petermolnar's repositories — wp-webmention-again: 8fcfa2e36b4da23be95ac50638746d36b57337a7

0.4: adding more polished parsing of incoming comments

Peter Molnar hello@petermolnar.eu

Mon, 22 Feb 2016 11:59:08 +0000

commit

8fcfa2e36b4da23be95ac50638746d36b57337a7

parent

5110346ce26c9b9b92a695fb323e467a3913df74

3 files changed, 80 insertions(+), 72 deletions(-)

jump to

readme.txt

receiver.php

wp-webmention-again.php

M readme.txt → readme.txt

@@ -4,7 +4,7 @@ Donate link: https://paypal.me/petermolnar/3
 Tags: webmention, pingback, indieweb
 Requires at least: 4.3
 Tested up to: 4.4.1
-Stable tag: 0.3
+Stable tag: 0.4
 License: GPLv3
 License URI: http://www.gnu.org/licenses/gpl-3.0.html
 Required minimum PHP version: 5.3
@@ -32,6 +32,12 @@
 * every A. indicates BIG changes.
 * every .B version indicates new features.
 * every ..C indicates bugfixes for A.B version.
+
+
+= 0.4 =
+*2016-02-22*
+
+* adding more polished parsing of incoming comments
 
 = 0.3 =
 *2016-01-14*

M receiver.php → receiver.php

@@ -242,7 +242,7 @@ $post_id = static::validate_local( $target );
 
 		if (! $post_id || 0 == $post_id ) {
 			status_header( 404 );
-			echo '"target" not found.';
+			echo '"target" POST not found.';
 			exit;
 		}
 
@@ -355,7 +355,7 @@ *
 	 * @return bool|array false on error; plain array or Mf2 parsed (and
 	 *   flattened ) array of remote content on success
 	 */
-	protected static function try_receive_remote ( &$post_id, &$source, &$target ) {
+	protected static function try_receive_remote ( $post_id, $source, $target ) {
 
 		$content = false;
 		$q = static::_wp_remote_get( $source );
@@ -425,10 +425,26 @@ * @param array $content (Mf2) array of remote content
 	 *
 	 * @return bool|int false on error; insterted comment ID on success
 	 */
-	protected static function try_parse_remote ( &$post_id, &$source, &$target, &$content ) {
+	protected static function try_parse_remote ( $post_id, $source, $target, $content ) {
+
+		// default
+		$c = array (
+			'comment_author'				=> $source,
+			'comment_author_url'		=> $source,
+			'comment_author_email'	=> '',
+			'comment_post_ID'				=> $post_id,
+			'comment_type'					=> 'webmention',
+			'comment_date'					=> date("Y-m-d H:i:s"),
+			'comment_date_gmt'			=> date("Y-m-d H:i:s"),
+			'comment_agent'					=> __CLASS__,
+			'comment_approved' 			=> 0,
+			'comment_content'				=> sprintf( __( 'This entry was webmentioned on <a href="%s">%s</a>.' ), $source, $source ),
+			'comment_meta'					=> array(),
+		);
 
+		// try to find the actual entry
 		$item = false;
-		$p_authors = array();
+		$p_authors = $items = $comments = array();
 
 		if ( isset( $content['items']['properties'] ) && isset( $content['items']['type'] ) ) {
 			$item = $content['items'];
@@ -452,27 +468,20 @@ $item = array_pop( $items );
 		elseif ( empty( $items ) && ! empty( $comments ) )
 			$item = array_pop( $comments );
 
+
+		// if the entry wasn't found, fall back to a regular mention
 		if (! $item || empty( $item )) {
 			static::debug('    no parseable h-entry found, saving as standard mention comment');
-			$c = array (
-				'comment_author'				=> $source,
-				'comment_author_url'		=> $source,
-				'comment_author_email'	=> '',
-				'comment_post_ID'				=> $post_id,
-				'comment_type'					=> 'webmention',
-				'comment_date'					=> date("Y-m-d H:i:s"),
-				'comment_date_gmt'			=> date("Y-m-d H:i:s"),
-				'comment_agent'					=> __CLASS__,
-				'comment_approved' 			=> 0,
-				'comment_content'				=> sprintf( __( 'This entry was webmentioned on <a href="%s">%s</a>.' ), $source, $source ),
-			);
 			return $c;
 		}
 
-		$author_url = $source;
-		// process author
-		$author_name = $avatar = $a = false;
+		// otherwise start parsing
+
+		// get real source if set
+		if ( isset ( $item['properties']['url'] ) && !empty( $item['properties']['url'] ) )
+			$c['comment_meta']['comment_url'] = ( is_array ( $item['properties']['url'] ) ) ? array_pop ( $item['properties']['url'] ) :  $item['properties']['url'];
 
+		// try to get author
 		if ( isset( $item['properties']['author'] ) ) {
 			$a = $item['properties']['author'];
 		}
@@ -480,77 +489,63 @@ elseif ( ! empty( $p_authors ) ) {
 			$a = array_pop( $p_authors );
 		}
 
+		// try to process author
 		if ( $a && isset( $a['properties'] ) ) {
 			$a = $a['properties'];
 
+			// author name
 			if ( isset($a['name']) && ! empty( $a['name'] ) )
-				$author_name = $a['name'];
+				$c['comment_author'] = $a['name'];
 
+			// avatar
 			$try_photos = array ('photo', 'avatar');
 			$p = false;
 			foreach ( $try_photos as $photo ) {
 				if (isset( $a[ $photo ]) && ! empty( $a[ $photo ] ) ) {
 					$p = $a[ $photo ];
 					if ( !empty( $p ) ) {
-						$avatar = $p;
+						$c['comment_meta']['avatar'] = $p;
 						break;
 					}
 				}
 			}
 
-			if ( isset($a['url']) && ! empty( $a['url'] ) )
-				$author_url = $a['url'];
+			// author url & author email
+			if ( isset($a['url']) && ! empty( $a['url'] ) ) {
+				if ( is_array ($a['url']) )
+					$c['comment_author_url'] = array_pop ( $a['url'] );
+				else
+					$c['comment_author_url'] = $a['url'];
+
+				$c['comment_author_email'] = static::try_get_author_email ( $c['comment_author_url'] );
+			}
 		}
 
-		// process type
-		$type = 'webmention';
-
 		foreach ( static::mftypes() as $k => $mapped ) {
 			if ( is_array( $item['properties'] ) && isset( $item['properties'][ $mapped ]) )
-				$type = $k;
+				$c['comment_type'] = $k;
 		}
 
 		//process content
-		$c = '';
+		$content = '';
 		if ( isset( $item['properties']['content'] ) && isset( $item['properties']['content']['html'] ) )
-			$c = $item['properties']['content']['html'];
+			$content = $item['properties']['content']['html'];
 		if ( isset( $item['properties']['content'] ) && isset( $item['properties']['content']['value'] ) )
-			$c = $item['properties']['content']['value'];
+			$content = $item['properties']['content']['value'];
 
-		$c = wp_filter_kses ( $c );
+		$c['comment_content'] = wp_filter_kses ( $content );
 
 		// REACJI
-		$emoji = EmojiRecognizer::isSingleEmoji( $c );
+		$emoji = EmojiRecognizer::isSingleEmoji( $content );
 
-		if ( $emoji ) {
-			static::debug( "wheeeee, reacji!" );
-			$type = 'reacji';
-			//static::register_reacji( $type );
-		}
+		if ( $emoji )
+			$c['comment_type'] = 'reacji';
 
 		// process date
 		if ( isset( $item['properties']['modified'] ) )
-			$date = strtotime( $item['properties']['modified'] );
+			$c['comment_date'] = date( "Y-m-d H:i:s", strtotime( $item['properties']['modified'] ));
 		elseif ( isset( $item['properties']['published'] ) )
-			$date = strtotime( $item['properties']['published'] );
-		else
-			$date = time();
-
-		$name = empty( $author_name ) ? $source : $author_name;
-
-		$c = array (
-			'comment_author'				=> $name,
-			'comment_author_url'		=> $author_url,
-			'comment_author_email'	=> static::try_get_author_email ( $author_url ),
-			'comment_post_ID'				=> $post_id,
-			'comment_type'					=> $type,
-			'comment_date'					=> date( "Y-m-d H:i:s", $date ),
-			'comment_date_gmt'			=> date( "Y-m-d H:i:s", $date ),
-			'comment_agent'					=> __CLASS__,
-			'comment_approved' 			=> 0,
-			'comment_content'				=> $c,
-			'comment_avatar'				=> $avatar,
-		);
+			$c['comment_date'] = date( "Y-m-d H:i:s", strtotime( $item['properties']['published'] ));
 
 		return $c;
 	}
@@ -558,21 +553,23 @@
 	/**
 	 * Comment inserter
 	 *
-	 * @param string &$post_id post ID
-	 * @param string &$source Originator URL
-	 * @param string &$target Target URL
-	 * @param mixed &$raw Raw format of the comment, like JSON response from the provider
-	 * @param array &$comment array formatted to match a WP Comment requirement
+	 * @param string $post_id post ID
+	 * @param string $source Originator URL
+	 * @param string $target Target URL
+	 * @param mixed $raw Raw format of the comment, like JSON response from the provider
+	 * @param array $comment array formatted to match a WP Comment requirement
 	 *
 	 */
-	protected static function insert_comment ( &$post_id, &$source, &$target, &$raw, &$comment ) {
+	protected static function insert_comment ( $post_id, $source, $target, $raw, $comment ) {
+
+		$comment = apply_filters ( 'wp_webmention_again_insert_comment', $comment, $post_id, $source, $target, $raw );
 
 		$comment_id = false;
 
 		$avatar = false;
-		if( isset( $comment['comment_avatar'] ) ) {
-			$avatar = $comment['comment_avatar'];
-			unset( $comment['comment_avatar'] );
+		if( isset( $comment['comment_meta'] ) ) {
+			$meta = $comment['comment_meta'];
+			unset( $comment['comment_meta'] );
 		}
 
 		// safety first
@@ -638,8 +635,11 @@
 		if ( $comment_id = wp_new_comment( $comment ) ) {
 
 			// add avatar for later use if present
-			if ( ! empty( $avatar ) )
-				update_comment_meta( $comment_id, 'avatar', $avatar );
+			if ( ! empty( $meta ) ) {
+				foreach ( $meta as $key => $m ) {
+					update_comment_meta( $comment_id, $key, $m );
+				}
+			}
 
 			// full raw response for the vote, just in case
 			update_comment_meta( $comment_id, 'webmention_source_mf2', $raw );
@@ -659,6 +659,7 @@ }
 
 		// re-add flood control
 		add_filter( 'check_comment_flood', 'check_comment_flood_db', 10, 3 );
+		do_action ( 'wp_webmention_again_insert_comment' );
 
 		static::debug( $r );

M wp-webmention-again.php → wp-webmention-again.php

@@ -3,7 +3,7 @@ /*
 Plugin Name: wp-webmention-again
 Plugin URI: https://github.com/petermolnar/wp-webmention-again
 Description:
-Version: 0.3
+Version: 0.4
 Author: Peter Molnar <hello@petermolnar.eu>
 Author URI: http://petermolnar.eu/
 License: GPLv3
@@ -452,8 +452,9 @@ * @return bool|int false on not found, int post_id on found
 	 */
 	protected static function validate_local ( $url ) {
 		// normalize url scheme, url_to_postid will take care of it anyway
-		$url = preg_replace( '/^https?:\/\//i', 'http://', $url );
-		return url_to_postid( $url );
+		$url = preg_replace( '/^https?:\/\//i', 'http://', strtolower($url) );
+		$postid = url_to_postid( $url );
+		return apply_filters ('wp_webmention_again_validate_local', $postid , $url );
 	}
 
 	/**