Untitled

'''
This imports your old WP posts to a static html version
'''

import sys
import MySQLdb as mysql

USER = ''
PASSWORD = ''
DATABASE = ''
lOST = ''

SAVE_TO = ''
BASE_URL = ''
POSTS_URL = ''
HEAD_TAG ='''
        <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
        <meta name="generator" content="WordPress 2 HTML v1.0" />
        <meta name="description" content="My old blog posts" />

        <link rel="stylesheet" type="text/css" media="screen" href="%s/css/posts.css" />
''' % POSTS_URL
OLDER_FIRST = True
DATE_FORMAT = '%d-%m-%Y' # day-month-4-digit year

def get_posts():
    global USER, PASSWORD, DATABASE, HOST
    try:
        conn = mysql.Connection(user=USER, passwd=PASSWORD, db=DATABASE)
    except:
        print 'Couldn\'t connect to database %s' % DATABASE
        sys.exit()

    cursor = conn.cursor();

    cursor.execute('SELECT * FROM wp_posts WHERE post_status = \'publish\'')

    results = cursor.fetchall()
    conn.close()
    return results

def write_html(content):
    global SAVE_TO, HEAD_TAG
    try:
        opened = open(SAVE_TO + str(content['id']) + '.html', 'w')
    except IOError,e:
        print 'Miserable failure, %s' % e
        sys.exit()

    # content is a dict with 3 keys: id, title and correctly parsed content.

    opened.write('''<html>
    <head>
        <title>%s</title>
        %s
    </head>
    <body>
        <h1>%s</h1>
%s
    </body>
</html>''' % (content['title'], HEAD_TAG, content['title'],
        content['content']))

    opened.close();

def sortify(set_of_posts):
    # set_of_posts is a list with all the posts fetched by get_posts()
    # this should return a sorted by date list of posts.
    # I have no idea if there's a built in way to do this or if this does
    # it properly, so bear with me.
    # I am forced to do this because I imported some posts from RSS on my WP
    # Install so I have mess up ids.
    global OLDER_FIRST

    datetime_objects = []
    for i in set_of_posts:
        datetime_objects.append(i[2])

    datetime_objects.sort()
    new_set = []

    for v in datetime_objects:
        for i in set_of_posts:
            if i[2] == v:
                new_set.append(i)
    if OLDER_FIRST:
        return reversed(new_set)
    else:
        return new_set

def htmlize():
    global BASE_URL, POSTS_URL, DATE_FORMAT
    posts = sortify(get_posts())
    links = ''
    for post in posts:
        not_parsed = post[4]
        content = '</p>\n\n<p>'.join(not_parsed.split('\r\n\r\n'))
        post_info = {'id': int(post[0]), 'title': post[5], 'content':
        content+'\n<p><a href="%s/index.html" alt="Return to index">Return to List of Posts' % POSTS_URL}
        write_html(post_info)
        links += '<p><a href="%s/%d.html" alt="Link to: %s">%s</a> posted on %s</p>\n' % (POSTS_URL, post_info['id'],post_info['title'],
                post_info['title'], post[2].strftime(DATE_FORMAT))


    # also let's make a nice page for it
    content = {'id': 'index', 'title': 'My old wordpress posts', 'content':
            links+'\n<h2><a href="%s" alt="Return to Main Home">Return to Homepage</a></p>' % BASE_URL}
    write_html(content);

if __name__ == '__main__':
    htmlize();