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 uses regular expressions. Fallback for installs without an XML parser.

[9] Fix | Delete

[10] Fix | Delete

class WXR_Parser_Regex {

[11] Fix | Delete

public $authors = array();

[12] Fix | Delete

public $posts = array();

[13] Fix | Delete

public $categories = array();

[14] Fix | Delete

public $tags = array();

[15] Fix | Delete

public $terms = array();

[16] Fix | Delete

public $base_url = '';

[17] Fix | Delete

public $base_blog_url = '';

[18] Fix | Delete

public $has_gzip;

[19] Fix | Delete

[20] Fix | Delete

function __construct() {

[21] Fix | Delete

$this->has_gzip = is_callable( 'gzopen' );

[22] Fix | Delete

}

[23] Fix | Delete

[24] Fix | Delete

function parse( $file ) {

[25] Fix | Delete

$wxr_version = false;

[26] Fix | Delete

$in_multiline = false;

[27] Fix | Delete

[28] Fix | Delete

$multiline_content = '';

[29] Fix | Delete

[30] Fix | Delete

$multiline_tags = array(

[31] Fix | Delete

'item' => array( 'posts', array( $this, 'process_post' ) ),

[32] Fix | Delete

'wp:category' => array( 'categories', array( $this, 'process_category' ) ),

[33] Fix | Delete

'wp:tag' => array( 'tags', array( $this, 'process_tag' ) ),

[34] Fix | Delete

'wp:term' => array( 'terms', array( $this, 'process_term' ) ),

[35] Fix | Delete

);

[36] Fix | Delete

[37] Fix | Delete

$fp = $this->fopen( $file, 'r' );

[38] Fix | Delete

if ( $fp ) {

[39] Fix | Delete

while ( ! $this->feof( $fp ) ) {

[40] Fix | Delete

$is_tag_line = false;

[41] Fix | Delete

$importline = rtrim( $this->fgets( $fp ) );

[42] Fix | Delete

[43] Fix | Delete

if ( ! $wxr_version && preg_match( '|<wp:wxr_version>(\d+\.\d+)</wp:wxr_version>|', $importline, $version ) ) {

[44] Fix | Delete

$wxr_version = $version[1];

[45] Fix | Delete

}

[46] Fix | Delete

[47] Fix | Delete

if ( false !== strpos( $importline, '<wp:base_site_url>' ) ) {

[48] Fix | Delete

preg_match( '|<wp:base_site_url>(.*?)</wp:base_site_url>|is', $importline, $url );

[49] Fix | Delete

$this->base_url = $url[1];

[50] Fix | Delete

continue;

[51] Fix | Delete

}

[52] Fix | Delete

[53] Fix | Delete

if ( false !== strpos( $importline, '<wp:base_blog_url>' ) ) {

[54] Fix | Delete

preg_match( '|<wp:base_blog_url>(.*?)</wp:base_blog_url>|is', $importline, $blog_url );

[55] Fix | Delete

$this->base_blog_url = $blog_url[1];

[56] Fix | Delete

continue;

[57] Fix | Delete

} elseif ( empty( $this->base_blog_url ) ) {

[58] Fix | Delete

$this->base_blog_url = $this->base_url;

[59] Fix | Delete

}

[60] Fix | Delete

[61] Fix | Delete

if ( false !== strpos( $importline, '<wp:author>' ) ) {

[62] Fix | Delete

preg_match( '|<wp:author>(.*?)</wp:author>|is', $importline, $author );

[63] Fix | Delete

$a = $this->process_author( $author[1] );

[64] Fix | Delete

$this->authors[ $a['author_login'] ] = $a;

[65] Fix | Delete

continue;

[66] Fix | Delete

}

[67] Fix | Delete

[68] Fix | Delete

foreach ( $multiline_tags as $tag => $handler ) {

[69] Fix | Delete

// Handle multi-line tags on a singular line

[70] Fix | Delete

$pos = strpos( $importline, "<$tag>" );

[71] Fix | Delete

$pos_closing = strpos( $importline, "</$tag>" );

[72] Fix | Delete

if ( preg_match( '|<' . $tag . '>(.*?)</' . $tag . '>|is', $importline, $matches ) ) {

[73] Fix | Delete

$this->{$handler[0]}[] = call_user_func( $handler[1], $matches[1] );

[74] Fix | Delete

[75] Fix | Delete

} elseif ( false !== $pos ) {

[76] Fix | Delete

// Take note of any content after the opening tag

[77] Fix | Delete

$multiline_content = trim( substr( $importline, $pos + strlen( $tag ) + 2 ) );

[78] Fix | Delete

[79] Fix | Delete

// We don't want to have this line added to `$is_multiline` below.

[80] Fix | Delete

$in_multiline = $tag;

[81] Fix | Delete

$is_tag_line = true;

[82] Fix | Delete

[83] Fix | Delete

} elseif ( false !== $pos_closing ) {

[84] Fix | Delete

$in_multiline = false;

[85] Fix | Delete

$multiline_content .= trim( substr( $importline, 0, $pos_closing ) );

[86] Fix | Delete

[87] Fix | Delete

$this->{$handler[0]}[] = call_user_func( $handler[1], $multiline_content );

[88] Fix | Delete

}

[89] Fix | Delete

}

[90] Fix | Delete

[91] Fix | Delete

if ( $in_multiline && ! $is_tag_line ) {

[92] Fix | Delete

$multiline_content .= $importline . "\n";

[93] Fix | Delete

}

[94] Fix | Delete

}

[95] Fix | Delete

[96] Fix | Delete

$this->fclose( $fp );

[97] Fix | Delete

}

[98] Fix | Delete

[99] Fix | Delete

if ( ! $wxr_version ) {

[100] 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' ) );

[101] Fix | Delete

}

[102] Fix | Delete

[103] Fix | Delete

return array(

[104] Fix | Delete

'authors' => $this->authors,

[105] Fix | Delete

'posts' => $this->posts,

[106] Fix | Delete

'categories' => $this->categories,

[107] Fix | Delete

'tags' => $this->tags,

[108] Fix | Delete

'terms' => $this->terms,

[109] Fix | Delete

'base_url' => $this->base_url,

[110] Fix | Delete

'base_blog_url' => $this->base_blog_url,

[111] Fix | Delete

'version' => $wxr_version,

[112] Fix | Delete

);

[113] Fix | Delete

}

[114] Fix | Delete

[115] Fix | Delete

function get_tag( $string, $tag ) {

[116] Fix | Delete

preg_match( "|<$tag.*?>(.*?)</$tag>|is", $string, $return );

[117] Fix | Delete

if ( isset( $return[1] ) ) {

[118] Fix | Delete

if ( substr( $return[1], 0, 9 ) == '<![CDATA[' ) {

[119] Fix | Delete

if ( strpos( $return[1], ']]]]><![CDATA[>' ) !== false ) {

[120] Fix | Delete

preg_match_all( '|<!\[CDATA\[(.*?)\]\]>|s', $return[1], $matches );

[121] Fix | Delete

$return = '';

[122] Fix | Delete

foreach ( $matches[1] as $match ) {

[123] Fix | Delete

$return .= $match;

[124] Fix | Delete

}

[125] Fix | Delete

} else {

[126] Fix | Delete

$return = preg_replace( '|^<!\[CDATA\[(.*)\]\]>$|s', '$1', $return[1] );

[127] Fix | Delete

}

[128] Fix | Delete

} else {

[129] Fix | Delete

$return = $return[1];

[130] Fix | Delete

}

[131] Fix | Delete

} else {

[132] Fix | Delete

$return = '';

[133] Fix | Delete

}

[134] Fix | Delete

return $return;

[135] Fix | Delete

}

[136] Fix | Delete

[137] Fix | Delete

function process_category( $c ) {

[138] Fix | Delete

$term = array(

[139] Fix | Delete

'term_id' => $this->get_tag( $c, 'wp:term_id' ),

[140] Fix | Delete

'cat_name' => $this->get_tag( $c, 'wp:cat_name' ),

[141] Fix | Delete

'category_nicename' => $this->get_tag( $c, 'wp:category_nicename' ),

[142] Fix | Delete

'category_parent' => $this->get_tag( $c, 'wp:category_parent' ),

[143] Fix | Delete

'category_description' => $this->get_tag( $c, 'wp:category_description' ),

[144] Fix | Delete

);

[145] Fix | Delete

[146] Fix | Delete

$term_meta = $this->process_meta( $c, 'wp:termmeta' );

[147] Fix | Delete

if ( ! empty( $term_meta ) ) {

[148] Fix | Delete

$term['termmeta'] = $term_meta;

[149] Fix | Delete

}

[150] Fix | Delete

[151] Fix | Delete

return $term;

[152] Fix | Delete

}

[153] Fix | Delete

[154] Fix | Delete

function process_tag( $t ) {

[155] Fix | Delete

$term = array(

[156] Fix | Delete

'term_id' => $this->get_tag( $t, 'wp:term_id' ),

[157] Fix | Delete

'tag_name' => $this->get_tag( $t, 'wp:tag_name' ),

[158] Fix | Delete

'tag_slug' => $this->get_tag( $t, 'wp:tag_slug' ),

[159] Fix | Delete

'tag_description' => $this->get_tag( $t, 'wp:tag_description' ),

[160] Fix | Delete

);

[161] Fix | Delete

[162] Fix | Delete

$term_meta = $this->process_meta( $t, 'wp:termmeta' );

[163] Fix | Delete

if ( ! empty( $term_meta ) ) {

[164] Fix | Delete

$term['termmeta'] = $term_meta;

[165] Fix | Delete

}

[166] Fix | Delete

[167] Fix | Delete

return $term;

[168] Fix | Delete

}

[169] Fix | Delete

[170] Fix | Delete

function process_term( $t ) {

[171] Fix | Delete

$term = array(

[172] Fix | Delete

'term_id' => $this->get_tag( $t, 'wp:term_id' ),

[173] Fix | Delete

'term_taxonomy' => $this->get_tag( $t, 'wp:term_taxonomy' ),

[174] Fix | Delete

'slug' => $this->get_tag( $t, 'wp:term_slug' ),

[175] Fix | Delete

'term_parent' => $this->get_tag( $t, 'wp:term_parent' ),

[176] Fix | Delete

'term_name' => $this->get_tag( $t, 'wp:term_name' ),

[177] Fix | Delete

'term_description' => $this->get_tag( $t, 'wp:term_description' ),

[178] Fix | Delete

);

[179] Fix | Delete

[180] Fix | Delete

$term_meta = $this->process_meta( $t, 'wp:termmeta' );

[181] Fix | Delete

if ( ! empty( $term_meta ) ) {

[182] Fix | Delete

$term['termmeta'] = $term_meta;

[183] Fix | Delete

}

[184] Fix | Delete

[185] Fix | Delete

return $term;

[186] Fix | Delete

}

[187] Fix | Delete

[188] Fix | Delete

function process_meta( $string, $tag ) {

[189] Fix | Delete

$parsed_meta = array();

[190] Fix | Delete

[191] Fix | Delete

preg_match_all( "|<$tag>(.+?)</$tag>|is", $string, $meta );

[192] Fix | Delete

[193] Fix | Delete

if ( ! isset( $meta[1] ) ) {

[194] Fix | Delete

return $parsed_meta;

[195] Fix | Delete

}

[196] Fix | Delete

[197] Fix | Delete

foreach ( $meta[1] as $m ) {

[198] Fix | Delete

$parsed_meta[] = array(

[199] Fix | Delete

'key' => $this->get_tag( $m, 'wp:meta_key' ),

[200] Fix | Delete

'value' => $this->get_tag( $m, 'wp:meta_value' ),

[201] Fix | Delete

);

[202] Fix | Delete

}

[203] Fix | Delete

[204] Fix | Delete

return $parsed_meta;

[205] Fix | Delete

}

[206] Fix | Delete

[207] Fix | Delete

function process_author( $a ) {

[208] Fix | Delete

return array(

[209] Fix | Delete

'author_id' => $this->get_tag( $a, 'wp:author_id' ),

[210] Fix | Delete

'author_login' => $this->get_tag( $a, 'wp:author_login' ),

[211] Fix | Delete

'author_email' => $this->get_tag( $a, 'wp:author_email' ),

[212] Fix | Delete

'author_display_name' => $this->get_tag( $a, 'wp:author_display_name' ),

[213] Fix | Delete

'author_first_name' => $this->get_tag( $a, 'wp:author_first_name' ),

[214] Fix | Delete

'author_last_name' => $this->get_tag( $a, 'wp:author_last_name' ),

[215] Fix | Delete

);

[216] Fix | Delete

}

[217] Fix | Delete

[218] Fix | Delete

function process_post( $post ) {

[219] Fix | Delete

$post_id = $this->get_tag( $post, 'wp:post_id' );

[220] Fix | Delete

$post_title = $this->get_tag( $post, 'title' );

[221] Fix | Delete

$post_date = $this->get_tag( $post, 'wp:post_date' );

[222] Fix | Delete

$post_date_gmt = $this->get_tag( $post, 'wp:post_date_gmt' );

[223] Fix | Delete

$comment_status = $this->get_tag( $post, 'wp:comment_status' );

[224] Fix | Delete

$ping_status = $this->get_tag( $post, 'wp:ping_status' );

[225] Fix | Delete

$status = $this->get_tag( $post, 'wp:status' );

[226] Fix | Delete

$post_name = $this->get_tag( $post, 'wp:post_name' );

[227] Fix | Delete

$post_parent = $this->get_tag( $post, 'wp:post_parent' );

[228] Fix | Delete

$menu_order = $this->get_tag( $post, 'wp:menu_order' );

[229] Fix | Delete

$post_type = $this->get_tag( $post, 'wp:post_type' );

[230] Fix | Delete

$post_password = $this->get_tag( $post, 'wp:post_password' );

[231] Fix | Delete

$is_sticky = $this->get_tag( $post, 'wp:is_sticky' );

[232] Fix | Delete

$guid = $this->get_tag( $post, 'guid' );

[233] Fix | Delete

$post_author = $this->get_tag( $post, 'dc:creator' );

[234] Fix | Delete

[235] Fix | Delete

$post_excerpt = $this->get_tag( $post, 'excerpt:encoded' );

[236] Fix | Delete

$post_excerpt = preg_replace_callback( '|<(/?[A-Z]+)|', array( &$this, '_normalize_tag' ), $post_excerpt );

[237] Fix | Delete

$post_excerpt = str_replace( '<br>', '<br />', $post_excerpt );

[238] Fix | Delete

$post_excerpt = str_replace( '<hr>', '<hr />', $post_excerpt );

[239] Fix | Delete

[240] Fix | Delete

$post_content = $this->get_tag( $post, 'content:encoded' );

[241] Fix | Delete

$post_content = preg_replace_callback( '|<(/?[A-Z]+)|', array( &$this, '_normalize_tag' ), $post_content );

[242] Fix | Delete

$post_content = str_replace( '<br>', '<br />', $post_content );

[243] Fix | Delete

$post_content = str_replace( '<hr>', '<hr />', $post_content );

[244] Fix | Delete

[245] Fix | Delete

$postdata = compact(

[246] Fix | Delete

'post_id',

[247] Fix | Delete

'post_author',

[248] Fix | Delete

'post_date',

[249] Fix | Delete

'post_date_gmt',

[250] Fix | Delete

'post_content',

[251] Fix | Delete

'post_excerpt',

[252] Fix | Delete

'post_title',

[253] Fix | Delete

'status',

[254] Fix | Delete

'post_name',

[255] Fix | Delete

'comment_status',

[256] Fix | Delete

'ping_status',

[257] Fix | Delete

'guid',

[258] Fix | Delete

'post_parent',

[259] Fix | Delete

'menu_order',

[260] Fix | Delete

'post_type',

[261] Fix | Delete

'post_password',

[262] Fix | Delete

'is_sticky'

[263] Fix | Delete

);

[264] Fix | Delete

[265] Fix | Delete

$attachment_url = $this->get_tag( $post, 'wp:attachment_url' );

[266] Fix | Delete

if ( $attachment_url ) {

[267] Fix | Delete

$postdata['attachment_url'] = $attachment_url;

[268] Fix | Delete

}

[269] Fix | Delete

[270] Fix | Delete

preg_match_all( '|<category domain="([^"]+?)" nicename="([^"]+?)">(.+?)</category>|is', $post, $terms, PREG_SET_ORDER );

[271] Fix | Delete

foreach ( $terms as $t ) {

[272] Fix | Delete

$post_terms[] = array(

[273] Fix | Delete

'slug' => $t[2],

[274] Fix | Delete

'domain' => $t[1],

[275] Fix | Delete

'name' => str_replace( array( '<![CDATA[', ']]>' ), '', $t[3] ),

[276] Fix | Delete

);

[277] Fix | Delete

}

[278] Fix | Delete

if ( ! empty( $post_terms ) ) {

[279] Fix | Delete

$postdata['terms'] = $post_terms;

[280] Fix | Delete

}

[281] Fix | Delete

[282] Fix | Delete

preg_match_all( '|<wp:comment>(.+?)</wp:comment>|is', $post, $comments );

[283] Fix | Delete

$comments = $comments[1];

[284] Fix | Delete

if ( $comments ) {

[285] Fix | Delete

foreach ( $comments as $comment ) {

[286] Fix | Delete

$post_comments[] = array(

[287] Fix | Delete

'comment_id' => $this->get_tag( $comment, 'wp:comment_id' ),

[288] Fix | Delete

'comment_author' => $this->get_tag( $comment, 'wp:comment_author' ),

[289] Fix | Delete

'comment_author_email' => $this->get_tag( $comment, 'wp:comment_author_email' ),

[290] Fix | Delete

'comment_author_IP' => $this->get_tag( $comment, 'wp:comment_author_IP' ),

[291] Fix | Delete

'comment_author_url' => $this->get_tag( $comment, 'wp:comment_author_url' ),

[292] Fix | Delete

'comment_date' => $this->get_tag( $comment, 'wp:comment_date' ),

[293] Fix | Delete

'comment_date_gmt' => $this->get_tag( $comment, 'wp:comment_date_gmt' ),

[294] Fix | Delete

'comment_content' => $this->get_tag( $comment, 'wp:comment_content' ),

[295] Fix | Delete

'comment_approved' => $this->get_tag( $comment, 'wp:comment_approved' ),

[296] Fix | Delete

'comment_type' => $this->get_tag( $comment, 'wp:comment_type' ),

[297] Fix | Delete

'comment_parent' => $this->get_tag( $comment, 'wp:comment_parent' ),

[298] Fix | Delete

'comment_user_id' => $this->get_tag( $comment, 'wp:comment_user_id' ),

[299] Fix | Delete

'commentmeta' => $this->process_meta( $comment, 'wp:commentmeta' ),

[300] Fix | Delete

);

[301] Fix | Delete

}

[302] Fix | Delete

}

[303] Fix | Delete

if ( ! empty( $post_comments ) ) {

[304] Fix | Delete

$postdata['comments'] = $post_comments;

[305] Fix | Delete

}

[306] Fix | Delete

[307] Fix | Delete

$post_meta = $this->process_meta( $post, 'wp:postmeta' );

[308] Fix | Delete

if ( ! empty( $post_meta ) ) {

[309] Fix | Delete

$postdata['postmeta'] = $post_meta;

[310] Fix | Delete

}

[311] Fix | Delete

[312] Fix | Delete

return $postdata;

[313] Fix | Delete

}

[314] Fix | Delete

[315] Fix | Delete

function _normalize_tag( $matches ) {

[316] Fix | Delete

return '<' . strtolower( $matches[1] );

[317] Fix | Delete

}

[318] Fix | Delete

[319] Fix | Delete

function fopen( $filename, $mode = 'r' ) {

[320] Fix | Delete

if ( $this->has_gzip ) {

[321] Fix | Delete

return gzopen( $filename, $mode );

[322] Fix | Delete

}

[323] Fix | Delete

return fopen( $filename, $mode );

[324] Fix | Delete

}

[325] Fix | Delete

[326] Fix | Delete

function feof( $fp ) {

[327] Fix | Delete

if ( $this->has_gzip ) {

[328] Fix | Delete

return gzeof( $fp );

[329] Fix | Delete

}

[330] Fix | Delete

return feof( $fp );

[331] Fix | Delete

}

[332] Fix | Delete

[333] Fix | Delete

function fgets( $fp, $len = 8192 ) {

[334] Fix | Delete

if ( $this->has_gzip ) {

[335] Fix | Delete

return gzgets( $fp, $len );

[336] Fix | Delete

}

[337] Fix | Delete

return fgets( $fp, $len );

[338] Fix | Delete

}

[339] Fix | Delete

[340] Fix | Delete

function fclose( $fp ) {

[341] Fix | Delete

if ( $this->has_gzip ) {

[342] Fix | Delete

return gzclose( $fp );

[343] Fix | Delete

}

[344] Fix | Delete

return fclose( $fp );

[345] Fix | Delete

}

[346] Fix | Delete

}

[347] Fix | Delete

[348] Fix | Delete