root/trunk/wp-admin/import/wordpress.php

<?php

class WP_Import {

    var $post_ids_processed = array ();
    var $orphans = array ();
    var $file;
    var $id;
    var $mtnames = array ();
    var $newauthornames = array ();
    var $allauthornames = array ();

    var $author_ids = array ();
    var $tags = array ();
    var $categories = array ();

    var $j = -1;
    var $fetch_attachments = false;
    var $url_remap = array ();

    function header() {
        echo '<div class="wrap">';
        echo '<h2>'.__('Import WordPress').'</h2>';
    }

    function footer() {
        echo '</div>';
    }

    function unhtmlentities($string) { // From php.net for < 4.3 compat
        $trans_tbl = get_html_translation_table(HTML_ENTITIES);
        $trans_tbl = array_flip($trans_tbl);
        return strtr($string, $trans_tbl);
    }

    function greet() {
        echo '<div class="narrow">';
        echo '<p>'.__('Howdy! Upload your WordPress eXtended RSS (WXR) file and we&#8217;ll import the posts, comments, custom fields, and categories into this blog.').'</p>';
        echo '<p>'.__('Choose a WordPress WXR file to upload, then click Upload file and import.').'</p>';
        wp_import_upload_form("admin.php?import=wordpress&amp;step=1");
        echo '</div>';
    }

    function get_tag( $string, $tag ) {
        global $wpdb;
        preg_match("|<$tag.*?>(.*?)</$tag>|is", $string, $return);
        $return = preg_replace('|^<!\[CDATA\[(.*)\]\]>$|s', '$1', $return[1]);
        $return = $wpdb->escape( trim( $return ) );
        return $return;
    }

    function has_gzip() {
        return is_callable('gzopen');
    }

    function fopen($filename, $mode='r') {
        if ( $this->has_gzip() )
            return gzopen($filename, $mode);
        return fopen($filename, $mode);
    }

    function feof($fp) {
        if ( $this->has_gzip() )
            return gzeof($fp);
        return feof($fp);
    }

    function fgets($fp, $len=8192) {
        if ( $this->has_gzip() )
            return gzgets($fp, $len);
        return fgets($fp, $len);
    }

    function fclose($fp) {
        if ( $this->has_gzip() )
            return gzclose($fp);
        return fclose($fp);
    }

    function get_entries($process_post_func=NULL) {
        set_magic_quotes_runtime(0);

        $doing_entry = false;
        $is_wxr_file = false;

        $fp = $this->fopen($this->file, 'r');
        if ($fp) {
            while ( !$this->feof($fp) ) {
                $importline = rtrim($this->fgets($fp));

                // this doesn't check that the file is perfectly valid but will at least confirm that it's not the wrong format altogether
                if ( !$is_wxr_file && preg_match('|xmlns:wp="http://wordpress[.]org/export/\d+[.]\d+/"|', $importline) )
                    $is_wxr_file = true;

                if ( false !== strpos($importline, '<wp:category>') ) {
                    preg_match('|<wp:category>(.*?)</wp:category>|is', $importline, $category);
                    $this->categories[] = $category[1];
                    continue;
                }
                if ( false !== strpos($importline, '<wp:tag>') ) {
                    preg_match('|<wp:tag>(.*?)</wp:tag>|is', $importline, $tag);
                    $this->tags[] = $tag[1];
                    continue;
                }
                if ( false !== strpos($importline, '<item>') ) {
                    $this->post = '';
                    $doing_entry = true;
                    continue;
                }
                if ( false !== strpos($importline, '</item>') ) {
                    $doing_entry = false;
                    if ($process_post_func)
                        call_user_func($process_post_func, $this->post);
                    continue;
                }
                if ( $doing_entry ) {
                    $this->post .= $importline . "\n";
                }
            }

            $this->fclose($fp);
        }

        return $is_wxr_file;

    }

    function get_wp_authors() {
        // We need to find unique values of author names, while preserving the order, so this function emulates the unique_value(); php function, without the sorting.
        $temp = $this->allauthornames;
        $authors[0] = array_shift($temp);
        $y = count($temp) + 1;
        for ($x = 1; $x < $y; $x ++) {
            $next = array_shift($temp);
            if (!(in_array($next, $authors)))
                array_push($authors, "$next");
        }

        return $authors;
    }

    function get_authors_from_post() {
        global $current_user;

        // this will populate $this->author_ids with a list of author_names => user_ids

        foreach ( $_POST['author_in'] as $i => $in_author_name ) {

            if ( !empty($_POST['user_select'][$i]) ) {
                // an existing user was selected in the dropdown list
                $user = get_userdata( intval($_POST['user_select'][$i]) );
                if ( isset($user->ID) )
                    $this->author_ids[$in_author_name] = $user->ID;
            }
            elseif ( $this->allow_create_users() ) {
                // nothing was selected in the dropdown list, so we'll use the name in the text field

                $new_author_name = trim($_POST['user_create'][$i]);
                // if the user didn't enter a name, assume they want to use the same name as in the import file
                if ( empty($new_author_name) )
                    $new_author_name = $in_author_name;

                $user_id = username_exists($new_author_name);
                if ( !$user_id ) {
                    $user_id = wp_create_user($new_author_name, wp_generate_password());
                }

                $this->author_ids[$in_author_name] = $user_id;
            }

            // failsafe: if the user_id was invalid, default to the current user
            if ( empty($this->author_ids[$in_author_name]) ) {
                $this->author_ids[$in_author_name] = intval($current_user->ID);
            }
        }

    }

    function wp_authors_form() {
?>
<h2><?php _e('Assign Authors'); ?></h2>
<p><?php _e('To make it easier for you to edit and save the imported posts and drafts, you may want to change the name of the author of the posts. For example, you may want to import all the entries as <code>admin</code>s entries.'); ?></p>
<?php
    if ( $this->allow_create_users() ) {
        echo '<p>'.__('If a new user is created by WordPress, a password will be randomly generated. Manually change the user\'s details if necessary.')."</p>\n";
    }


        $authors = $this->get_wp_authors();
        echo '<ol id="authors">';
        echo '<form action="?import=wordpress&amp;step=2&amp;id=' . $this->id . '" method="post">';
        wp_nonce_field('import-wordpress');
        $j = -1;
        foreach ($authors as $author) {
            ++ $j;
            echo '<li>'.__('Import author:').' <strong>'.$author.'</strong><br />';
            $this->users_form($j, $author);
            echo '</li>';
        }

        if ( $this->allow_fetch_attachments() ) {
?>
</ol>
<h2><?php _e('Import Attachments'); ?></h2>
<p>
    <input type="checkbox" value="1" name="attachments" id="import-attachments" />
    <label for="import-attachments"><?php _e('Download and import file attachments') ?></label>
</p>

<?php
        }

        echo '<input type="submit" value="'.attribute_escape( __('Submit') ).'">'.'<br />';
        echo '</form>';

    }

    function users_form($n, $author) {

        if ( $this->allow_create_users() ) {
            printf('<label>'.__('Create user %1$s or map to existing'), ' <input type="text" value="'.$author.'" name="'.'user_create['.intval($n).']'.'" maxlength="30"></label> <br />');
        }
        else {
            echo __('Map to existing').'<br />';
        }

        // keep track of $n => $author name
        echo '<input type="hidden" name="author_in['.intval($n).']" value="'.htmlspecialchars($author).'" />';

        $users = get_users_of_blog();
?><select name="user_select[<?php echo $n; ?>]">
    <option value="0"><?php _e('- Select -'); ?></option>
    <?php
        foreach ($users as $user) {
            echo '<option value="'.$user->user_id.'">'.$user->user_login.'</option>';
        }
?>
    </select>
    <?php
    }

    function select_authors() {
        $is_wxr_file = $this->get_entries(array(&$this, 'process_author'));
        if ( $is_wxr_file ) {
            $this->wp_authors_form();
        }
        else {
            echo '<h2>'.__('Invalid file').'</h2>';
            echo '<p>'.__('Please upload a valid WXR (WordPress eXtended RSS) export file.').'</p>';
        }
    }

    // fetch the user ID for a given author name, respecting the mapping preferences
    function checkauthor($author) {
        global $current_user;

        if ( !empty($this->author_ids[$author]) )
            return $this->author_ids[$author];

        // failsafe: map to the current user
        return $current_user->ID;
    }



    function process_categories() {
        global $wpdb;

        $cat_names = (array) get_terms('category', 'fields=names');

        while ( $c = array_shift($this->categories) ) {
            $cat_name = trim($this->get_tag( $c, 'wp:cat_name' ));

            // If the category exists we leave it alone
            if ( in_array($cat_name, $cat_names) )
                continue;

            $category_nicename    = $this->get_tag( $c, 'wp:category_nicename' );
            $posts_private        = (int) $this->get_tag( $c, 'wp:posts_private' );
            $links_private        = (int) $this->get_tag( $c, 'wp:links_private' );

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

            if ( empty($parent) )
                $category_parent = '0';
            else
                $category_parent = category_exists($parent);

            $catarr = compact('category_nicename', 'category_parent', 'posts_private', 'links_private', 'posts_private', 'cat_name');

            $cat_ID = wp_insert_category($catarr);
        }
    }

    function process_tags() {
        global $wpdb;

        $tag_names = (array) get_terms('post_tag', 'fields=names');

        while ( $c = array_shift($this->tags) ) {
            $tag_name = trim($this->get_tag( $c, 'wp:tag_name' ));

            // If the category exists we leave it alone
            if ( in_array($tag_name, $tag_names) )
                continue;

            $slug = $this->get_tag( $c, 'wp:tag_slug' );
            $description = $this->get_tag( $c, 'wp:tag_description' );

            $tagarr = compact('slug', 'description');

            $tag_ID = wp_insert_term($tag_name, 'post_tag', $tagarr);
        }
    }

    function process_author($post) {
        $author = $this->get_tag( $post, 'dc:creator' );
        if ($author)
            $this->allauthornames[] = $author;
    }

    function process_posts() {
        $i = -1;
        echo '<ol>';

        $this->get_entries(array(&$this, 'process_post'));

        echo '</ol>';

        wp_import_cleanup($this->id);
        do_action('import_done', 'wordpress');

        echo '<h3>'.sprintf(__('All done.').' <a href="%s">'.__('Have fun!').'</a>', get_option('home')).'</h3>';
    }

    function process_post($post) {
        global $wpdb;

        $post_ID = (int) $this->get_tag( $post, 'wp:post_id' );
          if ( $post_ID && !empty($this->post_ids_processed[$post_ID]) ) // Processed already
            return 0;
        
        set_time_limit( 60 );

        // There are only ever one of these
        $post_title     = $this->get_tag( $post, 'title' );
        $post_date      = $this->get_tag( $post, 'wp:post_date' );
        $post_date_gmt  = $this->get_tag( $post, 'wp:post_date_gmt' );
        $comment_status = $this->get_tag( $post, 'wp:comment_status' );
        $ping_status    = $this->get_tag( $post, 'wp:ping_status' );
        $post_status    = $this->get_tag( $post, 'wp:status' );
        $post_name      = $this->get_tag( $post, 'wp:post_name' );
        $post_parent    = $this->get_tag( $post, 'wp:post_parent' );
        $menu_order     = $this->get_tag( $post, 'wp:menu_order' );
        $post_type      = $this->get_tag( $post, 'wp:post_type' );
        $post_password  = $this->get_tag( $post, 'wp:post_password' );
        $guid           = $this->get_tag( $post, 'guid' );
        $post_author    = $this->get_tag( $post, 'dc:creator' );

        $post_excerpt = $this->get_tag( $post, 'excerpt:encoded' );
        $post_excerpt = preg_replace('|<(/?[A-Z]+)|e', "'<' . strtolower('$1')", $post_excerpt);
        $post_excerpt = str_replace('<br>', '<br />', $post_excerpt);
        $post_excerpt = str_replace('<hr>', '<hr />', $post_excerpt);

        $post_content = $this->get_tag( $post, 'content:encoded' );
        $post_content = preg_replace('|<(/?[A-Z]+)|e', "'<' . strtolower('$1')", $post_content);
        $post_content = str_replace('<br>', '<br />', $post_content);
        $post_content = str_replace('<hr>', '<hr />', $post_content);

        preg_match_all('|<category domain="tag">(.*?)</category>|is', $post, $tags);
        $tags = $tags[1];

        $tag_index = 0;
        foreach ($tags as $tag) {
            $tags[$tag_index] = $wpdb->escape($this->unhtmlentities(str_replace(array ('<![CDATA[', ']]>'), '', $tag)));
            $tag_index++;
        }

        preg_match_all('|<category>(.*?)</category>|is', $post, $categories);
        $categories = $categories[1];

        $cat_index = 0;
        foreach ($categories as $category) {
            $categories[$cat_index] = $wpdb->escape($this->unhtmlentities(str_replace(array ('<![CDATA[', ']]>'), '', $category)));
            $cat_index++;
        }

        $post_exists = post_exists($post_title, '', $post_date);

        if ( $post_exists ) {
            echo '<li>';
            printf(__('Post <em>%s</em> already exists.'), stripslashes($post_title));
        } else {

            // If it has parent, process parent first.
            $post_parent = (int) $post_parent;
            if ($post_parent) {
                // if we already know the parent, map it to the local ID
                if ( $parent = $this->post_ids_processed[$post_parent] ) {
                    $post_parent = $parent;  // new ID of the parent
                }
                else {
                    // record the parent for later
                    $this->orphans[intval($post_ID)] = $post_parent;
                }
            }

            echo '<li>';

            $post_author = $this->checkauthor($post_author); //just so that if a post already exists, new users are not created by checkauthor

            $postdata = compact('post_author', 'post_date', 'post_date_gmt', 'post_content', 'post_excerpt', 'post_title', 'post_status', 'post_name', 'comment_status', 'ping_status', 'guid', 'post_parent', 'menu_order', 'post_type', 'post_password');
            if ($post_type == 'attachment') {
                $remote_url = $this->get_tag( $post, 'wp:attachment_url' );
                if ( !$remote_url )
                    $remote_url = $guid;

                $comment_post_ID = $post_id = $this->process_attachment($postdata, $remote_url);
                if ( !$post_id or is_wp_error($post_id) )
                    return $post_id;
            }
            else {
                printf(__('Importing post <em>%s</em>...'), stripslashes($post_title));
                $comment_post_ID = $post_id = wp_insert_post($postdata);
            }

            if ( is_wp_error( $post_id ) )
                return $post_id;

            // Memorize old and new ID.
            if ( $post_id && $post_ID ) {
                $this->post_ids_processed[intval($post_ID)] = intval($post_id);
            }

            // Add categories.
            if (count($categories) > 0) {
                $post_cats = array();
                foreach ($categories as $category) {
                    $slug = sanitize_term_field('slug', $category, 0, 'category', 'db');
                    $cat = get_term_by('slug', $slug, 'category');
                    $cat_ID = 0;
                    if ( ! empty($cat) )
                        $cat_ID = $cat->term_id;
                    if ($cat_ID == 0) {
                        $category = $wpdb->escape($category);
                        $cat_ID = wp_insert_category(array('cat_name' => $category));
                    }
                    $post_cats[] = $cat_ID;
                }
                wp_set_post_categories($post_id, $post_cats);
            }

            // Add tags.
            if (count($tags) > 0) {
                $post_tags = array();
                foreach ($tags as $tag) {
                    $slug = sanitize_term_field('slug', $tag, 0, </