X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=application%2FLinkDB.php;h=9308164af685bd8830b6c1a4a68621fddaa9f548;hb=1fdb40fc169b42af7622610c4f088688de231118;hp=47dbcaf3667315a55e6c6c4caa747b90a75e1b8d;hpb=9f15ca9ee76bc36832f12a952005fdaf52e2559d;p=github%2Fshaarli%2FShaarli.git diff --git a/application/LinkDB.php b/application/LinkDB.php index 47dbcaf3..9308164a 100644 --- a/application/LinkDB.php +++ b/application/LinkDB.php @@ -6,37 +6,66 @@ * * Example: * $myLinks = new LinkDB(); - * echo $myLinks['20110826_161819']['title']; + * echo $myLinks[350]['title']; * foreach ($myLinks as $link) * echo $link['title'].' at url '.$link['url'].'; description:'.$link['description']; * * Available keys: + * - id: primary key, incremental integer identifier (persistent) * - description: description of the entry - * - linkdate: date of the creation of this entry, in the form YYYYMMDD_HHMMSS - * (e.g.'20110914_192317') + * - created: creation date of this entry, DateTime object. + * - updated: last modification date of this entry, DateTime object. * - 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. + * - shorturl Permalink smallhash * * Implements 3 interfaces: * - ArrayAccess: behaves like an associative array; * - Countable: there is a count() method; * - Iterator: usable in foreach () loops. + * + * ID mechanism: + * ArrayAccess is implemented in a way that will allow to access a link + * with the unique identifier ID directly with $link[ID]. + * Note that it's not the real key of the link array attribute. + * This mechanism is in place to have persistent link IDs, + * even though the internal array is reordered by date. + * Example: + * - DB: link #1 (2010-01-01) link #2 (2016-01-01) + * - Order: #2 #1 + * - Import links containing: link #3 (2013-01-01) + * - New DB: link #1 (2010-01-01) link #2 (2016-01-01) link #3 (2013-01-01) + * - Real order: #2 #3 #1 */ class LinkDB implements Iterator, Countable, ArrayAccess { + // Links are stored as a PHP serialized string + private $datastore; + + // 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; - // List of all recorded URLs (key=url, value=linkdate) - // for fast reserve search (url-->linkdate) + // List of all recorded URLs (key=url, value=link offset) + // for fast reserve search (url-->link offset) private $urls; - // List of linkdate keys (for the Iterator interface implementation) + /** + * @var array List of all links IDS mapped with their array offset. + * Map: id->offset. + */ + protected $ids; + + // List of offset keys (for the Iterator interface implementation) private $keys; // Position in the $this->keys array (for the Iterator interface) @@ -48,20 +77,45 @@ class LinkDB implements Iterator, Countable, ArrayAccess // Hide public links 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 $redirectorEncode Enable urlencode on redirected urls (default: true). */ - function __construct($isLoggedIn, $hidePublicLinks) + public function __construct( + $datastore, + $isLoggedIn, + $hidePublicLinks, + $redirector = '', + $redirectorEncode = true + ) { - // FIXME: do not access $GLOBALS, pass the datastore instead + $this->datastore = $datastore; $this->loggedIn = $isLoggedIn; $this->hidePublicLinks = $hidePublicLinks; - $this->checkDB(); - $this->readdb(); + $this->redirector = $redirector; + $this->redirectorEncode = $redirectorEncode === true; + $this->check(); + $this->read(); } /** @@ -81,14 +135,26 @@ class LinkDB implements Iterator, Countable, ArrayAccess if (!$this->loggedIn) { die('You are not authorized to add a link.'); } - if (empty($value['linkdate']) || empty($value['url'])) { - die('Internal Error: A link should always have a linkdate and URL.'); + if (!isset($value['id']) || empty($value['url'])) { + die('Internal Error: A link should always have an id and URL.'); + } + if (($offset !== null && ! is_int($offset)) || ! is_int($value['id'])) { + die('You must specify an integer as a key.'); } - if (empty($offset)) { - die('You must specify a key.'); + if ($offset !== null && $offset !== $value['id']) { + die('Array offset and link ID must be equal.'); + } + + // 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; + $this->urls[$value['url']] = $offset; + $this->ids[$value['id']] = $offset; } /** @@ -96,7 +162,7 @@ class LinkDB implements Iterator, Countable, ArrayAccess */ public function offsetExists($offset) { - return array_key_exists($offset, $this->links); + return array_key_exists($this->getLinkOffset($offset), $this->links); } /** @@ -108,9 +174,11 @@ class LinkDB implements Iterator, Countable, ArrayAccess // TODO: raise an exception die('You are not authorized to delete a link.'); } - $url = $this->links[$offset]['url']; + $realOffset = $this->getLinkOffset($offset); + $url = $this->links[$realOffset]['url']; unset($this->urls[$url]); - unset($this->links[$offset]); + unset($this->ids[$realOffset]); + unset($this->links[$realOffset]); } /** @@ -118,21 +186,22 @@ 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]; } @@ -140,7 +209,7 @@ class LinkDB implements Iterator, Countable, ArrayAccess /** * Iterator - Moves forward to next element */ - function next() + public function next() { ++$this->position; } @@ -150,17 +219,16 @@ 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); + $this->keys = array_keys($this->ids); $this->position = 0; } /** * Iterator - Checks if current position is valid */ - function valid() + public function valid() { return isset($this->keys[$this->position]); } @@ -170,105 +238,140 @@ class LinkDB implements Iterator, Countable, ArrayAccess * * If no DB file is found, creates a dummy DB. */ - private function checkDB() + private function check() { - if (file_exists($GLOBALS['config']['DATASTORE'])) { + if (file_exists($this->datastore)) { return; } // Create a dummy database for example $this->links = array(); $link = array( - 'title'=>'Shaarli - sebsauvage.net', - 'url'=>'http://sebsauvage.net/wiki/doku.php?id=php:shaarli', - 'description'=>'Welcome to Shaarli! This is a bookmark. To edit or delete me, you must first login.', + 'id' => 1, + '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. + +To learn how to use Shaarli, consult the link "Help/documentation" at the bottom of this page. + +You use the community supported version of the original Shaarli project, by Sebastien Sauvage.', 'private'=>0, - 'linkdate'=>'20110914_190000', + '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( + 'id' => 0, 'title'=>'My secret stuff... - Pastebin.com', 'url'=>'http://sebsauvage.net/paste/?8434b27936c09649#bR7XsXhoTiLcqCpQbmOpBi3rq2zzQUC5hBI7ZT1O3x8=', - 'description'=>'SShhhh!! I\'m a private link only YOU can see. You can delete me too.', + 'description'=>'Shhhh! I\'m a private link only YOU can see. You can delete me too.', 'private'=>1, - 'linkdate'=>'20110914_074522', - '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 - // TODO: raise an exception if the file is not write-able - file_put_contents( - // FIXME: do not use $GLOBALS - $GLOBALS['config']['DATASTORE'], - PHPPREFIX.base64_encode(gzdeflate(serialize($this->links))).PHPSUFFIX - ); + $this->write(); } /** * Reads database from disk to memory */ - private function readdb() + private function read() { - // Public links are hidden and user not logged in => nothing to show if ($this->hidePublicLinks && !$this->loggedIn) { $this->links = array(); return; } - // Read data - // Note that gzinflate is faster than gzuncompress. - // See: http://www.php.net/manual/en/function.gzdeflate.php#96439 - // FIXME: do not use $GLOBALS - $this->links = array(); + $this->links = FileUtils::readFlatDB($this->datastore, []); - if (file_exists($GLOBALS['config']['DATASTORE'])) { - $this->links = unserialize(gzinflate(base64_decode( - substr(file_get_contents($GLOBALS['config']['DATASTORE']), - strlen(PHPPREFIX), -strlen(PHPSUFFIX))))); - } + $toremove = array(); + foreach ($this->links as $key => &$link) { + if (! $this->loggedIn && $link['private'] != 0) { + // Transition for not upgraded databases. + $toremove[] = $key; + continue; + } - // If user is not logged in, filter private links. - if (!$this->loggedIn) { - $toremove = array(); - foreach ($this->links as $link) { - if ($link['private'] != 0) { - $toremove[] = $link['linkdate']; + // Sanitize data fields. + sanitizeLink($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'] .= $link['url']; } } - foreach ($toremove as $linkdate) { - unset($this->links[$linkdate]); + 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']); } } - // Keep the list of the mapping URLs-->linkdate up-to-date. - $this->urls = array(); - foreach ($this->links as $link) { - $this->urls[$link['url']] = $link['linkdate']; + // If user is not logged in, filter private links. + foreach ($toremove as $offset) { + unset($this->links[$offset]); } + + $this->reorder(); + } + + /** + * Saves the database from memory to disk + * + * @throws IOException the datastore is not writable + */ + private function write() + { + FileUtils::writeFlatDB($this->datastore, $this->links); } /** * Saves the database from memory to disk + * + * @param string $pageCacheDir page cache directory */ - public function savedb() + public function save($pageCacheDir) { if (!$this->loggedIn) { // TODO: raise an Exception instead die('You are not authorized to change the database.'); } - file_put_contents( - $GLOBALS['config']['DATASTORE'], - PHPPREFIX.base64_encode(gzdeflate(serialize($this->links))).PHPSUFFIX - ); - invalidateCaches(); + + $this->write(); + + invalidateCaches($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) { @@ -279,130 +382,119 @@ class LinkDB implements Iterator, Countable, ArrayAccess } /** - * 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'); - $keys = ['title', 'description', 'url', 'tags']; - - foreach ($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 string $visibility return only all/private/public links + * @param string $untaggedonly return only untagged links + * + * @return array filtered links, all links if no suitable filter was provided. */ - public function filterDay($day) + public function filterSearch($filterRequest = array(), $casesensitive = false, $visibility = 'all', $untaggedonly = false) { - // TODO: check input format - $filtered = array(); - foreach ($this->links as $l) { - if (startsWith($l['linkdate'], $day)) { - $filtered[$l['linkdate']] = $l; - } - } - ksort($filtered); - return $filtered; + // 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 article corresponding to a smallHash + * Returns the list tags appearing in the links with the given tags + * @param $filteringTags: tags selecting the links to consider + * @param $visibility: process only all/private/public links + * @return: a tag=>linksCount array */ - public function filterSmallHash($smallHash) + public function linksCountPerTag($filteringTags = [], $visibility = 'all') { - $filtered = array(); - foreach ($this->links as $l) { - if ($smallHash == smallHash($l['linkdate'])) { - // Yes, this is ugly and slow - $filtered[$l['linkdate']] = $l; - return $filtered; + $links = empty($filteringTags) ? $this->links : $this->filterSearch(['searchtags' => $filteringTags], false, $visibility); + $tags = array(); + $caseMapping = array(); + foreach ($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)]]++; } } - return $filtered; + // Sort tags by usage (most used tag first) + arsort($tags); + return $tags; } /** - * Returns the list of all tags - * Output: associative array key=tags, value=0 + * Rename or delete a tag across all links. + * + * @param string $from Tag to rename + * @param string $to New tag. If none is provided, the from tag will be deleted + * + * @return array|bool List of altered links or false on error */ - public function allTags() + public function renameTag($from, $to) { - $tags = array(); - foreach ($this->links as $link) { - foreach (explode(' ', $link['tags']) as $tag) { - if (!empty($tag)) { - $tags[$tag] = (empty($tags[$tag]) ? 1 : $tags[$tag] + 1); + 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']] = $value; } } - // Sort tags by usage (most used tag first) - arsort($tags); - return $tags; + + return $linksToAlter; } /** @@ -412,12 +504,64 @@ class LinkDB implements Iterator, Countable, ArrayAccess 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) { + return $a['created'] < $b['created'] ? 1 * $order : -1 * $order; + }); + + $this->urls = array(); + $this->ids = array(); + foreach ($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; + } } -?>