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('Welcome to Shaarli! This is your first public bookmark. To edit or delete me, you must first login.
 253
 254 To learn how to use Shaarli, consult the link "Documentation" at the bottom of this page.
 255
 256 You use the community supported version of the original Shaarli project, by Sebastien Sauvage.'),
 257             'private'=>0,
 258             'created'=> new DateTime(),
 259             'tags'=>'opensource software'
 260         );
 261         $link['shorturl'] = link_small_hash($link['created'], $link['id']);
 262         $this->links[1] = $link;
 263
 264         $link = array(
 265             'id' => 0,
 266             'title'=> t('My secret stuff... - Pastebin.com'),
 267             'url'=>'http://sebsauvage.net/paste/?8434b27936c09649#bR7XsXhoTiLcqCpQbmOpBi3rq2zzQUC5hBI7ZT1O3x8=',
 268             'description'=> t('Shhhh! I\'m a private link only YOU can see. You can delete me too.'),
 269             'private'=>1,
 270             'created'=> new DateTime('1 minute ago'),
 271             'tags'=>'secretstuff',
 272         );
 273         $link['shorturl'] = link_small_hash($link['created'], $link['id']);
 274         $this->links[0] = $link;
 275
 276         // Write database to disk
 277         $this->write();
 278     }
 279
 280     /**
 281      * Reads database from disk to memory
 282      */
 283     private function read()
 284     {
 285         // Public links are hidden and user not logged in => nothing to show
 286         if ($this->hidePublicLinks && !$this->loggedIn) {
 287             $this->links = array();
 288             return;
 289         }
 290
 291         $this->urls = [];
 292         $this->ids = [];
 293         $this->links = FileUtils::readFlatDB($this->datastore, []);
 294
 295         $toremove = array();
 296         foreach ($this->links as $key => &$link) {
 297             if (! $this->loggedIn && $link['private'] != 0) {
 298                 // Transition for not upgraded databases.
 299                 unset($this->links[$key]);
 300                 continue;
 301             }
 302
 303             // Sanitize data fields.
 304             sanitizeLink($link);
 305
 306             // Remove private tags if the user is not logged in.
 307             if (! $this->loggedIn) {
 308                 $link['tags'] = preg_replace('/(^|\s+)\.[^($|\s)]+\s*/', ' ', $link['tags']);
 309             }
 310
 311             // Do not use the redirector for internal links (Shaarli note URL starting with a '?').
 312             if (!empty($this->redirector) && !startsWith($link['url'], '?')) {
 313                 $link['real_url'] = $this->redirector;
 314                 if ($this->redirectorEncode) {
 315                     $link['real_url'] .= urlencode(unescape($link['url']));
 316                 } else {
 317                     $link['real_url'] .= $link['url'];
 318                 }
 319             } else {
 320                 $link['real_url'] = $link['url'];
 321             }
 322
 323             // To be able to load links before running the update, and prepare the update
 324             if (! isset($link['created'])) {
 325                 $link['id'] = $link['linkdate'];
 326                 $link['created'] = DateTime::createFromFormat(self::LINK_DATE_FORMAT, $link['linkdate']);
 327                 if (! empty($link['updated'])) {
 328                     $link['updated'] = DateTime::createFromFormat(self::LINK_DATE_FORMAT, $link['updated']);
 329                 }
 330                 $link['shorturl'] = smallHash($link['linkdate']);
 331             }
 332
 333             $this->urls[$link['url']] = $key;
 334             $this->ids[$link['id']] = $key;
 335         }
 336     }
 337
 338     /**
 339      * Saves the database from memory to disk
 340      *
 341      * @throws IOException the datastore is not writable
 342      */
 343     private function write()
 344     {
 345         $this->reorder();
 346         FileUtils::writeFlatDB($this->datastore, $this->links);
 347     }
 348
 349     /**
 350      * Saves the database from memory to disk
 351      *
 352      * @param string $pageCacheDir page cache directory
 353      */
 354     public function save($pageCacheDir)
 355     {
 356         if (!$this->loggedIn) {
 357             // TODO: raise an Exception instead
 358             die('You are not authorized to change the database.');
 359         }
 360
 361         $this->write();
 362
 363         invalidateCaches($pageCacheDir);
 364     }
 365
 366     /**
 367      * Returns the link for a given URL, or False if it does not exist.
 368      *
 369      * @param string $url URL to search for
 370      *
 371      * @return mixed the existing link if it exists, else 'false'
 372      */
 373     public function getLinkFromUrl($url)
 374     {
 375         if (isset($this->urls[$url])) {
 376             return $this->links[$this->urls[$url]];
 377         }
 378         return false;
 379     }
 380
 381     /**
 382      * Returns the shaare corresponding to a smallHash.
 383      *
 384      * @param string $request QUERY_STRING server parameter.
 385      *
 386      * @return array $filtered array containing permalink data.
 387      *
 388      * @throws LinkNotFoundException if the smallhash is malformed or doesn't match any link.
 389      */
 390     public function filterHash($request)
 391     {
 392         $request = substr($request, 0, 6);
 393         $linkFilter = new LinkFilter($this->links);
 394         return $linkFilter->filter(LinkFilter::$FILTER_HASH, $request);
 395     }
 396
 397     /**
 398      * Returns the list of articles for a given day.
 399      *
 400      * @param string $request day to filter. Format: YYYYMMDD.
 401      *
 402      * @return array list of shaare found.
 403      */
 404     public function filterDay($request)
 405     {
 406         $linkFilter = new LinkFilter($this->links);
 407         return $linkFilter->filter(LinkFilter::$FILTER_DAY, $request);
 408     }
 409
 410     /**
 411      * Filter links according to search parameters.
 412      *
 413      * @param array  $filterRequest Search request content. Supported keys:
 414      *                                - searchtags: list of tags
 415      *                                - searchterm: term search
 416      * @param bool   $casesensitive Optional: Perform case sensitive filter
 417      * @param string $visibility    return only all/private/public links
 418      * @param string $untaggedonly  return only untagged links
 419      *
 420      * @return array filtered links, all links if no suitable filter was provided.
 421      */
 422     public function filterSearch($filterRequest = array(), $casesensitive = false, $visibility = 'all', $untaggedonly = false)
 423     {
 424         // Filter link database according to parameters.
 425         $searchtags = isset($filterRequest['searchtags']) ? escape($filterRequest['searchtags']) : '';
 426         $searchterm = isset($filterRequest['searchterm']) ? escape($filterRequest['searchterm']) : '';
 427
 428         // Search tags + fullsearch - blank string parameter will return all links.
 429         $type = LinkFilter::$FILTER_TAG | LinkFilter::$FILTER_TEXT; // == "vuotext"
 430         $request = [$searchtags, $searchterm];
 431
 432         $linkFilter = new LinkFilter($this);
 433         return $linkFilter->filter($type, $request, $casesensitive, $visibility, $untaggedonly);
 434     }
 435
 436     /**
 437      * Returns the list tags appearing in the links with the given tags
 438      *
 439      * @param array $filteringTags tags selecting the links to consider
 440      * @param string $visibility   process only all/private/public links
 441      *
 442      * @return array tag => linksCount
 443      */
 444     public function linksCountPerTag($filteringTags = [], $visibility = 'all')
 445     {
 446         $links = $this->filterSearch(['searchtags' => $filteringTags], false, $visibility);
 447         $tags = [];
 448         $caseMapping = [];
 449         foreach ($links as $link) {
 450             foreach (preg_split('/\s+/', $link['tags'], 0, PREG_SPLIT_NO_EMPTY) as $tag) {
 451                 if (empty($tag)) {
 452                     continue;
 453                 }
 454                 // The first case found will be displayed.
 455                 if (!isset($caseMapping[strtolower($tag)])) {
 456                     $caseMapping[strtolower($tag)] = $tag;
 457                     $tags[$caseMapping[strtolower($tag)]] = 0;
 458                 }
 459                 $tags[$caseMapping[strtolower($tag)]]++;
 460             }
 461         }
 462
 463         /*
 464          * Formerly used arsort(), which doesn't define the sort behaviour for equal values.
 465          * Also, this function doesn't produce the same result between PHP 5.6 and 7.
 466          *
 467          * So we now use array_multisort() to sort tags by DESC occurrences,
 468          * then ASC alphabetically for equal values.
 469          *
 470          * @see https://github.com/shaarli/Shaarli/issues/1142
 471          */
 472         $keys = array_keys($tags);
 473         $tmpTags = array_combine($keys, $keys);
 474         array_multisort($tags, SORT_DESC, $tmpTags, SORT_ASC, $tags);
 475         return $tags;
 476     }
 477
 478     /**
 479      * Rename or delete a tag across all links.
 480      *
 481      * @param string $from Tag to rename
 482      * @param string $to   New tag. If none is provided, the from tag will be deleted
 483      *
 484      * @return array|bool List of altered links or false on error
 485      */
 486     public function renameTag($from, $to)
 487     {
 488         if (empty($from)) {
 489             return false;
 490         }
 491         $delete = empty($to);
 492         // True for case-sensitive tag search.
 493         $linksToAlter = $this->filterSearch(['searchtags' => $from], true);
 494         foreach ($linksToAlter as $key => &$value) {
 495             $tags = preg_split('/\s+/', trim($value['tags']));
 496             if (($pos = array_search($from, $tags)) !== false) {
 497                 if ($delete) {
 498                     unset($tags[$pos]); // Remove tag.
 499                 } else {
 500                     $tags[$pos] = trim($to);
 501                 }
 502                 $value['tags'] = trim(implode(' ', array_unique($tags)));
 503                 $this[$value['id']] = $value;
 504             }
 505         }
 506
 507         return $linksToAlter;
 508     }
 509
 510     /**
 511      * Returns the list of days containing articles (oldest first)
 512      * Output: An array containing days (in format YYYYMMDD).
 513      */
 514     public function days()
 515     {
 516         $linkDays = array();
 517         foreach ($this->links as $link) {
 518             $linkDays[$link['created']->format('Ymd')] = 0;
 519         }
 520         $linkDays = array_keys($linkDays);
 521         sort($linkDays);
 522
 523         return $linkDays;
 524     }
 525
 526     /**
 527      * Reorder links by creation date (newest first).
 528      *
 529      * Also update the urls and ids mapping arrays.
 530      *
 531      * @param string $order ASC|DESC
 532      */
 533     public function reorder($order = 'DESC')
 534     {
 535         $order = $order === 'ASC' ? -1 : 1;
 536         // Reorder array by dates.
 537         usort($this->links, function ($a, $b) use ($order) {
 538             if (isset($a['sticky']) && isset($b['sticky']) && $a['sticky'] !== $b['sticky']) {
 539                 return $a['sticky'] ? -1 : 1;
 540             }
 541             return $a['created'] < $b['created'] ? 1 * $order : -1 * $order;
 542         });
 543
 544         $this->urls = [];
 545         $this->ids = [];
 546         foreach ($this->links as $key => $link) {
 547             $this->urls[$link['url']] = $key;
 548             $this->ids[$link['id']] = $key;
 549         }
 550     }
 551
 552     /**
 553      * Return the next key for link creation.
 554      * E.g. If the last ID is 597, the next will be 598.
 555      *
 556      * @return int next ID.
 557      */
 558     public function getNextId()
 559     {
 560         if (!empty($this->ids)) {
 561             return max(array_keys($this->ids)) + 1;
 562         }
 563         return 0;
 564     }
 565
 566     /**
 567      * Returns a link offset in links array from its unique ID.
 568      *
 569      * @param int $id Persistent ID of a link.
 570      *
 571      * @return int Real offset in local array, or null if doesn't exist.
 572      */
 573     protected function getLinkOffset($id)
 574     {
 575         if (isset($this->ids[$id])) {
 576             return $this->ids[$id];
 577         }
 578         return null;
 579     }
 580 }