用于解析Wordpress导出的WXR格式的工具。可输出为PHP数组或者SQL语句。

WXRDecoder.php

<?php
class WXRDecoder
{
	protected $_parser;
	protected $_ns;

	public $posts=array();
	public $tags=array();

	public $post_table = 'tbl_posts';
	public $tag_table  = 'tbl_tags';

	public function __construct($filename='wordpress.xml')
	{
		$this->_parser = new SimpleXMLElement($filename, LIBXML_NSCLEAN | LIBXML_NOCDATA, true);
		$this->_ns = $this->_parser->getNamespaces(true);
	}

	public function set_post_table($table)
	{
		$this->post_table = $table;
	}

	public function set_tag_table($table)
	{
		$this->tag_table = $table;
	}

	public function parse_posts()
	{
		$items = $this->_parser->channel->item;

		foreach ($items as $item) {
			//非发布状态的文章直接跳过
			$status    = $item->children($this->_ns['wp'])->status;
			if ($status!='publish') {
				continue;
			}

			$title     = $item->title;
			$post_id   = $item->children($this->_ns['wp'])->post_id;
			$post_date = $item->children($this->_ns['wp'])->post_date;
			$type      = $item->children($this->_ns['wp'])->post_type;
			$content   = $item->children($this->_ns['content'])->encoded;

			//解析标签和分类（都统一合并到标签中）
			$tags = array();
			foreach ($item->category as $meta) {
				$t = ($meta=='Uncategorized') ? '未分类': $meta;
				$tags[] = (string)$t;
			}
			$tags = implode(',',array_unique($tags));

			$post = array('id'=>$post_id,'title'=>$title,'date'=>$post_date,
					'tags'=>$tags,'content'=>$content);
			$this->posts[(string)$post_id] = (object)$post;
		}

		sort($this->posts);
		return $this->posts;
	}

	public function parse_tags()
	{
		$items = $this->_parser->channel->children($this->_ns['wp'])->tag;

		$tags = array();
		foreach ($items as $item) {
			//非发布状态的文章直接跳过
			$tag = $item->children($this->_ns['wp'])->tag_name;
			$tags[] = (string)$tag;
		}
		$this->tags = array_unique($tags);

		sort($this->tags);
		return $this->tags;
	}

	public function parse_posts_sql()
	{
		$this->parse_posts();
		$sql = '';
		foreach ($this->posts as $p) {
			$sql .= "INSERT INTO `{$this->post_table}` (`id`,`title`,`date`,`tags`,`content`)";
			$sql .= " VALUES ('{$p->id}','{$p->title}','{$p->date}','{$p->tags}','{$p->content}');\n";
		}
		return $sql;
	}

	public function parse_tags_sql()
	{
		$this->parse_tags();
		$sql = '';
		foreach ($this->tags as $p) {
			$sql .= "INSERT INTO `{$this->tag_table}` (`name`) VALUES ('{$p}');\n";
		}
		return $sql;
	}
}

//$decoder = new WXRDecoder('gooth.wordpress.2013-07-10.xml');
//echo $decoder->parse_posts_sql();
//echo $decoder->parse_tags_sql();
?>