tags. Stopping short
    of a full HTML parser, we only un-wrap those sections not inside 
 tag pairs.
    (This code could be misled by things that look like 
 tags wrapped in HTML comments,
    but oh well.)
    */
    /*$pos = $lastpos = 0;
    while ($lastpos !== false && ($pos = strpos($post_content, '
', $lastpos)) !== false) {
        $post_content = substr($post_content, 0, $lastpos)
            . str_replace("\n", ' ', str_replace("\r", ' ', substr($post_content, $lastpos, $pos - $lastpos)))
            . substr($post_content, $pos);
        $lastpos = strpos($post_content, '
', $pos);
    }
    if ($lastpos !== false) {
        $post_content = substr($post_content, 0, $lastpos)
            . str_replace("\n", ' ', str_replace("\r", ' ', substr($post_content, $lastpos)));
    }
    */
    
    $post_content = addslashes($post_content);
    
    #$post_content = str_replace("\r", ' ', $post_content);
    #$post_content = str_replace("\n", ' ', $post_content);
    $post_date = addslashes($post->createDate);
    $post_title = addslashes($post->title);
    $post_modified = $kSetModDateField ? addslashes($post->modDate) : '';
    $post_name = '';
    if (isset($post->permalink) && strlen($post->permalink)) {
        $matches = array();
        if (preg_match('|/[0-9]{4}/[0-9]{2}/[0-9]{2}/([A-Za-z0-9_-]*)/?|', $post->permalink, $matches)) {
            $post_name = $matches[1];
            $post_name = mysql_escape_string($post_name);
        }
    }
        
    
    $categoryIDList = array();
    foreach ($post->categories as $categoryName) {
        if (isset($kExcludedCategories[$categoryName])) {
            continue;
        }
        $categoryID = $wpdb->get_var("SELECT cat_ID FROM $tablecategories WHERE cat_name = '".mysql_escape_string($categoryName)."'");
        if (!$categoryID) {
            if ($kTakeNoAction) {
                echo "Would have inserted new category '$categoryName'.";
                $categoryID = 0;
            }
            else {
                $categoryNiceName = sanitize_title($categoryName);
                $wpdb->query("INSERT INTO $tablecategories
                    (cat_name, category_nicename)
                  VALUES
                    ('".mysql_escape_string($categoryName)."','".mysql_escape_string($categoryNiceName)."')");
                $categoryID = $wpdb->get_var("SELECT LAST_INSERT_ID()");
            }
        }
        else {
            // category already exists; could update its nicename here if it tended not to be correct already.
            //$wpdb->query("UPDATE $tablecategories SET category_nicename = '".mysql_escape_string(sanitize_title($categoryName))."' WHERE cat_ID = ".intval($categoryID));
        }
        $categoryIDList[] = $categoryID;
    }
    
    
    // Quick-n-dirty check for dups:
    if ($kUpdatePostsIfNewer) {
        $dupcheck = $wpdb->get_results("SELECT ID,post_date,post_title,post_modified FROM $tableposts WHERE post_date='$post_date' AND post_title='$post_title' LIMIT 1",ARRAY_A);
    }
    else {
        $dupcheck = $wpdb->get_results("SELECT ID,post_date,post_title FROM $tableposts WHERE post_date='$post_date' AND post_title='$post_title' LIMIT 1",ARRAY_A);
    }
    if ($dupcheck[0]['ID']) {
        // post already exists
        if ($kUpdatePostsAlways || ($kUpdatePostsIfNewer && $kSetModDateField && $dupcheck[0]['post_modified'] < $post_modified)) {
            print "
\n\nUpdating post, ID = '" . $dupcheck[0]['ID'] . "'
\n";
            print "Timestamp: " . $post_date . "
\n";
            print "Post Title: '" . stripslashes($post_title) . "'
\n";
            if (!$kTakeNoAction) {
                $postID = $dupcheck[0]['ID'];
                $result = $wpdb->query("
                UPDATE $tableposts 
                    SET post_author = '$post_author_ID', post_date = '$post_date', 
                    ".($kSetModDateField ? "post_modified = '$post_modified', " : "")."
                    post_content='$post_content',
                    post_title = '$post_title', post_name = '$post_name' WHERE ID = ".intval($postID));
                //echo "DELETE FROM $tablepost2cat WHERE post_id = ".intval($postID);
                $result = $wpdb->query("DELETE FROM $tablepost2cat WHERE post_id = ".intval($postID));
                foreach ($categoryIDList as $categoryID) {
                    $result = $wpdb->query("
                        INSERT INTO $tablepost2cat 
                            (post_id, category_id)
                        VALUES 
                            (".intval($postID).",".intval($categoryID).")
                        ");
                }
            }
        }
        else {  
            print "
\n\nSkipping duplicate post, ID = '" . $dupcheck[0]['ID'] . "'
\n";
            print "Timestamp: " . $post_date . "
\n";
            print "Post Title: '" . stripslashes($post_title) . "'
\n";
        }
    }
    else {
        print "
\nInserting new post.
\n";
        print "Timestamp: " . $post_date . "
\n";
        print "Post Title: '" . stripslashes($post_title) . "'
\n";
        if (!$kTakeNoAction) {
            $result = $wpdb->query("
            INSERT INTO $tableposts 
                (post_author,post_date,post_content,post_title,post_name,post_category".($post_modified ? ",post_modified" : "").")
            VALUES 
                ('$post_author_ID','$post_date','$post_content','$post_title','$post_name','1'".($post_modified ? ",'$post_modified'" : "").")
            ");
            $postID = $wpdb->get_var("SELECT LAST_INSERT_ID();");
            if ($postID) {
                foreach ($categoryIDList as $categoryID) {
                    $result = $wpdb->query("
                    INSERT INTO $tablepost2cat 
                        (post_id, category_id)
                    VALUES 
                        (".intval($postID).",".intval($categoryID).")
                    ");
                }
            } 
        }
    }
}

// XML parsing code
function importRSSFile($filePath) {
    if (function_exists('xml_parser_create_ns')) {
        $xml_parser = xml_parser_create_ns('iso-8859-1',' ');    // space sep for namespace URI
    }
    else {
        $xml_parser = xml_parser_create();
    }
    // make sure to turn off case-folding; XML 1.0 is case-sensitive
    xml_parser_set_option($xml_parser, XML_OPTION_CASE_FOLDING, false);
    xml_set_element_handler($xml_parser, "startElement", "endElement");
    xml_set_character_data_handler($xml_parser, "characterData");
    if (!($fp = fopen($filePath, "r"))) {
        die("could not open XML input");
    }
    
    while ($data = fread($fp, 4096)) {
        if (!xml_parse($xml_parser, $data, feof($fp))) {
            die(sprintf("XML error: %s at line %d",
                        xml_error_string(xml_get_error_code($xml_parser)),
                        xml_get_current_line_number($xml_parser)));
        }
    }
    xml_parser_free($xml_parser);
    fclose($fp);
}

function importBlogArchive($dirPath) {
    $startYear = 1980;
    $endYear = intval(strftime('%Y'));
    for ($testYear = $startYear; $testYear <= $endYear; $testYear++) {
        for ($testMonth = 1; $testMonth <= 12; $testMonth++) {
            $rssFilePath = $dirPath.'/'.$testYear.'/'.($testMonth < 10 ? '0' : '').$testMonth.'.xml';
            if (is_file($rssFilePath)) {
                importRSSFile($rssFilePath);
            }
        }
    }
}


if (is_dir($path)) {
    importBlogArchive($path);
}
else {
    importRSSFile($path);
}

/*echo '
';
print_r($EZSQL_ERROR);
echo '
';
*/

?>