aboutsummaryrefslogtreecommitdiffhomepage
path: root/application/bookmark
diff options
context:
space:
mode:
Diffstat (limited to 'application/bookmark')
-rw-r--r--application/bookmark/LinkDB.php600
-rw-r--r--application/bookmark/LinkFilter.php449
-rw-r--r--application/bookmark/LinkUtils.php222
-rw-r--r--application/bookmark/exception/LinkNotFoundException.php15
4 files changed, 1286 insertions, 0 deletions
diff --git a/application/bookmark/LinkDB.php b/application/bookmark/LinkDB.php
new file mode 100644
index 00000000..c13a1141
--- /dev/null
+++ b/application/bookmark/LinkDB.php
@@ -0,0 +1,600 @@
1<?php
2
3namespace Shaarli\Bookmark;
4
5use ArrayAccess;
6use Countable;
7use DateTime;
8use Iterator;
9use Shaarli\Bookmark\Exception\LinkNotFoundException;
10use Shaarli\Exceptions\IOException;
11use Shaarli\FileUtils;
12
13/**
14 * Data storage for links.
15 *
16 * This object behaves like an associative array.
17 *
18 * Example:
19 * $myLinks = new LinkDB();
20 * echo $myLinks[350]['title'];
21 * foreach ($myLinks as $link)
22 * echo $link['title'].' at url '.$link['url'].'; description:'.$link['description'];
23 *
24 * Available keys:
25 * - id: primary key, incremental integer identifier (persistent)
26 * - description: description of the entry
27 * - created: creation date of this entry, DateTime object.
28 * - updated: last modification date of this entry, DateTime object.
29 * - private: Is this link private? 0=no, other value=yes
30 * - tags: tags attached to this entry (separated by spaces)
31 * - title Title of the link
32 * - url URL of the link. Used for displayable links (no redirector, relative, etc.).
33 * Can be absolute or relative.
34 * Relative URLs are permalinks (e.g.'?m-ukcw')
35 * - real_url Absolute processed URL.
36 * - shorturl Permalink smallhash
37 *
38 * Implements 3 interfaces:
39 * - ArrayAccess: behaves like an associative array;
40 * - Countable: there is a count() method;
41 * - Iterator: usable in foreach () loops.
42 *
43 * ID mechanism:
44 * ArrayAccess is implemented in a way that will allow to access a link
45 * with the unique identifier ID directly with $link[ID].
46 * Note that it's not the real key of the link array attribute.
47 * This mechanism is in place to have persistent link IDs,
48 * even though the internal array is reordered by date.
49 * Example:
50 * - DB: link #1 (2010-01-01) link #2 (2016-01-01)
51 * - Order: #2 #1
52 * - Import links containing: link #3 (2013-01-01)
53 * - New DB: link #1 (2010-01-01) link #2 (2016-01-01) link #3 (2013-01-01)
54 * - Real order: #2 #3 #1
55 */
56class LinkDB implements Iterator, Countable, ArrayAccess
57{
58 // Links are stored as a PHP serialized string
59 private $datastore;
60
61 // Link date storage format
62 const LINK_DATE_FORMAT = 'Ymd_His';
63
64 // List of links (associative array)
65 // - key: link date (e.g. "20110823_124546"),
66 // - value: associative array (keys: title, description...)
67 private $links;
68
69 // List of all recorded URLs (key=url, value=link offset)
70 // for fast reserve search (url-->link offset)
71 private $urls;
72
73 /**
74 * @var array List of all links IDS mapped with their array offset.
75 * Map: id->offset.
76 */
77 protected $ids;
78
79 // List of offset keys (for the Iterator interface implementation)
80 private $keys;
81
82 // Position in the $this->keys array (for the Iterator interface)
83 private $position;
84
85 // Is the user logged in? (used to filter private links)
86 private $loggedIn;
87
88 // Hide public links
89 private $hidePublicLinks;
90
91 // link redirector set in user settings.
92 private $redirector;
93
94 /**
95 * Set this to `true` to urlencode link behind redirector link, `false` to leave it untouched.
96 *
97 * Example:
98 * anonym.to needs clean URL while dereferer.org needs urlencoded URL.
99 *
100 * @var boolean $redirectorEncode parameter: true or false
101 */
102 private $redirectorEncode;
103
104 /**
105 * Creates a new LinkDB
106 *
107 * Checks if the datastore exists; else, attempts to create a dummy one.
108 *
109 * @param string $datastore datastore file path.
110 * @param boolean $isLoggedIn is the user logged in?
111 * @param boolean $hidePublicLinks if true all links are private.
112 * @param string $redirector link redirector set in user settings.
113 * @param boolean $redirectorEncode Enable urlencode on redirected urls (default: true).
114 */
115 public function __construct(
116 $datastore,
117 $isLoggedIn,
118 $hidePublicLinks,
119 $redirector = '',
120 $redirectorEncode = true
121 ) {
122
123 $this->datastore = $datastore;
124 $this->loggedIn = $isLoggedIn;
125 $this->hidePublicLinks = $hidePublicLinks;
126 $this->redirector = $redirector;
127 $this->redirectorEncode = $redirectorEncode === true;
128 $this->check();
129 $this->read();
130 }
131
132 /**
133 * Countable - Counts elements of an object
134 */
135 public function count()
136 {
137 return count($this->links);
138 }
139
140 /**
141 * ArrayAccess - Assigns a value to the specified offset
142 */
143 public function offsetSet($offset, $value)
144 {
145 // TODO: use exceptions instead of "die"
146 if (!$this->loggedIn) {
147 die(t('You are not authorized to add a link.'));
148 }
149 if (!isset($value['id']) || empty($value['url'])) {
150 die(t('Internal Error: A link should always have an id and URL.'));
151 }
152 if (($offset !== null && !is_int($offset)) || !is_int($value['id'])) {
153 die(t('You must specify an integer as a key.'));
154 }
155 if ($offset !== null && $offset !== $value['id']) {
156 die(t('Array offset and link ID must be equal.'));
157 }
158
159 // If the link exists, we reuse the real offset, otherwise new entry
160 $existing = $this->getLinkOffset($offset);
161 if ($existing !== null) {
162 $offset = $existing;
163 } else {
164 $offset = count($this->links);
165 }
166 $this->links[$offset] = $value;
167 $this->urls[$value['url']] = $offset;
168 $this->ids[$value['id']] = $offset;
169 }
170
171 /**
172 * ArrayAccess - Whether or not an offset exists
173 */
174 public function offsetExists($offset)
175 {
176 return array_key_exists($this->getLinkOffset($offset), $this->links);
177 }
178
179 /**
180 * ArrayAccess - Unsets an offset
181 */
182 public function offsetUnset($offset)
183 {
184 if (!$this->loggedIn) {
185 // TODO: raise an exception
186 die('You are not authorized to delete a link.');
187 }
188 $realOffset = $this->getLinkOffset($offset);
189 $url = $this->links[$realOffset]['url'];
190 unset($this->urls[$url]);
191 unset($this->ids[$realOffset]);
192 unset($this->links[$realOffset]);
193 }
194
195 /**
196 * ArrayAccess - Returns the value at specified offset
197 */
198 public function offsetGet($offset)
199 {
200 $realOffset = $this->getLinkOffset($offset);
201 return isset($this->links[$realOffset]) ? $this->links[$realOffset] : null;
202 }
203
204 /**
205 * Iterator - Returns the current element
206 */
207 public function current()
208 {
209 return $this[$this->keys[$this->position]];
210 }
211
212 /**
213 * Iterator - Returns the key of the current element
214 */
215 public function key()
216 {
217 return $this->keys[$this->position];
218 }
219
220 /**
221 * Iterator - Moves forward to next element
222 */
223 public function next()
224 {
225 ++$this->position;
226 }
227
228 /**
229 * Iterator - Rewinds the Iterator to the first element
230 *
231 * Entries are sorted by date (latest first)
232 */
233 public function rewind()
234 {
235 $this->keys = array_keys($this->ids);
236 $this->position = 0;
237 }
238
239 /**
240 * Iterator - Checks if current position is valid
241 */
242 public function valid()
243 {
244 return isset($this->keys[$this->position]);
245 }
246
247 /**
248 * Checks if the DB directory and file exist
249 *
250 * If no DB file is found, creates a dummy DB.
251 */
252 private function check()
253 {
254 if (file_exists($this->datastore)) {
255 return;
256 }
257
258 // Create a dummy database for example
259 $this->links = array();
260 $link = array(
261 'id' => 1,
262 'title' => t('The personal, minimalist, super-fast, database free, bookmarking service'),
263 'url' => 'https://shaarli.readthedocs.io',
264 'description' => t(
265 'Welcome to Shaarli! This is your first public bookmark. '
266 . 'To edit or delete me, you must first login.
267
268To learn how to use Shaarli, consult the link "Documentation" at the bottom of this page.
269
270You use the community supported version of the original Shaarli project, by Sebastien Sauvage.'
271 ),
272 'private' => 0,
273 'created' => new DateTime(),
274 'tags' => 'opensource software'
275 );
276 $link['shorturl'] = link_small_hash($link['created'], $link['id']);
277 $this->links[1] = $link;
278
279 $link = array(
280 'id' => 0,
281 'title' => t('My secret stuff... - Pastebin.com'),
282 'url' => 'http://sebsauvage.net/paste/?8434b27936c09649#bR7XsXhoTiLcqCpQbmOpBi3rq2zzQUC5hBI7ZT1O3x8=',
283 'description' => t('Shhhh! I\'m a private link only YOU can see. You can delete me too.'),
284 'private' => 1,
285 'created' => new DateTime('1 minute ago'),
286 'tags' => 'secretstuff',
287 );
288 $link['shorturl'] = link_small_hash($link['created'], $link['id']);
289 $this->links[0] = $link;
290
291 // Write database to disk
292 $this->write();
293 }
294
295 /**
296 * Reads database from disk to memory
297 */
298 private function read()
299 {
300 // Public links are hidden and user not logged in => nothing to show
301 if ($this->hidePublicLinks && !$this->loggedIn) {
302 $this->links = array();
303 return;
304 }
305
306 $this->urls = [];
307 $this->ids = [];
308 $this->links = FileUtils::readFlatDB($this->datastore, []);
309
310 $toremove = array();
311 foreach ($this->links as $key => &$link) {
312 if (!$this->loggedIn && $link['private'] != 0) {
313 // Transition for not upgraded databases.
314 unset($this->links[$key]);
315 continue;
316 }
317
318 // Sanitize data fields.
319 sanitizeLink($link);
320
321 // Remove private tags if the user is not logged in.
322 if (!$this->loggedIn) {
323 $link['tags'] = preg_replace('/(^|\s+)\.[^($|\s)]+\s*/', ' ', $link['tags']);
324 }
325
326 // Do not use the redirector for internal links (Shaarli note URL starting with a '?').
327 if (!empty($this->redirector) && !startsWith($link['url'], '?')) {
328 $link['real_url'] = $this->redirector;
329 if ($this->redirectorEncode) {
330 $link['real_url'] .= urlencode(unescape($link['url']));
331 } else {
332 $link['real_url'] .= $link['url'];
333 }
334 } else {
335 $link['real_url'] = $link['url'];
336 }
337
338 // To be able to load links before running the update, and prepare the update
339 if (!isset($link['created'])) {
340 $link['id'] = $link['linkdate'];
341 $link['created'] = DateTime::createFromFormat(self::LINK_DATE_FORMAT, $link['linkdate']);
342 if (!empty($link['updated'])) {
343 $link['updated'] = DateTime::createFromFormat(self::LINK_DATE_FORMAT, $link['updated']);
344 }
345 $link['shorturl'] = smallHash($link['linkdate']);
346 }
347
348 $this->urls[$link['url']] = $key;
349 $this->ids[$link['id']] = $key;
350 }
351 }
352
353 /**
354 * Saves the database from memory to disk
355 *
356 * @throws IOException the datastore is not writable
357 */
358 private function write()
359 {
360 $this->reorder();
361 FileUtils::writeFlatDB($this->datastore, $this->links);
362 }
363
364 /**
365 * Saves the database from memory to disk
366 *
367 * @param string $pageCacheDir page cache directory
368 */
369 public function save($pageCacheDir)
370 {
371 if (!$this->loggedIn) {
372 // TODO: raise an Exception instead
373 die('You are not authorized to change the database.');
374 }
375
376 $this->write();
377
378 invalidateCaches($pageCacheDir);
379 }
380
381 /**
382 * Returns the link for a given URL, or False if it does not exist.
383 *
384 * @param string $url URL to search for
385 *
386 * @return mixed the existing link if it exists, else 'false'
387 */
388 public function getLinkFromUrl($url)
389 {
390 if (isset($this->urls[$url])) {
391 return $this->links[$this->urls[$url]];
392 }
393 return false;
394 }
395
396 /**
397 * Returns the shaare corresponding to a smallHash.
398 *
399 * @param string $request QUERY_STRING server parameter.
400 *
401 * @return array $filtered array containing permalink data.
402 *
403 * @throws LinkNotFoundException if the smallhash is malformed or doesn't match any link.
404 */
405 public function filterHash($request)
406 {
407 $request = substr($request, 0, 6);
408 $linkFilter = new LinkFilter($this->links);
409 return $linkFilter->filter(LinkFilter::$FILTER_HASH, $request);
410 }
411
412 /**
413 * Returns the list of articles for a given day.
414 *
415 * @param string $request day to filter. Format: YYYYMMDD.
416 *
417 * @return array list of shaare found.
418 */
419 public function filterDay($request)
420 {
421 $linkFilter = new LinkFilter($this->links);
422 return $linkFilter->filter(LinkFilter::$FILTER_DAY, $request);
423 }
424
425 /**
426 * Filter links according to search parameters.
427 *
428 * @param array $filterRequest Search request content. Supported keys:
429 * - searchtags: list of tags
430 * - searchterm: term search
431 * @param bool $casesensitive Optional: Perform case sensitive filter
432 * @param string $visibility return only all/private/public links
433 * @param bool $untaggedonly return only untagged links
434 *
435 * @return array filtered links, all links if no suitable filter was provided.
436 */
437 public function filterSearch(
438 $filterRequest = array(),
439 $casesensitive = false,
440 $visibility = 'all',
441 $untaggedonly = false
442 ) {
443
444 // Filter link database according to parameters.
445 $searchtags = isset($filterRequest['searchtags']) ? escape($filterRequest['searchtags']) : '';
446 $searchterm = isset($filterRequest['searchterm']) ? escape($filterRequest['searchterm']) : '';
447
448 // Search tags + fullsearch - blank string parameter will return all links.
449 $type = LinkFilter::$FILTER_TAG | LinkFilter::$FILTER_TEXT; // == "vuotext"
450 $request = [$searchtags, $searchterm];
451
452 $linkFilter = new LinkFilter($this);
453 return $linkFilter->filter($type, $request, $casesensitive, $visibility, $untaggedonly);
454 }
455
456 /**
457 * Returns the list tags appearing in the links with the given tags
458 *
459 * @param array $filteringTags tags selecting the links to consider
460 * @param string $visibility process only all/private/public links
461 *
462 * @return array tag => linksCount
463 */
464 public function linksCountPerTag($filteringTags = [], $visibility = 'all')
465 {
466 $links = $this->filterSearch(['searchtags' => $filteringTags], false, $visibility);
467 $tags = [];
468 $caseMapping = [];
469 foreach ($links as $link) {
470 foreach (preg_split('/\s+/', $link['tags'], 0, PREG_SPLIT_NO_EMPTY) as $tag) {
471 if (empty($tag)) {
472 continue;
473 }
474 // The first case found will be displayed.
475 if (!isset($caseMapping[strtolower($tag)])) {
476 $caseMapping[strtolower($tag)] = $tag;
477 $tags[$caseMapping[strtolower($tag)]] = 0;
478 }
479 $tags[$caseMapping[strtolower($tag)]]++;
480 }
481 }
482
483 /*
484 * Formerly used arsort(), which doesn't define the sort behaviour for equal values.
485 * Also, this function doesn't produce the same result between PHP 5.6 and 7.
486 *
487 * So we now use array_multisort() to sort tags by DESC occurrences,
488 * then ASC alphabetically for equal values.
489 *
490 * @see https://github.com/shaarli/Shaarli/issues/1142
491 */
492 $keys = array_keys($tags);
493 $tmpTags = array_combine($keys, $keys);
494 array_multisort($tags, SORT_DESC, $tmpTags, SORT_ASC, $tags);
495 return $tags;
496 }
497
498 /**
499 * Rename or delete a tag across all links.
500 *
501 * @param string $from Tag to rename
502 * @param string $to New tag. If none is provided, the from tag will be deleted
503 *
504 * @return array|bool List of altered links or false on error
505 */
506 public function renameTag($from, $to)
507 {
508 if (empty($from)) {
509 return false;
510 }
511 $delete = empty($to);
512 // True for case-sensitive tag search.
513 $linksToAlter = $this->filterSearch(['searchtags' => $from], true);
514 foreach ($linksToAlter as $key => &$value) {
515 $tags = preg_split('/\s+/', trim($value['tags']));
516 if (($pos = array_search($from, $tags)) !== false) {
517 if ($delete) {
518 unset($tags[$pos]); // Remove tag.
519 } else {
520 $tags[$pos] = trim($to);
521 }
522 $value['tags'] = trim(implode(' ', array_unique($tags)));
523 $this[$value['id']] = $value;
524 }
525 }
526
527 return $linksToAlter;
528 }
529
530 /**
531 * Returns the list of days containing articles (oldest first)
532 * Output: An array containing days (in format YYYYMMDD).
533 */
534 public function days()
535 {
536 $linkDays = array();
537 foreach ($this->links as $link) {
538 $linkDays[$link['created']->format('Ymd')] = 0;
539 }
540 $linkDays = array_keys($linkDays);
541 sort($linkDays);
542
543 return $linkDays;
544 }
545
546 /**
547 * Reorder links by creation date (newest first).
548 *
549 * Also update the urls and ids mapping arrays.
550 *
551 * @param string $order ASC|DESC
552 */
553 public function reorder($order = 'DESC')
554 {
555 $order = $order === 'ASC' ? -1 : 1;
556 // Reorder array by dates.
557 usort($this->links, function ($a, $b) use ($order) {
558 if (isset($a['sticky']) && isset($b['sticky']) && $a['sticky'] !== $b['sticky']) {
559 return $a['sticky'] ? -1 : 1;
560 }
561 return $a['created'] < $b['created'] ? 1 * $order : -1 * $order;
562 });
563
564 $this->urls = [];
565 $this->ids = [];
566 foreach ($this->links as $key => $link) {
567 $this->urls[$link['url']] = $key;
568 $this->ids[$link['id']] = $key;
569 }
570 }
571
572 /**
573 * Return the next key for link creation.
574 * E.g. If the last ID is 597, the next will be 598.
575 *
576 * @return int next ID.
577 */
578 public function getNextId()
579 {
580 if (!empty($this->ids)) {
581 return max(array_keys($this->ids)) + 1;
582 }
583 return 0;
584 }
585
586 /**
587 * Returns a link offset in links array from its unique ID.
588 *
589 * @param int $id Persistent ID of a link.
590 *
591 * @return int Real offset in local array, or null if doesn't exist.
592 */
593 protected function getLinkOffset($id)
594 {
595 if (isset($this->ids[$id])) {
596 return $this->ids[$id];
597 }
598 return null;
599 }
600}
diff --git a/application/bookmark/LinkFilter.php b/application/bookmark/LinkFilter.php
new file mode 100644
index 00000000..9b966307
--- /dev/null
+++ b/application/bookmark/LinkFilter.php
@@ -0,0 +1,449 @@
1<?php
2
3namespace Shaarli\Bookmark;
4
5use Exception;
6use Shaarli\Bookmark\Exception\LinkNotFoundException;
7
8/**
9 * Class LinkFilter.
10 *
11 * Perform search and filter operation on link data list.
12 */
13class 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}
diff --git a/application/bookmark/LinkUtils.php b/application/bookmark/LinkUtils.php
new file mode 100644
index 00000000..de5b61cb
--- /dev/null
+++ b/application/bookmark/LinkUtils.php
@@ -0,0 +1,222 @@
1<?php
2
3use Shaarli\Bookmark\LinkDB;
4
5/**
6 * Get cURL callback function for CURLOPT_WRITEFUNCTION
7 *
8 * @param string $charset to extract from the downloaded page (reference)
9 * @param string $title to extract from the downloaded page (reference)
10 * @param string $curlGetInfo Optionally overrides curl_getinfo function
11 *
12 * @return Closure
13 */
14function get_curl_download_callback(&$charset, &$title, $curlGetInfo = 'curl_getinfo')
15{
16 $isRedirected = false;
17 /**
18 * cURL callback function for CURLOPT_WRITEFUNCTION (called during the download).
19 *
20 * While downloading the remote page, we check that the HTTP code is 200 and content type is 'html/text'
21 * Then we extract the title and the charset and stop the download when it's done.
22 *
23 * @param resource $ch cURL resource
24 * @param string $data chunk of data being downloaded
25 *
26 * @return int|bool length of $data or false if we need to stop the download
27 */
28 return function (&$ch, $data) use ($curlGetInfo, &$charset, &$title, &$isRedirected) {
29 $responseCode = $curlGetInfo($ch, CURLINFO_RESPONSE_CODE);
30 if (!empty($responseCode) && in_array($responseCode, [301, 302])) {
31 $isRedirected = true;
32 return strlen($data);
33 }
34 if (!empty($responseCode) && $responseCode !== 200) {
35 return false;
36 }
37 // After a redirection, the content type will keep the previous request value
38 // until it finds the next content-type header.
39 if (! $isRedirected || strpos(strtolower($data), 'content-type') !== false) {
40 $contentType = $curlGetInfo($ch, CURLINFO_CONTENT_TYPE);
41 }
42 if (!empty($contentType) && strpos($contentType, 'text/html') === false) {
43 return false;
44 }
45 if (!empty($contentType) && empty($charset)) {
46 $charset = header_extract_charset($contentType);
47 }
48 if (empty($charset)) {
49 $charset = html_extract_charset($data);
50 }
51 if (empty($title)) {
52 $title = html_extract_title($data);
53 }
54 // We got everything we want, stop the download.
55 if (!empty($responseCode) && !empty($contentType) && !empty($charset) && !empty($title)) {
56 return false;
57 }
58
59 return strlen($data);
60 };
61}
62
63/**
64 * Extract title from an HTML document.
65 *
66 * @param string $html HTML content where to look for a title.
67 *
68 * @return bool|string Extracted title if found, false otherwise.
69 */
70function html_extract_title($html)
71{
72 if (preg_match('!<title.*?>(.*?)</title>!is', $html, $matches)) {
73 return trim(str_replace("\n", '', $matches[1]));
74 }
75 return false;
76}
77
78/**
79 * Extract charset from HTTP header if it's defined.
80 *
81 * @param string $header HTTP header Content-Type line.
82 *
83 * @return bool|string Charset string if found (lowercase), false otherwise.
84 */
85function header_extract_charset($header)
86{
87 preg_match('/charset="?([^; ]+)/i', $header, $match);
88 if (! empty($match[1])) {
89 return strtolower(trim($match[1]));
90 }
91
92 return false;
93}
94
95/**
96 * Extract charset HTML content (tag <meta charset>).
97 *
98 * @param string $html HTML content where to look for charset.
99 *
100 * @return bool|string Charset string if found, false otherwise.
101 */
102function html_extract_charset($html)
103{
104 // Get encoding specified in HTML header.
105 preg_match('#<meta .*charset=["\']?([^";\'>/]+)["\']? */?>#Usi', $html, $enc);
106 if (!empty($enc[1])) {
107 return strtolower($enc[1]);
108 }
109
110 return false;
111}
112
113/**
114 * Count private links in given linklist.
115 *
116 * @param array|Countable $links Linklist.
117 *
118 * @return int Number of private links.
119 */
120function count_private($links)
121{
122 $cpt = 0;
123 foreach ($links as $link) {
124 if ($link['private']) {
125 $cpt += 1;
126 }
127 }
128
129 return $cpt;
130}
131
132/**
133 * In a string, converts URLs to clickable links.
134 *
135 * @param string $text input string.
136 * @param string $redirector if a redirector is set, use it to gerenate links.
137 * @param bool $urlEncode Use `urlencode()` on the URL after the redirector or not.
138 *
139 * @return string returns $text with all links converted to HTML links.
140 *
141 * @see Function inspired from http://www.php.net/manual/en/function.preg-replace.php#85722
142 */
143function text2clickable($text, $redirector = '', $urlEncode = true)
144{
145 $regex = '!(((?:https?|ftp|file)://|apt:|magnet:)\S+[a-z0-9\(\)]/?)!si';
146
147 if (empty($redirector)) {
148 return preg_replace($regex, '<a href="$1">$1</a>', $text);
149 }
150 // Redirector is set, urlencode the final URL.
151 return preg_replace_callback(
152 $regex,
153 function ($matches) use ($redirector, $urlEncode) {
154 $url = $urlEncode ? urlencode($matches[1]) : $matches[1];
155 return '<a href="' . $redirector . $url .'">'. $matches[1] .'</a>';
156 },
157 $text
158 );
159}
160
161/**
162 * Auto-link hashtags.
163 *
164 * @param string $description Given description.
165 * @param string $indexUrl Root URL.
166 *
167 * @return string Description with auto-linked hashtags.
168 */
169function hashtag_autolink($description, $indexUrl = '')
170{
171 /*
172 * To support unicode: http://stackoverflow.com/a/35498078/1484919
173 * \p{Pc} - to match underscore
174 * \p{N} - numeric character in any script
175 * \p{L} - letter from any language
176 * \p{Mn} - any non marking space (accents, umlauts, etc)
177 */
178 $regex = '/(^|\s)#([\p{Pc}\p{N}\p{L}\p{Mn}]+)/mui';
179 $replacement = '$1<a href="'. $indexUrl .'?addtag=$2" title="Hashtag $2">#$2</a>';
180 return preg_replace($regex, $replacement, $description);
181}
182
183/**
184 * This function inserts &nbsp; where relevant so that multiple spaces are properly displayed in HTML
185 * even in the absence of <pre> (This is used in description to keep text formatting).
186 *
187 * @param string $text input text.
188 *
189 * @return string formatted text.
190 */
191function space2nbsp($text)
192{
193 return preg_replace('/(^| ) /m', '$1&nbsp;', $text);
194}
195
196/**
197 * Format Shaarli's description
198 *
199 * @param string $description shaare's description.
200 * @param string $redirector if a redirector is set, use it to gerenate links.
201 * @param bool $urlEncode Use `urlencode()` on the URL after the redirector or not.
202 * @param string $indexUrl URL to Shaarli's index.
203
204 * @return string formatted description.
205 */
206function format_description($description, $redirector = '', $urlEncode = true, $indexUrl = '')
207{
208 return nl2br(space2nbsp(hashtag_autolink(text2clickable($description, $redirector, $urlEncode), $indexUrl)));
209}
210
211/**
212 * Generate a small hash for a link.
213 *
214 * @param DateTime $date Link creation date.
215 * @param int $id Link ID.
216 *
217 * @return string the small hash generated from link data.
218 */
219function link_small_hash($date, $id)
220{
221 return smallHash($date->format(LinkDB::LINK_DATE_FORMAT) . $id);
222}
diff --git a/application/bookmark/exception/LinkNotFoundException.php b/application/bookmark/exception/LinkNotFoundException.php
new file mode 100644
index 00000000..f9414428
--- /dev/null
+++ b/application/bookmark/exception/LinkNotFoundException.php
@@ -0,0 +1,15 @@
1<?php
2namespace Shaarli\Bookmark\Exception;
3
4use Exception;
5
6class LinkNotFoundException extends Exception
7{
8 /**
9 * LinkNotFoundException constructor.
10 */
11 public function __construct()
12 {
13 $this->message = t('The link you are trying to reach does not exist or has been deleted.');
14 }
15}
generated by cgit v1.2.3 (git 2.25.1) at 2024-05-26 23:31:27 +0000