application/LinkDB.php

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