<?php

/**
 * A base class for parsers.
 *
 * In the context of this plugin, a "parser" is a class that knows how to extract or modify
 * a specific type of links from a given piece of text. For example, there could be a "HTML Link"
 * parser that knows how to find and modify standard HTML links such as this one :
 * <a href="http://example.com/">Example</a>
 *
 * Other parsers could extract plaintext URLs or handle metadata fields.
 *
 * Each parser has a list of supported formats (e.g. "html", "plaintext", etc) and container types
 * (e.g. "post", "comment", "blogroll", etc). When something needs to be parsed, the involved
 * container class will look up the parsers that support the relevant format or the container's type,
 * and apply them to the to-be-parsed string.
 *
 * All sub-classes of blcParser should override at least the blcParser::parse() method.
 *
 * @see blcContainer::$fields
 *
 * @package Broken Link Checker
 * @access public
 */
class blcParser extends blcModule {

	var $parser_type;
	var $supported_formats    = array();
	var $supported_containers = array();

	/**
	 * Initialize the parser. Nothing much here.
	 *
	 * @return void
	 */
	function init() {
		parent::init();
		$this->parser_type = $this->module_id;
	}

	/**
	 * Called when the parser is activated.
	 *
	 * @return void
	 */
	function activated() {
		parent::activated();
		$this->resynch_relevant_containers();
	}

	/**
	 * Called when BLC is activated.
	 */
	function plugin_activated() {
		//Intentionally do nothing. BLC can not parse containers while it's inactive, so we can be
		//pretty sure that there are no already-parsed containers that need to be resynchronized.
	}

	/**
	 * Mark containers that this parser might be interested in as unparsed.
	 *
	 * @uses blcContainerHelper::mark_as_unsynched_where()
	 *
	 * @param bool $only_return If true, just return the list of formats and container types without actually modifying any synch. records.
	 * @return void|array Either nothing or an array in the form [ [format1=>timestamp1, ...], [container_type1=>timestamp1, ...] ]
	 */
	function resynch_relevant_containers( $only_return = false ) {
		global $blclog;
		$blclog->log( sprintf( '...... Parser "%s" is marking relevant items as unsynched', $this->module_id ) );

		$last_deactivated = $this->module_manager->get_last_deactivation_time( $this->module_id );

		$formats = array();
		foreach ( $this->supported_formats as $format ) {
			$formats[ $format ] = $last_deactivated;
		}

		$container_types = array();
		foreach ( $this->supported_containers as $container_type ) {
			$container_types[ $container_type ] = $last_deactivated;
		}

		if ( $only_return ) {
			return array( $formats, $container_types );
		} else {
			blcContainerHelper::mark_as_unsynched_where( $formats, $container_types );
		}
	}

	/**
	 * Parse a string for links.
	 *
	 * @param string $content The text to parse.
	 * @param string $base_url The base URL to use for normalizing relative URLs. If ommitted, the blog's root URL will be used.
	 * @param string $default_link_text
	 * @return array An array of new blcLinkInstance objects. The objects will include info about the links found, but not about the corresponding container entity.
	 */
	function parse( $content, $base_url = '', $default_link_text = '' ) {
		return array();
	}

	/**
	 * Change all links that have a certain URL to a new URL.
	 *
	 * @param string $content Look for links in this string.
	 * @param string $new_url Change the links to this URL.
	 * @param string $old_url The URL to look for.
	 * @param string $old_raw_url The raw, not-normalized URL of the links to look for. Optional.
	 *
	 * @return array|WP_Error If successful, the return value will be an associative array with two
	 * keys : 'content' - the modified content, and 'raw_url' - the new raw, non-normalized URL used
	 * for the modified links. In most cases, the returned raw_url will be equal to the new_url.
	 */
	function edit( $content, $new_url, $old_url, $old_raw_url ) {
		return new WP_Error(
			'not_implemented',
			sprintf( __( "Editing is not implemented in the '%s' parser", 'broken-link-checker' ), $this->parser_type )
		);
	}

	/**
	 * Remove all links that have a certain URL, leaving anchor text intact.
	 *
	 * @param string $content Look for links in this string.
	 * @param string $url The URL to look for.
	 * @param string $raw_url The raw, non-normalized version of the URL to look for. Optional.
	 * @return string Input string with all matching links removed.
	 */
	function unlink( $content, $url, $raw_url ) {
		return new WP_Error(
			'not_implemented',
			sprintf( __( "Unlinking is not implemented in the '%s' parser", 'broken-link-checker' ), $this->parser_type )
		);
	}

	/**
	 * Get the link text for printing in the "Broken Links" table.
	 * Sub-classes should override this method and display the link text in a way appropriate for the link type.
	 *
	 * @param blcLinkInstance $instance
	 * @param string $context
	 * @return string HTML
	 */
	function ui_get_link_text( $instance, $context = 'display' ) {
		return $instance->link_text;
	}

	/**
	 * Check if the parser supports editing the link text.
	 *
	 * @return bool
	 */
	public function is_link_text_editable() {
		return false;
	}

	/**
	 * Check if the parser supports editing the link URL.
	 *
	 * @return bool
	 */
	public function is_url_editable() {
		return true;
	}

	/**
	 * Turn a relative URL into an absolute one.
	 *
	 * WordPress 3.4 has WP_Http::make_absolute_url() which is well-tested but not as comprehensive
	 * as this implementation. For example, WP_Http doesn't properly handle directory traversal with "..",
	 * and it removes #anchors for no good reason. The BLC implementation deals with both pretty well.
	 *
	 * @param string $url Relative URL.
	 * @param string $base_url Base URL. If omitted, the blog's root URL will be used.
	 * @return string
	 */
	function relative2absolute( $url, $base_url = '' ) {
		if ( empty( $base_url ) ) {
			$base_url = home_url();
		}

		$p = @parse_url( $url );
		if ( ! $p ) {
			//URL is a malformed
			return false;
		}
		if ( isset( $p['scheme'] ) ) {
			return $url;
		}

		//If the relative URL is just a query string or anchor, simply attach it to the absolute URL and return
		$first_char = substr( $url, 0, 1 );
		if ( ( '?' == $first_char ) || ( '#' == $first_char ) ) {
			return $base_url . $url;
		}

		$parts = ( parse_url( $base_url ) );

		//Protocol-relative URLs start with "//". We just need to prepend the right protocol.
		if ( '//' === substr( $url, 0, 2 ) ) {
			$scheme = isset( $parts['scheme'] ) ? $parts['scheme'] : 'http';
			return $scheme . ':' . $url;
		}

		if ( substr( $url, 0, 1 ) == '/' ) {
			//Relative URL starts with a slash => ignore the base path and jump straight to the root.
			$path_segments = explode( '/', $url );
			array_shift( $path_segments );
		} else {
			if ( isset( $parts['path'] ) ) {
				$aparts = explode( '/', $parts['path'] );
				array_pop( $aparts );
				$aparts = array_filter( $aparts );
			} else {
				$aparts = array();
			}

			//Merge together the base path & the relative path
			$aparts = array_merge( $aparts, explode( '/', $url ) );

			//Filter the merged path
			$path_segments = array();
			foreach ( $aparts as $part ) {
				if ( '.' == $part ) {
					continue; //. = "this directory". It's basically a no-op, so we skip it.
				} elseif ( '..' == $part ) {
					array_pop( $path_segments );  //.. = one directory up. Remove the last seen path segment.
				} else {
					array_push( $path_segments, $part ); //Normal directory -> add it to the path.
				}
			}
		}
		$path = implode( '/', $path_segments );

		//Build the absolute URL.
		$url = '';
		if ( $parts['scheme'] ) {
			$url = "$parts[scheme]://";
		}
		if ( isset( $parts['user'] ) ) {
			$url .= $parts['user'];
			if ( isset( $parts['pass'] ) ) {
				$url .= ':' . $parts['pass'];
			}
			$url .= '@';
		}
		if ( isset( $parts['host'] ) ) {
			$url .= $parts['host'];
			if ( isset( $parts['port'] ) ) {
				$url .= ':' . $parts['port'];
			}
			$url .= '/';
		}
		$url .= $path;

		return $url;
	}

	/**
	 * Apply a callback function to all links found in a string and return the results.
	 *
	 * The first argument passed to the callback function will be an associative array
	 * of link data. If the optional $extra parameter is set, it will be passed as the
	 * second argument to the callback function.
	 *
	 * The link data array will contain at least these keys :
	 *  'href' - the URL of the link, as-is (i.e. without any sanitization or relative-to-absolute translation).
	 *  '#raw' - the raw link code, e.g. the entire '<a href="...">...</a>' tag of a HTML link.
	 *
	 * Sub-classes may also set additional keys.
	 *
	 * This method is currently used only internally, so sub-classes are not required
	 * to implement it.
	 *
	 * @param string $content A text string to parse for links.
	 * @param callback $callback Callback function to apply to all found links.
	 * @param mixed $extra If the optional $extra param. is supplied, it will be passed as the second parameter to the function $callback.
	 * @return array An array of all detected links after applying $callback to each of them.
	 */
	function map( $content, $callback, $extra = null ) {
		return array();
	}

	/**
	 * Modify all links found in a string using a callback function.
	 *
	 * The first argument passed to the callback function will be an associative array
	 * of link data. If the optional $extra parameter is set, it will be passed as the
	 * second argument to the callback function. See the map() method of this class for
	 * details on the first argument.
	 *
	 * The callback function should return either an associative array or a string. If
	 * a string is returned, the parser will replace the current link with the contents
	 * of that string. If an array is returned, the current link will be modified/rebuilt
	 * by substituting the new values for the old ones (e.g. returning array with the key
	 * 'href' set to 'http://example.com/' will replace the current link's URL with
	 * http://example.com/).
	 *
	 * This method is currently only used internally, so sub-classes are not required
	 * to implement it.
	 *
	 * @see blcParser::map()
	 *
	 * @param string $content A text string containing the links to edit.
	 * @param callback $callback Callback function used to modify the links.
	 * @param mixed $extra If supplied, $extra will be passed as the second parameter to the function $callback.
	 * @return string The modified input string.
	 */
	function multi_edit( $content, $callback, $extra = null ) {
		return $content; //No-op
	}
}

/**
 * A helper class for working with parsers. All its methods should be called statically.
 *
 * @see blcParser
 *
 * @package Broken Link Checker
 * @access public
 */
class blcParserHelper {

	/**
	 * Get the parser matching a parser type ID.
	 *
	 * @uses blcModuleManager::get_module()
	 *
	 * @param string $parser_type
	 * @return blcParser|null
	 */
	static function get_parser( $parser_type ) {
		$manager = blcModuleManager::getInstance();
		return $manager->get_module( $parser_type, true, 'parser' );
	}

	/**
	 * Get all parsers that support either the specified format or the container type.
	 * If a parser supports both, it will still be included only once.
	 *
	 * @param string $format
	 * @param string $container_type
	 * @return blcParser[]
	 */
	static function get_parsers( $format, $container_type ) {
		$found = array();

		//Retrieve a list of active parsers
		$manager        = blcModuleManager::getInstance();
		$active_parsers = $manager->get_modules_by_category( 'parser' );

		//Try each one
		foreach ( $active_parsers as $module_id => $module_data ) {
			$parser = $manager->get_module( $module_id ); //Will autoload if necessary
			if ( ! $parser ) {
				continue;
			}

			if ( in_array( $format, $parser->supported_formats ) || in_array( $container_type, $parser->supported_containers ) ) {
				array_push( $found, $parser );
			}
		}

		return $found;
	}
}