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