Extract users and post from a Wordpress data export XML file and insert them into PyroCMS

wp_to_pyro.php

<?php
////////////////////////////////////////////////////////////////////////////////
//
// Wordpress -> PyroCMS
//
 
//
// Config
 
$filename              = "<WP EXPORT FILE>";
 
$db_schema             = "<SCHEMA>";
$site_prefix           = "<SITE PREFIX>";   // i.e. "default"
$db_users_default_pass = "<ENCODED_PASS>";  // i.e. "bd0fc5f02742c4e2a94f7fc6baa0d9844b69544a" = m00m00
$db_users_default_salt = "<SALT>";          // i.e. "ef7ab2"                                   = ^ salt

////////////////////////////////////////////////////////////////////////////////
//
// Setup
//

//
// DB
include_once "ezsql_core.php";
include_once "ezsql.php";

$db = new ezSQL_mysql("root", "root", $db_schema, "localhost");

//
// File IO
$fh   = fopen($filename, "r");
$data = fread($fh, filesize($filename));
fclose($fh);

//
// XML
define("NS_WP",      "http://wordpress.org/export/1.2/");
define("NS_DC",      "http://purl.org/dc/elements/1.1/");
define("NS_CONTENT", "http://purl.org/rss/1.0/modules/content/");

$doc = new DOMDocument();
$doc->loadXML($data);

//
// Variables
$author_map   = array();
$tag_map      = array();
$category_map = array();
$titles       = array();

////////////////////////////////////////////////////////////////////////////////
//
// Authors
//

$author_list = $doc->getElementsByTagName("author");

foreach($author_list as $author)
{
    $author_login      = $author->getElementsByTagNameNS(NS_WP, "author_login")->item(0)->nodeValue;
    $author_email      = $author->getElementsByTagNameNS(NS_WP, "author_email")->item(0)->nodeValue;
    $author_username   = $author->getElementsByTagNameNS(NS_WP, "author_display_name")->item(0)->nodeValue;
    $author_first_name = $author->getElementsByTagNameNS(NS_WP, "author_first_name")->item(0)->nodeValue;
    $author_last_name  = $author->getElementsByTagNameNS(NS_WP, "author_last_name")->item(0)->nodeValue;
    
    $created_on   = time();
    
    //
    // Create User
    $db->query("INSERT INTO `{$db_schema}`.`{$site_prefix}_users` (email, password, salt, group_id, ip_address, active, activation_code, created_on, last_login, username, forgotten_password_code, remember_code) " .
               "VALUES('{$author_email}', '{$db_users_default_pass}', '{$db_users_default_salt}', 2, '127.0.0.1', 1, NULL, {$created_on}, 0, '{$author_login}', NULL, NULL);");

    $user_id = mysql_insert_id();
    
    //
    // Create Profile
    $created_date = date("Y-m-d H:i:s");
    
    $db->query("INSERT INTO `{$db_schema}`.`{$site_prefix}_profiles` (`created`,`updated`,`created_by`,`ordering_count`,`user_id`,`display_name`,`first_name`,`last_name`,`company`,`lang`,`bio`,`dob`) " .
               "VALUES('{$created_date}', NULL, '1', '1', '{$user_id}', '{$author_username}', '{$author_first_name}', '{$author_last_name}', NULL, 'en', NULL, '0');");
    
    $author_map[$author_login] = $user_id;
}

////////////////////////////////////////////////////////////////////////////////
//
// Posts
//

$item_list = $doc->getElementsByTagName("item");

foreach($item_list as $item)
{
    //
    // Handle Post
    $post_title   = mysql_real_escape_string(htmlentities($item->getElementsByTagName("title")->item(0)->nodeValue));
    $post_title   = !empty($post_title) ? $post_title : "Untitled";
    $post_content = mysql_real_escape_string(htmlentities($item->getElementsByTagNameNS(NS_CONTENT, "encoded")->item(0)->nodeValue));
    $post_date    = strtotime($item->getElementsByTagNameNS(NS_WP, "post_date")->item(0)->nodeValue);
    //
    $post_creator = $item->getElementsByTagNameNS(NS_DC, "creator")->item(0)->nodeValue;
    $post_creator = isset($author_map[$post_creator]) ? $author_map[$post_creator] : 1;
    //
    $post_status  = $item->getElementsByTagNameNS(NS_WP, "status")->item(0)->nodeValue;
    $post_slug    = uniqid("post_");
    
    if($post_status === "publish")
    {
        $post_slug = $item->getElementsByTagNameNS(NS_WP, "post_name")->item(0)->nodeValue;
    }
    // Ignore Trash ?
    else if($post_status === "trash")
        continue;
    
    //
    // Handle Tags/Categories
    $category_list   = $item->getElementsByTagName("category");
    $post_categories = array();
    $post_tags       = array();
    
    if($category_list->length > 0)
    {
        foreach($category_list as $category)
        {
            $tag_domain = $category->getAttribute("domain");
            $tag_slug   = $category->getAttribute("nicename");
            $tag_value  = $category->nodeValue;
            
            if($tag_domain === "post_tag") 
            {                
                if(!isset($tag_map[$tag_slug]))
                {
                    $db->query("INSERT INTO `$db_schema`.`{$site_prefix}_keywords` (`name`) VALUES ('{$tag_value}');");
                    $tag_id = mysql_insert_id();
                    $tag_map[$tag_slug] = $tag_id;
                }
                else
                    $tag_id = $tag_map[$tag_slug];
                
                $post_tags[] = array(
                    "id"   => $tag_id,
                    "slug" => $tag_slug
                );
            }
            else
            {
                if(!isset($category_map[$tag_slug]))
                {
                    $db->query("INSERT INTO `$db_schema`.`{$site_prefix}_blog_categories` (`slug`, `title`) VALUES ('{$tag_slug}', '{$tag_value}');");
                    $category_id = mysql_insert_id();
                    $category_map[$tag_slug] = $category_id;
                }
                else
                    $category_id = $category_map[$tag_slug];
                
                $post_categories[] = $category_id;
            }
        }
    }
    
    //
    // Handle applied keywords (Tags)
    $post_tags_hash = "";
    if(count($post_tags) > 0)
    {
        $post_tags_hash = uniqid("tags_");
        
        $query = "INSERT INTO `$db_schema`.`{$site_prefix}_keywords_applied` (`hash`, `keyword_id`) VALUES ";
        foreach($post_tags as $i=>$tag)
        {
            if($i > 0) $query .= ",";
            $query .= "('{$post_tags_hash}', {$tag["id"]})";
        }
        $query .= ";";
        $db->query($query);
    }
    
    //
    // Check unique titles
    $found_title = FALSE;

    for($i = 0; $i < count($titles); $i++)
    {
        $clean_title = trim(strtolower($post_title));
        if($clean_title === $titles[$i]["value"])
        {
            $post_title .= "_" . $titles[$i]["count"];
            $titles[$i]["count"]++;
            break;
        }
    }
    if($found_title === FALSE)
        $titles[] = array("name" => $post_title, "value" => trim(strtolower($post_title)), "count" => 1);

    //
    // Defaults
    $post_category = isset($post_categories[0]) ? $post_categories[0] : 1;
    $post_status   = $post_status === "publish" ? "live" : "draft";
    
    
    //
    // Insert
    $db->query("INSERT INTO `{$db_schema}`.`{$site_prefix}_blog` (`title`,`slug`,`category_id`,`attachment`,`intro`,`body`,`parsed`,`keywords`,`author_id`,`created_on`,`updated_on`,`comments_enabled`,`status`,`type`,`preview_hash`) ".
               "VALUES('{$post_title}', '{$post_slug}', $post_category, '', '{$post_content}','{$post_content}', '', '{$post_tags_hash}', $post_creator, $post_date, $post_date, 1, '{$post_status}', 'wysiwyg-advanced', '');");
}

echo "Done!";