X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=application%2FLinkDB.php;h=87a53e2e99563edf6749f52c763053599b40a399;hb=f3d2f257946e2a3c8791c1ba99b379acbe934fec;hp=9f4d3e3cc7b4fabb23750fda02427ad23612b14d;hpb=195acf9f0931aa3b1a6db8832ab52a260d11090d;p=github%2Fshaarli%2FShaarli.git diff --git a/application/LinkDB.php b/application/LinkDB.php index 9f4d3e3c..87a53e2e 100644 --- a/application/LinkDB.php +++ b/application/LinkDB.php @@ -1,4 +1,7 @@ '; + // Link date storage format + const LINK_DATE_FORMAT = 'Ymd_His'; // List of links (associative array) // - key: link date (e.g. "20110823_124546"), // - value: associative array (keys: title, description...) - private $_links; + private $links; + + // List of all recorded URLs (key=url, value=link offset) + // for fast reserve search (url-->link offset) + private $urls; - // List of all recorded URLs (key=url, value=linkdate) - // for fast reserve search (url-->linkdate) - private $_urls; + /** + * @var array List of all links IDS mapped with their array offset. + * Map: id->offset. + */ + protected $ids; - // List of linkdate keys (for the Iterator interface implementation) - private $_keys; + // List of offset keys (for the Iterator interface implementation) + 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; + 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 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 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, $redirector = '') - { - $this->_datastore = $datastore; - $this->_loggedIn = $isLoggedIn; - $this->_hidePublicLinks = $hidePublicLinks; - $this->_redirector = $redirector; - $this->_checkDB(); - $this->_readDB(); + public function __construct( + $datastore, + $isLoggedIn, + $hidePublicLinks, + $redirector = '', + $redirectorEncode = true + ) { + $this->datastore = $datastore; + $this->loggedIn = $isLoggedIn; + $this->hidePublicLinks = $hidePublicLinks; + $this->redirector = $redirector; + $this->redirectorEncode = $redirectorEncode === true; + $this->check(); + $this->read(); } /** @@ -87,7 +125,7 @@ class LinkDB implements Iterator, Countable, ArrayAccess */ public function count() { - return count($this->_links); + return count($this->links); } /** @@ -96,17 +134,29 @@ class LinkDB implements Iterator, Countable, ArrayAccess public function offsetSet($offset, $value) { // TODO: use exceptions instead of "die" - if (!$this->_loggedIn) { - die('You are not authorized to add a link.'); + if (!$this->loggedIn) { + die(t('You are not authorized to add a link.')); + } + if (!isset($value['id']) || empty($value['url'])) { + die(t('Internal Error: A link should always have an id and URL.')); + } + if (($offset !== null && ! is_int($offset)) || ! is_int($value['id'])) { + die(t('You must specify an integer as a key.')); } - if (empty($value['linkdate']) || empty($value['url'])) { - die('Internal Error: A link should always have a linkdate and URL.'); + if ($offset !== null && $offset !== $value['id']) { + die(t('Array offset and link ID must be equal.')); } - if (empty($offset)) { - die('You must specify a key.'); + + // If the link exists, we reuse the real offset, otherwise new entry + $existing = $this->getLinkOffset($offset); + if ($existing !== null) { + $offset = $existing; + } else { + $offset = count($this->links); } - $this->_links[$offset] = $value; - $this->_urls[$value['url']]=$offset; + $this->links[$offset] = $value; + $this->urls[$value['url']] = $offset; + $this->ids[$value['id']] = $offset; } /** @@ -114,7 +164,7 @@ class LinkDB implements Iterator, Countable, ArrayAccess */ public function offsetExists($offset) { - return array_key_exists($offset, $this->_links); + return array_key_exists($this->getLinkOffset($offset), $this->links); } /** @@ -122,13 +172,15 @@ class LinkDB implements Iterator, Countable, ArrayAccess */ public function offsetUnset($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]); + $realOffset = $this->getLinkOffset($offset); + $url = $this->links[$realOffset]['url']; + unset($this->urls[$url]); + unset($this->ids[$realOffset]); + unset($this->links[$realOffset]); } /** @@ -136,31 +188,32 @@ class LinkDB implements Iterator, Countable, ArrayAccess */ public function offsetGet($offset) { - return isset($this->_links[$offset]) ? $this->_links[$offset] : null; + $realOffset = $this->getLinkOffset($offset); + return isset($this->links[$realOffset]) ? $this->links[$realOffset] : null; } /** * Iterator - Returns the current element */ - function current() + public function current() { - return $this->_links[$this->_keys[$this->_position]]; + return $this[$this->keys[$this->position]]; } /** * Iterator - Returns the key of the current element */ - function key() + public function key() { - return $this->_keys[$this->_position]; + return $this->keys[$this->position]; } /** * Iterator - Moves forward to next element */ - function next() + public function next() { - ++$this->_position; + ++$this->position; } /** @@ -168,19 +221,18 @@ class LinkDB implements Iterator, Countable, ArrayAccess * * Entries are sorted by date (latest first) */ - function rewind() + public function rewind() { - $this->_keys = array_keys($this->_links); - rsort($this->_keys); - $this->_position = 0; + $this->keys = array_keys($this->ids); + $this->position = 0; } /** * Iterator - Checks if current position is valid */ - function valid() + public function valid() { - return isset($this->_keys[$this->_position]); + return isset($this->keys[$this->position]); } /** @@ -188,98 +240,104 @@ class LinkDB implements Iterator, Countable, ArrayAccess * * If no DB file is found, creates a dummy DB. */ - private function _checkDB() + private function check() { - if (file_exists($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', - 'description'=>'Welcome to Shaarli! This is your first public bookmark. To edit or delete me, you must first login. + 'id' => 1, + 'title'=> t('The personal, minimalist, super-fast, database free, bookmarking service'), + 'url'=>'https://shaarli.readthedocs.io', + 'description'=>t( + 'Welcome to Shaarli! This is your first public bookmark. ' + .'To edit or delete me, you must first login. -To learn how to use Shaarli, consult the link "Help/documentation" at the bottom of this page. +To learn how to use Shaarli, consult the link "Documentation" at the bottom of this page. -You use the community supported version of the original Shaarli project, by Sebastien Sauvage.', +You use the community supported version of the original Shaarli project, by Sebastien Sauvage.' + ), 'private'=>0, - 'linkdate'=> date('Ymd_His'), + 'created'=> new DateTime(), 'tags'=>'opensource software' ); - $this->_links[$link['linkdate']] = $link; + $link['shorturl'] = link_small_hash($link['created'], $link['id']); + $this->links[1] = $link; $link = array( - 'title'=>'My secret stuff... - Pastebin.com', + 'id' => 0, + 'title'=> t('My secret stuff... - Pastebin.com'), 'url'=>'http://sebsauvage.net/paste/?8434b27936c09649#bR7XsXhoTiLcqCpQbmOpBi3rq2zzQUC5hBI7ZT1O3x8=', - 'description'=>'Shhhh! I\'m a private link only YOU can see. You can delete me too.', + 'description'=> t('Shhhh! I\'m a private link only YOU can see. You can delete me too.'), 'private'=>1, - 'linkdate'=> date('Ymd_His', strtotime('-1 minute')), - 'tags'=>'secretstuff' + 'created'=> new DateTime('1 minute ago'), + 'tags'=>'secretstuff', ); - $this->_links[$link['linkdate']] = $link; + $link['shorturl'] = link_small_hash($link['created'], $link['id']); + $this->links[0] = $link; // Write database to disk - $this->writeDB(); + $this->write(); } /** * Reads database from disk to memory */ - private function _readDB() + private function read() { - // Public links are hidden and user not logged in => nothing to show - if ($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->urls = []; + $this->ids = []; + $this->links = FileUtils::readFlatDB($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) { - $toremove = array(); - foreach ($this->_links as $link) { - if ($link['private'] != 0) { - $toremove[] = $link['linkdate']; - } - } - foreach ($toremove as $linkdate) { - unset($this->_links[$linkdate]); + $toremove = array(); + foreach ($this->links as $key => &$link) { + if (! $this->loggedIn && $link['private'] != 0) { + // Transition for not upgraded databases. + unset($this->links[$key]); + continue; } - } - - $this->_urls = array(); - foreach ($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) { - $link['tags'] = preg_replace('/(^| )\.[^($| )]+/', '', $link['tags']); + 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'] = $this->_redirector . urlencode($link['url']); - } - else { + 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'] = $link['url']; } + + // To be able to load links before running the update, and prepare the update + if (! isset($link['created'])) { + $link['id'] = $link['linkdate']; + $link['created'] = DateTime::createFromFormat(self::LINK_DATE_FORMAT, $link['linkdate']); + if (! empty($link['updated'])) { + $link['updated'] = DateTime::createFromFormat(self::LINK_DATE_FORMAT, $link['updated']); + } + $link['shorturl'] = smallHash($link['linkdate']); + } + + $this->urls[$link['url']] = $key; + $this->ids[$link['id']] = $key; } } @@ -288,21 +346,10 @@ You use the community supported version of the original Shaarli project, by Seba * * @throws IOException the datastore is not writable */ - private function writeDB() + private function write() { - if (is_file($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 - ); - + $this->reorder(); + FileUtils::writeFlatDB($this->datastore, $this->links); } /** @@ -310,14 +357,14 @@ You use the community supported version of the original Shaarli project, by Seba * * @param string $pageCacheDir page cache directory */ - public function savedb($pageCacheDir) + public function save($pageCacheDir) { - if (!$this->_loggedIn) { + if (!$this->loggedIn) { // TODO: raise an Exception instead die('You are not authorized to change the database.'); } - $this->writeDB(); + $this->write(); invalidateCaches($pageCacheDir); } @@ -331,48 +378,145 @@ You use the community supported version of the original Shaarli project, by Seba */ 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; } /** - * Filter links. + * Returns the shaare corresponding to a smallHash. * - * @param string $type Type of filter. - * @param mixed $request Search request, string or array. - * @param bool $casesensitive Optional: Perform case sensitive filter - * @param bool $privateonly Optional: Returns private links only if true. + * @param string $request QUERY_STRING server parameter. * - * @return array filtered links + * @return array $filtered array containing permalink data. + * + * @throws LinkNotFoundException if the smallhash is malformed or doesn't match any link. */ - public function filter($type = '', $request = '', $casesensitive = false, $privateonly = false) + public function filterHash($request) { - $linkFilter = new LinkFilter($this->_links); - $requestFilter = is_array($request) ? implode(' ', $request) : $request; - return $linkFilter->filter($type, trim($requestFilter), $casesensitive, $privateonly); + $request = substr($request, 0, 6); + $linkFilter = new LinkFilter($this->links); + return $linkFilter->filter(LinkFilter::$FILTER_HASH, $request); } /** - * Returns the list of all tags - * Output: associative array key=tags, value=0 + * Returns the list of articles for a given day. + * + * @param string $request day to filter. Format: YYYYMMDD. + * + * @return array list of shaare found. */ - public function allTags() + public function filterDay($request) { - $tags = array(); - foreach ($this->_links as $link) { - foreach (explode(' ', $link['tags']) as $tag) { - if (!empty($tag)) { - $tags[$tag] = (empty($tags[$tag]) ? 1 : $tags[$tag] + 1); + $linkFilter = new LinkFilter($this->links); + return $linkFilter->filter(LinkFilter::$FILTER_DAY, $request); + } + + /** + * Filter links according to search parameters. + * + * @param array $filterRequest Search request content. Supported keys: + * - searchtags: list of tags + * - searchterm: term search + * @param bool $casesensitive Optional: Perform case sensitive filter + * @param string $visibility return only all/private/public links + * @param string $untaggedonly return only untagged links + * + * @return array filtered links, all links if no suitable filter was provided. + */ + public function filterSearch( + $filterRequest = array(), + $casesensitive = false, + $visibility = 'all', + $untaggedonly = false + ) { + // Filter link database according to parameters. + $searchtags = isset($filterRequest['searchtags']) ? escape($filterRequest['searchtags']) : ''; + $searchterm = isset($filterRequest['searchterm']) ? escape($filterRequest['searchterm']) : ''; + + // Search tags + fullsearch - blank string parameter will return all links. + $type = LinkFilter::$FILTER_TAG | LinkFilter::$FILTER_TEXT; // == "vuotext" + $request = [$searchtags, $searchterm]; + + $linkFilter = new LinkFilter($this); + return $linkFilter->filter($type, $request, $casesensitive, $visibility, $untaggedonly); + } + + /** + * Returns the list tags appearing in the links with the given tags + * + * @param array $filteringTags tags selecting the links to consider + * @param string $visibility process only all/private/public links + * + * @return array tag => linksCount + */ + public function linksCountPerTag($filteringTags = [], $visibility = 'all') + { + $links = $this->filterSearch(['searchtags' => $filteringTags], false, $visibility); + $tags = []; + $caseMapping = []; + foreach ($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($tag)]]++; } } - // Sort tags by usage (most used tag first) - arsort($tags); + + /* + * Formerly used arsort(), which doesn't define the sort behaviour for equal values. + * Also, this function doesn't produce the same result between PHP 5.6 and 7. + * + * So we now use array_multisort() to sort tags by DESC occurrences, + * then ASC alphabetically for equal values. + * + * @see https://github.com/shaarli/Shaarli/issues/1142 + */ + $keys = array_keys($tags); + $tmpTags = array_combine($keys, $keys); + array_multisort($tags, SORT_DESC, $tmpTags, SORT_ASC, $tags); return $tags; } + /** + * Rename or delete a tag across all links. + * + * @param string $from Tag to rename + * @param string $to New tag. If none is provided, the from tag will be deleted + * + * @return array|bool List of altered links or false on error + */ + public function renameTag($from, $to) + { + if (empty($from)) { + return false; + } + $delete = empty($to); + // True for case-sensitive tag search. + $linksToAlter = $this->filterSearch(['searchtags' => $from], true); + foreach ($linksToAlter as $key => &$value) { + $tags = preg_split('/\s+/', trim($value['tags'])); + if (($pos = array_search($from, $tags)) !== false) { + if ($delete) { + unset($tags[$pos]); // Remove tag. + } else { + $tags[$pos] = trim($to); + } + $value['tags'] = trim(implode(' ', array_unique($tags))); + $this[$value['id']] = $value; + } + } + + return $linksToAlter; + } + /** * Returns the list of days containing articles (oldest first) * Output: An array containing days (in format YYYYMMDD). @@ -380,12 +524,67 @@ 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) { - $linkDays[substr($day, 0, 8)] = 0; + foreach ($this->links as $link) { + $linkDays[$link['created']->format('Ymd')] = 0; } $linkDays = array_keys($linkDays); sort($linkDays); return $linkDays; } + + /** + * Reorder links by creation date (newest first). + * + * Also update the urls and ids mapping arrays. + * + * @param string $order ASC|DESC + */ + public function reorder($order = 'DESC') + { + $order = $order === 'ASC' ? -1 : 1; + // Reorder array by dates. + usort($this->links, function ($a, $b) use ($order) { + if (isset($a['sticky']) && isset($b['sticky']) && $a['sticky'] !== $b['sticky']) { + return $a['sticky'] ? -1 : 1; + } + return $a['created'] < $b['created'] ? 1 * $order : -1 * $order; + }); + + $this->urls = []; + $this->ids = []; + foreach ($this->links as $key => $link) { + $this->urls[$link['url']] = $key; + $this->ids[$link['id']] = $key; + } + } + + /** + * Return the next key for link creation. + * E.g. If the last ID is 597, the next will be 598. + * + * @return int next ID. + */ + public function getNextId() + { + if (!empty($this->ids)) { + return max(array_keys($this->ids)) + 1; + } + return 0; + } + + /** + * Returns a link offset in links array from its unique ID. + * + * @param int $id Persistent ID of a link. + * + * @return int Real offset in local array, or null if doesn't exist. + */ + protected function getLinkOffset($id) + { + if (isset($this->ids[$id])) { + return $this->ids[$id]; + } + return null; + } }