</span>";
break;
case "class":
$css = get_option("relevanssi_class");
if (!$css) $css = "relevanssi-query-term";
$start_emp = "<span class='$css'>";
$end_emp = "</span>";
break;
default:
return $excerpt;
}
$start_emp_token = "*[/";
$end_emp_token = "\]*";
mb_internal_encoding("UTF-8");
$terms = array_keys(relevanssi_tokenize($query, false));
$phrases = relevanssi_extract_phrases(stripslashes($query));
$non_phrase_terms = array();
foreach ($phrases as $phrase) {
$phrase_terms = array_keys(relevanssi_tokenize($phrase, false));
foreach ($terms as $term) {
if (!in_array($term, $phrase_terms)) {
$non_phrase_terms[] = $term;
}
}
$terms = $non_phrase_terms;
$terms[] = $phrase;
}
usort($terms, 'relevanssi_strlen_sort');
foreach ($terms as $term) {
$excerpt = preg_replace("/(\b$term|$term\b)(?!([^<]+)?>)/iu", $start_emp_token . '\\1' . $end_emp_token, $excerpt);
// thanks to http://pureform.wordpress.com/2008/01/04/matching-a-word-characters-outside-of-html-tags/
}
$excerpt = relevanssi_remove_nested_highlights($excerpt, $start_emp_token, $end_emp_token);
$excerpt = str_replace($start_emp_token, $start_emp, $excerpt);
$excerpt = str_replace($end_emp_token, $end_emp, $excerpt);
$excerpt = str_replace($end_emp . $start_emp, "", $excerpt);
if (function_exists('mb_ereg_replace')) {
$pattern = $end_emp . '\s*' . $start_emp;
$excerpt = mb_ereg_replace($pattern, " ", $excerpt);
}
return $excerpt;
}
