freekrai/migrate.php

## migrate.php
<?php
/*
Move second crack files to wordpress:

	1) Create a folder called "migrate", and upload this file to that folder.
	2) Upload the PHPMarkdownExtra and PHP Smartypants libraries to this folder as well.
	3) Move the posts folder from your second crack site into this folder.
	4) In your browser, run: http://mysite/migrate/migrate.php and let it run.

	This will move all posts in your posts folder into wordpress, it doesn't touch pages, only posts.

One final note, this worked for me perfectly, you may have to play around with it a little for your own set up.
*/
$fdir = dirname(__FILE__);
require_once($fdir . '/PHPMarkdownExtra/markdown.php');
require_once($fdir . '/PHPSmartyPants/smartypants.php');
set_time_limit(0);
ignore_user_abort( true );
date_default_timezone_set('America/Vancouver');
require('../wp-load.php');

$posts = array();

$folder = "posts";
//	/$year/$month
$all_files = filelist( "posts/", true);

#	echo "<pre>".print_r( $all_files,true )."</pre>";
$i = 0;
foreach($all_files as $fullpath=>$v){
	$orig = new Post($fullpath);
	$row = $orig->array_for_template();
#	echo "<pre>".print_r($row,true)."</pre><hr />";
	$post = array();
	$perma = $row['post-permalink'];
	$post['post_status'] = 'publish';
	$post['post_category'] = catList( $orig->tags );
	$post['post_date'] = date('Y-m-d H:i:s',$orig->timestamp);
	$post['post_title'] = $row['post-title'];
	$post['post_content'] = $row['post-body'];
	$post['body'] = $orig->body;
	$post['guid'] = $row['post-absolute-permalink'];
	if( !empty($orig->headers['link']) ){
		$post['link'] = $orig->headers['link'];
	}
//	$post['tags'] = $row->tags;
	$posts[ $perma ] = $post;
	$i++;
}
#echo "<pre>".print_r($posts,true)."</pre><hr />";
foreach ($posts as $id=>$post) {
	$tags = $post['tags'];
	$link = $post['link'];
	$body = $post['body'];
	unset( $post['body'] );
	unset( $post['link'] );
	unset( $post['tags'] );
	$pid = wp_insert_post($post);
	if( !empty($pid) ){
		add_post_meta($pid,'orig',$body);
		add_post_meta($pid, 'orig_id', $id);
		add_post_meta($pid, 'tags', $tags);
		add_post_meta($pid, 'link', $link);
	}
}

	function catList($makes){
		$taxIDs = array();
		foreach($makes as $make) {
			$make = trim($make);
			$make = str_replace("  "," ",$make);
			if( empty($make) )	continue;
			$term = get_term_by( 'name', $make, 'category' );
			if( $term->term_id ){
				$taxIDs[] = $term->term_id;
			}else{
				$term = wp_insert_term(
					$make,
					'category',
					array(
						'description'=> $make,
						'slug' => $make,
						'parent'=> 0
					)
				);
				if( is_array($term) ){
					$taxIDs[] = $term['term_id'];
				}else{
				}
			}
		}
		return $taxIDs;
	}

    function filelist($dir, $parse_info_in_text_files = true, $use_cached_info = array(), &$deleted_files = array())
    {
        $out = array();
        if (is_dir($dir)) {
            if ($dh = opendir($dir)) {
                while ( ($file = readdir($dh) ) !== false) {
                    if ($file[0] == '.') continue;
                    $fullpath = $dir . '/' . $file;
                    if (is_dir($fullpath)) {
                        $out = array_merge($out, filelist($fullpath, $parse_info_in_text_files, $use_cached_info, $deleted_files));
                    } else {
                        if (isset($deleted_files[$fullpath])) unset($deleted_files[$fullpath]);

                        $new_stat_prefix = filemtime($fullpath) . '-' . filesize($fullpath);

                        if (isset($use_cached_info[$fullpath])) {
                            $stat_prefix = substr($use_cached_info[$fullpath], 0, strpos($use_cached_info[$fullpath], '|'));
                            if ($stat_prefix == $new_stat_prefix) continue;
                        }

                        $tags = '';
                        $type = '';
                        if ($parse_info_in_text_files && substr($fullpath, -(strlen('.txt'))) == '.txt') {
                            $post = new Post($fullpath);
                            $tags = implode(',', $post->tags);
                            $type = $post->type;
                            $expected_fname = $post->expected_source_filename();
#echo $expected_fname."<br />";
                            if ($expected_fname != $fullpath && ! $post->is_draft) {
                                $fullpath = $expected_fname;
                            }
                        }
                        $out[$fullpath] = $new_stat_prefix . '|' . $type . '|' . $tags;
                    }
                }
                closedir($dh);
            }
        }
        return $out;
    }

    class Post{
	    public $type;
	    public $tags = array();
	    public $expected_source_filename;
	    public $is_draft;
	    public $source_filename;
	    public $timestamp;
	    public $year;
	    public $month;
	    public $day;
	    public $offset_in_day;
	    public $slug;
	    public $body;
	    public $headers = array();
	    public $publish_now;
	    public $title;

	    public function __construct($source_filename, $is_draft = -1 /* auto */){
	        $this->source_filename = $source_filename;
	        $this->is_draft = ($is_draft === -1 ? (false !== strpos($source_filename, 'drafts/') || false !== strpos($source_filename, 'pages/')) : $is_draft);
	        $this->timestamp = filemtime($source_filename);
	        $segments = explode("\n\n", trim(file_get_contents($source_filename)), 2);
	        if (! isset($segments[1])) $segments[1] = '';
	        if (count($segments) > 1) {
	            $headers = explode("\n", $segments[0]);
	            $has_title_yet = false;
	            foreach ($headers as $header) {
	                if (isset($header[0]) && $header[0] == '=') {
	                    $has_title_yet = true;
	                    continue;
	                }

	                if (! $has_title_yet) {
	                    $has_title_yet = true;
	                    $this->title = $header;
	                    continue;
	                }

	                if ($this->is_draft && strtolower($header) == 'publish-now') {
	                    $this->publish_now = true;
	                    continue;
	                }

	                $fields = explode(':', $header, 2);
	                if (count($fields) < 2) continue;
	                $fname = strtolower($fields[0]);
	                $fields[1] = trim($fields[1]);
	                if ($fname == 'tags') {
	                    $this->tags = $this->parse_tag_str($fields[1]);
	                } else if ($fname == 'type') {
	                    $this->type = str_replace('|', ' ', $fields[1]);
	                } else if ($fname == 'published') {
	                    $this->timestamp = strtotime($fields[1]);
	                } else {
	                    $this->headers[$fname] = $fields[1];
	                }

	                if (isset($this->headers['link'])) $this->type = 'link';
	            }
	            array_shift($segments);
	        }

	        $this->body = isset($segments[0]) ? $segments[0] : '';
#	        $this->body = SmartyPants( Markdown( $this->body ) );
	        $filename = basename($source_filename);
	        $filename_datestr = substr($filename, 0, 8);
	        if (is_numeric($filename_datestr)) {
	            $this->year = intval(substr($filename_datestr, 0, 4));
	            $this->month = intval(substr($filename_datestr, 4, 2));
	            $this->day = intval(substr($filename_datestr, 6, 2));
	            $this->offset_in_day = intval(substr($filename, 9, 2));
	            $this->slug = ltrim(substr($filename, 11, -(strlen('.txt'))), '-');
	        } else {
	            $this->year = intval(idate('Y', $this->timestamp));
	            $this->month = intval(idate('m', $this->timestamp));
	            $this->day = intval(idate('d', $this->timestamp));
	            $posts_in_ym = Updater::posts_in_year_month($this->year, $this->month);
	            $this->offset_in_day = isset($posts_in_ym[$this->day]) ? count($posts_in_ym[$this->day]) + 1 : 1;
	            $this->slug = substring_before($filename, '.');
	        }
#			echo "----> ".$source_filename." ---- ".$this->expected_source_filename()."<br />";
	        if (! $this->is_draft && $source_filename != $this->expected_source_filename()) {
#	            die("Expected filename mismatch:\nfilename: $source_filename\nexpected: " . $this->expected_source_filename());
	        }
	    }
	    public function array_for_template(){
	        $tags = array();
	        foreach ($this->tags as $tag) { $tags[$tag] = array('post-tag' => $tag); }

	        // Convert relative image references to absolute so index pages work
	        $base_uri = '/' . $this->year . '/' . str_pad($this->month, 2, '0', STR_PAD_LEFT) . '/' . str_pad($this->day, 2, '0', STR_PAD_LEFT);

	        return array_merge(
	            $this->headers,
	            array(
	                'post-title' => html_entity_decode(SmartyPants($this->title), ENT_QUOTES, 'UTF-8'),
	                'post-slug' => $this->slug,
	                'post-timestamp' => $this->timestamp,
	                'post-rss-date' => date('D, d M Y H:i:s T', $this->timestamp),
	                'post-body' => SmartyPants(Markdown($this->body)),
	                'post-tags' => $tags,
	                'post-type' => $this->type,
	                'post-permalink' => $base_uri . '/' . $this->slug,
	                'post-permalink-or-link' => isset($this->headers['link']) && $this->headers['link'] ? $this->headers['link'] : $base_uri . '/' . $this->slug,
	                'post-absolute-permalink' => rtrim('http://rogerstringer.com', '/') . $base_uri . '/' . $this->slug,
	                'post-absolute-permalink-or-link' => rtrim('http://rogerstringer.com', '/') . (isset($this->headers['link']) && $this->headers['link'] ? $this->headers['link'] : $base_uri . '/' . $this->slug),
	                'post-is-first-on-date' => $this->is_first_post_on_this_date ? 'yes' : '',
	            )
	        );
	    }
		public function expected_source_filename($for_new_file = false){
			$padded_month = str_pad($this->month, 2, '0', STR_PAD_LEFT);
			$padded_day = str_pad($this->day, 2, '0', STR_PAD_LEFT);

			$prefix = 'posts/' . $this->year . '/' . $padded_month . '/' . $this->year . $padded_month . $padded_day . '-';
			if ($for_new_file) $padded_offset = self::next_sequence_number_for_day($prefix);
			else $padded_offset = str_pad($this->offset_in_day, 2, '0', STR_PAD_LEFT);
			if ($for_new_file) {
				$slug = trim(preg_replace('/[^a-z0-9-]+/ms', '-', strtolower($this->slug)), '-');
				if (! $slug) $slug = 'post';
			} else {
				$slug = $this->slug;
			}
#			echo( $prefix . $padded_offset . '-' . strtolower($slug) . '.txt' );
			return $prefix . $padded_offset . '-' . strtolower($slug) . '.txt';
		}
		public static function next_sequence_number_for_day($filename_prefix){
			$max = 0;
			foreach (glob($filename_prefix . '*' . '.txt') as $filename) {
				$n = intval(substr(substring_after($filename, $filename_prefix), 0, 2));
				if ($n > $max) $max = $n;
			}
			return str_pad($max + 1, 2, '0', STR_PAD_LEFT);
		}

	    public function parse_tag_str($tag_str){
			// Tags are comma-separated, and any spaces between multiple words in a tag will be converted to underscores for URLs
			if (! strlen($tag_str)) return array();
			$tags = array_map('trim', explode(',', strtolower($tag_str)));
			$tags = str_replace(' ', '_', $tags);
			$tags = array_unique(array_filter($tags, 'strlen'));
			sort($tags);
			return $tags;
		}
    }

?>
	<?php
	/*
	Move second crack files to wordpress:

	1) Create a folder called "migrate", and upload this file to that folder.
	2) Upload the PHPMarkdownExtra and PHP Smartypants libraries to this folder as well.
	3) Move the posts folder from your second crack site into this folder.
	4) In your browser, run: http://mysite/migrate/migrate.php and let it run.

	This will move all posts in your posts folder into wordpress, it doesn't touch pages, only posts.

	One final note, this worked for me perfectly, you may have to play around with it a little for your own set up.
	*/
	$fdir = dirname(__FILE__);
	require_once($fdir . '/PHPMarkdownExtra/markdown.php');
	require_once($fdir . '/PHPSmartyPants/smartypants.php');
	set_time_limit(0);
	ignore_user_abort( true );
	date_default_timezone_set('America/Vancouver');
	require('../wp-load.php');

	$posts = array();

	$folder = "posts";
	// /$year/$month
	$all_files = filelist( "posts/", true);

	# echo "<pre>".print_r( $all_files,true )."</pre>";
	$i = 0;
	foreach($all_files as $fullpath=>$v){
	$orig = new Post($fullpath);
	$row = $orig->array_for_template();
	# echo "<pre>".print_r($row,true)."</pre><hr />";
	$post = array();
	$perma = $row['post-permalink'];
	$post['post_status'] = 'publish';
	$post['post_category'] = catList( $orig->tags );
	$post['post_date'] = date('Y-m-d H:i:s',$orig->timestamp);
	$post['post_title'] = $row['post-title'];
	$post['post_content'] = $row['post-body'];
	$post['body'] = $orig->body;
	$post['guid'] = $row['post-absolute-permalink'];
	if( !empty($orig->headers['link']) ){
	$post['link'] = $orig->headers['link'];
	}
	// $post['tags'] = $row->tags;
	$posts[ $perma ] = $post;
	$i++;
	}
	#echo "<pre>".print_r($posts,true)."</pre><hr />";
	foreach ($posts as $id=>$post) {
	$tags = $post['tags'];
	$link = $post['link'];
	$body = $post['body'];
	unset( $post['body'] );
	unset( $post['link'] );
	unset( $post['tags'] );
	$pid = wp_insert_post($post);
	if( !empty($pid) ){
	add_post_meta($pid,'orig',$body);
	add_post_meta($pid, 'orig_id', $id);
	add_post_meta($pid, 'tags', $tags);
	add_post_meta($pid, 'link', $link);
	}
	}

	function catList($makes){
	$taxIDs = array();
	foreach($makes as $make) {
	$make = trim($make);
	$make = str_replace(" "," ",$make);
	if( empty($make) ) continue;
	$term = get_term_by( 'name', $make, 'category' );
	if( $term->term_id ){
	$taxIDs[] = $term->term_id;
	}else{
	$term = wp_insert_term(
	$make,
	'category',
	array(
	'description'=> $make,
	'slug' => $make,
	'parent'=> 0
	)
	);
	if( is_array($term) ){
	$taxIDs[] = $term['term_id'];
	}else{
	}
	}
	}
	return $taxIDs;
	}

	function filelist($dir, $parse_info_in_text_files = true, $use_cached_info = array(), &$deleted_files = array())
	{
	$out = array();
	if (is_dir($dir)) {
	if ($dh = opendir($dir)) {
	while ( ($file = readdir($dh) ) !== false) {
	if ($file[0] == '.') continue;
	$fullpath = $dir . '/' . $file;
	if (is_dir($fullpath)) {
	$out = array_merge($out, filelist($fullpath, $parse_info_in_text_files, $use_cached_info, $deleted_files));
	} else {
	if (isset($deleted_files[$fullpath])) unset($deleted_files[$fullpath]);

	$new_stat_prefix = filemtime($fullpath) . '-' . filesize($fullpath);

	if (isset($use_cached_info[$fullpath])) {
	$stat_prefix = substr($use_cached_info[$fullpath], 0, strpos($use_cached_info[$fullpath], '\|'));
	if ($stat_prefix == $new_stat_prefix) continue;
	}

	$tags = '';
	$type = '';
	if ($parse_info_in_text_files && substr($fullpath, -(strlen('.txt'))) == '.txt') {
	$post = new Post($fullpath);
	$tags = implode(',', $post->tags);
	$type = $post->type;
	$expected_fname = $post->expected_source_filename();
	#echo $expected_fname."<br />";
	if ($expected_fname != $fullpath && ! $post->is_draft) {
	$fullpath = $expected_fname;
	}
	}
	$out[$fullpath] = $new_stat_prefix . '\|' . $type . '\|' . $tags;
	}
	}
	closedir($dh);
	}
	}
	return $out;
	}

	class Post{
	public $type;
	public $tags = array();
	public $expected_source_filename;
	public $is_draft;
	public $source_filename;
	public $timestamp;
	public $year;
	public $month;
	public $day;
	public $offset_in_day;
	public $slug;
	public $body;
	public $headers = array();
	public $publish_now;
	public $title;

	public function __construct($source_filename, $is_draft = -1 /* auto */){
	$this->source_filename = $source_filename;
	$this->is_draft = ($is_draft === -1 ? (false !== strpos($source_filename, 'drafts/') \|\| false !== strpos($source_filename, 'pages/')) : $is_draft);
	$this->timestamp = filemtime($source_filename);
	$segments = explode("\n\n", trim(file_get_contents($source_filename)), 2);
	if (! isset($segments[1])) $segments[1] = '';
	if (count($segments) > 1) {
	$headers = explode("\n", $segments[0]);
	$has_title_yet = false;
	foreach ($headers as $header) {
	if (isset($header[0]) && $header[0] == '=') {
	$has_title_yet = true;
	continue;
	}

	if (! $has_title_yet) {
	$has_title_yet = true;
	$this->title = $header;
	continue;
	}

	if ($this->is_draft && strtolower($header) == 'publish-now') {
	$this->publish_now = true;
	continue;
	}

	$fields = explode(':', $header, 2);
	if (count($fields) < 2) continue;
	$fname = strtolower($fields[0]);
	$fields[1] = trim($fields[1]);
	if ($fname == 'tags') {
	$this->tags = $this->parse_tag_str($fields[1]);
	} else if ($fname == 'type') {
	$this->type = str_replace('\|', ' ', $fields[1]);
	} else if ($fname == 'published') {
	$this->timestamp = strtotime($fields[1]);
	} else {
	$this->headers[$fname] = $fields[1];
	}

	if (isset($this->headers['link'])) $this->type = 'link';
	}
	array_shift($segments);
	}

	$this->body = isset($segments[0]) ? $segments[0] : '';
	# $this->body = SmartyPants( Markdown( $this->body ) );
	$filename = basename($source_filename);
	$filename_datestr = substr($filename, 0, 8);
	if (is_numeric($filename_datestr)) {
	$this->year = intval(substr($filename_datestr, 0, 4));
	$this->month = intval(substr($filename_datestr, 4, 2));
	$this->day = intval(substr($filename_datestr, 6, 2));
	$this->offset_in_day = intval(substr($filename, 9, 2));
	$this->slug = ltrim(substr($filename, 11, -(strlen('.txt'))), '-');
	} else {
	$this->year = intval(idate('Y', $this->timestamp));
	$this->month = intval(idate('m', $this->timestamp));
	$this->day = intval(idate('d', $this->timestamp));
	$posts_in_ym = Updater::posts_in_year_month($this->year, $this->month);
	$this->offset_in_day = isset($posts_in_ym[$this->day]) ? count($posts_in_ym[$this->day]) + 1 : 1;
	$this->slug = substring_before($filename, '.');
	}
	# echo "----> ".$source_filename." ---- ".$this->expected_source_filename()."<br />";
	if (! $this->is_draft && $source_filename != $this->expected_source_filename()) {
	# die("Expected filename mismatch:\nfilename: $source_filename\nexpected: " . $this->expected_source_filename());
	}
	}
	public function array_for_template(){
	$tags = array();
	foreach ($this->tags as $tag) { $tags[$tag] = array('post-tag' => $tag); }

	// Convert relative image references to absolute so index pages work
	$base_uri = '/' . $this->year . '/' . str_pad($this->month, 2, '0', STR_PAD_LEFT) . '/' . str_pad($this->day, 2, '0', STR_PAD_LEFT);

	return array_merge(
	$this->headers,
	array(
	'post-title' => html_entity_decode(SmartyPants($this->title), ENT_QUOTES, 'UTF-8'),
	'post-slug' => $this->slug,
	'post-timestamp' => $this->timestamp,
	'post-rss-date' => date('D, d M Y H:i:s T', $this->timestamp),
	'post-body' => SmartyPants(Markdown($this->body)),
	'post-tags' => $tags,
	'post-type' => $this->type,
	'post-permalink' => $base_uri . '/' . $this->slug,
	'post-permalink-or-link' => isset($this->headers['link']) && $this->headers['link'] ? $this->headers['link'] : $base_uri . '/' . $this->slug,
	'post-absolute-permalink' => rtrim('http://rogerstringer.com', '/') . $base_uri . '/' . $this->slug,
	'post-absolute-permalink-or-link' => rtrim('http://rogerstringer.com', '/') . (isset($this->headers['link']) && $this->headers['link'] ? $this->headers['link'] : $base_uri . '/' . $this->slug),
	'post-is-first-on-date' => $this->is_first_post_on_this_date ? 'yes' : '',
	)
	);
	}
	public function expected_source_filename($for_new_file = false){
	$padded_month = str_pad($this->month, 2, '0', STR_PAD_LEFT);
	$padded_day = str_pad($this->day, 2, '0', STR_PAD_LEFT);

	$prefix = 'posts/' . $this->year . '/' . $padded_month . '/' . $this->year . $padded_month . $padded_day . '-';
	if ($for_new_file) $padded_offset = self::next_sequence_number_for_day($prefix);
	else $padded_offset = str_pad($this->offset_in_day, 2, '0', STR_PAD_LEFT);
	if ($for_new_file) {
	$slug = trim(preg_replace('/[^a-z0-9-]+/ms', '-', strtolower($this->slug)), '-');
	if (! $slug) $slug = 'post';
	} else {
	$slug = $this->slug;
	}
	# echo( $prefix . $padded_offset . '-' . strtolower($slug) . '.txt' );
	return $prefix . $padded_offset . '-' . strtolower($slug) . '.txt';
	}
	public static function next_sequence_number_for_day($filename_prefix){
	$max = 0;
	foreach (glob($filename_prefix . '*' . '.txt') as $filename) {
	$n = intval(substr(substring_after($filename, $filename_prefix), 0, 2));
	if ($n > $max) $max = $n;
	}
	return str_pad($max + 1, 2, '0', STR_PAD_LEFT);
	}

	public function parse_tag_str($tag_str){
	// Tags are comma-separated, and any spaces between multiple words in a tag will be converted to underscores for URLs
	if (! strlen($tag_str)) return array();
	$tags = array_map('trim', explode(',', strtolower($tag_str)));
	$tags = str_replace(' ', '_', $tags);
	$tags = array_unique(array_filter($tags, 'strlen'));
	sort($tags);
	return $tags;
	}
	}

	?>