Fix for #664 - Missing source url attribute in RSS feeds

[github/wallabag/wallabag.git] / inc / 3rdparty / libraries / content-extractor / SiteConfig.php

<?php\r
/**\r
 * Site Config\r
 * \r
 * Each instance of this class should hold extraction patterns and other directives\r
 * for a website. See ContentExtractor class to see how it's used.\r
 * \r
 * @version 0.7\r
 * @date 2012-08-27\r
 * @author Keyvan Minoukadeh\r
 * @copyright 2012 Keyvan Minoukadeh\r
 * @license http://www.gnu.org/licenses/agpl-3.0.html AGPL v3\r
 */\r
\r
class SiteConfig\r
{\r
	// Use first matching element as title (0 or more xpath expressions)\r
	public $title = array();\r
	\r
	// Use first matching element as body (0 or more xpath expressions)\r
	public $body = array();\r
	\r
	// Use first matching element as author (0 or more xpath expressions)\r
	public $author = array();\r
	\r
	// Use first matching element as date (0 or more xpath expressions)\r
	public $date = array();\r
	\r
	// Strip elements matching these xpath expressions (0 or more)\r
	public $strip = array();\r
	\r
	// Strip elements which contain these strings (0 or more) in the id or class attribute \r
	public $strip_id_or_class = array();\r
	\r
	// Strip images which contain these strings (0 or more) in the src attribute \r
	public $strip_image_src = array();\r
	\r
	// Additional HTTP headers to send\r
	// NOT YET USED\r
	public $http_header = array();\r
	\r
	// Process HTML with tidy before creating DOM (bool or null if undeclared)\r
	public $tidy = null;\r
	\r
	protected $default_tidy = true; // used if undeclared\r
	\r
	// Autodetect title/body if xpath expressions fail to produce results.\r
	// Note that this applies to title and body separately, ie. \r
	//   * if we get a body match but no title match, this option will determine whether we autodetect title \r
	//   * if neither match, this determines whether we autodetect title and body.\r
	// Also note that this only applies when there is at least one xpath expression in title or body, ie.\r
	//   * if title and body are both empty (no xpath expressions), this option has no effect (both title and body will be auto-detected)\r
	//   * if there's an xpath expression for title and none for body, body will be auto-detected and this option will determine whether we auto-detect title if the xpath expression for it fails to produce results.\r
	// Usage scenario: you want to extract something specific from a set of URLs, e.g. a table, and if the table is not found, you want to ignore the entry completely. Auto-detection is unlikely to succeed here, so you construct your patterns and set this option to false. Another scenario may be a site where auto-detection has proven to fail (or worse, picked up the wrong content).\r
	// bool or null if undeclared\r
	public $autodetect_on_failure = null;\r
	protected $default_autodetect_on_failure = true; // used if undeclared\r
	\r
	// Clean up content block - attempt to remove elements that appear to be superfluous\r
	// bool or null if undeclared\r
	public $prune = null;\r
	protected $default_prune = true; // used if undeclared\r
	\r
	// Test URL - if present, can be used to test the config above\r
	public $test_url = array();\r
	\r
	// Single-page link - should identify a link element or URL pointing to the page holding the entire article\r
	// This is useful for sites which split their articles across multiple pages. Links to such pages tend to \r
	// display the first page with links to the other pages at the bottom. Often there is also a link to a page\r
	// which displays the entire article on one page (e.g. 'print view').\r
	// This should be an XPath expression identifying the link to that page. If present and we find a match,\r
	// we will retrieve that page and the rest of the options in this config will be applied to the new page.\r
	public $single_page_link = array();\r
	\r
	public $next_page_link = array();\r
	\r
	// Single-page link in feed? - same as above, but patterns applied to item description HTML taken from feed\r
	public $single_page_link_in_feed = array();\r
	\r
	// Which parser to use for turning raw HTML into a DOMDocument (either 'libxml' or 'html5lib')\r
	// string or null if undeclared\r
	public $parser = null;\r
	protected $default_parser = 'libxml'; // used if undeclared\r
	\r
	// Strings to search for in HTML before processing begins (used with $replace_string)\r
	public $find_string = array();\r
	// Strings to replace those found in $find_string before HTML processing begins\r
	public $replace_string = array();\r
	\r
	// the options below cannot be set in the config files which this class represents\r
	\r
	//public $cache_in_apc = false; // used to decide if we should cache in apc or not\r
	public $cache_key = null;\r
	public static $debug = false;\r
	protected static $apc = false;\r
	protected static $config_path;\r
	protected static $config_path_fallback;\r
	protected static $config_cache = array();\r
	const HOSTNAME_REGEX = '/^(([a-zA-Z0-9-]*[a-zA-Z0-9])\.)*([A-Za-z0-9-]*[A-Za-z0-9])$/';\r
	\r
	protected static function debug($msg) {\r
		if (self::$debug) {\r
			//$mem = round(memory_get_usage()/1024, 2);\r
			//$memPeak = round(memory_get_peak_usage()/1024, 2);\r
			echo '* ',$msg;\r
			//echo ' - mem used: ',$mem," (peak: $memPeak)\n";\r
			echo "\n";\r
			ob_flush();\r
			flush();\r
		}\r
	}\r
	\r
	// enable APC caching of certain site config files?\r
	// If enabled the following site config files will be \r
	// cached in APC cache (when requested for first time):\r
	// * anything in site_config/custom/ and its corresponding file in site_config/standard/\r
	// * the site config files associated with HTML fingerprints\r
	// * the global site config file\r
	// returns true if enabled, false otherwise\r
	public static function use_apc($apc=true) {\r
		if (!function_exists('apc_add')) {\r
			if ($apc) self::debug('APC will not be used (function apc_add does not exist)');\r
			return false;\r
		}\r
		self::$apc = $apc;\r
		return $apc;\r
	}\r
	\r
	// return bool or null\r
	public function tidy($use_default=true) {\r
		if ($use_default) return (isset($this->tidy)) ? $this->tidy : $this->default_tidy;\r
		return $this->tidy;\r
	}\r
	\r
	// return bool or null\r
	public function prune($use_default=true) {\r
		if ($use_default) return (isset($this->prune)) ? $this->prune : $this->default_prune;\r
		return $this->prune;\r
	}\r
	\r
	// return string or null\r
	public function parser($use_default=true) {\r
		if ($use_default) return (isset($this->parser)) ? $this->parser : $this->default_parser;\r
		return $this->parser;\r
	}\r
\r
	// return bool or null\r
	public function autodetect_on_failure($use_default=true) {\r
		if ($use_default) return (isset($this->autodetect_on_failure)) ? $this->autodetect_on_failure : $this->default_autodetect_on_failure;\r
		return $this->autodetect_on_failure;\r
	}\r
	\r
	public static function set_config_path($path, $fallback=null) {\r
		self::$config_path = $path;\r
		self::$config_path_fallback = $fallback;\r
	}\r
	\r
	public static function add_to_cache($key, SiteConfig $config, $use_apc=true) {\r
		$key = strtolower($key);\r
		if (substr($key, 0, 4) == 'www.') $key = substr($key, 4);\r
		if ($config->cache_key) $key = $config->cache_key;\r
		self::$config_cache[$key] = $config;\r
		if (self::$apc && $use_apc) {\r
			self::debug("Adding site config to APC cache with key sc.$key");\r
			apc_add("sc.$key", $config);\r
		}\r
		self::debug("Cached site config with key $key");\r
	}\r
	\r
	public static function is_cached($key) {\r
		$key = strtolower($key);\r
		if (substr($key, 0, 4) == 'www.') $key = substr($key, 4);\r
		if (array_key_exists($key, self::$config_cache)) {\r
			return true;\r
		} elseif (self::$apc && (bool)apc_fetch("sc.$key")) {\r
			return true;\r
		}\r
		return false;\r
	}\r
	\r
	public function append(SiteConfig $newconfig) {\r
		// check for commands where we accept multiple statements (no test_url)\r
		foreach (array('title', 'body', 'author', 'date', 'strip', 'strip_id_or_class', 'strip_image_src', 'single_page_link', 'single_page_link_in_feed', 'next_page_link', 'http_header', 'find_string', 'replace_string') as $var) {\r
			// append array elements for this config variable from $newconfig to this config\r
			//$this->$var = $this->$var + $newconfig->$var;\r
			$this->$var = array_unique(array_merge($this->$var, $newconfig->$var));\r
		}\r
		// check for single statement commands\r
		// we do not overwrite existing non null values\r
		foreach (array('tidy', 'prune', 'parser', 'autodetect_on_failure') as $var) {\r
			if ($this->$var === null) $this->$var = $newconfig->$var;\r
		}\r
	}\r
	\r
	// returns SiteConfig instance if an appropriate one is found, false otherwise\r
	// if $exact_host_match is true, we will not look for wildcard config matches\r
	// by default if host is 'test.example.org' we will look for and load '.example.org.txt' if it exists\r
	public static function build($host, $exact_host_match=false) {\r
		$host = strtolower($host);\r
		if (substr($host, 0, 4) == 'www.') $host = substr($host, 4);\r
		if (!$host || (strlen($host) > 200) || !preg_match(self::HOSTNAME_REGEX, ltrim($host, '.'))) return false;\r
		// check for site configuration\r
		$try = array($host);\r
		// should we look for wildcard matches \r
		if (!$exact_host_match) {\r
			$split = explode('.', $host);\r
			if (count($split) > 1) {\r
				array_shift($split);\r
				$try[] = '.'.implode('.', $split);\r
			}\r
		}\r
		\r
		// look for site config file in primary folder\r
		self::debug(". looking for site config for $host in primary folder");\r
		foreach ($try as $h) {\r
			if (array_key_exists($h, self::$config_cache)) {\r
				self::debug("... site config for $h already loaded in this request");\r
				return self::$config_cache[$h];\r
			} elseif (self::$apc && ($sconfig = apc_fetch("sc.$h"))) {\r
				self::debug("... site config for $h in APC cache");\r
				return $sconfig;\r
			} elseif (file_exists(self::$config_path."/$h.txt")) {\r
				self::debug("... found site config ($h.txt)");\r
				$file_primary = self::$config_path."/$h.txt";\r
				$matched_name = $h;\r
				break;\r
			}\r
		}\r
		\r
		// if we found site config, process it\r
		if (isset($file_primary)) {\r
			$config_lines = file($file_primary, FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES);\r
			if (!$config_lines || !is_array($config_lines)) return false;\r
			$config = self::build_from_array($config_lines);\r
			// if APC caching is available and enabled, mark this for cache\r
			//$config->cache_in_apc = true;\r
			$config->cache_key = $matched_name;\r
			\r
			// if autodetec on failure is off (on by default) we do not need to look\r
			// in secondary folder\r
			if (!$config->autodetect_on_failure()) {\r
				self::debug('... autodetect on failure is disabled (no other site config files will be loaded)');\r
				return $config;\r
			}\r
		}\r
		\r
		// look for site config file in secondary folder\r
		if (isset(self::$config_path_fallback)) {\r
			self::debug(". looking for site config for $host in secondary folder");\r
			foreach ($try as $h) {\r
				if (file_exists(self::$config_path_fallback."/$h.txt")) {\r
					self::debug("... found site config in secondary folder ($h.txt)");\r
					$file_secondary = self::$config_path_fallback."/$h.txt";\r
					$matched_name = $h;\r
					break;\r
				}\r
			}\r
			if (!isset($file_secondary)) {\r
				self::debug("... no site config match in secondary folder");\r
			}\r
		}\r
		\r
		// return false if no config file found\r
		if (!isset($file_primary) && !isset($file_secondary)) {\r
			self::debug("... no site config match for $host");\r
			return false;\r
		}\r
		\r
		// return primary config if secondary not found\r
		if (!isset($file_secondary) && isset($config)) {\r
			return $config;\r
		}\r
		\r
		// process secondary config file\r
		$config_lines = file($file_secondary, FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES);\r
		if (!$config_lines || !is_array($config_lines)) {\r
			// failed to process secondary\r
			if (isset($config)) {\r
				// return primary config\r
				return $config;\r
			} else {\r
				return false;\r
			}\r
		}\r
		\r
		// merge with primary and return\r
		if (isset($config)) {\r
			self::debug('. merging config files');\r
			$config->append(self::build_from_array($config_lines));\r
			return $config;\r
		} else {\r
			// return just secondary\r
			$config = self::build_from_array($config_lines);\r
			// if APC caching is available and enabled, mark this for cache\r
			//$config->cache_in_apc = true;\r
			$config->cache_key = $matched_name;\r
			return $config;\r
		}\r
	}\r
	\r
	public static function build_from_array(array $lines) {\r
		$config = new SiteConfig();\r
		foreach ($lines as $line) {\r
			$line = trim($line);\r
			\r
			// skip comments, empty lines\r
			if ($line == '' || $line[0] == '#') continue;\r
			\r
			// get command\r
			$command = explode(':', $line, 2);\r
			// if there's no colon ':', skip this line\r
			if (count($command) != 2) continue;\r
			$val = trim($command[1]);\r
			$command = trim($command[0]);\r
			if ($command == '' || $val == '') continue;\r
			\r
			// check for commands where we accept multiple statements\r
			if (in_array($command, array('title', 'body', 'author', 'date', 'strip', 'strip_id_or_class', 'strip_image_src', 'single_page_link', 'single_page_link_in_feed', 'next_page_link', 'http_header', 'test_url', 'find_string', 'replace_string'))) {\r
				array_push($config->$command, $val);\r
			// check for single statement commands that evaluate to true or false\r
			} elseif (in_array($command, array('tidy', 'prune', 'autodetect_on_failure'))) {\r
				$config->$command = ($val == 'yes');\r
			// check for single statement commands stored as strings\r
			} elseif (in_array($command, array('parser'))) {\r
				$config->$command = $val;\r
			// check for replace_string(find): replace\r
			} elseif ((substr($command, -1) == ')') && preg_match('!^([a-z0-9_]+)\((.*?)\)$!i', $command, $match)) {\r
				if (in_array($match[1], array('replace_string'))) {\r
					$command = $match[1];\r
					array_push($config->find_string, $match[2]);\r
					array_push($config->$command, $val);\r
				}\r
			}\r
		}\r
		return $config;\r
	}\r
}\r
?>