Convert movable type text exported from hatena diary to markdown.

hatena_diary2markdown.php

<?php

// decoder

function decodePost($data)
{
    $lines    = explode("\n", $data);
    $readMeta = true;
    $post     = array();
    $body     = array();
    $meta     = null;

    foreach ($lines as $line) {
        if (preg_match('/^-----$/', $line, $matches)) {
            $readMeta = true;
        } elseif ($readMeta && preg_match('/^(?P<meta>[\w\s]+?)\:$/', $line, $matches)) {
            $readMeta = false;
            $meta     = strtolower($matches['meta']);
        } elseif ($readMeta && preg_match('/^(?P<meta>[\w\s]+?)\: (?P<value>.*)$/', $line, $matches)) {
            $key   = strtolower($matches['meta']);
            $value = $matches['value'];

            if (array_key_exists($key, $post)) {
                if (is_array($post[$key])) {
                    $post[$key][] = $value;
                } else {
                    $old = $post[$key];
                    $post[$key] = array($old, $value);
                }
            } else {
                $post[$key] = $value;
            }
        } elseif ($meta === 'body') {
            $body[] = $line;
        }
    }

    $post['body'] = implode("\n", $body);

    return $post;
}

function getPosts($path)
{
    $contents     = file_get_contents($path);
    $postContents = preg_split("/\n--------\n/", $contents);
    $posts        = array();

    foreach ($postContents as $data) {
        $post = getPost($data);

        if ($post !== null) {
            $posts[] = $post;
        }
    }

    return $posts;
}

function getPost($data)
{
    $data = trim($data);
    if (empty($data)) {
        return null;
    }

    $post = decodePost($data);

    // post process
    // category
    if (array_key_exists('category', $post)) {
        $post['category'] = (array)$post['category'];
    } else {
        $post['category'] = array();
    }

    // to lower case
    $post['category'] = categoryToLower($post['category']);

    // date time
    // "11/23/2009 00:09:52 PM"
    if (isset($post['date'])) {
        $post['datetime'] = \DateTime::createFromFormat('m/d/Y h:i:s A', $post['date']);
    }

    return $post;
}

function categoryToLower(array $categories)
{
    $lowers = array();

    foreach ($categories as $category) {
        $lowers[] = strtolower($category);
    }

    return $lowers;
}

// converter

function toMarkdown(array $post)
{
    $meta = toMetaMarkdown($post);

    $dir = '_tmp';
    if (!is_dir($dir)) {
        mkdir($dir);
    }
    $body = toMarkdownBody($dir, $post['body']);

    return sprintf("%s\n%s", $meta, $body);
}

function toMetaMarkdown(array $post)
{
    $template = "---
layout: post
title: %s
date: %s
comments: false
categories: %s
published: false
---
";
    if (!empty($post['category'])) {
        $categories = sprintf('[%s]', implode(', ', $post['category']));
    } else {
        $categories = '';
    }

    return sprintf($template, $post['title'], $post['datetime']->format('Y-m-d H:i'), $categories);
}

function toMarkdownBody($dir, $body)
{
    $bodyHtml         = 'body.html';
    $bodyPath         = $dir . "/" . $bodyHtml;
    file_put_contents($bodyPath, $body);

    $filename         = 'body.markdown';
    $bodyMarkdownPath = $dir . "/" . $filename;
    $cmd              = sprintf('cd %s; pandoc -f html -t markdown %s -o %s', $dir, $bodyHtml, $filename);

    exec($cmd, $output, $returnCode);

    unlink($bodyPath);

    if ($returnCode !== 0 || !empty($output)) {
        throw new \RuntimeException('pandoc failure.');
    }

    $body = file_get_contents($bodyMarkdownPath);

    unlink($bodyMarkdownPath);

    return $body;
}

// dumper

function dump($dir, array $post)
{
    $markdown = toMarkdown($post);
    $filename = sprintf('%s-%s.markdown', $post['datetime']->format('Y-m-d'), $post['datetime']->format('YmdHis'));
    $path     = $dir . '/' . $filename;

    file_put_contents($path, $markdown);
}

function dumpAll($dir, array $posts)
{
    foreach ($posts as $post) {
        dump($dir, $post);
    }
}

// run

$dir = '_posts';

if (!is_dir($dir)) {
    mkdir($dir);
}

$file = 'movable_type.txt';
$posts = getPosts($file);

dumpAll($dir, $posts);