Parsing HTML using querypath library and creating a JSON feed of all blog posts

parse-html-json-blog-posts.php

<?php

require 'querypath/src/qp.php';

$posts = array();

$map = array(
		'kl.' => ' ',
		'.' => ' ',
		'des' => 'dec',
		'mai' => 'may',
		'okt' => 'oct'
	);

foreach ( glob('posts/*.html') as $file ) {
	$html = file_get_contents( $file );

	$date = current( explode( '_', basename( $file ) ) );

	if ( ! is_numeric( $date ) || $date < 100000 )
		continue;

	$tags = array();
	foreach( htmlqp( $html, '#main .meta .tags a' ) as $tag )
		$tags[] = $tag->text();
	
	$comments = array();
	foreach( htmlqp( $html, '#comments .comment' ) as $comment )
		$comments[] = array(
				'comment_author' => strip_tags( $comment->find('.meta h4')->text() ),
				'comment_author_url' => $comment->find('.meta h4 a')->attr('href'),
				'comment_content' => $comment->find('.content')->text(),
				'comment_date' => date( 'r', strtotime( str_replace( array_keys($map), array_values($map), $comment->find('.meta p')->text() ) ) ),
			);

	$post = array(
			'post_title' => htmlqp($html, '.entry h2')->text(),
			'post_content' => htmlqp( $html, '.content' )->children()->html(),
			'post_date' => date( 'r', $date ),
			'categories' => htmlqp( $html, '#main .meta:eq(3) li:first a' )->text(),
			'tags' => $tags,
			'comments' => $comments
		);

	$posts[] = $post;

	/*

	$item = array(
			sprintf( '<title>%s</title>', $post['post_title'] ),
			sprintf( '<pubDate>%s</pubDate>', $post['post_date'] ),
			sprintf( '<content:encoded><![CDATA[%s]]></content:encoded>', $post['post_content'] ),
			sprintf( '<category domain="category"><![CDATA[%s]]></category>', $post['categories'] ),
			'<wp:post_type>post</wp:post_type>'
		);

	foreach ( $post['tags'] as $tag )
		$item[] = sprintf( '<category domain="post_tag"><![CDATA[%s]]></category>', $tag );

	foreach ( $post['comments'] as $comment )
		$item[] = sprintf(
				'<wp:comment>
					 <wp:comment_author><![CDATA[%s]]></wp:comment_author>
					 <wp:comment_author_url>%s</wp:comment_author_url>
					 <wp:comment_date>%s</wp:comment_date>
					 <wp:comment_content><![CDATA[%s]]></wp:comment_content>
				</wp:comment>',
				$comment['comment_author'],
				$comment['comment_author_url'],
				$comment['comment_date'],
				$comment['comment_content']
			);

	$items[] = sprintf( '<item>%s</item>', implode( "\n", $item ) );
	*/
}

file_put_contents( 'export.json', json_encode( $posts ) );

echo 'DONE';

/*
// RSS export
file_put_contents( 
	'export.xml', 
	sprintf( 
		'<?xml version="1.0" encoding="UTF-8"?>
		<rss>
			<wp:wxr_version>1.2</wp:wxr_version>
			<channel>%s</channel>
		</rss>', 
		implode( "\n", $items ) 
	) 
);
*/