]>
git.immae.eu Git - github/shaarli/Shaarli.git/blob - application/LinkFilter.php
6 * Perform search and filter operation on link data list.
11 * @var string permalinks.
13 public static $FILTER_HASH = 'permalink';
16 * @var string text search.
18 public static $FILTER_TEXT = 'fulltext';
21 * @var string tag filter.
23 public static $FILTER_TAG = 'tags';
26 * @var string filter by day.
28 public static $FILTER_DAY = 'FILTER_DAY';
31 * @var string Allowed characters for hashtags (regex syntax).
33 public static $HASHTAG_CHARS = '\p{Pc}\p{N}\p{L}\p{Mn}';
36 * @var LinkDB all available links.
41 * @param LinkDB $links initialization.
43 public function __construct($links)
45 $this->links
= $links;
49 * Filter links according to parameters.
51 * @param string $type Type of filter (eg. tags, permalink, etc.).
52 * @param mixed $request Filter content.
53 * @param bool $casesensitive Optional: Perform case sensitive filter if true.
54 * @param string $visibility Optional: return only all/private/public links
55 * @param string $untaggedonly Optional: return only untagged links. Applies only if $type includes FILTER_TAG
57 * @return array filtered link list.
59 public function filter($type, $request, $casesensitive = false, $visibility = 'all', $untaggedonly = false)
61 if (! in_array($visibility, ['all', 'public', 'private'])) {
66 case self
::$FILTER_HASH:
67 return $this->filterSmallHash($request);
68 case self
::$FILTER_TAG | self
::$FILTER_TEXT: // == "vuotext"
69 $noRequest = empty($request) || (empty($request[0]) && empty($request[1]));
72 return $this->filterUntagged($visibility);
74 return $this->noFilter($visibility);
77 $filtered = $this->filterUntagged($visibility);
79 $filtered = $this->links
;
81 if (!empty($request[0])) {
82 $filtered = (new LinkFilter($filtered))->filterTags($request[0], $casesensitive, $visibility);
84 if (!empty($request[1])) {
85 $filtered = (new LinkFilter($filtered))->filterFulltext($request[1], $visibility);
88 case self
::$FILTER_TEXT:
89 return $this->filterFulltext($request, $visibility);
90 case self
::$FILTER_TAG:
92 return $this->filterUntagged($visibility);
94 return $this->filterTags($request, $casesensitive, $visibility);
96 case self
::$FILTER_DAY:
97 return $this->filterDay($request);
99 return $this->noFilter($visibility);
104 * Unknown filter, but handle private only.
106 * @param string $visibility Optional: return only all/private/public links
108 * @return array filtered links.
110 private function noFilter($visibility = 'all')
112 if ($visibility === 'all') {
117 foreach ($this->links
as $key => $value) {
118 if ($value['private'] && $visibility === 'private') {
120 } else if (! $value['private'] && $visibility === 'public') {
129 * Returns the shaare corresponding to a smallHash.
131 * @param string $smallHash permalink hash.
133 * @return array $filtered array containing permalink data.
135 * @throws LinkNotFoundException if the smallhash doesn't match any link.
137 private function filterSmallHash($smallHash)
140 foreach ($this->links
as $key => $l) {
141 if ($smallHash == $l['shorturl']) {
142 // Yes, this is ugly and slow
143 $filtered[$key] = $l;
148 if (empty($filtered)) {
149 throw new LinkNotFoundException();
156 * Returns the list of links corresponding to a full-text search
159 * - in the URLs, title and description;
160 * - are case-insensitive;
161 * - terms surrounded by quotes " are exact terms search.
162 * - terms starting with a dash - are excluded (except exact terms).
165 * print_r($mydb->filterFulltext('hollandais'));
167 * mb_convert_case($val, MB_CASE_LOWER, 'UTF-8')
168 * - allows to perform searches on Unicode text
169 * - see https://github.com/shaarli/Shaarli/issues/75 for examples
171 * @param string $searchterms search query.
172 * @param string $visibility Optional: return only all/private/public links.
174 * @return array search results.
176 private function filterFulltext($searchterms, $visibility = 'all')
178 if (empty($searchterms)) {
179 return $this->noFilter($visibility);
183 $search = mb_convert_case(html_entity_decode($searchterms), MB_CASE_LOWER
, 'UTF-8');
184 $exactRegex = '/"([^"]+)"/';
185 // Retrieve exact search terms.
186 preg_match_all($exactRegex, $search, $exactSearch);
187 $exactSearch = array_values(array_filter($exactSearch[1]));
189 // Remove exact search terms to get AND terms search.
190 $explodedSearchAnd = explode(' ', trim(preg_replace($exactRegex, '', $search)));
191 $explodedSearchAnd = array_values(array_filter($explodedSearchAnd));
193 // Filter excluding terms and update andSearch.
194 $excludeSearch = array();
195 $andSearch = array();
196 foreach ($explodedSearchAnd as $needle) {
197 if ($needle[0] == '-' && strlen($needle) > 1) {
198 $excludeSearch[] = substr($needle, 1);
200 $andSearch[] = $needle;
204 $keys = array('title', 'description', 'url', 'tags');
206 // Iterate over every stored link.
207 foreach ($this->links
as $id => $link) {
209 // ignore non private links when 'privatonly' is on.
210 if ($visibility !== 'all') {
211 if (! $link['private'] && $visibility === 'private') {
213 } else if ($link['private'] && $visibility === 'public') {
218 // Concatenate link fields to search across fields.
219 // Adds a '\' separator for exact search terms.
221 foreach ($keys as $key) {
222 $content .= mb_convert_case($link[$key], MB_CASE_LOWER
, 'UTF-8') . '\\';
228 // First, we look for exact term search
229 for ($i = 0; $i < count($exactSearch) && $found; $i++
) {
230 $found = strpos($content, $exactSearch[$i]) !== false;
233 // Iterate over keywords, if keyword is not found,
234 // no need to check for the others. We want all or nothing.
235 for ($i = 0; $i < count($andSearch) && $found; $i++
) {
236 $found = strpos($content, $andSearch[$i]) !== false;
240 for ($i = 0; $i < count($excludeSearch) && $found; $i++
) {
241 $found = strpos($content, $excludeSearch[$i]) === false;
245 $filtered[$id] = $link;
253 * generate a regex fragment out of a tag
254 * @param string $tag to to generate regexs from. may start with '-' to negate, contain '*' as wildcard
255 * @return string generated regex fragment
257 private static function tag2regex($tag)
260 if(!$len || $tag === "-" || $tag === "*"){
261 // nothing to search, return empty regex
264 if($tag[0] === "-") {
266 $i = 1; // use offset to start after '-' character
267 $regex = '(?!'; // create negative lookahead
269 $i = 0; // start at first character
270 $regex = '(?='; // use positive lookahead
272 $regex .= '.*(?:^| )'; // before tag may only be a space or the beginning
273 // iterate over string, separating it into placeholder and content
274 for(; $i < $len; $i++
){
275 if($tag[$i] === '*'){
279 // regular characters
280 $offset = strpos($tag, '*', $i);
281 if($offset === false){
282 // no placeholder found, set offset to end of string
285 // subtract one, as we want to get before the placeholder or end of string
287 // we got a tag name that we want to search for. escape any regex characters to prevent conflicts.
288 $regex .= preg_quote(substr($tag, $i, $offset - $i +
1), '/');
293 $regex .= '(?:$| ))'; // after the tag may only be a space or the end
298 * Returns the list of links associated with a given list of tags
300 * You can specify one or more tags, separated by space or a comma, e.g.
301 * print_r($mydb->filterTags('linux programming'));
303 * @param string $tags list of tags separated by commas or blank spaces.
304 * @param bool $casesensitive ignore case if false.
305 * @param string $visibility Optional: return only all/private/public links.
307 * @return array filtered links.
309 public function filterTags($tags, $casesensitive = false, $visibility = 'all')
311 // get single tags (we may get passed an array, even though the docs say different)
313 if(!is_array($tags)) {
314 // we got an input string, split tags
315 $inputTags = preg_split('/(?:\s+)|,/', $inputTags, -1, PREG_SPLIT_NO_EMPTY
);
318 if(!count($inputTags)){
320 return $this->noFilter($visibility);
323 // build regex from all tags
324 $re = '/^' . implode(array_map("self::tag2regex", $inputTags)) . '.*$/';
325 if(!$casesensitive) {
326 // make regex case insensitive
330 // create resulting array
333 // iterate over each link
334 foreach ($this->links
as $key => $link) {
335 // check level of visibility
336 // ignore non private links when 'privateonly' is on.
337 if ($visibility !== 'all') {
338 if (! $link['private'] && $visibility === 'private') {
340 } else if ($link['private'] && $visibility === 'public') {
344 $search = $link['tags']; // build search string, start with tags of current link
345 if(strlen(trim($link['description'])) && strpos($link['description'], '#') !== false){
346 // description given and at least one possible tag found
348 // find all tags in the form of #tag in the description
350 '/(?<![' . self
::$HASHTAG_CHARS . '])#([' . self
::$HASHTAG_CHARS . ']+?)\b/sm',
351 $link['description'],
354 if(count($descTags[1])){
355 // there were some tags in the description, add them to the search string
356 $search .= ' ' . implode(' ', $descTags[1]);
359 // match regular expression with search string
360 if(!preg_match($re, $search)){
361 // this entry does _not_ match our regex
364 $filtered[$key] = $link;
370 * Return only links without any tag.
372 * @param string $visibility return only all/private/public links.
374 * @return array filtered links.
376 public function filterUntagged($visibility)
379 foreach ($this->links
as $key => $link) {
380 if ($visibility !== 'all') {
381 if (! $link['private'] && $visibility === 'private') {
383 } else if ($link['private'] && $visibility === 'public') {
388 if (empty(trim($link['tags']))) {
389 $filtered[$key] = $link;
397 * Returns the list of articles for a given day, chronologically sorted
399 * Day must be in the form 'YYYYMMDD' (e.g. '20120125'), e.g.
400 * print_r($mydb->filterDay('20120125'));
402 * @param string $day day to filter.
404 * @return array all link matching given day.
406 * @throws Exception if date format is invalid.
408 public function filterDay($day)
410 if (! checkDateFormat('Ymd', $day)) {
411 throw new Exception('Invalid date format');
415 foreach ($this->links
as $key => $l) {
416 if ($l['created']->format('Ymd') == $day) {
417 $filtered[$key] = $l;
422 return array_reverse($filtered, true);
426 * Convert a list of tags (str) to an array. Also
427 * - handle case sensitivity.
428 * - accepts spaces commas as separator.
430 * @param string $tags string containing a list of tags.
431 * @param bool $casesensitive will convert everything to lowercase if false.
433 * @return array filtered tags string.
435 public static function tagsStrToArray($tags, $casesensitive)
437 // We use UTF-8 conversion to handle various graphemes (i.e. cyrillic, or greek)
438 $tagsOut = $casesensitive ? $tags : mb_convert_case($tags, MB_CASE_LOWER
, 'UTF-8');
439 $tagsOut = str_replace(',', ' ', $tagsOut);
441 return preg_split('/\s+/', $tagsOut, -1, PREG_SPLIT_NO_EMPTY
);
445 class LinkNotFoundException
extends Exception
448 * LinkNotFoundException constructor.
450 public function __construct()
452 $this->message
= t('The link you are trying to reach does not exist or has been deleted.');