Change wordpress post links with octocpress post links

migratr.rb

require 'time'

SLUG_TITLE = 0
LINK = 1
DATE = 2

FIND_WP_POST_LINKS = 'grep "http://codebrane.com/blog/?p=" /Users/alistair/dev/codebrane/codebrane.com/codebrane.com.octopress/source/_posts/*'
FIND_WP_POST_IN_EXPORT = 'grep -C 1 "<link>http://codebrane.com/blog/?p=__POST_ID__</link>" ../codebrane.com.wp.export/codebraneblog.wordpress.2013-06-23.xml'
FAR = "find /Users/alistair/dev/codebrane/codebrane.com/codebrane.com.octopress/source/_posts/ -type f | xargs perl -pi -e 's/http:\\/\\/codebrane.com\\/blog\\/\\?p=__WP_POST_ID__\"/__OCTOPRESS_LINK__\"/g'"

# Find all Wordpress style post links in the Octopress posts
wp_post_links_output = IO.popen(FIND_WP_POST_LINKS)
while (line = wp_post_links_output.gets)
  # pull the post id from the link:
  # (http://codebrane.com/blog/?p=1499) or (http://codebrane.com/blog/?p=1499 )
  # line.strip.scan(/(http:\/\/codebrane.com\/blog\/\?p=[0-9]+ ?\))/) {|match|
  line.strip.scan(/(http:\/\/codebrane.com\/blog\/\?p=[0-9]+ ?)/) {|match|
    # post_id = match[0].gsub(/.*p=([0-9]+) ?\).*/, '\1')
    post_id = match[0].gsub(/.*p=([0-9]+) ?.*/, '\1')
  
    # find the Wordpress post link in the original Wordpress export and
    # take the line on either side of it:
    # <title>Coffee break Shibboleth IdP install with custom certificate, login page and LDAP</title>
    # <link>http://codebrane.com/blog/?p=2899</link> <-- match this line
    # <pubDate>Thu, 07 Feb 2013 09:05:08 +0000</pubDate>
    command = FIND_WP_POST_IN_EXPORT.gsub(/__POST_ID__/, post_id)
    output = IO.popen(command)
    lines = Array.new
    match_count = 0
    while (wp_line = output.gets)
      # remove the xml elements
      lines[lines.length] = wp_line.strip.gsub(/<.*>(.*)<\/.*>/, '\1')
    end
  
    # convert the post title to a slug
    lines[SLUG_TITLE] = lines[SLUG_TITLE].downcase.strip.gsub(/:/, '').gsub(/ +/, '-').gsub(/[^\w-]/, '')
    date = Time.parse(lines[DATE])
    slug_date = date.strftime("%Y/%m/%d")
    octopress_post_link = "http://codebrane.com/blog/#{slug_date}/#{lines[SLUG_TITLE]}"
  
    # replace the Wordpress post link in the Octopress posts with the Octoporess post link
    puts "replacing #{post_id} with #{octopress_post_link}"
    system(FAR.gsub(/__WP_POST_ID__/, post_id).gsub(/__OCTOPRESS_LINK__/, octopress_post_link.gsub(/\//, '\\/')))
  }
end