wp-admin/import/rss.php

   1 <?php
   2
   3 class RSS_Import {
   4
   5         var $posts = array ();
   6         var $file;
   7
   8         function header() {
   9                 echo '<div class="wrap">';
  10                 echo '<h2>'.__('Import RSS').'</h2>';
  11         }
  12
  13         function footer() {
  14                 echo '</div>';
  15         }
  16
  17         function unhtmlentities($string) { // From php.net for < 4.3 compat
  18                 $trans_tbl = get_html_translation_table(HTML_ENTITIES);
  19                 $trans_tbl = array_flip($trans_tbl);
  20                 return strtr($string, $trans_tbl);
  21         }
  22
  23         function greet() {
  24                 echo '<p>'.__('Howdy! This importer allows you to extract posts from any RSS 2.0 file into your blog. This is useful if you want to import your posts from a system that is not handled by a custom import tool. Pick an RSS file to upload and click Import.').'</p>';
  25                 wp_import_upload_form("admin.php?import=rss&amp;step=1");
  26         }
  27
  28         function get_posts() {
  29                 global $wpdb;
  30
  31                 set_magic_quotes_runtime(0);
  32                 $datalines = file($this->file); // Read the file into an array
  33                 $importdata = implode('', $datalines); // squish it
  34                 $importdata = str_replace(array ("\r\n", "\r"), "\n", $importdata);
  35
  36                 preg_match_all('|<item>(.*?)</item>|is', $importdata, $this->posts);
  37                 $this->posts = $this->posts[1];
  38                 $index = 0;
  39                 foreach ($this->posts as $post) {
  40                         preg_match('|<title>(.*?)</title>|is', $post, $post_title);
  41                         $post_title = str_replace(array('<![CDATA[', ']]>'), '', $wpdb->escape( trim($post_title[1]) ));
  42
  43                         preg_match('|<pubdate>(.*?)</pubdate>|is', $post, $post_date_gmt);
  44
  45                         if ($post_date_gmt) {
  46                                 $post_date_gmt = strtotime($post_date_gmt[1]);
  47                         } else {
  48                                 // if we don't already have something from pubDate
  49                                 preg_match('|<dc:date>(.*?)</dc:date>|is', $post, $post_date_gmt);
  50                                 $post_date_gmt = preg_replace('|([-+])([0-9]+):([0-9]+)$|', '\1\2\3', $post_date_gmt[1]);
  51                                 $post_date_gmt = str_replace('T', ' ', $post_date_gmt);
  52                                 $post_date_gmt = strtotime($post_date_gmt);
  53                         }
  54
  55                         $post_date_gmt = gmdate('Y-m-d H:i:s', $post_date_gmt);
  56                         $post_date = get_date_from_gmt( $post_date_gmt );
  57
  58                         preg_match_all('|<category>(.*?)</category>|is', $post, $categories);
  59                         $categories = $categories[1];
  60
  61                         if (!$categories) {
  62                                 preg_match_all('|<dc:subject>(.*?)</dc:subject>|is', $post, $categories);
  63                                 $categories = $categories[1];
  64                         }
  65
  66                         $cat_index = 0;
  67                         foreach ($categories as $category) {
  68                                 $categories[$cat_index] = $wpdb->escape($this->unhtmlentities($category));
  69                                 $cat_index++;
  70                         }
  71
  72                         preg_match('|<guid.+?>(.*?)</guid>|is', $post, $guid);
  73                         if ($guid)
  74                                 $guid = $wpdb->escape(trim($guid[1]));
  75                         else
  76                                 $guid = '';
  77
  78                         preg_match('|<content:encoded>(.*?)</content:encoded>|is', $post, $post_content);
  79                         $post_content = str_replace(array ('<![CDATA[', ']]>'), '', $wpdb->escape(trim($post_content[1])));
  80
  81                         if (!$post_content) {
  82                                 // This is for feeds that put content in description
  83                                 preg_match('|<description>(.*?)</description>|is', $post, $post_content);
  84                                 $post_content = $wpdb->escape($this->unhtmlentities(trim($post_content[1])));
  85                         }
  86
  87                         // Clean up content
  88                         $post_content = preg_replace('|<(/?[A-Z]+)|e', "'<' . strtolower('$1')", $post_content);
  89                         $post_content = str_replace('<br>', '<br />', $post_content);
  90                         $post_content = str_replace('<hr>', '<hr />', $post_content);
  91
  92                         $post_author = 1;
  93                         $post_status = 'publish';
  94                         $this->posts[$index] = compact('post_author', 'post_date', 'post_date_gmt', 'post_content', 'post_title', 'post_status', 'guid', 'categories');
  95                         $index++;
  96                 }
  97         }
  98
  99         function import_posts() {
 100                 echo '<ol>';
 101
 102                 foreach ($this->posts as $post) {
 103                         echo "<li>".__('Importing post...');
 104
 105                         extract($post);
 106
 107                         if ($post_id = post_exists($post_title, $post_content, $post_date)) {
 108                                 _e('Post already imported');
 109                         } else {
 110                                 $post_id = wp_insert_post($post);
 111                                 if (!$post_id) {
 112                                         _e("Couldn't get post ID");
 113                                         return;
 114                                 }
 115
 116                                 if (0 != count($categories))
 117                                         wp_create_categories($categories, $post_id);
 118                                 _e('Done !');
 119                         }
 120                         echo '</li>';
 121                 }
 122
 123                 echo '</ol>';
 124
 125         }
 126
 127         function import() {
 128                 $file = wp_import_handle_upload();
 129                 if ( isset($file['error']) ) {
 130                         echo $file['error'];
 131                         return;
 132                 }
 133
 134                 $this->file = $file['file'];
 135                 $this->get_posts();
 136                 $this->import_posts();
 137                 wp_import_cleanup($file['id']);
 138
 139                 echo '<h3>';
 140                 printf(__('All done. <a href="%s">Have fun!</a>'), get_option('home'));
 141                 echo '</h3>';
 142         }
 143
 144         function dispatch() {
 145                 if (empty ($_GET['step']))
 146                         $step = 0;
 147                 else
 148                         $step = (int) $_GET['step'];
 149
 150                 $this->header();
 151
 152                 switch ($step) {
 153                         case 0 :
 154                                 $this->greet();
 155                                 break;
 156                         case 1 :
 157                                 check_admin_referer('import-upload');
 158                                 $this->import();
 159                                 break;
 160                 }
 161
 162                 $this->footer();
 163         }
 164
 165         function RSS_Import() {
 166                 // Nothing.
 167         }
 168 }
 169
 170 $rss_import = new RSS_Import();
 171
 172 register_importer('rss', __('RSS'), __('Import posts from an RSS feed'), array ($rss_import, 'dispatch'));
 173 ?>