From ca4138d0217f699050f80aa6bfb37e121ebec9a6 Mon Sep 17 00:00:00 2001 From: Alexandre Alapetite Date: Sat, 22 Apr 2017 12:48:26 +0200 Subject: Update to Favicon 1.2.0 https://github.com/ArthurHoaro/favicon/releases/tag/v1.2.0 --- lib/Favicon/DataAccess.php | 53 +++++++------ lib/Favicon/Favicon.php | 172 ++++++++++++++++++++++++++++++------------ lib/Favicon/FaviconDLType.php | 23 ++++++ 3 files changed, 173 insertions(+), 75 deletions(-) create mode 100644 lib/Favicon/FaviconDLType.php (limited to 'lib/Favicon') diff --git a/lib/Favicon/DataAccess.php b/lib/Favicon/DataAccess.php index 4c1a29541..2bfdf640e 100644 --- a/lib/Favicon/DataAccess.php +++ b/lib/Favicon/DataAccess.php @@ -9,33 +9,32 @@ namespace Favicon; **/ class DataAccess { public function retrieveUrl($url) { - $this->set_context(); - return @file_get_contents($url); + $this->set_context(); + return @file_get_contents($url); } - + public function retrieveHeader($url) { - $this->set_context(); - $headers = @get_headers($url, 1); - return is_array($headers) ? array_change_key_case($headers) : array(); - } - - public function saveCache($file, $data) { - file_put_contents($file, $data); - } - - public function readCache($file) { - return file_get_contents($file); - } - - private function set_context() { - stream_context_set_default( - array( - 'http' => array( - 'method' => 'GET', - 'timeout' => 10, - 'header' => "User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:20.0; Favicon; +https://github.com/ArthurHoaro/favicon) Gecko/20100101 Firefox/32.0\r\n", - ) - ) - ); + $this->set_context(); + return @get_headers($url, TRUE); } -} + + public function saveCache($file, $data) { + file_put_contents($file, $data); + } + + public function readCache($file) { + return file_get_contents($file); + } + + private function set_context() { + stream_context_set_default( + array( + 'http' => array( + 'method' => 'GET', + 'timeout' => 10, + 'header' => "User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:20.0; Favicon; +https://github.com/ArthurHoaro/favicon) Gecko/20100101 Firefox/32.0\r\n", + ) + ) + ); + } +} \ No newline at end of file diff --git a/lib/Favicon/Favicon.php b/lib/Favicon/Favicon.php index 85d2ef19b..f39adc7d7 100644 --- a/lib/Favicon/Favicon.php +++ b/lib/Favicon/Favicon.php @@ -4,6 +4,8 @@ namespace Favicon; class Favicon { + protected static $TYPE_CACHE_URL = 'url'; + protected static $TYPE_CACHE_IMG = 'img'; protected $url = ''; protected $cacheDir; protected $cacheTimeout; @@ -16,18 +18,24 @@ class Favicon } $this->cacheDir = __DIR__ . '/../../resources/cache'; + $this->cacheTimeout = 604800; $this->dataAccess = new DataAccess(); } + /** + * Set cache settings: + * - dir: cache directory + * - timeout: in seconds + * + * @param array $args + */ public function cache($args = array()) { if (isset($args['dir'])) { $this->cacheDir = $args['dir']; } if (!empty($args['timeout'])) { - $this->cacheTimeout = $args['timeout']; - } else { - $this->cacheTimeout = 0; + $this->cacheTimeout = $args['timeout']; } } @@ -89,9 +97,6 @@ class Favicon $loop = TRUE; while ($loop && $max_loop-- > 0) { $headers = $this->dataAccess->retrieveHeader($url); - if (empty($headers)) { - return false; - } $exploded = explode(' ', $headers[0]); if( !isset($exploded[1]) ) { @@ -102,7 +107,7 @@ class Favicon switch ($status) { case '301': case '302': - $url = isset($headers['location']) ? $headers['location'] : ''; + $url = $headers['Location']; break; default: $loop = FALSE; @@ -120,9 +125,16 @@ class Favicon /** * Find remote (or cached) favicon - * @return favicon URL, false if nothing was found - **/ - public function get($url = '') + * + * @param string $url to look for a favicon + * @param int $type type of retrieval (FaviconDLType): + * - HOTLINK_URL: returns remote URL + * - DL_FILE_PATH: returns file path of the favicon downloaded locally + * - RAW_IMAGE: returns the favicon image binary string + * + * @return string|bool favicon URL, false if nothing was found + */ + public function get($url = '', $type = FaviconDLType::HOTLINK_URL) { // URLs passed to this method take precedence. if (!empty($url)) { @@ -130,25 +142,30 @@ class Favicon } // Get the base URL without the path for clearer concatenations. - $original = rtrim($this->baseUrl($this->url, true), '/'); - $url = rtrim($this->endRedirect($this->baseUrl($this->url, false)), '/'); - - if(($favicon = $this->checkCache($url)) || ($favicon = $this->getFavicon($url))) { - $base = true; - } - elseif(($favicon = $this->checkCache($original)) || ($favicon = $this->getFavicon($original, false))) { - $base = false; + $url = rtrim($this->baseUrl($this->url, true), '/'); + $original = $url; + if (($favicon = $this->checkCache($original, self::$TYPE_CACHE_URL)) === false + && ! $favicon = $this->getFavicon($original, false) + ) { + $url = rtrim($this->endRedirect($this->baseUrl($this->url, false)), '/'); + if (($favicon = $this->checkCache($url, self::$TYPE_CACHE_URL)) === false + && ! $favicon = $this->getFavicon($url) + ) { + $url = $original; + } } - else - return false; - - // Save cache if necessary - $cache = $this->cacheDir . '/' . md5($base ? $url : $original); - if ($this->cacheTimeout && !file_exists($cache) || (is_writable($cache) && time() - filemtime($cache) > $this->cacheTimeout)) { - $this->dataAccess->saveCache($cache, $favicon); + + $this->saveCache($url, $favicon, self::$TYPE_CACHE_URL); + + switch ($type) { + case FaviconDLType::DL_FILE_PATH: + return $this->getImage($url, $favicon, false); + case FaviconDLType::RAW_IMAGE: + return $this->getImage($url, $favicon, true); + case FaviconDLType::HOTLINK_URL: + default: + return empty($favicon) ? false : $favicon; } - - return $favicon; } private function getFavicon($url, $checkDefault = true) { @@ -179,13 +196,54 @@ class Favicon // Sometimes people lie, so check the status. // And sometimes, it's not even an image. Sneaky bastards! // If cacheDir isn't writable, that's not our problem - if ($favicon && is_writable($this->cacheDir) && extension_loaded('fileinfo') && !$this->checkImageMType($favicon)) { + if ($favicon && is_writable($this->cacheDir) && !$this->checkImageMType($favicon)) { $favicon = false; } return $favicon; } - + + /** + * Find remote favicon and return it as an image + */ + private function getImage($url, $faviconUrl = '', $image = false) + { + if (empty($faviconUrl)) { + return false; + } + + $favicon = $this->checkCache($url, self::$TYPE_CACHE_IMG); + // Favicon not found in the cache + if( $favicon === false ) { + $favicon = $this->dataAccess->retrieveUrl($faviconUrl); + // Definitely not found + if (!$this->checkImageMTypeContent($favicon)) { + return false; + } else { + $this->saveCache($url, $favicon, self::$TYPE_CACHE_IMG); + } + } + + if( $image ) { + return $favicon; + } + else + return self::$TYPE_CACHE_IMG . md5($url); + } + + /** + * Display data as a PNG Favicon, then exit + * @param $data + */ + private function displayFavicon($data) { + header('Content-Type: image/png'); + header('Cache-Control: private, max-age=10800, pre-check=10800'); + header('Pragma: private'); + header('Expires: ' . date(DATE_RFC822,strtotime('7 day'))); + echo $data; + exit; + } + private function getInPage($url) { $html = $this->dataAccess->retrieveUrl("{$url}/"); preg_match('!.*!ims', $html, $match); @@ -197,7 +255,7 @@ class Favicon $head = $match[0]; $dom = new \DOMDocument(); - // Use error supression, because the HTML might be too malformed. + // Use error suppression, because the HTML might be too malformed. if (@$dom->loadHTML($head)) { $links = $dom->getElementsByTagName('link'); foreach ($links as $link) { @@ -212,33 +270,51 @@ class Favicon } return false; } - - private function checkCache($url) { + + private function checkCache($url, $type) { if ($this->cacheTimeout) { - $cache = $this->cacheDir . '/' . md5($url); - if (file_exists($cache) && is_readable($cache) && (time() - filemtime($cache) < $this->cacheTimeout)) { + $cache = $this->cacheDir . '/'. $type . md5($url); + if (file_exists($cache) && is_readable($cache) + && ($this->cacheTimeout === -1 || time() - filemtime($cache) < $this->cacheTimeout) + ) { return $this->dataAccess->readCache($cache); } - } + } return false; } - + + /** + * Will save data in cacheDir if the directory writable and any previous cache is expired (cacheTimeout) + * @param $url + * @param $data + * @param $type + * @return string cache file path + */ + private function saveCache($url, $data, $type) { + // Save cache if necessary + $cache = $this->cacheDir . '/'. $type . md5($url); + if ($this->cacheTimeout && !file_exists($cache) + || (is_writable($cache) && $this->cacheTimeout !== -1 && time() - filemtime($cache) > $this->cacheTimeout) + ) { + $this->dataAccess->saveCache($cache, $data); + } + return $cache; + } + private function checkImageMType($url) { - $tmpFile = $this->cacheDir . '/tmp.ico'; $fileContent = $this->dataAccess->retrieveUrl($url); - $this->dataAccess->saveCache($tmpFile, $fileContent); - $isImage = true; - try { - $finfo = finfo_open(FILEINFO_MIME_TYPE); - $isImage = strpos(finfo_file($finfo, $tmpFile), 'image') !== false; - finfo_close($finfo); - } catch (Exception $e) { - } + return $this->checkImageMTypeContent($fileContent); + } + + private function checkImageMTypeContent($content) { + if(empty($content)) return false; + + $fInfo = finfo_open(FILEINFO_MIME_TYPE); + $isImage = strpos(finfo_buffer($fInfo, $content), 'image') !== false; + finfo_close($fInfo); - unlink($tmpFile); - return $isImage; } @@ -291,7 +367,7 @@ class Favicon } /** - * @param DataAccess $dataAccess + * @param DataAccess|\PHPUnit_Framework_MockObject_MockObject $dataAccess */ public function setDataAccess($dataAccess) { diff --git a/lib/Favicon/FaviconDLType.php b/lib/Favicon/FaviconDLType.php new file mode 100644 index 000000000..6da525a7f --- /dev/null +++ b/lib/Favicon/FaviconDLType.php @@ -0,0 +1,23 @@ + Date: Sat, 22 Apr 2017 13:35:51 +0200 Subject: Re-apply Favicon FreshRSS patches Patches sent upstream https://github.com/ArthurHoaro/favicon/pull/5 --- lib/Favicon/DataAccess.php | 3 ++- lib/Favicon/Favicon.php | 17 ++++++++++++----- 2 files changed, 14 insertions(+), 6 deletions(-) (limited to 'lib/Favicon') diff --git a/lib/Favicon/DataAccess.php b/lib/Favicon/DataAccess.php index 2bfdf640e..838df66c3 100644 --- a/lib/Favicon/DataAccess.php +++ b/lib/Favicon/DataAccess.php @@ -15,7 +15,8 @@ class DataAccess { public function retrieveHeader($url) { $this->set_context(); - return @get_headers($url, TRUE); + $headers = @get_headers($url, 1); + return is_array($headers) ? array_change_key_case($headers) : array(); } public function saveCache($file, $data) { diff --git a/lib/Favicon/Favicon.php b/lib/Favicon/Favicon.php index f39adc7d7..7e6c42d6e 100644 --- a/lib/Favicon/Favicon.php +++ b/lib/Favicon/Favicon.php @@ -97,6 +97,9 @@ class Favicon $loop = TRUE; while ($loop && $max_loop-- > 0) { $headers = $this->dataAccess->retrieveHeader($url); + if (empty($headers)) { + return false; + } $exploded = explode(' ', $headers[0]); if( !isset($exploded[1]) ) { @@ -107,7 +110,7 @@ class Favicon switch ($status) { case '301': case '302': - $url = $headers['Location']; + $url = isset($headers['location']) ? $headers['location'] : ''; break; default: $loop = FALSE; @@ -196,7 +199,7 @@ class Favicon // Sometimes people lie, so check the status. // And sometimes, it's not even an image. Sneaky bastards! // If cacheDir isn't writable, that's not our problem - if ($favicon && is_writable($this->cacheDir) && !$this->checkImageMType($favicon)) { + if ($favicon && is_writable($this->cacheDir) && extension_loaded('fileinfo') && !$this->checkImageMType($favicon)) { $favicon = false; } @@ -311,9 +314,13 @@ class Favicon private function checkImageMTypeContent($content) { if(empty($content)) return false; - $fInfo = finfo_open(FILEINFO_MIME_TYPE); - $isImage = strpos(finfo_buffer($fInfo, $content), 'image') !== false; - finfo_close($fInfo); + $isImage = true; + try { + $fInfo = finfo_open(FILEINFO_MIME_TYPE); + $isImage = strpos(finfo_buffer($fInfo, $content), 'image') !== false; + finfo_close($fInfo); + } catch (Exception $e) { + } return $isImage; } -- cgit v1.2.3 From 57f1e9d6570e86f1bc679ed4897437ae4b14a2b5 Mon Sep 17 00:00:00 2001 From: Alexandre Alapetite Date: Sat, 22 Apr 2017 23:36:41 +0200 Subject: Fix some bugs of Favicon 1.2 https://github.com/ArthurHoaro/favicon/pull/6 https://github.com/ArthurHoaro/favicon/pull/7 --- lib/Favicon/Favicon.php | 18 ++++++++++++++---- lib/favicons.php | 4 ++-- 2 files changed, 16 insertions(+), 6 deletions(-) (limited to 'lib/Favicon') diff --git a/lib/Favicon/Favicon.php b/lib/Favicon/Favicon.php index 7e6c42d6e..39f45f139 100644 --- a/lib/Favicon/Favicon.php +++ b/lib/Favicon/Favicon.php @@ -188,7 +188,10 @@ class Favicon // See if it's specified in a link tag in domain url. if (!$favicon) { - $favicon = $this->getInPage($url); + $favicon = trim($this->getInPage($url)); + } + if (substr($favicon, 0, 2) === '//') { + $favicon = 'https:' . $favicon; } // Make sure the favicon is an absolute URL. @@ -264,9 +267,15 @@ class Favicon foreach ($links as $link) { if ($link->hasAttribute('rel') && strtolower($link->getAttribute('rel')) == 'shortcut icon') { return $link->getAttribute('href'); - } elseif ($link->hasAttribute('rel') && strtolower($link->getAttribute('rel')) == 'icon') { + } + } + foreach ($links as $link) { + if ($link->hasAttribute('rel') && strtolower($link->getAttribute('rel')) == 'icon') { return $link->getAttribute('href'); - } elseif ($link->hasAttribute('href') && strpos($link->getAttribute('href'), 'favicon') !== FALSE) { + } + } + foreach ($links as $link) { + if ($link->hasAttribute('href') && strpos($link->getAttribute('href'), 'favicon') !== FALSE) { return $link->getAttribute('href'); } } @@ -319,7 +328,8 @@ class Favicon $fInfo = finfo_open(FILEINFO_MIME_TYPE); $isImage = strpos(finfo_buffer($fInfo, $content), 'image') !== false; finfo_close($fInfo); - } catch (Exception $e) { + } catch (\Exception $e) { + error_log('Favicon checkImageMTypeContent error: ' . $e->getMessage()); } return $isImage; diff --git a/lib/favicons.php b/lib/favicons.php index 0e80aa145..93e570d27 100644 --- a/lib/favicons.php +++ b/lib/favicons.php @@ -12,10 +12,10 @@ function download_favicon($website, $dest) { syslog(LOG_INFO, 'FreshRSS Favicon discovery GET ' . $website); $favicon_getter = new \Favicon\Favicon(); - $tmpPath = realpath(TMP_PATH) . '/'; + $tmpPath = realpath(TMP_PATH); $favicon_getter->setCacheDir($tmpPath); $favicon_path = $favicon_getter->get($website, \Favicon\FaviconDLType::DL_FILE_PATH); - return ($favicon_path != false && @rename($tmpPath . $favicon_path, $dest)) || + return ($favicon_path != false && @rename($tmpPath . '/' . $favicon_path, $dest)) || @copy($default_favicon, $dest); } -- cgit v1.2.3 From f483a5e95bc78cfb42e6fd90436c9f1c60e0242b Mon Sep 17 00:00:00 2001 From: Alexandre Alapetite Date: Sun, 23 Apr 2017 01:59:09 +0200 Subject: Favicon fix redirects https://github.com/ArthurHoaro/favicon/pull/5/commits/92b42591591b4282261f21ed5ffa553f5e987a9d --- lib/Favicon/DataAccess.php | 2 ++ lib/Favicon/Favicon.php | 3 +++ lib/favicons.php | 1 + 3 files changed, 6 insertions(+) (limited to 'lib/Favicon') diff --git a/lib/Favicon/DataAccess.php b/lib/Favicon/DataAccess.php index 838df66c3..1445e9343 100644 --- a/lib/Favicon/DataAccess.php +++ b/lib/Favicon/DataAccess.php @@ -32,6 +32,8 @@ class DataAccess { array( 'http' => array( 'method' => 'GET', + 'follow_location' => 0, + 'max_redirects' => 1, 'timeout' => 10, 'header' => "User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:20.0; Favicon; +https://github.com/ArthurHoaro/favicon) Gecko/20100101 Firefox/32.0\r\n", ) diff --git a/lib/Favicon/Favicon.php b/lib/Favicon/Favicon.php index 39f45f139..c026d8a95 100644 --- a/lib/Favicon/Favicon.php +++ b/lib/Favicon/Favicon.php @@ -111,6 +111,9 @@ class Favicon case '301': case '302': $url = isset($headers['location']) ? $headers['location'] : ''; + if (is_array($url)) { + $url = end($url); + } break; default: $loop = FALSE; diff --git a/lib/favicons.php b/lib/favicons.php index 93e570d27..d8baba342 100644 --- a/lib/favicons.php +++ b/lib/favicons.php @@ -14,6 +14,7 @@ function download_favicon($website, $dest) { $favicon_getter = new \Favicon\Favicon(); $tmpPath = realpath(TMP_PATH); $favicon_getter->setCacheDir($tmpPath); + $favicon_getter->setCacheTimeout(-1); $favicon_path = $favicon_getter->get($website, \Favicon\FaviconDLType::DL_FILE_PATH); return ($favicon_path != false && @rename($tmpPath . '/' . $favicon_path, $dest)) || -- cgit v1.2.3