/**
* Returns the pages that use a given media file
*
* Does a quick lookup with the fulltext index, then
* evaluates the instructions of the found pages
*
* Aborts after $max found results
*/
function ft_mediause($id, $max)
{
if (!$max) {
$max = 1;
}
// need to find at least one
$result = array();
// quick lookup of the mediafile
// FIXME use metadata key lookup
$media = noNS($id);
$matches = idx_lookup(idx_tokenizer($media));
$docs = array_keys(ft_resultCombine(array_values($matches)));
if (!count($docs)) {
return $result;
}
// go through all found pages
$found = 0;
$pcre = preg_quote($media, '/');
foreach ($docs as $doc) {
$ns = getNS($doc);
preg_match_all('/\\{\\{([^|}]*' . $pcre . '[^|}]*)(|[^}]+)?\\}\\}/i', rawWiki($doc), $matches);
foreach ($matches[1] as $img) {
$img = trim($img);
if (preg_match('/^https?:\\/\\//i', $img)) {
continue;
}
// skip external images
list($img) = explode('?', $img);
// remove any parameters
resolve_mediaid($ns, $img, $exists);
// resolve the possibly relative img
if ($img == $id) {
// we have a match
$result[] = $doc;
$found++;
break;
}
}
if ($found >= $max) {
break;
}
}
sort($result);
return $result;
}
/**
* Transforms given search term into intermediate representation
*
* This function is used in ft_queryParser() and not for general purpose use.
*
* @author Kazutaka Miyasaka <*****@*****.**>
*/
function ft_termParser($term, &$stopwords, $consider_asian = true, $phrase_mode = false)
{
$parsed = '';
if ($consider_asian) {
// successive asian characters need to be searched as a phrase
$words = preg_split('/(' . IDX_ASIAN . '+)/u', $term, -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY);
foreach ($words as $word) {
if (preg_match('/' . IDX_ASIAN . '/u', $word)) {
$phrase_mode = true;
}
$parsed .= ft_termParser($word, $stopwords, false, $phrase_mode);
}
} else {
$term_noparen = str_replace(array('(', ')'), ' ', $term);
$words = idx_tokenizer($term_noparen, $stopwords, true);
// W_: no need to highlight
if (empty($words)) {
$parsed = '()';
// important: do not remove
} elseif ($words[0] === $term) {
$parsed = '(W+:' . $words[0] . ')';
} elseif ($phrase_mode) {
$term_encoded = str_replace(array('(', ')'), array('OP', 'CP'), $term);
$parsed = '((W_:' . implode(')(W_:', $words) . ')(P+:' . $term_encoded . '))';
} else {
$parsed = '((W+:' . implode(')(W+:', $words) . '))';
}
}
return $parsed;
}
/**
* Split a page into words
*
* Returns an array of word counts, false if an error occurred.
* Array is keyed on the word length, then the word index.
*
* @author Andreas Gohr <*****@*****.**>
* @author Christopher Smith <*****@*****.**>
*/
function idx_getPageWords($page)
{
global $conf;
$swfile = DOKU_INC . 'inc/lang/' . $conf['lang'] . '/stopwords.txt';
if (@file_exists($swfile)) {
$stopwords = file($swfile);
} else {
$stopwords = array();
}
$body = '';
$data = array($page, $body);
$evt = new Doku_Event('INDEXER_PAGE_ADD', $data);
if ($evt->advise_before()) {
$data[1] .= rawWiki($page);
}
$evt->advise_after();
unset($evt);
list($page, $body) = $data;
$body = strtr($body, "\r\n\t", ' ');
$tokens = explode(' ', $body);
$tokens = array_count_values($tokens);
// count the frequency of each token
// ensure the deaccented or romanised page names of internal links are added to the token array
// (this is necessary for the backlink function -- there maybe a better way!)
if ($conf['deaccent']) {
$links = p_get_metadata($page, 'relation references');
if (!empty($links)) {
$tmp = join(' ', array_keys($links));
// make a single string
$tmp = strtr($tmp, ':', ' ');
// replace namespace separator with a space
$link_tokens = array_unique(explode(' ', $tmp));
// break into tokens
foreach ($link_tokens as $link_token) {
if (isset($tokens[$link_token])) {
continue;
}
$tokens[$link_token] = 1;
}
}
}
$words = array();
foreach ($tokens as $word => $count) {
$arr = idx_tokenizer($word, $stopwords);
$arr = array_count_values($arr);
foreach ($arr as $w => $c) {
$l = wordlen($w);
if (isset($words[$l])) {
$words[$l][$w] = $c * $count + (isset($words[$l][$w]) ? $words[$l][$w] : 0);
} else {
$words[$l] = array($w => $c * $count);
}
}
}
// arrive here with $words = array(wordlen => array(word => frequency))
$index = array();
//resulting index
foreach (array_keys($words) as $wlen) {
$word_idx = idx_getIndex('w', $wlen);
foreach ($words[$wlen] as $word => $freq) {
$wid = array_search("{$word}\n", $word_idx);
if (!is_int($wid)) {
$wid = count($word_idx);
$word_idx[] = "{$word}\n";
}
if (!isset($index[$wlen])) {
$index[$wlen] = array();
}
$index[$wlen][$wid] = $freq;
}
// save back word index
if (!idx_saveIndex('w', $wlen, $word_idx)) {
trigger_error("Failed to write word index", E_USER_ERROR);
return false;
}
}
return $index;
}
/**
* Builds an array of search words from a query
*
* @todo support OR and parenthesises?
*/
function ft_queryParser($query)
{
global $conf;
$swfile = DOKU_INC . 'inc/lang/' . $conf['lang'] . '/stopwords.txt';
if (@file_exists($swfile)) {
$stopwords = file($swfile);
} else {
$stopwords = array();
}
$q = array();
$q['query'] = $query;
$q['ns'] = array();
$q['phrases'] = array();
$q['words'] = array();
$q['and'] = array();
$q['not'] = array();
// strip namespace from query
if (preg_match('/([^@]*)@(.*)/', $query, $match)) {
$query = $match[1];
$q['ns'] = explode('@', preg_replace("/ /", '', $match[2]));
}
// handle phrase searches
while (preg_match('/"(.*?)"/', $query, $match)) {
$q['phrases'][] = $match[1];
$q['and'] = array_merge($q['and'], idx_tokenizer($match[0], $stopwords));
$query = preg_replace('/"(.*?)"/', '', $query, 1);
}
$words = explode(' ', $query);
foreach ($words as $w) {
if ($w[0] == '-') {
$token = idx_tokenizer($w, $stopwords, true);
if (count($token)) {
$q['not'] = array_merge($q['not'], $token);
}
} else {
// asian "words" need to be searched as phrases
if (@preg_match_all('/((' . IDX_ASIAN . ')+)/u', $w, $matches)) {
$q['phrases'] = array_merge($q['phrases'], $matches[1]);
}
$token = idx_tokenizer($w, $stopwords, true);
if (count($token)) {
$q['and'] = array_merge($q['and'], $token);
$q['words'] = array_merge($q['words'], $token);
}
}
}
return $q;
}
