X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=application%2FLinkDB.php;h=2d42c51420bd015941e3392b724807706c3f5d4c;hb=628b97cbdf276785eb9ff4f7a124e81e67d2f76c;hp=463aa47ef5ef10ecac945206c77e9de48e823e8b;hpb=a3b1b4ae709bd3c5a8ec6fe29aa36fd01c52bca5;p=github%2Fshaarli%2FShaarli.git diff --git a/application/LinkDB.php b/application/LinkDB.php index 463aa47e..2d42c514 100644 --- a/application/LinkDB.php +++ b/application/LinkDB.php @@ -12,13 +12,16 @@ * * Available keys: * - description: description of the entry - * - linkdate: date of the creation of this entry, in the form YYYYMMDD_HHMMSS + * - linkdate: creation date of this entry, format: YYYYMMDD_HHMMSS * (e.g.'20110914_192317') + * - updated: last modification date of this entry, format: YYYYMMDD_HHMMSS * - private: Is this link private? 0=no, other value=yes * - tags: tags attached to this entry (separated by spaces) * - title Title of the link - * - url URL of the link. Can be absolute or relative. + * - url URL of the link. Used for displayable links (no redirector, relative, etc.). + * Can be absolute or relative. * Relative URLs are permalinks (e.g.'?m-ukcw') + * - real_url Absolute processed URL. * * Implements 3 interfaces: * - ArrayAccess: behaves like an associative array; @@ -28,7 +31,10 @@ class LinkDB implements Iterator, Countable, ArrayAccess { // Links are stored as a PHP serialized string - private $_datastore; + private $datastore; + + // Link date storage format + const LINK_DATE_FORMAT = 'Ymd_His'; // Datastore PHP prefix protected static $phpPrefix = 'linkdate) - private $_urls; + private $urls; // List of linkdate keys (for the Iterator interface implementation) - private $_keys; + private $keys; - // Position in the $this->_keys array (for the Iterator interface) - private $_position; + // Position in the $this->keys array (for the Iterator interface) + private $position; // Is the user logged in? (used to filter private links) - private $_loggedIn; + private $loggedIn; // Hide public links - private $_hidePublicLinks; + private $hidePublicLinks; + + // link redirector set in user settings. + private $redirector; + + /** + * Set this to `true` to urlencode link behind redirector link, `false` to leave it untouched. + * + * Example: + * anonym.to needs clean URL while dereferer.org needs urlencoded URL. + * + * @var boolean $redirectorEncode parameter: true or false + */ + private $redirectorEncode; /** * Creates a new LinkDB * * Checks if the datastore exists; else, attempts to create a dummy one. * - * @param $isLoggedIn is the user logged in? + * @param string $datastore datastore file path. + * @param boolean $isLoggedIn is the user logged in? + * @param boolean $hidePublicLinks if true all links are private. + * @param string $redirector link redirector set in user settings. + * @param boolean $redirectorEncode Enable urlencode on redirected urls (default: true). */ - function __construct($datastore, $isLoggedIn, $hidePublicLinks) + function __construct( + $datastore, + $isLoggedIn, + $hidePublicLinks, + $redirector = '', + $redirectorEncode = true + ) { - $this->_datastore = $datastore; - $this->_loggedIn = $isLoggedIn; - $this->_hidePublicLinks = $hidePublicLinks; - $this->_checkDB(); - $this->_readDB(); + $this->datastore = $datastore; + $this->loggedIn = $isLoggedIn; + $this->hidePublicLinks = $hidePublicLinks; + $this->redirector = $redirector; + $this->redirectorEncode = $redirectorEncode === true; + $this->checkDB(); + $this->readDB(); } /** @@ -78,7 +109,7 @@ class LinkDB implements Iterator, Countable, ArrayAccess */ public function count() { - return count($this->_links); + return count($this->links); } /** @@ -87,7 +118,7 @@ class LinkDB implements Iterator, Countable, ArrayAccess public function offsetSet($offset, $value) { // TODO: use exceptions instead of "die" - if (!$this->_loggedIn) { + if (!$this->loggedIn) { die('You are not authorized to add a link.'); } if (empty($value['linkdate']) || empty($value['url'])) { @@ -96,8 +127,8 @@ class LinkDB implements Iterator, Countable, ArrayAccess if (empty($offset)) { die('You must specify a key.'); } - $this->_links[$offset] = $value; - $this->_urls[$value['url']]=$offset; + $this->links[$offset] = $value; + $this->urls[$value['url']]=$offset; } /** @@ -105,7 +136,7 @@ class LinkDB implements Iterator, Countable, ArrayAccess */ public function offsetExists($offset) { - return array_key_exists($offset, $this->_links); + return array_key_exists($offset, $this->links); } /** @@ -113,13 +144,13 @@ class LinkDB implements Iterator, Countable, ArrayAccess */ public function offsetUnset($offset) { - if (!$this->_loggedIn) { + if (!$this->loggedIn) { // TODO: raise an exception die('You are not authorized to delete a link.'); } - $url = $this->_links[$offset]['url']; - unset($this->_urls[$url]); - unset($this->_links[$offset]); + $url = $this->links[$offset]['url']; + unset($this->urls[$url]); + unset($this->links[$offset]); } /** @@ -127,7 +158,7 @@ class LinkDB implements Iterator, Countable, ArrayAccess */ public function offsetGet($offset) { - return isset($this->_links[$offset]) ? $this->_links[$offset] : null; + return isset($this->links[$offset]) ? $this->links[$offset] : null; } /** @@ -135,7 +166,7 @@ class LinkDB implements Iterator, Countable, ArrayAccess */ function current() { - return $this->_links[$this->_keys[$this->_position]]; + return $this->links[$this->keys[$this->position]]; } /** @@ -143,7 +174,7 @@ class LinkDB implements Iterator, Countable, ArrayAccess */ function key() { - return $this->_keys[$this->_position]; + return $this->keys[$this->position]; } /** @@ -151,7 +182,7 @@ class LinkDB implements Iterator, Countable, ArrayAccess */ function next() { - ++$this->_position; + ++$this->position; } /** @@ -161,9 +192,9 @@ class LinkDB implements Iterator, Countable, ArrayAccess */ function rewind() { - $this->_keys = array_keys($this->_links); - rsort($this->_keys); - $this->_position = 0; + $this->keys = array_keys($this->links); + rsort($this->keys); + $this->position = 0; } /** @@ -171,7 +202,7 @@ class LinkDB implements Iterator, Countable, ArrayAccess */ function valid() { - return isset($this->_keys[$this->_position]); + return isset($this->keys[$this->position]); } /** @@ -179,14 +210,14 @@ class LinkDB implements Iterator, Countable, ArrayAccess * * If no DB file is found, creates a dummy DB. */ - private function _checkDB() + private function checkDB() { - if (file_exists($this->_datastore)) { + if (file_exists($this->datastore)) { return; } // Create a dummy database for example - $this->_links = array(); + $this->links = array(); $link = array( 'title'=>' Shaarli: the personal, minimalist, super-fast, no-database delicious clone', 'url'=>'https://github.com/shaarli/Shaarli/wiki', @@ -199,7 +230,7 @@ You use the community supported version of the original Shaarli project, by Seba 'linkdate'=> date('Ymd_His'), 'tags'=>'opensource software' ); - $this->_links[$link['linkdate']] = $link; + $this->links[$link['linkdate']] = $link; $link = array( 'title'=>'My secret stuff... - Pastebin.com', @@ -209,62 +240,96 @@ You use the community supported version of the original Shaarli project, by Seba 'linkdate'=> date('Ymd_His', strtotime('-1 minute')), 'tags'=>'secretstuff' ); - $this->_links[$link['linkdate']] = $link; + $this->links[$link['linkdate']] = $link; // Write database to disk - // TODO: raise an exception if the file is not write-able - file_put_contents( - $this->_datastore, - self::$phpPrefix.base64_encode(gzdeflate(serialize($this->_links))).self::$phpSuffix - ); + $this->writeDB(); } /** * Reads database from disk to memory */ - private function _readDB() + private function readDB() { // Public links are hidden and user not logged in => nothing to show - if ($this->_hidePublicLinks && !$this->_loggedIn) { - $this->_links = array(); + if ($this->hidePublicLinks && !$this->loggedIn) { + $this->links = array(); return; } // Read data // Note that gzinflate is faster than gzuncompress. // See: http://www.php.net/manual/en/function.gzdeflate.php#96439 - $this->_links = array(); + $this->links = array(); - if (file_exists($this->_datastore)) { - $this->_links = unserialize(gzinflate(base64_decode( - substr(file_get_contents($this->_datastore), + if (file_exists($this->datastore)) { + $this->links = unserialize(gzinflate(base64_decode( + substr(file_get_contents($this->datastore), strlen(self::$phpPrefix), -strlen(self::$phpSuffix))))); } // If user is not logged in, filter private links. - if (!$this->_loggedIn) { + if (!$this->loggedIn) { $toremove = array(); - foreach ($this->_links as $link) { + foreach ($this->links as $link) { if ($link['private'] != 0) { $toremove[] = $link['linkdate']; } } foreach ($toremove as $linkdate) { - unset($this->_links[$linkdate]); + unset($this->links[$linkdate]); } } - // Keep the list of the mapping URLs-->linkdate up-to-date. - $this->_urls = array(); - foreach ($this->_links as $link) { - $this->_urls[$link['url']] = $link['linkdate']; + $this->urls = array(); + foreach ($this->links as &$link) { + // Keep the list of the mapping URLs-->linkdate up-to-date. + $this->urls[$link['url']] = $link['linkdate']; + + // Sanitize data fields. + sanitizeLink($link); + + // Remove private tags if the user is not logged in. + if (! $this->loggedIn) { + $link['tags'] = preg_replace('/(^|\s+)\.[^($|\s)]+\s*/', ' ', $link['tags']); + } + + // Do not use the redirector for internal links (Shaarli note URL starting with a '?'). + if (!empty($this->redirector) && !startsWith($link['url'], '?')) { + $link['real_url'] = $this->redirector; + if ($this->redirectorEncode) { + $link['real_url'] .= urlencode(unescape($link['url'])); + } else { + $link['real_url'] .= $link['url']; + } + } + else { + $link['real_url'] = $link['url']; + } } + } - // Escape links data - foreach($this->_links as &$link) { - sanitizeLink($link); + /** + * Saves the database from memory to disk + * + * @throws IOException the datastore is not writable + */ + private function writeDB() + { + if (is_file($this->datastore) && !is_writeable($this->datastore)) { + // The datastore exists but is not writeable + throw new IOException($this->datastore); + } else if (!is_file($this->datastore) && !is_writeable(dirname($this->datastore))) { + // The datastore does not exist and its parent directory is not writeable + throw new IOException(dirname($this->datastore)); } + + file_put_contents( + $this->datastore, + self::$phpPrefix.base64_encode(gzdeflate(serialize($this->links))).self::$phpSuffix + ); + } /** @@ -274,137 +339,99 @@ You use the community supported version of the original Shaarli project, by Seba */ public function savedb($pageCacheDir) { - if (!$this->_loggedIn) { + if (!$this->loggedIn) { // TODO: raise an Exception instead die('You are not authorized to change the database.'); } - file_put_contents( - $this->_datastore, - self::$phpPrefix.base64_encode(gzdeflate(serialize($this->_links))).self::$phpSuffix - ); + + $this->writeDB(); + invalidateCaches($pageCacheDir); } /** * Returns the link for a given URL, or False if it does not exist. + * + * @param string $url URL to search for + * + * @return mixed the existing link if it exists, else 'false' */ public function getLinkFromUrl($url) { - if (isset($this->_urls[$url])) { - return $this->_links[$this->_urls[$url]]; + if (isset($this->urls[$url])) { + return $this->links[$this->urls[$url]]; } return false; } /** - * Returns the list of links corresponding to a full-text search + * Returns the shaare corresponding to a smallHash. * - * Searches: - * - in the URLs, title and description; - * - are case-insensitive. + * @param string $request QUERY_STRING server parameter. * - * Example: - * print_r($mydb->filterFulltext('hollandais')); + * @return array $filtered array containing permalink data. * - * mb_convert_case($val, MB_CASE_LOWER, 'UTF-8') - * - allows to perform searches on Unicode text - * - see https://github.com/shaarli/Shaarli/issues/75 for examples + * @throws LinkNotFoundException if the smallhash is malformed or doesn't match any link. */ - public function filterFulltext($searchterms) + public function filterHash($request) { - // FIXME: explode(' ',$searchterms) and perform a AND search. - // FIXME: accept double-quotes to search for a string "as is"? - $filtered = array(); - $search = mb_convert_case($searchterms, MB_CASE_LOWER, 'UTF-8'); - $keys = array('title', 'description', 'url', 'tags'); - - foreach ($this->_links as $link) { - $found = false; - - foreach ($keys as $key) { - if (strpos(mb_convert_case($link[$key], MB_CASE_LOWER, 'UTF-8'), - $search) !== false) { - $found = true; - } - } - - if ($found) { - $filtered[$link['linkdate']] = $link; - } - } - krsort($filtered); - return $filtered; + $request = substr($request, 0, 6); + $linkFilter = new LinkFilter($this->links); + return $linkFilter->filter(LinkFilter::$FILTER_HASH, $request); } /** - * Returns the list of links associated with a given list of tags + * Returns the list of articles for a given day. * - * You can specify one or more tags, separated by space or a comma, e.g. - * print_r($mydb->filterTags('linux programming')); + * @param string $request day to filter. Format: YYYYMMDD. + * + * @return array list of shaare found. */ - public function filterTags($tags, $casesensitive=false) - { - // Same as above, we use UTF-8 conversion to handle various graphemes (i.e. cyrillic, or greek) - // FIXME: is $casesensitive ever true? - $t = str_replace( - ',', ' ', - ($casesensitive ? $tags : mb_convert_case($tags, MB_CASE_LOWER, 'UTF-8')) - ); - - $searchtags = explode(' ', $t); - $filtered = array(); - - foreach ($this->_links as $l) { - $linktags = explode( - ' ', - ($casesensitive ? $l['tags']:mb_convert_case($l['tags'], MB_CASE_LOWER, 'UTF-8')) - ); - - if (count(array_intersect($linktags, $searchtags)) == count($searchtags)) { - $filtered[$l['linkdate']] = $l; - } - } - krsort($filtered); - return $filtered; + public function filterDay($request) { + $linkFilter = new LinkFilter($this->links); + return $linkFilter->filter(LinkFilter::$FILTER_DAY, $request); } - /** - * Returns the list of articles for a given day, chronologically sorted + * Filter links according to search parameters. * - * Day must be in the form 'YYYYMMDD' (e.g. '20120125'), e.g. - * print_r($mydb->filterDay('20120125')); + * @param array $filterRequest Search request content. Supported keys: + * - searchtags: list of tags + * - searchterm: term search + * @param bool $casesensitive Optional: Perform case sensitive filter + * @param bool $privateonly Optional: Returns private links only if true. + * + * @return array filtered links, all links if no suitable filter was provided. */ - public function filterDay($day) + public function filterSearch($filterRequest = array(), $casesensitive = false, $privateonly = false) { - if (! checkDateFormat('Ymd', $day)) { - throw new Exception('Invalid date format'); + // Filter link database according to parameters. + $searchtags = !empty($filterRequest['searchtags']) ? escape($filterRequest['searchtags']) : ''; + $searchterm = !empty($filterRequest['searchterm']) ? escape($filterRequest['searchterm']) : ''; + + // Search tags + fullsearch. + if (! empty($searchtags) && ! empty($searchterm)) { + $type = LinkFilter::$FILTER_TAG | LinkFilter::$FILTER_TEXT; + $request = array($searchtags, $searchterm); } - - $filtered = array(); - foreach ($this->_links as $l) { - if (startsWith($l['linkdate'], $day)) { - $filtered[$l['linkdate']] = $l; - } + // Search by tags. + elseif (! empty($searchtags)) { + $type = LinkFilter::$FILTER_TAG; + $request = $searchtags; } - ksort($filtered); - return $filtered; - } - - /** - * Returns the article corresponding to a smallHash - */ - public function filterSmallHash($smallHash) - { - $filtered = array(); - foreach ($this->_links as $l) { - if ($smallHash == smallHash($l['linkdate'])) { - // Yes, this is ugly and slow - $filtered[$l['linkdate']] = $l; - return $filtered; - } + // Fulltext search. + elseif (! empty($searchterm)) { + $type = LinkFilter::$FILTER_TEXT; + $request = $searchterm; + } + // Otherwise, display without filtering. + else { + $type = ''; + $request = ''; } - return $filtered; + + $linkFilter = new LinkFilter($this->links); + return $linkFilter->filter($type, $request, $casesensitive, $privateonly); } /** @@ -414,11 +441,18 @@ You use the community supported version of the original Shaarli project, by Seba public function allTags() { $tags = array(); - foreach ($this->_links as $link) { - foreach (explode(' ', $link['tags']) as $tag) { - if (!empty($tag)) { - $tags[$tag] = (empty($tags[$tag]) ? 1 : $tags[$tag] + 1); + $caseMapping = array(); + foreach ($this->links as $link) { + foreach (preg_split('/\s+/', $link['tags'], 0, PREG_SPLIT_NO_EMPTY) as $tag) { + if (empty($tag)) { + continue; + } + // The first case found will be displayed. + if (!isset($caseMapping[strtolower($tag)])) { + $caseMapping[strtolower($tag)] = $tag; + $tags[$caseMapping[strtolower($tag)]] = 0; } + $tags[$caseMapping[strtolower($tag)]]++; } } // Sort tags by usage (most used tag first) @@ -433,11 +467,12 @@ You use the community supported version of the original Shaarli project, by Seba public function days() { $linkDays = array(); - foreach (array_keys($this->_links) as $day) { + foreach (array_keys($this->links) as $day) { $linkDays[substr($day, 0, 8)] = 0; } $linkDays = array_keys($linkDays); sort($linkDays); + return $linkDays; } }