aboutsummaryrefslogtreecommitdiffhomepage
path: root/application/bookmark/LinkDB.php
diff options
context:
space:
mode:
Diffstat (limited to 'application/bookmark/LinkDB.php')
-rw-r--r--application/bookmark/LinkDB.php600
1 files changed, 600 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}