wp2jekyll.py

#!/usr/bin/python

# wp2jekyll.py - convert a Wordpress dump (MySQL) to "pure" text for Jekyll plataform
#
# author: tiago maluta
# e-mail: [email protected]

import MySQLdb

# data

db_server=""
db_user=""
db_pass=""
db_name = ""
db_query = "SELECT post_date, post_title, post_content FROM wp_posts WHERE post_status=\"publish\" and post_type=\"post\";"

msg = "<small>[disclaimer: generated automatically by wp2jekyll.py</small>\n\n"

def main():

    db = MySQLdb.connect(db_server,db_user,db_pass,db_name)
    cursor = db.cursor()
    cursor.execute(db_query)

    while True:
        data = cursor.fetchone()
        if data is None:
            break
        convert(data)

    db.close()

def convert(data):
    post_date = data[0]
    # unfortunately I'm having problems with non ascii chars. 
    post_title = (" ".join(data[1].split())).decode('utf-8','ignore')
    post_content = data[2].decode('utf-8','ignore')

    jekyll_postheader = "---\nlayout: post\ntitle: %s\n---\n\n%s\n\n" % (post_title.replace(":",""), msg) 

    # avoid naming problem in fs
    post_title = post_title.replace("/","-") 
    post_title = post_title.replace("!","")
    post_title = post_title.replace("(","")
    post_title = post_title.replace(")","")
    post_title = post_title.replace(".","")
    post_title = post_title.replace("?","")

    # template: YYYY-MM-DD-title.markdown

    filename = post_date.strftime("%Y-%m-%d")+"-"+post_title.replace(" ","-")+".markdown"

    print filename
    f = open(filename,'w')
    f.write(jekyll_postheader)
    f.write(post_content)
    f.close()

if __name__ == "__main__":
    main()