Edit File by line

<?php

[0] Fix | Delete

/**

[1] Fix | Delete

* WordPress eXtended RSS file parser implementations

[2] Fix | Delete

[3] Fix | Delete

* @package WordPress

[4] Fix | Delete

* @subpackage Importer

[5] Fix | Delete

[6] Fix | Delete

[7] Fix | Delete

/**

[8] Fix | Delete

* WXR Parser that makes use of the SimpleXML PHP extension.

[9] Fix | Delete

[10] Fix | Delete

class WXR_Parser_SimpleXML {

[11] Fix | Delete

function parse( $file ) {

[12] Fix | Delete

$authors = array();

[13] Fix | Delete

$posts = array();

[14] Fix | Delete

$categories = array();

[15] Fix | Delete

$tags = array();

[16] Fix | Delete

$terms = array();

[17] Fix | Delete

[18] Fix | Delete

$internal_errors = libxml_use_internal_errors( true );

[19] Fix | Delete

[20] Fix | Delete

$dom = new DOMDocument;

[21] Fix | Delete

$old_value = null;

[22] Fix | Delete

if ( function_exists( 'libxml_disable_entity_loader' ) && PHP_VERSION_ID < 80000 ) {

[23] Fix | Delete

$old_value = libxml_disable_entity_loader( true );

[24] Fix | Delete

}

[25] Fix | Delete

$success = $dom->loadXML( file_get_contents( $file ) );

[26] Fix | Delete

if ( ! is_null( $old_value ) ) {

[27] Fix | Delete

libxml_disable_entity_loader( $old_value );

[28] Fix | Delete

}

[29] Fix | Delete

[30] Fix | Delete

if ( ! $success || isset( $dom->doctype ) ) {

[31] Fix | Delete

return new WP_Error( 'SimpleXML_parse_error', __( 'There was an error when reading this WXR file', 'wordpress-importer' ), libxml_get_errors() );

[32] Fix | Delete

}

[33] Fix | Delete

[34] Fix | Delete

$xml = simplexml_import_dom( $dom );

[35] Fix | Delete

unset( $dom );

[36] Fix | Delete

[37] Fix | Delete

// halt if loading produces an error

[38] Fix | Delete

if ( ! $xml ) {

[39] Fix | Delete

return new WP_Error( 'SimpleXML_parse_error', __( 'There was an error when reading this WXR file', 'wordpress-importer' ), libxml_get_errors() );

[40] Fix | Delete

}

[41] Fix | Delete

[42] Fix | Delete

$wxr_version = $xml->xpath( '/rss/channel/wp:wxr_version' );

[43] Fix | Delete

if ( ! $wxr_version ) {

[44] Fix | Delete

return new WP_Error( 'WXR_parse_error', __( 'This does not appear to be a WXR file, missing/invalid WXR version number', 'wordpress-importer' ) );

[45] Fix | Delete

}

[46] Fix | Delete

[47] Fix | Delete

$wxr_version = (string) trim( $wxr_version[0] );

[48] Fix | Delete

// confirm that we are dealing with the correct file format

[49] Fix | Delete

if ( ! preg_match( '/^\d+\.\d+$/', $wxr_version ) ) {

[50] Fix | Delete

return new WP_Error( 'WXR_parse_error', __( 'This does not appear to be a WXR file, missing/invalid WXR version number', 'wordpress-importer' ) );

[51] Fix | Delete

}

[52] Fix | Delete

[53] Fix | Delete

$base_url = $xml->xpath( '/rss/channel/wp:base_site_url' );

[54] Fix | Delete

$base_url = (string) trim( isset( $base_url[0] ) ? $base_url[0] : '' );

[55] Fix | Delete

[56] Fix | Delete

$base_blog_url = $xml->xpath( '/rss/channel/wp:base_blog_url' );

[57] Fix | Delete

if ( $base_blog_url ) {

[58] Fix | Delete

$base_blog_url = (string) trim( $base_blog_url[0] );

[59] Fix | Delete

} else {

[60] Fix | Delete

$base_blog_url = $base_url;

[61] Fix | Delete

}

[62] Fix | Delete

[63] Fix | Delete

$namespaces = $xml->getDocNamespaces();

[64] Fix | Delete

if ( ! isset( $namespaces['wp'] ) ) {

[65] Fix | Delete

$namespaces['wp'] = 'http://wordpress.org/export/1.1/';

[66] Fix | Delete

}

[67] Fix | Delete

if ( ! isset( $namespaces['excerpt'] ) ) {

[68] Fix | Delete

$namespaces['excerpt'] = 'http://wordpress.org/export/1.1/excerpt/';

[69] Fix | Delete

}

[70] Fix | Delete

[71] Fix | Delete

// grab authors

[72] Fix | Delete

foreach ( $xml->xpath( '/rss/channel/wp:author' ) as $author_arr ) {

[73] Fix | Delete

$a = $author_arr->children( $namespaces['wp'] );

[74] Fix | Delete

$login = (string) $a->author_login;

[75] Fix | Delete

$authors[ $login ] = array(

[76] Fix | Delete

'author_id' => (int) $a->author_id,

[77] Fix | Delete

'author_login' => $login,

[78] Fix | Delete

'author_email' => (string) $a->author_email,

[79] Fix | Delete

'author_display_name' => (string) $a->author_display_name,

[80] Fix | Delete

'author_first_name' => (string) $a->author_first_name,

[81] Fix | Delete

'author_last_name' => (string) $a->author_last_name,

[82] Fix | Delete

);

[83] Fix | Delete

}

[84] Fix | Delete

[85] Fix | Delete

// grab cats, tags and terms

[86] Fix | Delete

foreach ( $xml->xpath( '/rss/channel/wp:category' ) as $term_arr ) {

[87] Fix | Delete

$t = $term_arr->children( $namespaces['wp'] );

[88] Fix | Delete

$category = array(

[89] Fix | Delete

'term_id' => (int) $t->term_id,

[90] Fix | Delete

'category_nicename' => (string) $t->category_nicename,

[91] Fix | Delete

'category_parent' => (string) $t->category_parent,

[92] Fix | Delete

'cat_name' => (string) $t->cat_name,

[93] Fix | Delete

'category_description' => (string) $t->category_description,

[94] Fix | Delete

);

[95] Fix | Delete

[96] Fix | Delete

foreach ( $t->termmeta as $meta ) {

[97] Fix | Delete

$category['termmeta'][] = array(

[98] Fix | Delete

'key' => (string) $meta->meta_key,

[99] Fix | Delete

'value' => (string) $meta->meta_value,

[100] Fix | Delete

);

[101] Fix | Delete

}

[102] Fix | Delete

[103] Fix | Delete

$categories[] = $category;

[104] Fix | Delete

}

[105] Fix | Delete

[106] Fix | Delete

foreach ( $xml->xpath( '/rss/channel/wp:tag' ) as $term_arr ) {

[107] Fix | Delete

$t = $term_arr->children( $namespaces['wp'] );

[108] Fix | Delete

$tag = array(

[109] Fix | Delete

'term_id' => (int) $t->term_id,

[110] Fix | Delete

'tag_slug' => (string) $t->tag_slug,

[111] Fix | Delete

'tag_name' => (string) $t->tag_name,

[112] Fix | Delete

'tag_description' => (string) $t->tag_description,

[113] Fix | Delete

);

[114] Fix | Delete

[115] Fix | Delete

foreach ( $t->termmeta as $meta ) {

[116] Fix | Delete

$tag['termmeta'][] = array(

[117] Fix | Delete

'key' => (string) $meta->meta_key,

[118] Fix | Delete

'value' => (string) $meta->meta_value,

[119] Fix | Delete

);

[120] Fix | Delete

}

[121] Fix | Delete

[122] Fix | Delete

$tags[] = $tag;

[123] Fix | Delete

}

[124] Fix | Delete

[125] Fix | Delete

foreach ( $xml->xpath( '/rss/channel/wp:term' ) as $term_arr ) {

[126] Fix | Delete

$t = $term_arr->children( $namespaces['wp'] );

[127] Fix | Delete

$term = array(

[128] Fix | Delete

'term_id' => (int) $t->term_id,

[129] Fix | Delete

'term_taxonomy' => (string) $t->term_taxonomy,

[130] Fix | Delete

'slug' => (string) $t->term_slug,

[131] Fix | Delete

'term_parent' => (string) $t->term_parent,

[132] Fix | Delete

'term_name' => (string) $t->term_name,

[133] Fix | Delete

'term_description' => (string) $t->term_description,

[134] Fix | Delete

);

[135] Fix | Delete

[136] Fix | Delete

foreach ( $t->termmeta as $meta ) {

[137] Fix | Delete

$term['termmeta'][] = array(

[138] Fix | Delete

'key' => (string) $meta->meta_key,

[139] Fix | Delete

'value' => (string) $meta->meta_value,

[140] Fix | Delete

);

[141] Fix | Delete

}

[142] Fix | Delete

[143] Fix | Delete

$terms[] = $term;

[144] Fix | Delete

}

[145] Fix | Delete

[146] Fix | Delete

// grab posts

[147] Fix | Delete

foreach ( $xml->channel->item as $item ) {

[148] Fix | Delete

$post = array(

[149] Fix | Delete

'post_title' => (string) $item->title,

[150] Fix | Delete

'guid' => (string) $item->guid,

[151] Fix | Delete

);

[152] Fix | Delete

[153] Fix | Delete

$dc = $item->children( 'http://purl.org/dc/elements/1.1/' );

[154] Fix | Delete

$post['post_author'] = (string) $dc->creator;

[155] Fix | Delete

[156] Fix | Delete

$content = $item->children( 'http://purl.org/rss/1.0/modules/content/' );

[157] Fix | Delete

$excerpt = $item->children( $namespaces['excerpt'] );

[158] Fix | Delete

$post['post_content'] = (string) $content->encoded;

[159] Fix | Delete

$post['post_excerpt'] = (string) $excerpt->encoded;

[160] Fix | Delete

[161] Fix | Delete

$wp = $item->children( $namespaces['wp'] );

[162] Fix | Delete

$post['post_id'] = (int) $wp->post_id;

[163] Fix | Delete

$post['post_date'] = (string) $wp->post_date;

[164] Fix | Delete

$post['post_date_gmt'] = (string) $wp->post_date_gmt;

[165] Fix | Delete

$post['comment_status'] = (string) $wp->comment_status;

[166] Fix | Delete

$post['ping_status'] = (string) $wp->ping_status;

[167] Fix | Delete

$post['post_name'] = (string) $wp->post_name;

[168] Fix | Delete

$post['status'] = (string) $wp->status;

[169] Fix | Delete

$post['post_parent'] = (int) $wp->post_parent;

[170] Fix | Delete

$post['menu_order'] = (int) $wp->menu_order;

[171] Fix | Delete

$post['post_type'] = (string) $wp->post_type;

[172] Fix | Delete

$post['post_password'] = (string) $wp->post_password;

[173] Fix | Delete

$post['is_sticky'] = (int) $wp->is_sticky;

[174] Fix | Delete

[175] Fix | Delete

if ( isset( $wp->attachment_url ) ) {

[176] Fix | Delete

$post['attachment_url'] = (string) $wp->attachment_url;

[177] Fix | Delete

}

[178] Fix | Delete

[179] Fix | Delete

foreach ( $item->category as $c ) {

[180] Fix | Delete

$att = $c->attributes();

[181] Fix | Delete

if ( isset( $att['nicename'] ) ) {

[182] Fix | Delete

$post['terms'][] = array(

[183] Fix | Delete

'name' => (string) $c,

[184] Fix | Delete

'slug' => (string) $att['nicename'],

[185] Fix | Delete

'domain' => (string) $att['domain'],

[186] Fix | Delete

);

[187] Fix | Delete

}

[188] Fix | Delete

}

[189] Fix | Delete

[190] Fix | Delete

foreach ( $wp->postmeta as $meta ) {

[191] Fix | Delete

$post['postmeta'][] = array(

[192] Fix | Delete

'key' => (string) $meta->meta_key,

[193] Fix | Delete

'value' => (string) $meta->meta_value,

[194] Fix | Delete

);

[195] Fix | Delete

}

[196] Fix | Delete

[197] Fix | Delete

foreach ( $wp->comment as $comment ) {

[198] Fix | Delete

$meta = array();

[199] Fix | Delete

if ( isset( $comment->commentmeta ) ) {

[200] Fix | Delete

foreach ( $comment->commentmeta as $m ) {

[201] Fix | Delete

$meta[] = array(

[202] Fix | Delete

'key' => (string) $m->meta_key,

[203] Fix | Delete

'value' => (string) $m->meta_value,

[204] Fix | Delete

);

[205] Fix | Delete

}

[206] Fix | Delete

}

[207] Fix | Delete

[208] Fix | Delete

$post['comments'][] = array(

[209] Fix | Delete

'comment_id' => (int) $comment->comment_id,

[210] Fix | Delete

'comment_author' => (string) $comment->comment_author,

[211] Fix | Delete

'comment_author_email' => (string) $comment->comment_author_email,

[212] Fix | Delete

'comment_author_IP' => (string) $comment->comment_author_IP,

[213] Fix | Delete

'comment_author_url' => (string) $comment->comment_author_url,

[214] Fix | Delete

'comment_date' => (string) $comment->comment_date,

[215] Fix | Delete

'comment_date_gmt' => (string) $comment->comment_date_gmt,

[216] Fix | Delete

'comment_content' => (string) $comment->comment_content,

[217] Fix | Delete

'comment_approved' => (string) $comment->comment_approved,

[218] Fix | Delete

'comment_type' => (string) $comment->comment_type,

[219] Fix | Delete

'comment_parent' => (string) $comment->comment_parent,

[220] Fix | Delete

'comment_user_id' => (int) $comment->comment_user_id,

[221] Fix | Delete

'commentmeta' => $meta,

[222] Fix | Delete

);

[223] Fix | Delete

}

[224] Fix | Delete

[225] Fix | Delete

$posts[] = $post;

[226] Fix | Delete

}

[227] Fix | Delete

[228] Fix | Delete

return array(

[229] Fix | Delete

'authors' => $authors,

[230] Fix | Delete

'posts' => $posts,

[231] Fix | Delete

'categories' => $categories,

[232] Fix | Delete

'tags' => $tags,

[233] Fix | Delete

'terms' => $terms,

[234] Fix | Delete

'base_url' => $base_url,

[235] Fix | Delete

'base_blog_url' => $base_blog_url,

[236] Fix | Delete

'version' => $wxr_version,

[237] Fix | Delete

);

[238] Fix | Delete

}

[239] Fix | Delete

}

[240] Fix | Delete

[241] Fix | Delete