]>
Commit | Line | Data |
---|---|---|
822bffce A |
1 | <?php |
2 | ||
6696729b V |
3 | namespace Shaarli\Bookmark; |
4 | ||
5 | use Exception; | |
336a28fa | 6 | use Shaarli\Bookmark\Exception\BookmarkNotFoundException; |
f24896b2 | 7 | |
822bffce A |
8 | /** |
9 | * Class LinkFilter. | |
10 | * | |
11 | * Perform search and filter operation on link data list. | |
12 | */ | |
336a28fa | 13 | class BookmarkFilter |
822bffce A |
14 | { |
15 | /** | |
16 | * @var string permalinks. | |
17 | */ | |
6696729b | 18 | public static $FILTER_HASH = 'permalink'; |
822bffce A |
19 | |
20 | /** | |
21 | * @var string text search. | |
22 | */ | |
6696729b | 23 | public static $FILTER_TEXT = 'fulltext'; |
822bffce A |
24 | |
25 | /** | |
26 | * @var string tag filter. | |
27 | */ | |
6696729b | 28 | public static $FILTER_TAG = 'tags'; |
822bffce A |
29 | |
30 | /** | |
31 | * @var string filter by day. | |
32 | */ | |
6696729b | 33 | public static $FILTER_DAY = 'FILTER_DAY'; |
822bffce | 34 | |
336a28fa A |
35 | /** |
36 | * @var string filter by day. | |
37 | */ | |
38 | public static $DEFAULT = 'NO_FILTER'; | |
39 | ||
40 | /** @var string Visibility: all */ | |
41 | public static $ALL = 'all'; | |
42 | ||
43 | /** @var string Visibility: public */ | |
44 | public static $PUBLIC = 'public'; | |
45 | ||
46 | /** @var string Visibility: private */ | |
47 | public static $PRIVATE = 'private'; | |
48 | ||
9ccca401 A |
49 | /** |
50 | * @var string Allowed characters for hashtags (regex syntax). | |
51 | */ | |
52 | public static $HASHTAG_CHARS = '\p{Pc}\p{N}\p{L}\p{Mn}'; | |
53 | ||
822bffce | 54 | /** |
336a28fa | 55 | * @var Bookmark[] all available bookmarks. |
822bffce | 56 | */ |
336a28fa | 57 | private $bookmarks; |
822bffce A |
58 | |
59 | /** | |
336a28fa | 60 | * @param Bookmark[] $bookmarks initialization. |
822bffce | 61 | */ |
336a28fa | 62 | public function __construct($bookmarks) |
822bffce | 63 | { |
336a28fa | 64 | $this->bookmarks = $bookmarks; |
822bffce A |
65 | } |
66 | ||
67 | /** | |
336a28fa | 68 | * Filter bookmarks according to parameters. |
822bffce A |
69 | * |
70 | * @param string $type Type of filter (eg. tags, permalink, etc.). | |
528a6f8a | 71 | * @param mixed $request Filter content. |
822bffce | 72 | * @param bool $casesensitive Optional: Perform case sensitive filter if true. |
336a28fa A |
73 | * @param string $visibility Optional: return only all/private/public bookmarks |
74 | * @param bool $untaggedonly Optional: return only untagged bookmarks. Applies only if $type includes FILTER_TAG | |
75 | * | |
76 | * @return Bookmark[] filtered bookmark list. | |
822bffce | 77 | * |
336a28fa | 78 | * @throws BookmarkNotFoundException |
822bffce | 79 | */ |
f210d94f | 80 | public function filter($type, $request, $casesensitive = false, $visibility = 'all', $untaggedonly = false) |
822bffce | 81 | { |
6696729b | 82 | if (!in_array($visibility, ['all', 'public', 'private'])) { |
7f96d9ec A |
83 | $visibility = 'all'; |
84 | } | |
85 | ||
f211e417 | 86 | switch ($type) { |
822bffce A |
87 | case self::$FILTER_HASH: |
88 | return $this->filterSmallHash($request); | |
f210d94f LC |
89 | case self::$FILTER_TAG | self::$FILTER_TEXT: // == "vuotext" |
90 | $noRequest = empty($request) || (empty($request[0]) && empty($request[1])); | |
91 | if ($noRequest) { | |
92 | if ($untaggedonly) { | |
93 | return $this->filterUntagged($visibility); | |
c51fae92 | 94 | } |
f210d94f | 95 | return $this->noFilter($visibility); |
c51fae92 | 96 | } |
f210d94f LC |
97 | if ($untaggedonly) { |
98 | $filtered = $this->filterUntagged($visibility); | |
99 | } else { | |
336a28fa | 100 | $filtered = $this->bookmarks; |
f210d94f LC |
101 | } |
102 | if (!empty($request[0])) { | |
336a28fa | 103 | $filtered = (new BookmarkFilter($filtered))->filterTags($request[0], $casesensitive, $visibility); |
f210d94f LC |
104 | } |
105 | if (!empty($request[1])) { | |
336a28fa | 106 | $filtered = (new BookmarkFilter($filtered))->filterFulltext($request[1], $visibility); |
f210d94f LC |
107 | } |
108 | return $filtered; | |
822bffce | 109 | case self::$FILTER_TEXT: |
7f96d9ec | 110 | return $this->filterFulltext($request, $visibility); |
822bffce | 111 | case self::$FILTER_TAG: |
f210d94f LC |
112 | if ($untaggedonly) { |
113 | return $this->filterUntagged($visibility); | |
114 | } else { | |
115 | return $this->filterTags($request, $casesensitive, $visibility); | |
116 | } | |
822bffce A |
117 | case self::$FILTER_DAY: |
118 | return $this->filterDay($request); | |
822bffce | 119 | default: |
7f96d9ec | 120 | return $this->noFilter($visibility); |
822bffce A |
121 | } |
122 | } | |
123 | ||
124 | /** | |
125 | * Unknown filter, but handle private only. | |
126 | * | |
336a28fa | 127 | * @param string $visibility Optional: return only all/private/public bookmarks |
822bffce | 128 | * |
336a28fa | 129 | * @return Bookmark[] filtered bookmarks. |
822bffce | 130 | */ |
7f96d9ec | 131 | private function noFilter($visibility = 'all') |
822bffce | 132 | { |
7f96d9ec | 133 | if ($visibility === 'all') { |
336a28fa | 134 | return $this->bookmarks; |
822bffce A |
135 | } |
136 | ||
137 | $out = array(); | |
336a28fa A |
138 | foreach ($this->bookmarks as $key => $value) { |
139 | if ($value->isPrivate() && $visibility === 'private') { | |
7f96d9ec | 140 | $out[$key] = $value; |
336a28fa | 141 | } elseif (!$value->isPrivate() && $visibility === 'public') { |
01878a75 | 142 | $out[$key] = $value; |
822bffce A |
143 | } |
144 | } | |
145 | ||
822bffce A |
146 | return $out; |
147 | } | |
148 | ||
149 | /** | |
150 | * Returns the shaare corresponding to a smallHash. | |
151 | * | |
152 | * @param string $smallHash permalink hash. | |
153 | * | |
154 | * @return array $filtered array containing permalink data. | |
528a6f8a | 155 | * |
336a28fa | 156 | * @throws \Shaarli\Bookmark\Exception\BookmarkNotFoundException if the smallhash doesn't match any link. |
822bffce A |
157 | */ |
158 | private function filterSmallHash($smallHash) | |
159 | { | |
336a28fa A |
160 | foreach ($this->bookmarks as $key => $l) { |
161 | if ($smallHash == $l->getShortUrl()) { | |
822bffce | 162 | // Yes, this is ugly and slow |
336a28fa | 163 | return [$key => $l]; |
822bffce A |
164 | } |
165 | } | |
528a6f8a | 166 | |
336a28fa | 167 | throw new BookmarkNotFoundException(); |
822bffce A |
168 | } |
169 | ||
170 | /** | |
336a28fa | 171 | * Returns the list of bookmarks corresponding to a full-text search |
822bffce A |
172 | * |
173 | * Searches: | |
174 | * - in the URLs, title and description; | |
bedd176a A |
175 | * - are case-insensitive; |
176 | * - terms surrounded by quotes " are exact terms search. | |
177 | * - terms starting with a dash - are excluded (except exact terms). | |
822bffce A |
178 | * |
179 | * Example: | |
180 | * print_r($mydb->filterFulltext('hollandais')); | |
181 | * | |
182 | * mb_convert_case($val, MB_CASE_LOWER, 'UTF-8') | |
183 | * - allows to perform searches on Unicode text | |
184 | * - see https://github.com/shaarli/Shaarli/issues/75 for examples | |
185 | * | |
186 | * @param string $searchterms search query. | |
336a28fa | 187 | * @param string $visibility Optional: return only all/private/public bookmarks. |
822bffce A |
188 | * |
189 | * @return array search results. | |
190 | */ | |
7f96d9ec | 191 | private function filterFulltext($searchterms, $visibility = 'all') |
822bffce | 192 | { |
c51fae92 | 193 | if (empty($searchterms)) { |
7f96d9ec | 194 | return $this->noFilter($visibility); |
c51fae92 A |
195 | } |
196 | ||
522b278b | 197 | $filtered = array(); |
ebd8075a | 198 | $search = mb_convert_case(html_entity_decode($searchterms), MB_CASE_LOWER, 'UTF-8'); |
bedd176a A |
199 | $exactRegex = '/"([^"]+)"/'; |
200 | // Retrieve exact search terms. | |
201 | preg_match_all($exactRegex, $search, $exactSearch); | |
202 | $exactSearch = array_values(array_filter($exactSearch[1])); | |
203 | ||
204 | // Remove exact search terms to get AND terms search. | |
205 | $explodedSearchAnd = explode(' ', trim(preg_replace($exactRegex, '', $search))); | |
206 | $explodedSearchAnd = array_values(array_filter($explodedSearchAnd)); | |
207 | ||
208 | // Filter excluding terms and update andSearch. | |
209 | $excludeSearch = array(); | |
210 | $andSearch = array(); | |
211 | foreach ($explodedSearchAnd as $needle) { | |
212 | if ($needle[0] == '-' && strlen($needle) > 1) { | |
213 | $excludeSearch[] = substr($needle, 1); | |
214 | } else { | |
215 | $andSearch[] = $needle; | |
216 | } | |
217 | } | |
218 | ||
822bffce | 219 | // Iterate over every stored link. |
336a28fa A |
220 | foreach ($this->bookmarks as $id => $link) { |
221 | // ignore non private bookmarks when 'privatonly' is on. | |
7f96d9ec | 222 | if ($visibility !== 'all') { |
336a28fa | 223 | if (!$link->isPrivate() && $visibility === 'private') { |
7f96d9ec | 224 | continue; |
336a28fa | 225 | } elseif ($link->isPrivate() && $visibility === 'public') { |
7f96d9ec A |
226 | continue; |
227 | } | |
822bffce A |
228 | } |
229 | ||
522b278b A |
230 | // Concatenate link fields to search across fields. |
231 | // Adds a '\' separator for exact search terms. | |
336a28fa A |
232 | $content = mb_convert_case($link->getTitle(), MB_CASE_LOWER, 'UTF-8') .'\\'; |
233 | $content .= mb_convert_case($link->getDescription(), MB_CASE_LOWER, 'UTF-8') .'\\'; | |
234 | $content .= mb_convert_case($link->getUrl(), MB_CASE_LOWER, 'UTF-8') .'\\'; | |
235 | $content .= mb_convert_case($link->getTagsString(), MB_CASE_LOWER, 'UTF-8') .'\\'; | |
522b278b A |
236 | |
237 | // Be optimistic | |
238 | $found = true; | |
239 | ||
240 | // First, we look for exact term search | |
241 | for ($i = 0; $i < count($exactSearch) && $found; $i++) { | |
242 | $found = strpos($content, $exactSearch[$i]) !== false; | |
243 | } | |
244 | ||
245 | // Iterate over keywords, if keyword is not found, | |
246 | // no need to check for the others. We want all or nothing. | |
247 | for ($i = 0; $i < count($andSearch) && $found; $i++) { | |
248 | $found = strpos($content, $andSearch[$i]) !== false; | |
249 | } | |
250 | ||
251 | // Exclude terms. | |
252 | for ($i = 0; $i < count($excludeSearch) && $found; $i++) { | |
253 | $found = strpos($content, $excludeSearch[$i]) === false; | |
254 | } | |
255 | ||
822bffce | 256 | if ($found) { |
01878a75 | 257 | $filtered[$id] = $link; |
822bffce A |
258 | } |
259 | } | |
260 | ||
822bffce A |
261 | return $filtered; |
262 | } | |
263 | ||
341527ba WE |
264 | /** |
265 | * generate a regex fragment out of a tag | |
6696729b | 266 | * |
341527ba | 267 | * @param string $tag to to generate regexs from. may start with '-' to negate, contain '*' as wildcard |
6696729b | 268 | * |
341527ba WE |
269 | * @return string generated regex fragment |
270 | */ | |
271 | private static function tag2regex($tag) | |
272 | { | |
273 | $len = strlen($tag); | |
f211e417 | 274 | if (!$len || $tag === "-" || $tag === "*") { |
341527ba WE |
275 | // nothing to search, return empty regex |
276 | return ''; | |
277 | } | |
f211e417 | 278 | if ($tag[0] === "-") { |
341527ba WE |
279 | // query is negated |
280 | $i = 1; // use offset to start after '-' character | |
281 | $regex = '(?!'; // create negative lookahead | |
282 | } else { | |
283 | $i = 0; // start at first character | |
284 | $regex = '(?='; // use positive lookahead | |
285 | } | |
286 | $regex .= '.*(?:^| )'; // before tag may only be a space or the beginning | |
287 | // iterate over string, separating it into placeholder and content | |
f211e417 V |
288 | for (; $i < $len; $i++) { |
289 | if ($tag[$i] === '*') { | |
341527ba WE |
290 | // placeholder found |
291 | $regex .= '[^ ]*?'; | |
292 | } else { | |
293 | // regular characters | |
294 | $offset = strpos($tag, '*', $i); | |
f211e417 | 295 | if ($offset === false) { |
341527ba WE |
296 | // no placeholder found, set offset to end of string |
297 | $offset = $len; | |
298 | } | |
299 | // subtract one, as we want to get before the placeholder or end of string | |
300 | $offset -= 1; | |
301 | // we got a tag name that we want to search for. escape any regex characters to prevent conflicts. | |
302 | $regex .= preg_quote(substr($tag, $i, $offset - $i + 1), '/'); | |
303 | // move $i on | |
304 | $i = $offset; | |
305 | } | |
306 | } | |
307 | $regex .= '(?:$| ))'; // after the tag may only be a space or the end | |
308 | return $regex; | |
309 | } | |
310 | ||
822bffce | 311 | /** |
336a28fa | 312 | * Returns the list of bookmarks associated with a given list of tags |
822bffce A |
313 | * |
314 | * You can specify one or more tags, separated by space or a comma, e.g. | |
315 | * print_r($mydb->filterTags('linux programming')); | |
316 | * | |
317 | * @param string $tags list of tags separated by commas or blank spaces. | |
318 | * @param bool $casesensitive ignore case if false. | |
336a28fa | 319 | * @param string $visibility Optional: return only all/private/public bookmarks. |
822bffce | 320 | * |
336a28fa | 321 | * @return array filtered bookmarks. |
822bffce | 322 | */ |
7f96d9ec | 323 | public function filterTags($tags, $casesensitive = false, $visibility = 'all') |
822bffce | 324 | { |
341527ba WE |
325 | // get single tags (we may get passed an array, even though the docs say different) |
326 | $inputTags = $tags; | |
f211e417 | 327 | if (!is_array($tags)) { |
341527ba WE |
328 | // we got an input string, split tags |
329 | $inputTags = preg_split('/(?:\s+)|,/', $inputTags, -1, PREG_SPLIT_NO_EMPTY); | |
330 | } | |
331 | ||
f211e417 | 332 | if (!count($inputTags)) { |
341527ba | 333 | // no input tags |
7f96d9ec | 334 | return $this->noFilter($visibility); |
c51fae92 A |
335 | } |
336 | ||
336a28fa A |
337 | // If we only have public visibility, we can't look for hidden tags |
338 | if ($visibility === self::$PUBLIC) { | |
339 | $inputTags = array_values(array_filter($inputTags, function ($tag) { | |
340 | return ! startsWith($tag, '.'); | |
341 | })); | |
342 | ||
343 | if (empty($inputTags)) { | |
344 | return []; | |
345 | } | |
346 | } | |
347 | ||
341527ba WE |
348 | // build regex from all tags |
349 | $re = '/^' . implode(array_map("self::tag2regex", $inputTags)) . '.*$/'; | |
f211e417 | 350 | if (!$casesensitive) { |
341527ba WE |
351 | // make regex case insensitive |
352 | $re .= 'i'; | |
21979ff1 | 353 | } |
822bffce | 354 | |
341527ba | 355 | // create resulting array |
336a28fa | 356 | $filtered = []; |
341527ba WE |
357 | |
358 | // iterate over each link | |
336a28fa | 359 | foreach ($this->bookmarks as $key => $link) { |
341527ba | 360 | // check level of visibility |
336a28fa | 361 | // ignore non private bookmarks when 'privateonly' is on. |
7f96d9ec | 362 | if ($visibility !== 'all') { |
336a28fa | 363 | if (!$link->isPrivate() && $visibility === 'private') { |
7f96d9ec | 364 | continue; |
336a28fa | 365 | } elseif ($link->isPrivate() && $visibility === 'public') { |
7f96d9ec A |
366 | continue; |
367 | } | |
822bffce | 368 | } |
336a28fa A |
369 | $search = $link->getTagsString(); // build search string, start with tags of current link |
370 | if (strlen(trim($link->getDescription())) && strpos($link->getDescription(), '#') !== false) { | |
341527ba WE |
371 | // description given and at least one possible tag found |
372 | $descTags = array(); | |
373 | // find all tags in the form of #tag in the description | |
374 | preg_match_all( | |
375 | '/(?<![' . self::$HASHTAG_CHARS . '])#([' . self::$HASHTAG_CHARS . ']+?)\b/sm', | |
336a28fa | 376 | $link->getDescription(), |
341527ba WE |
377 | $descTags |
378 | ); | |
f211e417 | 379 | if (count($descTags[1])) { |
341527ba WE |
380 | // there were some tags in the description, add them to the search string |
381 | $search .= ' ' . implode(' ', $descTags[1]); | |
21979ff1 | 382 | } |
341527ba WE |
383 | }; |
384 | // match regular expression with search string | |
f211e417 | 385 | if (!preg_match($re, $search)) { |
341527ba WE |
386 | // this entry does _not_ match our regex |
387 | continue; | |
21979ff1 | 388 | } |
341527ba | 389 | $filtered[$key] = $link; |
822bffce | 390 | } |
822bffce A |
391 | return $filtered; |
392 | } | |
393 | ||
7d86f40b | 394 | /** |
336a28fa | 395 | * Return only bookmarks without any tag. |
7d86f40b | 396 | * |
336a28fa | 397 | * @param string $visibility return only all/private/public bookmarks. |
7d86f40b | 398 | * |
336a28fa | 399 | * @return array filtered bookmarks. |
7d86f40b A |
400 | */ |
401 | public function filterUntagged($visibility) | |
402 | { | |
403 | $filtered = []; | |
336a28fa | 404 | foreach ($this->bookmarks as $key => $link) { |
7d86f40b | 405 | if ($visibility !== 'all') { |
336a28fa | 406 | if (!$link->isPrivate() && $visibility === 'private') { |
7d86f40b | 407 | continue; |
336a28fa | 408 | } elseif ($link->isPrivate() && $visibility === 'public') { |
7d86f40b A |
409 | continue; |
410 | } | |
411 | } | |
412 | ||
336a28fa | 413 | if (empty(trim($link->getTagsString()))) { |
7d86f40b A |
414 | $filtered[$key] = $link; |
415 | } | |
416 | } | |
417 | ||
418 | return $filtered; | |
419 | } | |
420 | ||
822bffce A |
421 | /** |
422 | * Returns the list of articles for a given day, chronologically sorted | |
423 | * | |
424 | * Day must be in the form 'YYYYMMDD' (e.g. '20120125'), e.g. | |
425 | * print_r($mydb->filterDay('20120125')); | |
426 | * | |
427 | * @param string $day day to filter. | |
428 | * | |
429 | * @return array all link matching given day. | |
430 | * | |
431 | * @throws Exception if date format is invalid. | |
432 | */ | |
433 | public function filterDay($day) | |
434 | { | |
6696729b | 435 | if (!checkDateFormat('Ymd', $day)) { |
822bffce A |
436 | throw new Exception('Invalid date format'); |
437 | } | |
438 | ||
69e29ff6 | 439 | $filtered = []; |
336a28fa A |
440 | foreach ($this->bookmarks as $key => $l) { |
441 | if ($l->getCreated()->format('Ymd') == $day) { | |
01878a75 | 442 | $filtered[$key] = $l; |
822bffce A |
443 | } |
444 | } | |
01878a75 A |
445 | |
446 | // sort by date ASC | |
447 | return array_reverse($filtered, true); | |
822bffce A |
448 | } |
449 | ||
450 | /** | |
451 | * Convert a list of tags (str) to an array. Also | |
452 | * - handle case sensitivity. | |
453 | * - accepts spaces commas as separator. | |
822bffce A |
454 | * |
455 | * @param string $tags string containing a list of tags. | |
456 | * @param bool $casesensitive will convert everything to lowercase if false. | |
457 | * | |
458 | * @return array filtered tags string. | |
7f96d9ec | 459 | */ |
21979ff1 | 460 | public static function tagsStrToArray($tags, $casesensitive) |
822bffce A |
461 | { |
462 | // We use UTF-8 conversion to handle various graphemes (i.e. cyrillic, or greek) | |
463 | $tagsOut = $casesensitive ? $tags : mb_convert_case($tags, MB_CASE_LOWER, 'UTF-8'); | |
464 | $tagsOut = str_replace(',', ' ', $tagsOut); | |
465 | ||
b3051a6a | 466 | return preg_split('/\s+/', $tagsOut, -1, PREG_SPLIT_NO_EMPTY); |
822bffce A |
467 | } |
468 | } |