3 declare(strict_types
=1);
5 namespace Shaarli\Bookmark
;
8 use Shaarli\Bookmark\Exception\BookmarkNotFoundException
;
13 * Perform search and filter operation on link data list.
18 * @var string permalinks.
20 public static $FILTER_HASH = 'permalink';
23 * @var string text search.
25 public static $FILTER_TEXT = 'fulltext';
28 * @var string tag filter.
30 public static $FILTER_TAG = 'tags';
33 * @var string filter by day.
35 public static $FILTER_DAY = 'FILTER_DAY';
38 * @var string filter by day.
40 public static $DEFAULT = 'NO_FILTER';
42 /** @var string Visibility: all */
43 public static $ALL = 'all';
45 /** @var string Visibility: public */
46 public static $PUBLIC = 'public';
48 /** @var string Visibility: private */
49 public static $PRIVATE = 'private';
52 * @var string Allowed characters for hashtags (regex syntax).
54 public static $HASHTAG_CHARS = '\p{Pc}\p{N}\p{L}\p{Mn}';
57 * @var Bookmark[] all available bookmarks.
62 * @param Bookmark[] $bookmarks initialization.
64 public function __construct($bookmarks)
66 $this->bookmarks
= $bookmarks;
70 * Filter bookmarks according to parameters.
72 * @param string $type Type of filter (eg. tags, permalink, etc.).
73 * @param mixed $request Filter content.
74 * @param bool $casesensitive Optional: Perform case sensitive filter if true.
75 * @param string $visibility Optional: return only all/private/public bookmarks
76 * @param bool $untaggedonly Optional: return only untagged bookmarks. Applies only if $type includes FILTER_TAG
78 * @return Bookmark[] filtered bookmark list.
80 * @throws BookmarkNotFoundException
82 public function filter(
85 bool $casesensitive = false,
86 string $visibility = 'all',
87 bool $untaggedonly = false
89 if (!in_array($visibility, ['all', 'public', 'private'])) {
94 case self
::$FILTER_HASH:
95 return $this->filterSmallHash($request);
96 case self
::$FILTER_TAG | self
::$FILTER_TEXT: // == "vuotext"
97 $noRequest = empty($request) || (empty($request[0]) && empty($request[1]));
100 return $this->filterUntagged($visibility);
102 return $this->noFilter($visibility);
105 $filtered = $this->filterUntagged($visibility);
107 $filtered = $this->bookmarks
;
109 if (!empty($request[0])) {
110 $filtered = (new BookmarkFilter($filtered))->filterTags($request[0], $casesensitive, $visibility);
112 if (!empty($request[1])) {
113 $filtered = (new BookmarkFilter($filtered))->filterFulltext($request[1], $visibility);
116 case self
::$FILTER_TEXT:
117 return $this->filterFulltext($request, $visibility);
118 case self
::$FILTER_TAG:
120 return $this->filterUntagged($visibility);
122 return $this->filterTags($request, $casesensitive, $visibility);
124 case self
::$FILTER_DAY:
125 return $this->filterDay($request, $visibility);
127 return $this->noFilter($visibility);
132 * Unknown filter, but handle private only.
134 * @param string $visibility Optional: return only all/private/public bookmarks
136 * @return Bookmark[] filtered bookmarks.
138 private function noFilter(string $visibility = 'all')
140 if ($visibility === 'all') {
141 return $this->bookmarks
;
145 foreach ($this->bookmarks
as $key => $value) {
146 if ($value->isPrivate() && $visibility === 'private') {
148 } elseif (!$value->isPrivate() && $visibility === 'public') {
157 * Returns the shaare corresponding to a smallHash.
159 * @param string $smallHash permalink hash.
161 * @return Bookmark[] $filtered array containing permalink data.
163 * @throws BookmarkNotFoundException if the smallhash doesn't match any link.
165 private function filterSmallHash(string $smallHash)
167 foreach ($this->bookmarks
as $key => $l) {
168 if ($smallHash == $l->getShortUrl()) {
169 // Yes, this is ugly and slow
174 throw new BookmarkNotFoundException();
178 * Returns the list of bookmarks corresponding to a full-text search
181 * - in the URLs, title and description;
182 * - are case-insensitive;
183 * - terms surrounded by quotes " are exact terms search.
184 * - terms starting with a dash - are excluded (except exact terms).
187 * print_r($mydb->filterFulltext('hollandais'));
189 * mb_convert_case($val, MB_CASE_LOWER, 'UTF-8')
190 * - allows to perform searches on Unicode text
191 * - see https://github.com/shaarli/Shaarli/issues/75 for examples
193 * @param string $searchterms search query.
194 * @param string $visibility Optional: return only all/private/public bookmarks.
196 * @return Bookmark[] search results.
198 private function filterFulltext(string $searchterms, string $visibility = 'all')
200 if (empty($searchterms)) {
201 return $this->noFilter($visibility);
205 $search = mb_convert_case(html_entity_decode($searchterms), MB_CASE_LOWER
, 'UTF-8');
206 $exactRegex = '/"([^"]+)"/';
207 // Retrieve exact search terms.
208 preg_match_all($exactRegex, $search, $exactSearch);
209 $exactSearch = array_values(array_filter($exactSearch[1]));
211 // Remove exact search terms to get AND terms search.
212 $explodedSearchAnd = explode(' ', trim(preg_replace($exactRegex, '', $search)));
213 $explodedSearchAnd = array_values(array_filter($explodedSearchAnd));
215 // Filter excluding terms and update andSearch.
216 $excludeSearch = array();
217 $andSearch = array();
218 foreach ($explodedSearchAnd as $needle) {
219 if ($needle[0] == '-' && strlen($needle) > 1) {
220 $excludeSearch[] = substr($needle, 1);
222 $andSearch[] = $needle;
226 // Iterate over every stored link.
227 foreach ($this->bookmarks
as $id => $link) {
228 // ignore non private bookmarks when 'privatonly' is on.
229 if ($visibility !== 'all') {
230 if (!$link->isPrivate() && $visibility === 'private') {
232 } elseif ($link->isPrivate() && $visibility === 'public') {
237 // Concatenate link fields to search across fields.
238 // Adds a '\' separator for exact search terms.
239 $content = mb_convert_case($link->getTitle(), MB_CASE_LOWER
, 'UTF-8') .'\\';
240 $content .= mb_convert_case($link->getDescription(), MB_CASE_LOWER
, 'UTF-8') .'\\';
241 $content .= mb_convert_case($link->getUrl(), MB_CASE_LOWER
, 'UTF-8') .'\\';
242 $content .= mb_convert_case($link->getTagsString(), MB_CASE_LOWER
, 'UTF-8') .'\\';
247 // First, we look for exact term search
248 for ($i = 0; $i < count($exactSearch) && $found; $i++
) {
249 $found = strpos($content, $exactSearch[$i]) !== false;
252 // Iterate over keywords, if keyword is not found,
253 // no need to check for the others. We want all or nothing.
254 for ($i = 0; $i < count($andSearch) && $found; $i++
) {
255 $found = strpos($content, $andSearch[$i]) !== false;
259 for ($i = 0; $i < count($excludeSearch) && $found; $i++
) {
260 $found = strpos($content, $excludeSearch[$i]) === false;
264 $filtered[$id] = $link;
272 * generate a regex fragment out of a tag
274 * @param string $tag to to generate regexs from. may start with '-' to negate, contain '*' as wildcard
276 * @return string generated regex fragment
278 private static function tag2regex(string $tag): string
281 if (!$len || $tag === "-" || $tag === "*") {
282 // nothing to search, return empty regex
285 if ($tag[0] === "-") {
287 $i = 1; // use offset to start after '-' character
288 $regex = '(?!'; // create negative lookahead
290 $i = 0; // start at first character
291 $regex = '(?='; // use positive lookahead
293 $regex .= '.*(?:^| )'; // before tag may only be a space or the beginning
294 // iterate over string, separating it into placeholder and content
295 for (; $i < $len; $i++
) {
296 if ($tag[$i] === '*') {
300 // regular characters
301 $offset = strpos($tag, '*', $i);
302 if ($offset === false) {
303 // no placeholder found, set offset to end of string
306 // subtract one, as we want to get before the placeholder or end of string
308 // we got a tag name that we want to search for. escape any regex characters to prevent conflicts.
309 $regex .= preg_quote(substr($tag, $i, $offset - $i +
1), '/');
314 $regex .= '(?:$| ))'; // after the tag may only be a space or the end
319 * Returns the list of bookmarks associated with a given list of tags
321 * You can specify one or more tags, separated by space or a comma, e.g.
322 * print_r($mydb->filterTags('linux programming'));
324 * @param string|array $tags list of tags, separated by commas or blank spaces if passed as string.
325 * @param bool $casesensitive ignore case if false.
326 * @param string $visibility Optional: return only all/private/public bookmarks.
328 * @return Bookmark[] filtered bookmarks.
330 public function filterTags($tags, bool $casesensitive = false, string $visibility = 'all')
332 // get single tags (we may get passed an array, even though the docs say different)
334 if (!is_array($tags)) {
335 // we got an input string, split tags
336 $inputTags = preg_split('/(?:\s+)|,/', $inputTags, -1, PREG_SPLIT_NO_EMPTY
);
339 if (!count($inputTags)) {
341 return $this->noFilter($visibility);
344 // If we only have public visibility, we can't look for hidden tags
345 if ($visibility === self
::$PUBLIC) {
346 $inputTags = array_values(array_filter($inputTags, function ($tag) {
347 return ! startsWith($tag, '.');
350 if (empty($inputTags)) {
355 // build regex from all tags
356 $re = '/^' . implode(array_map("self::tag2regex", $inputTags)) . '.*$/';
357 if (!$casesensitive) {
358 // make regex case insensitive
362 // create resulting array
365 // iterate over each link
366 foreach ($this->bookmarks
as $key => $link) {
367 // check level of visibility
368 // ignore non private bookmarks when 'privateonly' is on.
369 if ($visibility !== 'all') {
370 if (!$link->isPrivate() && $visibility === 'private') {
372 } elseif ($link->isPrivate() && $visibility === 'public') {
376 $search = $link->getTagsString(); // build search string, start with tags of current link
377 if (strlen(trim($link->getDescription())) && strpos($link->getDescription(), '#') !== false) {
378 // description given and at least one possible tag found
380 // find all tags in the form of #tag in the description
382 '/(?<![' . self
::$HASHTAG_CHARS . '])#([' . self
::$HASHTAG_CHARS . ']+?)\b/sm',
383 $link->getDescription(),
386 if (count($descTags[1])) {
387 // there were some tags in the description, add them to the search string
388 $search .= ' ' . implode(' ', $descTags[1]);
391 // match regular expression with search string
392 if (!preg_match($re, $search)) {
393 // this entry does _not_ match our regex
396 $filtered[$key] = $link;
402 * Return only bookmarks without any tag.
404 * @param string $visibility return only all/private/public bookmarks.
406 * @return Bookmark[] filtered bookmarks.
408 public function filterUntagged(string $visibility)
411 foreach ($this->bookmarks
as $key => $link) {
412 if ($visibility !== 'all') {
413 if (!$link->isPrivate() && $visibility === 'private') {
415 } elseif ($link->isPrivate() && $visibility === 'public') {
420 if (empty(trim($link->getTagsString()))) {
421 $filtered[$key] = $link;
429 * Returns the list of articles for a given day, chronologically sorted
431 * Day must be in the form 'YYYYMMDD' (e.g. '20120125'), e.g.
432 * print_r($mydb->filterDay('20120125'));
434 * @param string $day day to filter.
435 * @param string $visibility return only all/private/public bookmarks.
437 * @return Bookmark[] all link matching given day.
439 * @throws Exception if date format is invalid.
441 public function filterDay(string $day, string $visibility)
443 if (!checkDateFormat('Ymd', $day)) {
444 throw new Exception('Invalid date format');
448 foreach ($this->bookmarks
as $key => $bookmark) {
449 if ($visibility === static::$PUBLIC && $bookmark->isPrivate()) {
453 if ($bookmark->getCreated()->format('Ymd') == $day) {
454 $filtered[$key] = $bookmark;
459 return array_reverse($filtered, true);
463 * Convert a list of tags (str) to an array. Also
464 * - handle case sensitivity.
465 * - accepts spaces commas as separator.
467 * @param string $tags string containing a list of tags.
468 * @param bool $casesensitive will convert everything to lowercase if false.
470 * @return string[] filtered tags string.
472 public static function tagsStrToArray(string $tags, bool $casesensitive): array
474 // We use UTF-8 conversion to handle various graphemes (i.e. cyrillic, or greek)
475 $tagsOut = $casesensitive ? $tags : mb_convert_case($tags, MB_CASE_LOWER
, 'UTF-8');
476 $tagsOut = str_replace(',', ' ', $tagsOut);
478 return preg_split('/\s+/', $tagsOut, -1, PREG_SPLIT_NO_EMPTY
);