A Blogger's backup file to WordPress' WXR converter. Only tested with posts and comments, and NOT with pages. May not be efficient for huge blogs since the script keep all content in the memory during conversion. Released as public domain.

blogger-to-wordpress.py

# Blogger's backup file to WordPress' WXR converter.
#
# Only tested with posts and comments, and NOT with pages.
# May not be efficient for huge blogs since the script keep
# all content in the memory during conversion.
#
# Released as public domain.
#
# Please note that I converted the labels in Blogspot
# as tags in WordPress. I also hardcoded two categories for the
# WordPress posts. Adjust these first to suit your need.

import sys
from datetime import datetime
from xml.dom.minidom import parse, parseString
from xml.dom import Node
import cgi

import dateutil.parser

inp = sys.argv[1]

def d(*msg):
    print >>sys.stderr, ' '.join(map(str, msg))

class Blog(object):
    class Author(object):
        name = None
        email = None
        uri = None

    class Entry(object):
        entry_id = None
        url = None
        permalink = None
        title = None
        title_type = None
        content = None
        content_type = None
        published = None
        updated = None
        author = None

    class Post(Entry):
        draft = False

        def __init__(self):
            self.labels = []
            self.comments = []

    class Comment(Entry):
        pass

    author = Author()
    posts = []

class BlogParser(object):
    def __init__(self, atom_file):
        self.atom_file = atom_file

    def parse(self):
        self.blog = Blog()
        dom = parse(open(self.atom_file))

        feed = None
        for child in dom.childNodes:
            if child.nodeName == 'feed':
                feed = child
                break

        if feed is not None:
            self.parse_metadata(feed)
            self.parse_entries(feed)

        return self.blog

    def get_text(self, el):
        if el.nodeType == Node.TEXT_NODE:
            return el.nodeValue

        value = []

        if el.nodeType == Node.ELEMENT_NODE:
            for child in el.childNodes:
                value.append(self.get_text(child))

        return ''.join(value)

    def parse_date(self, txt):
        return dateutil.parser.parse(txt)

    def parse_metadata(self, feed):
        for child in feed.childNodes:
            name = child.nodeName.split(':')[-1]

            if name == 'entry':
                break

            if name == 'id':
                self.blog.blog_id = self.get_text(child)
            elif name == 'updated':
                self.blog.updated = self.parse_date(self.get_text(child))
            elif name == 'title':
                self.blog.title = self.get_text(child)
            elif name == 'author':
                self.blog.author = self.parse_author(child)

    def parse_author(self, author):
        data = Blog.Author()

        for child in author.childNodes:
            name = child.nodeName.split(':')[-1]

            if name == 'name':
                data.name = self.get_text(child)
            elif name == 'uri':
                data.uri = self.get_text(child)
            elif name == 'email':
                data.email = self.get_text(child)

        return data

    def parse_entries(self, feed):
        self.posts = []
        self.comments = []
        self.post_ids = {}
        self.comment_ids = {}

        for child in feed.childNodes:
            if child.nodeName != 'entry':
                continue
            self.parse_entry(child)

        self.assign_comments()

        self.blog.posts = self.posts

        total = len(self.blog.posts)
        d('total posts:', total)
        d('total comments:', len(self.comments))
        for i, post in enumerate(self.blog.posts):
            d('(%d / %d) -> %d: %s' % (i+1, total, len(post.comments), post.title))

    def assign_comments(self):
        i = 0
        for comment in self.comments:
            entry_id = comment.post_entry_id
            if entry_id not in self.post_ids:
                continue

            post = self.post_ids[entry_id]
            post.comments.append(comment)

            d('%s. comment: %s -> %s' % (i+1, id(comment), id(post)))
            i += 1

    def parse_category(self, category):
        scheme = category.attributes['scheme'].nodeValue
        term = category.attributes['term'].nodeValue
        return scheme, term

    def get_kind(self, entry):
        for child in entry.childNodes:
            if child.nodeName == 'category':
                scheme, term = self.parse_category(child)
                if scheme == 'http://schemas.google.com/g/2005#kind':
                    return term

    def parse_entry(self, entry):
        kind = self.get_kind(entry)
        if kind == 'http://schemas.google.com/blogger/2008/kind#post':
            post = self.parse_post(entry)
            self.posts.append(post)
            self.post_ids[post.entry_id] = post
        elif kind == 'http://schemas.google.com/blogger/2008/kind#comment':
            comment = self.parse_comment(entry)
            self.comments.append(comment)
            self.comment_ids[comment.entry_id] = comment

    def get_draft(self, control):
        for child in control.childNodes:
            name = child.nodeName.split(':')[-1]
            if name == 'draft':
                return self.get_text(child) == 'yes'
        return False

    def parse_entry_common(self, entry, target):
        for child in entry.childNodes:
            name = child.nodeName.split(':')[-1]
            ns = child.namespaceURI

            if name == 'id':
                target.entry_id = self.get_text(child)
            elif name == 'published':
                target.published = self.parse_date(self.get_text(child))
            elif name == 'updated':
                target.updated = self.parse_date(self.get_text(child))
            elif name == 'title':
                target.title = self.get_text(child)
                target.title_type = child.attributes['type'].nodeValue
            elif name == 'content':
                target.content = self.get_text(child)
                target.content_type = child.attributes['type'].nodeValue
            elif name == 'author':
                target.author = self.parse_author(child)
            elif name == 'link':
                rel = child.attributes['rel'].nodeValue
                href = child.attributes['href'].nodeValue

                if rel == 'self':
                    target.permalink = href
                elif rel == 'alternate':
                    target.url = href

    def parse_post(self, entry):
        post = Blog.Post()
        self.parse_entry_common(entry, post)

        for child in entry.childNodes:
            name = child.nodeName.split(':')[-1]
            ns = child.namespaceURI

            if name == 'category':
                scheme, term = self.parse_category(child)
                if scheme == 'http://www.blogger.com/atom/ns#':
                    post.labels.append(term)
            elif ns == 'http://purl.org/atom/app#' and name == 'control':
                post.draft = self.get_draft(child)

        return post

    def parse_comment(self, entry):
        comment = Blog.Comment()
        self.parse_entry_common(entry, comment)

        for child in entry.childNodes:
            name = child.nodeName.split(':')[-1]
            ns = child.namespaceURI

            if ns == 'http://purl.org/syndication/thread/1.0' and name == 'in-reply-to':
                ref = child.attributes['ref'].nodeValue
                comment.post_entry_id = ref

        return comment

class WXRWriter(object):
    comment_status = 'open'

    def __init__(self, blog):
        self.blog = blog

    def write(self):
        self.post_id = 0
        self.comment_id = 0

        doc = self.get_header() + self.get_entries() + self.get_footer()
        doc = [line.strip() for line in doc]
        doc = '\n'.join(doc)
        return unicode(doc).encode('utf-8')

    def get_header(self):
        res = []
        res.append('<?xml version="1.0" encoding="UTF-8" ?>')
        res.append('<rss version="2.0"')
        res.append('     xmlns:excerpt="http://wordpress.org/export/1.2/excerpt/"')
        res.append('     xmlns:content="http://purl.org/rss/1.0/modules/content/"')
        res.append('     xmlns:wfw="http://wellformedweb.org/CommentAPI/"')
        res.append('     xmlns:dc="http://purl.org/dc/elements/1.1/"')
        res.append('     xmlns:wp="http://wordpress.org/export/1.2/">')

        res.append('<channel>')
        res.append('<title>%s</title>' % self.blog.title)
        res.append('<wp:wxr_version>1.2</wp:wxr_version>')

        return res

    def get_footer(self):
        res = []
        res.append('</channel>')
        res.append('</rss>')
        return res

    def get_entries(self):
        res = []

        for post in self.blog.posts:
            res += self.get_post(post)

        return res

    def get_date(self, ts):
        return ts.strftime("%a, %d %b %Y %H:%M:%S +0000")

    def get_date_wp(self, ts):
        return ts.strftime("%Y-%m-%d %H:%M:%S")

    def escape(self, s):
        return s
        return cgi.escape(s).encode('ascii', 'xmlcharrefreplace')

    def get_comment(self, comment):
        status = 1

        res = []

        self.comment_id += 1

        res.append('  <wp:comment>')
        res.append('    <wp:comment_id>%s</wp:comment_id>' % self.comment_id)
        if comment.author.name:
            res.append('    <wp:comment_author>%s</wp:comment_author>' % comment.author.name)
        if comment.author.email:
            res.append('    <wp:comment_author_email>%s</wp:comment_author_email>' % comment.author.email)
        if comment.author.uri:
            res.append('    <wp:comment_author_url>%s</wp:comment_author_url>' % comment.author.uri)
        res.append('    <wp:comment_author_IP>%s</wp:comment_author_IP>' % '')
        res.append('    <wp:comment_date>%s</wp:comment_date>' % self.get_date_wp(comment.published))
        res.append('    <wp:comment_date_gmt>%s</wp:comment_date_gmt>' % self.get_date_wp(comment.published))
        res.append('    <wp:comment_content><![CDATA[%s]]></wp:comment_content>' % self.escape(comment.content))
        res.append('    <wp:comment_approved>%s</wp:comment_approved>' % status)
        res.append('    <wp:commentmeta>')
        res.append('      <wp:meta_key>blogger_id</wp:meta_key>')
        res.append('      <wp:meta_value>%s</wp:meta_value>' % comment.entry_id)
        res.append('    </wp:commentmeta>')
        if comment.permalink:
            res.append('    <wp:commentmeta>')
            res.append('      <wp:meta_key>blogger_permalink</wp:meta_key>')
            res.append('      <wp:meta_value>%s</wp:meta_value>' % comment.permalink)
            res.append('    </wp:commentmeta>')
        if comment.url:
            res.append('    <wp:commentmeta>')
            res.append('      <wp:meta_key>blogger_url</wp:meta_key>')
            res.append('      <wp:meta_value>%s</wp:meta_value>' % comment.url)
            res.append('    </wp:commentmeta>')
        res.append('  </wp:comment>')

        return res

    def get_post(self, post):
        if post.content.strip() == '':
            return []

        res = []

        slug = None
        if post.url is not None:
            slug = post.url.split('/')[-1]
            slug = slug[:-5]

        status = 'publish'
        if post.draft:
            status = 'draft'

        self.post_id += 1

        res.append('<item>')
        res.append('  <title>%s</title>' % post.title)
        res.append('  <pubDate>%s</pubDate>' % self.get_date(post.published))
        res.append('  <dc:creator>%s</dc:creator>' % post.author.name)
        res.append('  <guid isPermaLink="true">%s</guid>' % post.permalink)
        res.append('  <description></description/>')
        res.append('  <content:encoded><![CDATA[%s]]></content:encoded>' % self.escape(post.content))
        res.append('  <excerpt:encoded><![CDATA[%s]]></excerpt:encoded>' % self.escape(post.content))
        res.append('  <wp:post_id>%s</wp:post_id>' % self.post_id)
        res.append('  <wp:post_date>%s</wp:post_date>' % self.get_date_wp(post.published))
        res.append('  <wp:post_date_gmt>%s</wp:post_date_gmt>' % self.get_date_wp(post.published))
        res.append('  <wp:comment_status>%s</wp:comment_status>' % self.comment_status)
        res.append('  <wp:ping_status>closed</wp:ping_status>')
        if slug:
            res.append('  <wp:post_name>%s</wp:post_name>' % slug)
        res.append('  <wp:status>%s</wp:status>' % status)
        res.append('  <wp:post_parent>0</wp:post_parent>')
        res.append('  <wp:menu_order>0</wp:menu_order>')
        res.append('  <wp:post_type>post</wp:post_type>')
        res.append('  <wp:post_password></wp:post_password>')
        res.append('  <wp:is_sticky>0</wp:is_sticky>')
        res.append('  <category domain="category" nicename="id"><![CDATA[Bahasa Indonesia]]></category>')
        res.append('  <category domain="category" nicename="hacking"><![CDATA[Hacking]]></category>')
        for label in post.labels:
            res.append('  <category domain="post_tag" nicename="%s"><![CDATA[%s]]></category>' % (label, label))
        res.append('  <wp:postmeta>')
        res.append('    <wp:meta_key>blogger_id</wp:meta_key>')
        res.append('    <wp:meta_value>%s</wp:meta_value>' % post.entry_id)
        res.append('  </wp:postmeta>')
        if post.permalink:
            res.append('  <wp:postmeta>')
            res.append('    <wp:meta_key>blogger_permalink</wp:meta_key>')
            res.append('    <wp:meta_value>%s</wp:meta_value>' % post.permalink)
            res.append('  </wp:postmeta>')
        if post.url:
            res.append('  <wp:postmeta>')
            res.append('    <wp:meta_key>blogger_url</wp:meta_key>')
            res.append('    <wp:meta_value>%s</wp:meta_value>' % post.url)
            res.append('  </wp:postmeta>')

        for comment in post.comments:
            res += self.get_comment(comment)

        res.append('</item>')
        return res

p = BlogParser(inp)
blog = p.parse()

writer = WXRWriter(blog)
xml = writer.write()

print xml

# f = open(out, 'w')
# f.write(xml)
# f.close()

@adem0x - just convert that str to unicode on line 24 and you're good.

Also there's a typo in the script. Replace line 361 with the following:

 res.append('  <description></description>')

(basically remove the / at the end of the closing tag) and you're good.