wp/wp-content/plugins/wordpress-importer/parsers.php
author ymh <ymh.work@gmail.com>
Wed, 06 Nov 2013 03:21:17 +0000
changeset 0 d970ebf37754
child 7 cf61fcea0001
permissions -rw-r--r--
first import
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
0
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
     1
<?php
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
     2
/**
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
     3
 * WordPress eXtended RSS file parser implementations
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
     4
 *
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
     5
 * @package WordPress
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
     6
 * @subpackage Importer
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
     7
 */
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
     8
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
     9
/**
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    10
 * WordPress Importer class for managing parsing of WXR files.
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    11
 */
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    12
class WXR_Parser {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    13
	function parse( $file ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    14
		// Attempt to use proper XML parsers first
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    15
		if ( extension_loaded( 'simplexml' ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    16
			$parser = new WXR_Parser_SimpleXML;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    17
			$result = $parser->parse( $file );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    18
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    19
			// If SimpleXML succeeds or this is an invalid WXR file then return the results
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    20
			if ( ! is_wp_error( $result ) || 'SimpleXML_parse_error' != $result->get_error_code() )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    21
				return $result;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    22
		} else if ( extension_loaded( 'xml' ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    23
			$parser = new WXR_Parser_XML;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    24
			$result = $parser->parse( $file );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    25
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    26
			// If XMLParser succeeds or this is an invalid WXR file then return the results
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    27
			if ( ! is_wp_error( $result ) || 'XML_parse_error' != $result->get_error_code() )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    28
				return $result;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    29
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    30
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    31
		// We have a malformed XML file, so display the error and fallthrough to regex
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    32
		if ( isset($result) && defined('IMPORT_DEBUG') && IMPORT_DEBUG ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    33
			echo '<pre>';
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    34
			if ( 'SimpleXML_parse_error' == $result->get_error_code() ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    35
				foreach  ( $result->get_error_data() as $error )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    36
					echo $error->line . ':' . $error->column . ' ' . esc_html( $error->message ) . "\n";
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    37
			} else if ( 'XML_parse_error' == $result->get_error_code() ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    38
				$error = $result->get_error_data();
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    39
				echo $error[0] . ':' . $error[1] . ' ' . esc_html( $error[2] );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    40
			}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    41
			echo '</pre>';
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    42
			echo '<p><strong>' . __( 'There was an error when reading this WXR file', 'wordpress-importer' ) . '</strong><br />';
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    43
			echo __( 'Details are shown above. The importer will now try again with a different parser...', 'wordpress-importer' ) . '</p>';
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    44
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    45
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    46
		// use regular expressions if nothing else available or this is bad XML
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    47
		$parser = new WXR_Parser_Regex;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    48
		return $parser->parse( $file );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    49
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    50
}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    51
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    52
/**
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    53
 * WXR Parser that makes use of the SimpleXML PHP extension.
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    54
 */
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    55
class WXR_Parser_SimpleXML {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    56
	function parse( $file ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    57
		$authors = $posts = $categories = $tags = $terms = array();
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    58
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    59
		$internal_errors = libxml_use_internal_errors(true);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    60
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    61
		$dom = new DOMDocument;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    62
		$old_value = null;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    63
		if ( function_exists( 'libxml_disable_entity_loader' ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    64
			$old_value = libxml_disable_entity_loader( true );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    65
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    66
		$success = $dom->loadXML( file_get_contents( $file ) );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    67
		if ( ! is_null( $old_value ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    68
			libxml_disable_entity_loader( $old_value );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    69
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    70
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    71
		if ( ! $success || isset( $dom->doctype ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    72
			return new WP_Error( 'SimpleXML_parse_error', __( 'There was an error when reading this WXR file', 'wordpress-importer' ), libxml_get_errors() );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    73
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    74
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    75
		$xml = simplexml_import_dom( $dom );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    76
		unset( $dom );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    77
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    78
		// halt if loading produces an error
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    79
		if ( ! $xml )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    80
			return new WP_Error( 'SimpleXML_parse_error', __( 'There was an error when reading this WXR file', 'wordpress-importer' ), libxml_get_errors() );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    81
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    82
		$wxr_version = $xml->xpath('/rss/channel/wp:wxr_version');
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    83
		if ( ! $wxr_version )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    84
			return new WP_Error( 'WXR_parse_error', __( 'This does not appear to be a WXR file, missing/invalid WXR version number', 'wordpress-importer' ) );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    85
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    86
		$wxr_version = (string) trim( $wxr_version[0] );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    87
		// confirm that we are dealing with the correct file format
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    88
		if ( ! preg_match( '/^\d+\.\d+$/', $wxr_version ) )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    89
			return new WP_Error( 'WXR_parse_error', __( 'This does not appear to be a WXR file, missing/invalid WXR version number', 'wordpress-importer' ) );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    90
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    91
		$base_url = $xml->xpath('/rss/channel/wp:base_site_url');
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    92
		$base_url = (string) trim( $base_url[0] );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    93
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    94
		$namespaces = $xml->getDocNamespaces();
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    95
		if ( ! isset( $namespaces['wp'] ) )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    96
			$namespaces['wp'] = 'http://wordpress.org/export/1.1/';
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    97
		if ( ! isset( $namespaces['excerpt'] ) )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    98
			$namespaces['excerpt'] = 'http://wordpress.org/export/1.1/excerpt/';
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
    99
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   100
		// grab authors
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   101
		foreach ( $xml->xpath('/rss/channel/wp:author') as $author_arr ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   102
			$a = $author_arr->children( $namespaces['wp'] );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   103
			$login = (string) $a->author_login;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   104
			$authors[$login] = array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   105
				'author_id' => (int) $a->author_id,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   106
				'author_login' => $login,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   107
				'author_email' => (string) $a->author_email,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   108
				'author_display_name' => (string) $a->author_display_name,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   109
				'author_first_name' => (string) $a->author_first_name,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   110
				'author_last_name' => (string) $a->author_last_name
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   111
			);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   112
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   113
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   114
		// grab cats, tags and terms
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   115
		foreach ( $xml->xpath('/rss/channel/wp:category') as $term_arr ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   116
			$t = $term_arr->children( $namespaces['wp'] );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   117
			$categories[] = array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   118
				'term_id' => (int) $t->term_id,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   119
				'category_nicename' => (string) $t->category_nicename,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   120
				'category_parent' => (string) $t->category_parent,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   121
				'cat_name' => (string) $t->cat_name,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   122
				'category_description' => (string) $t->category_description
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   123
			);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   124
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   125
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   126
		foreach ( $xml->xpath('/rss/channel/wp:tag') as $term_arr ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   127
			$t = $term_arr->children( $namespaces['wp'] );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   128
			$tags[] = array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   129
				'term_id' => (int) $t->term_id,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   130
				'tag_slug' => (string) $t->tag_slug,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   131
				'tag_name' => (string) $t->tag_name,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   132
				'tag_description' => (string) $t->tag_description
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   133
			);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   134
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   135
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   136
		foreach ( $xml->xpath('/rss/channel/wp:term') as $term_arr ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   137
			$t = $term_arr->children( $namespaces['wp'] );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   138
			$terms[] = array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   139
				'term_id' => (int) $t->term_id,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   140
				'term_taxonomy' => (string) $t->term_taxonomy,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   141
				'slug' => (string) $t->term_slug,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   142
				'term_parent' => (string) $t->term_parent,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   143
				'term_name' => (string) $t->term_name,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   144
				'term_description' => (string) $t->term_description
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   145
			);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   146
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   147
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   148
		// grab posts
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   149
		foreach ( $xml->channel->item as $item ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   150
			$post = array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   151
				'post_title' => (string) $item->title,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   152
				'guid' => (string) $item->guid,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   153
			);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   154
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   155
			$dc = $item->children( 'http://purl.org/dc/elements/1.1/' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   156
			$post['post_author'] = (string) $dc->creator;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   157
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   158
			$content = $item->children( 'http://purl.org/rss/1.0/modules/content/' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   159
			$excerpt = $item->children( $namespaces['excerpt'] );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   160
			$post['post_content'] = (string) $content->encoded;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   161
			$post['post_excerpt'] = (string) $excerpt->encoded;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   162
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   163
			$wp = $item->children( $namespaces['wp'] );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   164
			$post['post_id'] = (int) $wp->post_id;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   165
			$post['post_date'] = (string) $wp->post_date;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   166
			$post['post_date_gmt'] = (string) $wp->post_date_gmt;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   167
			$post['comment_status'] = (string) $wp->comment_status;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   168
			$post['ping_status'] = (string) $wp->ping_status;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   169
			$post['post_name'] = (string) $wp->post_name;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   170
			$post['status'] = (string) $wp->status;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   171
			$post['post_parent'] = (int) $wp->post_parent;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   172
			$post['menu_order'] = (int) $wp->menu_order;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   173
			$post['post_type'] = (string) $wp->post_type;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   174
			$post['post_password'] = (string) $wp->post_password;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   175
			$post['is_sticky'] = (int) $wp->is_sticky;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   176
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   177
			if ( isset($wp->attachment_url) )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   178
				$post['attachment_url'] = (string) $wp->attachment_url;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   179
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   180
			foreach ( $item->category as $c ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   181
				$att = $c->attributes();
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   182
				if ( isset( $att['nicename'] ) )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   183
					$post['terms'][] = array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   184
						'name' => (string) $c,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   185
						'slug' => (string) $att['nicename'],
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   186
						'domain' => (string) $att['domain']
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   187
					);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   188
			}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   189
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   190
			foreach ( $wp->postmeta as $meta ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   191
				$post['postmeta'][] = array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   192
					'key' => (string) $meta->meta_key,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   193
					'value' => (string) $meta->meta_value
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   194
				);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   195
			}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   196
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   197
			foreach ( $wp->comment as $comment ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   198
				$meta = array();
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   199
				if ( isset( $comment->commentmeta ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   200
					foreach ( $comment->commentmeta as $m ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   201
						$meta[] = array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   202
							'key' => (string) $m->meta_key,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   203
							'value' => (string) $m->meta_value
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   204
						);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   205
					}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   206
				}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   207
			
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   208
				$post['comments'][] = array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   209
					'comment_id' => (int) $comment->comment_id,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   210
					'comment_author' => (string) $comment->comment_author,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   211
					'comment_author_email' => (string) $comment->comment_author_email,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   212
					'comment_author_IP' => (string) $comment->comment_author_IP,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   213
					'comment_author_url' => (string) $comment->comment_author_url,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   214
					'comment_date' => (string) $comment->comment_date,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   215
					'comment_date_gmt' => (string) $comment->comment_date_gmt,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   216
					'comment_content' => (string) $comment->comment_content,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   217
					'comment_approved' => (string) $comment->comment_approved,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   218
					'comment_type' => (string) $comment->comment_type,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   219
					'comment_parent' => (string) $comment->comment_parent,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   220
					'comment_user_id' => (int) $comment->comment_user_id,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   221
					'commentmeta' => $meta,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   222
				);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   223
			}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   224
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   225
			$posts[] = $post;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   226
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   227
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   228
		return array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   229
			'authors' => $authors,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   230
			'posts' => $posts,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   231
			'categories' => $categories,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   232
			'tags' => $tags,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   233
			'terms' => $terms,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   234
			'base_url' => $base_url,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   235
			'version' => $wxr_version
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   236
		);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   237
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   238
}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   239
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   240
/**
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   241
 * WXR Parser that makes use of the XML Parser PHP extension.
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   242
 */
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   243
class WXR_Parser_XML {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   244
	var $wp_tags = array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   245
		'wp:post_id', 'wp:post_date', 'wp:post_date_gmt', 'wp:comment_status', 'wp:ping_status', 'wp:attachment_url',
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   246
		'wp:status', 'wp:post_name', 'wp:post_parent', 'wp:menu_order', 'wp:post_type', 'wp:post_password',
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   247
		'wp:is_sticky', 'wp:term_id', 'wp:category_nicename', 'wp:category_parent', 'wp:cat_name', 'wp:category_description',
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   248
		'wp:tag_slug', 'wp:tag_name', 'wp:tag_description', 'wp:term_taxonomy', 'wp:term_parent',
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   249
		'wp:term_name', 'wp:term_description', 'wp:author_id', 'wp:author_login', 'wp:author_email', 'wp:author_display_name',
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   250
		'wp:author_first_name', 'wp:author_last_name',
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   251
	);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   252
	var $wp_sub_tags = array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   253
		'wp:comment_id', 'wp:comment_author', 'wp:comment_author_email', 'wp:comment_author_url',
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   254
		'wp:comment_author_IP',	'wp:comment_date', 'wp:comment_date_gmt', 'wp:comment_content',
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   255
		'wp:comment_approved', 'wp:comment_type', 'wp:comment_parent', 'wp:comment_user_id',
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   256
	);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   257
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   258
	function parse( $file ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   259
		$this->wxr_version = $this->in_post = $this->cdata = $this->data = $this->sub_data = $this->in_tag = $this->in_sub_tag = false;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   260
		$this->authors = $this->posts = $this->term = $this->category = $this->tag = array();
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   261
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   262
		$xml = xml_parser_create( 'UTF-8' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   263
		xml_parser_set_option( $xml, XML_OPTION_SKIP_WHITE, 1 );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   264
		xml_parser_set_option( $xml, XML_OPTION_CASE_FOLDING, 0 );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   265
		xml_set_object( $xml, $this );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   266
		xml_set_character_data_handler( $xml, 'cdata' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   267
		xml_set_element_handler( $xml, 'tag_open', 'tag_close' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   268
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   269
		if ( ! xml_parse( $xml, file_get_contents( $file ), true ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   270
			$current_line = xml_get_current_line_number( $xml );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   271
			$current_column = xml_get_current_column_number( $xml );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   272
			$error_code = xml_get_error_code( $xml );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   273
			$error_string = xml_error_string( $error_code );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   274
			return new WP_Error( 'XML_parse_error', 'There was an error when reading this WXR file', array( $current_line, $current_column, $error_string ) );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   275
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   276
		xml_parser_free( $xml );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   277
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   278
		if ( ! preg_match( '/^\d+\.\d+$/', $this->wxr_version ) )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   279
			return new WP_Error( 'WXR_parse_error', __( 'This does not appear to be a WXR file, missing/invalid WXR version number', 'wordpress-importer' ) );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   280
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   281
		return array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   282
			'authors' => $this->authors,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   283
			'posts' => $this->posts,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   284
			'categories' => $this->category,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   285
			'tags' => $this->tag,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   286
			'terms' => $this->term,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   287
			'base_url' => $this->base_url,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   288
			'version' => $this->wxr_version
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   289
		);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   290
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   291
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   292
	function tag_open( $parse, $tag, $attr ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   293
		if ( in_array( $tag, $this->wp_tags ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   294
			$this->in_tag = substr( $tag, 3 );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   295
			return;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   296
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   297
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   298
		if ( in_array( $tag, $this->wp_sub_tags ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   299
			$this->in_sub_tag = substr( $tag, 3 );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   300
			return;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   301
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   302
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   303
		switch ( $tag ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   304
			case 'category':
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   305
				if ( isset($attr['domain'], $attr['nicename']) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   306
					$this->sub_data['domain'] = $attr['domain'];
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   307
					$this->sub_data['slug'] = $attr['nicename'];
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   308
				}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   309
				break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   310
			case 'item': $this->in_post = true;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   311
			case 'title': if ( $this->in_post ) $this->in_tag = 'post_title'; break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   312
			case 'guid': $this->in_tag = 'guid'; break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   313
			case 'dc:creator': $this->in_tag = 'post_author'; break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   314
			case 'content:encoded': $this->in_tag = 'post_content'; break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   315
			case 'excerpt:encoded': $this->in_tag = 'post_excerpt'; break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   316
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   317
			case 'wp:term_slug': $this->in_tag = 'slug'; break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   318
			case 'wp:meta_key': $this->in_sub_tag = 'key'; break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   319
			case 'wp:meta_value': $this->in_sub_tag = 'value'; break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   320
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   321
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   322
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   323
	function cdata( $parser, $cdata ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   324
		if ( ! trim( $cdata ) )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   325
			return;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   326
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   327
		$this->cdata .= trim( $cdata );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   328
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   329
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   330
	function tag_close( $parser, $tag ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   331
		switch ( $tag ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   332
			case 'wp:comment':
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   333
				unset( $this->sub_data['key'], $this->sub_data['value'] ); // remove meta sub_data
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   334
				if ( ! empty( $this->sub_data ) )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   335
					$this->data['comments'][] = $this->sub_data;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   336
				$this->sub_data = false;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   337
				break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   338
			case 'wp:commentmeta':
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   339
				$this->sub_data['commentmeta'][] = array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   340
					'key' => $this->sub_data['key'],
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   341
					'value' => $this->sub_data['value']
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   342
				);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   343
				break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   344
			case 'category':
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   345
				if ( ! empty( $this->sub_data ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   346
					$this->sub_data['name'] = $this->cdata;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   347
					$this->data['terms'][] = $this->sub_data;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   348
				}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   349
				$this->sub_data = false;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   350
				break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   351
			case 'wp:postmeta':
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   352
				if ( ! empty( $this->sub_data ) )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   353
					$this->data['postmeta'][] = $this->sub_data;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   354
				$this->sub_data = false;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   355
				break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   356
			case 'item':
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   357
				$this->posts[] = $this->data;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   358
				$this->data = false;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   359
				break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   360
			case 'wp:category':
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   361
			case 'wp:tag':
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   362
			case 'wp:term':
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   363
				$n = substr( $tag, 3 );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   364
				array_push( $this->$n, $this->data );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   365
				$this->data = false;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   366
				break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   367
			case 'wp:author':
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   368
				if ( ! empty($this->data['author_login']) )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   369
					$this->authors[$this->data['author_login']] = $this->data;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   370
				$this->data = false;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   371
				break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   372
			case 'wp:base_site_url':
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   373
				$this->base_url = $this->cdata;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   374
				break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   375
			case 'wp:wxr_version':
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   376
				$this->wxr_version = $this->cdata;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   377
				break;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   378
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   379
			default:
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   380
				if ( $this->in_sub_tag ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   381
					$this->sub_data[$this->in_sub_tag] = ! empty( $this->cdata ) ? $this->cdata : '';
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   382
					$this->in_sub_tag = false;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   383
				} else if ( $this->in_tag ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   384
					$this->data[$this->in_tag] = ! empty( $this->cdata ) ? $this->cdata : '';
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   385
					$this->in_tag = false;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   386
				}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   387
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   388
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   389
		$this->cdata = false;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   390
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   391
}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   392
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   393
/**
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   394
 * WXR Parser that uses regular expressions. Fallback for installs without an XML parser.
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   395
 */
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   396
class WXR_Parser_Regex {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   397
	var $authors = array();
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   398
	var $posts = array();
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   399
	var $categories = array();
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   400
	var $tags = array();
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   401
	var $terms = array();
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   402
	var $base_url = '';
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   403
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   404
	function WXR_Parser_Regex() {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   405
		$this->__construct();
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   406
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   407
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   408
	function __construct() {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   409
		$this->has_gzip = is_callable( 'gzopen' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   410
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   411
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   412
	function parse( $file ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   413
		$wxr_version = $in_post = false;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   414
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   415
		$fp = $this->fopen( $file, 'r' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   416
		if ( $fp ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   417
			while ( ! $this->feof( $fp ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   418
				$importline = rtrim( $this->fgets( $fp ) );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   419
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   420
				if ( ! $wxr_version && preg_match( '|<wp:wxr_version>(\d+\.\d+)</wp:wxr_version>|', $importline, $version ) )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   421
					$wxr_version = $version[1];
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   422
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   423
				if ( false !== strpos( $importline, '<wp:base_site_url>' ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   424
					preg_match( '|<wp:base_site_url>(.*?)</wp:base_site_url>|is', $importline, $url );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   425
					$this->base_url = $url[1];
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   426
					continue;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   427
				}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   428
				if ( false !== strpos( $importline, '<wp:category>' ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   429
					preg_match( '|<wp:category>(.*?)</wp:category>|is', $importline, $category );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   430
					$this->categories[] = $this->process_category( $category[1] );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   431
					continue;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   432
				}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   433
				if ( false !== strpos( $importline, '<wp:tag>' ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   434
					preg_match( '|<wp:tag>(.*?)</wp:tag>|is', $importline, $tag );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   435
					$this->tags[] = $this->process_tag( $tag[1] );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   436
					continue;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   437
				}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   438
				if ( false !== strpos( $importline, '<wp:term>' ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   439
					preg_match( '|<wp:term>(.*?)</wp:term>|is', $importline, $term );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   440
					$this->terms[] = $this->process_term( $term[1] );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   441
					continue;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   442
				}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   443
				if ( false !== strpos( $importline, '<wp:author>' ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   444
					preg_match( '|<wp:author>(.*?)</wp:author>|is', $importline, $author );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   445
					$a = $this->process_author( $author[1] );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   446
					$this->authors[$a['author_login']] = $a;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   447
					continue;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   448
				}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   449
				if ( false !== strpos( $importline, '<item>' ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   450
					$post = '';
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   451
					$in_post = true;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   452
					continue;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   453
				}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   454
				if ( false !== strpos( $importline, '</item>' ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   455
					$in_post = false;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   456
					$this->posts[] = $this->process_post( $post );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   457
					continue;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   458
				}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   459
				if ( $in_post ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   460
					$post .= $importline . "\n";
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   461
				}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   462
			}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   463
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   464
			$this->fclose($fp);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   465
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   466
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   467
		if ( ! $wxr_version )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   468
			return new WP_Error( 'WXR_parse_error', __( 'This does not appear to be a WXR file, missing/invalid WXR version number', 'wordpress-importer' ) );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   469
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   470
		return array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   471
			'authors' => $this->authors,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   472
			'posts' => $this->posts,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   473
			'categories' => $this->categories,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   474
			'tags' => $this->tags,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   475
			'terms' => $this->terms,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   476
			'base_url' => $this->base_url,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   477
			'version' => $wxr_version
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   478
		);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   479
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   480
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   481
	function get_tag( $string, $tag ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   482
		preg_match( "|<$tag.*?>(.*?)</$tag>|is", $string, $return );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   483
		if ( isset( $return[1] ) ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   484
			if ( substr( $return[1], 0, 9 ) == '<![CDATA[' ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   485
				if ( strpos( $return[1], ']]]]><![CDATA[>' ) !== false ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   486
					preg_match_all( '|<!\[CDATA\[(.*?)\]\]>|s', $return[1], $matches );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   487
					$return = '';
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   488
					foreach( $matches[1] as $match )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   489
						$return .= $match;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   490
				} else {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   491
					$return = preg_replace( '|^<!\[CDATA\[(.*)\]\]>$|s', '$1', $return[1] );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   492
				}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   493
			} else {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   494
				$return = $return[1];
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   495
			}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   496
		} else {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   497
			$return = '';
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   498
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   499
		return $return;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   500
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   501
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   502
	function process_category( $c ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   503
		return array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   504
			'term_id' => $this->get_tag( $c, 'wp:term_id' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   505
			'cat_name' => $this->get_tag( $c, 'wp:cat_name' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   506
			'category_nicename'	=> $this->get_tag( $c, 'wp:category_nicename' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   507
			'category_parent' => $this->get_tag( $c, 'wp:category_parent' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   508
			'category_description' => $this->get_tag( $c, 'wp:category_description' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   509
		);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   510
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   511
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   512
	function process_tag( $t ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   513
		return array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   514
			'term_id' => $this->get_tag( $t, 'wp:term_id' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   515
			'tag_name' => $this->get_tag( $t, 'wp:tag_name' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   516
			'tag_slug' => $this->get_tag( $t, 'wp:tag_slug' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   517
			'tag_description' => $this->get_tag( $t, 'wp:tag_description' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   518
		);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   519
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   520
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   521
	function process_term( $t ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   522
		return array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   523
			'term_id' => $this->get_tag( $t, 'wp:term_id' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   524
			'term_taxonomy' => $this->get_tag( $t, 'wp:term_taxonomy' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   525
			'slug' => $this->get_tag( $t, 'wp:term_slug' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   526
			'term_parent' => $this->get_tag( $t, 'wp:term_parent' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   527
			'term_name' => $this->get_tag( $t, 'wp:term_name' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   528
			'term_description' => $this->get_tag( $t, 'wp:term_description' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   529
		);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   530
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   531
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   532
	function process_author( $a ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   533
		return array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   534
			'author_id' => $this->get_tag( $a, 'wp:author_id' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   535
			'author_login' => $this->get_tag( $a, 'wp:author_login' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   536
			'author_email' => $this->get_tag( $a, 'wp:author_email' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   537
			'author_display_name' => $this->get_tag( $a, 'wp:author_display_name' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   538
			'author_first_name' => $this->get_tag( $a, 'wp:author_first_name' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   539
			'author_last_name' => $this->get_tag( $a, 'wp:author_last_name' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   540
		);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   541
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   542
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   543
	function process_post( $post ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   544
		$post_id        = $this->get_tag( $post, 'wp:post_id' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   545
		$post_title     = $this->get_tag( $post, 'title' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   546
		$post_date      = $this->get_tag( $post, 'wp:post_date' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   547
		$post_date_gmt  = $this->get_tag( $post, 'wp:post_date_gmt' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   548
		$comment_status = $this->get_tag( $post, 'wp:comment_status' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   549
		$ping_status    = $this->get_tag( $post, 'wp:ping_status' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   550
		$status         = $this->get_tag( $post, 'wp:status' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   551
		$post_name      = $this->get_tag( $post, 'wp:post_name' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   552
		$post_parent    = $this->get_tag( $post, 'wp:post_parent' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   553
		$menu_order     = $this->get_tag( $post, 'wp:menu_order' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   554
		$post_type      = $this->get_tag( $post, 'wp:post_type' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   555
		$post_password  = $this->get_tag( $post, 'wp:post_password' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   556
		$is_sticky      = $this->get_tag( $post, 'wp:is_sticky' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   557
		$guid           = $this->get_tag( $post, 'guid' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   558
		$post_author    = $this->get_tag( $post, 'dc:creator' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   559
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   560
		$post_excerpt = $this->get_tag( $post, 'excerpt:encoded' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   561
		$post_excerpt = preg_replace_callback( '|<(/?[A-Z]+)|', array( &$this, '_normalize_tag' ), $post_excerpt );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   562
		$post_excerpt = str_replace( '<br>', '<br />', $post_excerpt );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   563
		$post_excerpt = str_replace( '<hr>', '<hr />', $post_excerpt );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   564
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   565
		$post_content = $this->get_tag( $post, 'content:encoded' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   566
		$post_content = preg_replace_callback( '|<(/?[A-Z]+)|', array( &$this, '_normalize_tag' ), $post_content );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   567
		$post_content = str_replace( '<br>', '<br />', $post_content );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   568
		$post_content = str_replace( '<hr>', '<hr />', $post_content );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   569
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   570
		$postdata = compact( 'post_id', 'post_author', 'post_date', 'post_date_gmt', 'post_content', 'post_excerpt',
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   571
			'post_title', 'status', 'post_name', 'comment_status', 'ping_status', 'guid', 'post_parent',
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   572
			'menu_order', 'post_type', 'post_password', 'is_sticky'
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   573
		);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   574
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   575
		$attachment_url = $this->get_tag( $post, 'wp:attachment_url' );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   576
		if ( $attachment_url )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   577
			$postdata['attachment_url'] = $attachment_url;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   578
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   579
		preg_match_all( '|<category domain="([^"]+?)" nicename="([^"]+?)">(.+?)</category>|is', $post, $terms, PREG_SET_ORDER );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   580
		foreach ( $terms as $t ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   581
			$post_terms[] = array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   582
				'slug' => $t[2],
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   583
				'domain' => $t[1],
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   584
				'name' => str_replace( array( '<![CDATA[', ']]>' ), '', $t[3] ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   585
			);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   586
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   587
		if ( ! empty( $post_terms ) ) $postdata['terms'] = $post_terms;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   588
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   589
		preg_match_all( '|<wp:comment>(.+?)</wp:comment>|is', $post, $comments );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   590
		$comments = $comments[1];
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   591
		if ( $comments ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   592
			foreach ( $comments as $comment ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   593
				preg_match_all( '|<wp:commentmeta>(.+?)</wp:commentmeta>|is', $comment, $commentmeta );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   594
				$commentmeta = $commentmeta[1];
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   595
				$c_meta = array();
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   596
				foreach ( $commentmeta as $m ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   597
					$c_meta[] = array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   598
						'key' => $this->get_tag( $m, 'wp:meta_key' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   599
						'value' => $this->get_tag( $m, 'wp:meta_value' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   600
					);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   601
				}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   602
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   603
				$post_comments[] = array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   604
					'comment_id' => $this->get_tag( $comment, 'wp:comment_id' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   605
					'comment_author' => $this->get_tag( $comment, 'wp:comment_author' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   606
					'comment_author_email' => $this->get_tag( $comment, 'wp:comment_author_email' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   607
					'comment_author_IP' => $this->get_tag( $comment, 'wp:comment_author_IP' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   608
					'comment_author_url' => $this->get_tag( $comment, 'wp:comment_author_url' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   609
					'comment_date' => $this->get_tag( $comment, 'wp:comment_date' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   610
					'comment_date_gmt' => $this->get_tag( $comment, 'wp:comment_date_gmt' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   611
					'comment_content' => $this->get_tag( $comment, 'wp:comment_content' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   612
					'comment_approved' => $this->get_tag( $comment, 'wp:comment_approved' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   613
					'comment_type' => $this->get_tag( $comment, 'wp:comment_type' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   614
					'comment_parent' => $this->get_tag( $comment, 'wp:comment_parent' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   615
					'comment_user_id' => $this->get_tag( $comment, 'wp:comment_user_id' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   616
					'commentmeta' => $c_meta,
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   617
				);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   618
			}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   619
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   620
		if ( ! empty( $post_comments ) ) $postdata['comments'] = $post_comments;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   621
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   622
		preg_match_all( '|<wp:postmeta>(.+?)</wp:postmeta>|is', $post, $postmeta );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   623
		$postmeta = $postmeta[1];
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   624
		if ( $postmeta ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   625
			foreach ( $postmeta as $p ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   626
				$post_postmeta[] = array(
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   627
					'key' => $this->get_tag( $p, 'wp:meta_key' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   628
					'value' => $this->get_tag( $p, 'wp:meta_value' ),
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   629
				);
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   630
			}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   631
		}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   632
		if ( ! empty( $post_postmeta ) ) $postdata['postmeta'] = $post_postmeta;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   633
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   634
		return $postdata;
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   635
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   636
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   637
	function _normalize_tag( $matches ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   638
		return '<' . strtolower( $matches[1] );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   639
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   640
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   641
	function fopen( $filename, $mode = 'r' ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   642
		if ( $this->has_gzip )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   643
			return gzopen( $filename, $mode );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   644
		return fopen( $filename, $mode );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   645
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   646
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   647
	function feof( $fp ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   648
		if ( $this->has_gzip )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   649
			return gzeof( $fp );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   650
		return feof( $fp );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   651
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   652
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   653
	function fgets( $fp, $len = 8192 ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   654
		if ( $this->has_gzip )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   655
			return gzgets( $fp, $len );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   656
		return fgets( $fp, $len );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   657
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   658
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   659
	function fclose( $fp ) {
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   660
		if ( $this->has_gzip )
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   661
			return gzclose( $fp );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   662
		return fclose( $fp );
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   663
	}
d970ebf37754 first import
ymh <ymh.work@gmail.com>
parents:
diff changeset
   664
}