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 $myLinks['20110826_161819']['title'];
  10  *    foreach ($myLinks as $link)
  11  *       echo $link['title'].' at url '.$link['url'].'; description:'.$link['description'];
  12  *
  13  * Available keys:
  14  *  - description: description of the entry
  15  *  - linkdate: date of the creation of this entry, in the form YYYYMMDD_HHMMSS
  16  *              (e.g.'20110914_192317')
  17  *  - private:  Is this link private? 0=no, other value=yes
  18  *  - tags:     tags attached to this entry (separated by spaces)
  19  *  - title     Title of the link
  20  *  - url       URL of the link. Can be absolute or relative.
  21  *              Relative URLs are permalinks (e.g.'?m-ukcw')
  22  *
  23  * Implements 3 interfaces:
  24  *  - ArrayAccess: behaves like an associative array;
  25  *  - Countable:   there is a count() method;
  26  *  - Iterator:    usable in foreach () loops.
  27  */
  28 class LinkDB implements Iterator, Countable, ArrayAccess
  29 {
  30     // Links are stored as a PHP serialized string
  31     private $_datastore;
  32
  33     // Datastore PHP prefix
  34     protected static $phpPrefix = '<?php /* ';
  35
  36     // Datastore PHP suffix
  37     protected static $phpSuffix = ' */ ?>';
  38
  39     // List of links (associative array)
  40     //  - key:   link date (e.g. "20110823_124546"),
  41     //  - value: associative array (keys: title, description...)
  42     private $_links;
  43
  44     // List of all recorded URLs (key=url, value=linkdate)
  45     // for fast reserve search (url-->linkdate)
  46     private $_urls;
  47
  48     // List of linkdate keys (for the Iterator interface implementation)
  49     private $_keys;
  50
  51     // Position in the $this->_keys array (for the Iterator interface)
  52     private $_position;
  53
  54     // Is the user logged in? (used to filter private links)
  55     private $_loggedIn;
  56
  57     // Hide public links
  58     private $_hidePublicLinks;
  59
  60     /**
  61      * Creates a new LinkDB
  62      *
  63      * Checks if the datastore exists; else, attempts to create a dummy one.
  64      *
  65      * @param $isLoggedIn is the user logged in?
  66      */
  67     function __construct($datastore, $isLoggedIn, $hidePublicLinks)
  68     {
  69         $this->_datastore = $datastore;
  70         $this->_loggedIn = $isLoggedIn;
  71         $this->_hidePublicLinks = $hidePublicLinks;
  72         $this->_checkDB();
  73         $this->_readDB();
  74     }
  75
  76     /**
  77      * Countable - Counts elements of an object
  78      */
  79     public function count()
  80     {
  81         return count($this->_links);
  82     }
  83
  84     /**
  85      * ArrayAccess - Assigns a value to the specified offset
  86      */
  87     public function offsetSet($offset, $value)
  88     {
  89         // TODO: use exceptions instead of "die"
  90         if (!$this->_loggedIn) {
  91             die('You are not authorized to add a link.');
  92         }
  93         if (empty($value['linkdate']) || empty($value['url'])) {
  94             die('Internal Error: A link should always have a linkdate and URL.');
  95         }
  96         if (empty($offset)) {
  97             die('You must specify a key.');
  98         }
  99         $this->_links[$offset] = $value;
 100         $this->_urls[$value['url']]=$offset;
 101     }
 102
 103     /**
 104      * ArrayAccess - Whether or not an offset exists
 105      */
 106     public function offsetExists($offset)
 107     {
 108         return array_key_exists($offset, $this->_links);
 109     }
 110
 111     /**
 112      * ArrayAccess - Unsets an offset
 113      */
 114     public function offsetUnset($offset)
 115     {
 116         if (!$this->_loggedIn) {
 117             // TODO: raise an exception
 118             die('You are not authorized to delete a link.');
 119         }
 120         $url = $this->_links[$offset]['url'];
 121         unset($this->_urls[$url]);
 122         unset($this->_links[$offset]);
 123     }
 124
 125     /**
 126      * ArrayAccess - Returns the value at specified offset
 127      */
 128     public function offsetGet($offset)
 129     {
 130         return isset($this->_links[$offset]) ? $this->_links[$offset] : null;
 131     }
 132
 133     /**
 134      * Iterator - Returns the current element
 135      */
 136     function current()
 137     {
 138         return $this->_links[$this->_keys[$this->_position]];
 139     }
 140
 141     /**
 142      * Iterator - Returns the key of the current element
 143      */
 144     function key()
 145     {
 146         return $this->_keys[$this->_position];
 147     }
 148
 149     /**
 150      * Iterator - Moves forward to next element
 151      */
 152     function next()
 153     {
 154         ++$this->_position;
 155     }
 156
 157     /**
 158      * Iterator - Rewinds the Iterator to the first element
 159      *
 160      * Entries are sorted by date (latest first)
 161      */
 162     function rewind()
 163     {
 164         $this->_keys = array_keys($this->_links);
 165         rsort($this->_keys);
 166         $this->_position = 0;
 167     }
 168
 169     /**
 170      * Iterator - Checks if current position is valid
 171      */
 172     function valid()
 173     {
 174         return isset($this->_keys[$this->_position]);
 175     }
 176
 177     /**
 178      * Checks if the DB directory and file exist
 179      *
 180      * If no DB file is found, creates a dummy DB.
 181      */
 182     private function _checkDB()
 183     {
 184         if (file_exists($this->_datastore)) {
 185             return;
 186         }
 187
 188         // Create a dummy database for example
 189         $this->_links = array();
 190         $link = array(
 191             'title'=>' Shaarli: the personal, minimalist, super-fast, no-database delicious clone',
 192             'url'=>'https://github.com/shaarli/Shaarli/wiki',
 193             'description'=>'Welcome to Shaarli! This is your first public bookmark. To edit or delete me, you must first login.
 194
 195 To learn how to use Shaarli, consult the link "Help/documentation" at the bottom of this page.
 196
 197 You use the community supported version of the original Shaarli project, by Sebastien Sauvage.',
 198             'private'=>0,
 199             'linkdate'=> date('Ymd_His'),
 200             'tags'=>'opensource software'
 201         );
 202         $this->_links[$link['linkdate']] = $link;
 203
 204         $link = array(
 205             'title'=>'My secret stuff... - Pastebin.com',
 206             'url'=>'http://sebsauvage.net/paste/?8434b27936c09649#bR7XsXhoTiLcqCpQbmOpBi3rq2zzQUC5hBI7ZT1O3x8=',
 207             'description'=>'Shhhh! I\'m a private link only YOU can see. You can delete me too.',
 208             'private'=>1,
 209             'linkdate'=> date('Ymd_His', strtotime('-1 minute')),
 210             'tags'=>'secretstuff'
 211         );
 212         $this->_links[$link['linkdate']] = $link;
 213
 214         // Write database to disk
 215         // TODO: raise an exception if the file is not write-able
 216         file_put_contents(
 217             $this->_datastore,
 218             self::$phpPrefix.base64_encode(gzdeflate(serialize($this->_links))).self::$phpSuffix
 219         );
 220     }
 221
 222     /**
 223      * Reads database from disk to memory
 224      */
 225     private function _readDB()
 226     {
 227
 228         // Public links are hidden and user not logged in => nothing to show
 229         if ($this->_hidePublicLinks && !$this->_loggedIn) {
 230             $this->_links = array();
 231             return;
 232         }
 233
 234         // Read data
 235         // Note that gzinflate is faster than gzuncompress.
 236         // See: http://www.php.net/manual/en/function.gzdeflate.php#96439
 237         $this->_links = array();
 238
 239         if (file_exists($this->_datastore)) {
 240             $this->_links = unserialize(gzinflate(base64_decode(
 241                 substr(file_get_contents($this->_datastore),
 242                        strlen(self::$phpPrefix), -strlen(self::$phpSuffix)))));
 243         }
 244
 245         // If user is not logged in, filter private links.
 246         if (!$this->_loggedIn) {
 247             $toremove = array();
 248             foreach ($this->_links as $link) {
 249                 if ($link['private'] != 0) {
 250                     $toremove[] = $link['linkdate'];
 251                 }
 252             }
 253             foreach ($toremove as $linkdate) {
 254                 unset($this->_links[$linkdate]);
 255             }
 256         }
 257
 258         // Keep the list of the mapping URLs-->linkdate up-to-date.
 259         $this->_urls = array();
 260         foreach ($this->_links as $link) {
 261             $this->_urls[$link['url']] = $link['linkdate'];
 262         }
 263
 264         // Escape links data
 265         foreach($this->_links as &$link) {
 266             sanitizeLink($link);
 267         }
 268     }
 269
 270     /**
 271      * Saves the database from memory to disk
 272      *
 273      * @param string $pageCacheDir page cache directory
 274      */
 275     public function savedb($pageCacheDir)
 276     {
 277         if (!$this->_loggedIn) {
 278             // TODO: raise an Exception instead
 279             die('You are not authorized to change the database.');
 280         }
 281         file_put_contents(
 282             $this->_datastore,
 283             self::$phpPrefix.base64_encode(gzdeflate(serialize($this->_links))).self::$phpSuffix
 284         );
 285         invalidateCaches($pageCacheDir);
 286     }
 287
 288     /**
 289      * Returns the link for a given URL, or False if it does not exist.
 290      *
 291      * @param string $url URL to search for
 292      *
 293      * @return mixed the existing link if it exists, else 'false'
 294      */
 295     public function getLinkFromUrl($url)
 296     {
 297         if (isset($this->_urls[$url])) {
 298             return $this->_links[$this->_urls[$url]];
 299         }
 300         return false;
 301     }
 302
 303     /**
 304      * Returns the list of links corresponding to a full-text search
 305      *
 306      * Searches:
 307      *  - in the URLs, title and description;
 308      *  - are case-insensitive.
 309      *
 310      * Example:
 311      *    print_r($mydb->filterFulltext('hollandais'));
 312      *
 313      * mb_convert_case($val, MB_CASE_LOWER, 'UTF-8')
 314      *  - allows to perform searches on Unicode text
 315      *  - see https://github.com/shaarli/Shaarli/issues/75 for examples
 316      */
 317     public function filterFulltext($searchterms)
 318     {
 319         // FIXME: explode(' ',$searchterms) and perform a AND search.
 320         // FIXME: accept double-quotes to search for a string "as is"?
 321         $filtered = array();
 322         $search = mb_convert_case($searchterms, MB_CASE_LOWER, 'UTF-8');
 323         $keys = array('title', 'description', 'url', 'tags');
 324
 325         foreach ($this->_links as $link) {
 326             $found = false;
 327
 328             foreach ($keys as $key) {
 329                 if (strpos(mb_convert_case($link[$key], MB_CASE_LOWER, 'UTF-8'),
 330                            $search) !== false) {
 331                     $found = true;
 332                 }
 333             }
 334
 335             if ($found) {
 336                 $filtered[$link['linkdate']] = $link;
 337             }
 338         }
 339         krsort($filtered);
 340         return $filtered;
 341     }
 342
 343     /**
 344      * Returns the list of links associated with a given list of tags
 345      *
 346      * You can specify one or more tags, separated by space or a comma, e.g.
 347      *  print_r($mydb->filterTags('linux programming'));
 348      */
 349     public function filterTags($tags, $casesensitive=false)
 350     {
 351         // Same as above, we use UTF-8 conversion to handle various graphemes (i.e. cyrillic, or greek)
 352         // FIXME: is $casesensitive ever true?
 353         $t = str_replace(
 354             ',', ' ',
 355             ($casesensitive ? $tags : mb_convert_case($tags, MB_CASE_LOWER, 'UTF-8'))
 356         );
 357
 358         $searchtags = explode(' ', $t);
 359         $filtered = array();
 360
 361         foreach ($this->_links as $l) {
 362             $linktags = explode(
 363                 ' ',
 364                 ($casesensitive ? $l['tags']:mb_convert_case($l['tags'], MB_CASE_LOWER, 'UTF-8'))
 365             );
 366
 367             if (count(array_intersect($linktags, $searchtags)) == count($searchtags)) {
 368                 $filtered[$l['linkdate']] = $l;
 369             }
 370         }
 371         krsort($filtered);
 372         return $filtered;
 373     }
 374
 375
 376     /**
 377      * Returns the list of articles for a given day, chronologically sorted
 378      *
 379      * Day must be in the form 'YYYYMMDD' (e.g. '20120125'), e.g.
 380      *  print_r($mydb->filterDay('20120125'));
 381      */
 382     public function filterDay($day)
 383     {
 384         if (! checkDateFormat('Ymd', $day)) {
 385             throw new Exception('Invalid date format');
 386         }
 387
 388         $filtered = array();
 389         foreach ($this->_links as $l) {
 390             if (startsWith($l['linkdate'], $day)) {
 391                 $filtered[$l['linkdate']] = $l;
 392             }
 393         }
 394         ksort($filtered);
 395         return $filtered;
 396     }
 397
 398     /**
 399      * Returns the article corresponding to a smallHash
 400      */
 401     public function filterSmallHash($smallHash)
 402     {
 403         $filtered = array();
 404         foreach ($this->_links as $l) {
 405             if ($smallHash == smallHash($l['linkdate'])) {
 406                 // Yes, this is ugly and slow
 407                 $filtered[$l['linkdate']] = $l;
 408                 return $filtered;
 409             }
 410         }
 411         return $filtered;
 412     }
 413
 414     /**
 415      * Returns the list of all tags
 416      * Output: associative array key=tags, value=0
 417      */
 418     public function allTags()
 419     {
 420         $tags = array();
 421         foreach ($this->_links as $link) {
 422             foreach (explode(' ', $link['tags']) as $tag) {
 423                 if (!empty($tag)) {
 424                     $tags[$tag] = (empty($tags[$tag]) ? 1 : $tags[$tag] + 1);
 425                 }
 426             }
 427         }
 428         // Sort tags by usage (most used tag first)
 429         arsort($tags);
 430         return $tags;
 431     }
 432
 433     /**
 434      * Returns the list of days containing articles (oldest first)
 435      * Output: An array containing days (in format YYYYMMDD).
 436      */
 437     public function days()
 438     {
 439         $linkDays = array();
 440         foreach (array_keys($this->_links) as $day) {
 441             $linkDays[substr($day, 0, 8)] = 0;
 442         }
 443         $linkDays = array_keys($linkDays);
 444         sort($linkDays);
 445         return $linkDays;
 446     }
 447 }