From 822bffced8212e7f34bcb2ad063b31a78bd57bdb Mon Sep 17 00:00:00 2001 From: ArthurHoaro Date: Sun, 27 Dec 2015 10:08:20 +0100 Subject: Link filter refactoring * introduce class LinkFilter to handle link filter operation (and lighten LinkDB). * handle 'private only' in filtering. * update template to prefill search fields with current search terms. * coding style. * unit test (mostly move from LinkDB to LinkFilter). PS: preparation for #358 #315 and 'AND' search. --- application/LinkDB.php | 120 +++------------------ application/LinkFilter.php | 259 +++++++++++++++++++++++++++++++++++++++++++++ application/Utils.php | 12 ++- 3 files changed, 281 insertions(+), 110 deletions(-) create mode 100644 application/LinkFilter.php (limited to 'application') diff --git a/application/LinkDB.php b/application/LinkDB.php index 51fa926d..be7d9016 100644 --- a/application/LinkDB.php +++ b/application/LinkDB.php @@ -62,6 +62,11 @@ class LinkDB implements Iterator, Countable, ArrayAccess // link redirector set in user settings. private $_redirector; + /** + * @var LinkFilter instance. + */ + private $linkFilter; + /** * Creates a new LinkDB * @@ -80,6 +85,7 @@ class LinkDB implements Iterator, Countable, ArrayAccess $this->_redirector = $redirector; $this->_checkDB(); $this->_readDB(); + $this->linkFilter = new LinkFilter($this->_links); } /** @@ -334,114 +340,18 @@ You use the community supported version of the original Shaarli project, by Seba } /** - * Returns the list of links corresponding to a full-text search + * Filter links. * - * Searches: - * - in the URLs, title and description; - * - are case-insensitive. - * - * Example: - * print_r($mydb->filterFulltext('hollandais')); - * - * mb_convert_case($val, MB_CASE_LOWER, 'UTF-8') - * - allows to perform searches on Unicode text - * - see https://github.com/shaarli/Shaarli/issues/75 for examples - */ - public function filterFulltext($searchterms) - { - // FIXME: explode(' ',$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; - } - - /** - * Returns the list of links associated with a given list of tags + * @param string $type Type of filter. + * @param mixed $request Search request, string or array. + * @param bool $casesensitive Optional: Perform case sensitive filter + * @param bool $privateonly Optional: Returns private links only if true. * - * You can specify one or more tags, separated by space or a comma, e.g. - * print_r($mydb->filterTags('linux programming')); + * @return array filtered links */ - 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 $filtered; - } - - - /** - * Returns the list of articles for a given day, chronologically sorted - * - * Day must be in the form 'YYYYMMDD' (e.g. '20120125'), e.g. - * print_r($mydb->filterDay('20120125')); - */ - public function filterDay($day) - { - if (! checkDateFormat('Ymd', $day)) { - throw new Exception('Invalid date format'); - } - - $filtered = array(); - foreach ($this->_links as $l) { - if (startsWith($l['linkdate'], $day)) { - $filtered[$l['linkdate']] = $l; - } - } - ksort($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; - } - } - return $filtered; + public function filter($type, $request, $casesensitive = false, $privateonly = false) { + $requestFilter = is_array($request) ? implode(' ', $request) : $request; + return $this->linkFilter->filter($type, $requestFilter, $casesensitive, $privateonly); } /** diff --git a/application/LinkFilter.php b/application/LinkFilter.php new file mode 100644 index 00000000..cf647371 --- /dev/null +++ b/application/LinkFilter.php @@ -0,0 +1,259 @@ +links = $links; + } + + /** + * Filter links according to parameters. + * + * @param string $type Type of filter (eg. tags, permalink, etc.). + * @param string $request Filter content. + * @param bool $casesensitive Optional: Perform case sensitive filter if true. + * @param bool $privateonly Optional: Only returns private links if true. + * + * @return array filtered link list. + */ + public function filter($type, $request, $casesensitive = false, $privateonly = false) + { + switch($type) { + case self::$FILTER_HASH: + return $this->filterSmallHash($request); + break; + case self::$FILTER_TEXT: + return $this->filterFulltext($request, $privateonly); + break; + case self::$FILTER_TAG: + return $this->filterTags($request, $casesensitive, $privateonly); + break; + case self::$FILTER_DAY: + return $this->filterDay($request); + break; + default: + return $this->noFilter($privateonly); + } + } + + /** + * Unknown filter, but handle private only. + * + * @param bool $privateonly returns private link only if true. + * + * @return array filtered links. + */ + private function noFilter($privateonly = false) + { + if (! $privateonly) { + krsort($this->links); + return $this->links; + } + + $out = array(); + foreach ($this->links as $value) { + if ($value['private']) { + $out[$value['linkdate']] = $value; + } + } + + krsort($out); + return $out; + } + + /** + * Returns the shaare corresponding to a smallHash. + * + * @param string $smallHash permalink hash. + * + * @return array $filtered array containing permalink data. + */ + private function filterSmallHash($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; + } + } + return $filtered; + } + + /** + * Returns the list of links corresponding to a full-text search + * + * Searches: + * - in the URLs, title and description; + * - are case-insensitive. + * + * Example: + * print_r($mydb->filterFulltext('hollandais')); + * + * 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 + * + * @param string $searchterms search query. + * @param bool $privateonly return only private links if true. + * + * @return array search results. + */ + private function filterFulltext($searchterms, $privateonly = false) + { + // FIXME: explode(' ',$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'); + $explodedSearch = explode(' ', trim($search)); + $keys = array('title', 'description', 'url', 'tags'); + + // Iterate over every stored link. + foreach ($this->links as $link) { + $found = false; + + // ignore non private links when 'privatonly' is on. + if (! $link['private'] && $privateonly === true) { + continue; + } + + // Iterate over searchable link fields. + foreach ($keys as $key) { + // Search full expression. + if (strpos( + mb_convert_case($link[$key], MB_CASE_LOWER, 'UTF-8'), + $search + ) !== false) { + $found = true; + } + + if ($found) { + break; + } + } + + if ($found) { + $filtered[$link['linkdate']] = $link; + } + } + + krsort($filtered); + return $filtered; + } + + /** + * Returns the list of links associated with a given list of tags + * + * You can specify one or more tags, separated by space or a comma, e.g. + * print_r($mydb->filterTags('linux programming')); + * + * @param string $tags list of tags separated by commas or blank spaces. + * @param bool $casesensitive ignore case if false. + * @param bool $privateonly returns private links only. + * + * @return array filtered links. + */ + public function filterTags($tags, $casesensitive = false, $privateonly = false) + { + $searchtags = $this->tagsStrToArray($tags, $casesensitive); + $filtered = array(); + + foreach ($this->links as $l) { + // ignore non private links when 'privatonly' is on. + if (! $l['private'] && $privateonly === true) { + continue; + } + + $linktags = $this->tagsStrToArray($l['tags'], $casesensitive); + + if (count(array_intersect($linktags, $searchtags)) == count($searchtags)) { + $filtered[$l['linkdate']] = $l; + } + } + krsort($filtered); + return $filtered; + } + + /** + * Returns the list of articles for a given day, chronologically sorted + * + * Day must be in the form 'YYYYMMDD' (e.g. '20120125'), e.g. + * print_r($mydb->filterDay('20120125')); + * + * @param string $day day to filter. + * + * @return array all link matching given day. + * + * @throws Exception if date format is invalid. + */ + public function filterDay($day) + { + if (! checkDateFormat('Ymd', $day)) { + throw new Exception('Invalid date format'); + } + + $filtered = array(); + foreach ($this->links as $l) { + if (startsWith($l['linkdate'], $day)) { + $filtered[$l['linkdate']] = $l; + } + } + ksort($filtered); + return $filtered; + } + + /** + * Convert a list of tags (str) to an array. Also + * - handle case sensitivity. + * - accepts spaces commas as separator. + * - remove private tags for loggedout users. + * + * @param string $tags string containing a list of tags. + * @param bool $casesensitive will convert everything to lowercase if false. + * + * @return array filtered tags string. + */ + public function tagsStrToArray($tags, $casesensitive) + { + // We use UTF-8 conversion to handle various graphemes (i.e. cyrillic, or greek) + $tagsOut = $casesensitive ? $tags : mb_convert_case($tags, MB_CASE_LOWER, 'UTF-8'); + $tagsOut = str_replace(',', ' ', $tagsOut); + + return explode(' ', trim($tagsOut)); + } +} diff --git a/application/Utils.php b/application/Utils.php index f84f70e4..aeaef9ff 100644 --- a/application/Utils.php +++ b/application/Utils.php @@ -72,12 +72,14 @@ function sanitizeLink(&$link) /** * Checks if a string represents a valid date + + * @param string $format The expected DateTime format of the string + * @param string $string A string-formatted date + * + * @return bool whether the string is a valid date * - * @param string a string-formatted date - * @param format the expected DateTime format of the string - * @return whether the string is a valid date - * @see http://php.net/manual/en/class.datetime.php - * @see http://php.net/manual/en/datetime.createfromformat.php + * @see http://php.net/manual/en/class.datetime.php + * @see http://php.net/manual/en/datetime.createfromformat.php */ function checkDateFormat($format, $string) { -- cgit v1.2.3 From 2c75f8e780e674ddb42c935b54ed6c39925ba07c Mon Sep 17 00:00:00 2001 From: ArthurHoaro Date: Sun, 3 Jan 2016 15:29:15 +0100 Subject: Fixes #426 - Do not filter with blank tags. --- application/LinkDB.php | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'application') diff --git a/application/LinkDB.php b/application/LinkDB.php index be7d9016..16848519 100644 --- a/application/LinkDB.php +++ b/application/LinkDB.php @@ -351,7 +351,7 @@ You use the community supported version of the original Shaarli project, by Seba */ public function filter($type, $request, $casesensitive = false, $privateonly = false) { $requestFilter = is_array($request) ? implode(' ', $request) : $request; - return $this->linkFilter->filter($type, $requestFilter, $casesensitive, $privateonly); + return $this->linkFilter->filter($type, trim($requestFilter), $casesensitive, $privateonly); } /** -- cgit v1.2.3