Extracts issues of a project on Google Code, and transforms them to BitBucket issues package.

google-issues-to-bitbucket.rb

#!/usr/bin/env ruby
# vim: et sw=2 sts=2

require 'rexml/document'
require 'open-uri'
require 'nokogiri'
require 'json'

class Issue
  def initialize(entry)
    @id = REXML::XPath.first(entry, 'issues:id/text()').value.to_i
    @title = REXML::XPath.first(entry, 'title/text()').value
    @author = REXML::XPath.first(entry, 'author/name/text()').value
    @labels = REXML::XPath.each(entry, 'issues:label/text()').map {|t| t.value} .to_a
    @state = REXML::XPath.first(entry, 'issues:state/text()').value
    @status = REXML::XPath.first(entry, 'issues:status/text()').value
    @published = REXML::XPath.first(entry, 'published/text()').value
    @updated = REXML::XPath.first(entry, 'updated/text()').value
    @content_html = REXML::XPath.first(entry, 'content/text()').value
    @comments_url = REXML::XPath.first(entry, "link[@rel = 'replies']").attribute('href').value()
    owner = REXML::XPath.first(entry, 'issues:owner/issues:username/text()')
    if owner
      @owner = owner.value
    end
  end

  def to_issue_obj(mapping)
    {
      'assignee' => mapping.user(self.owner),
      'component' => self.component,
      'content' => self.content.gsub(/^/, '    '),
      'content_updated_on' => nil,
      'created_on' => self.published,
      'edited_on' => nil,
      'id' => self.id,
      'kind' => mapping.kind(self.type),
      'milestone' => self.milestone,
      'priority' => mapping.priority(self.priority),
      'reporter' => mapping.user(self.author),
      'status' => mapping.status(self.status),
      'title' => self.title,
      'updated_on' => self.updated,
      'version' => nil,
      'watchers' => [mapping.user(self.author)].uniq,
    }
  end

  def self.read_issues(project_name)
    url = "https://code.google.com/feeds/issues/p/#{project_name}/issues/full"
    issues = []
    while true
      open(url) { |atom|
        doc = REXML::Document.new(atom)
        REXML::XPath.each(doc, '/feed/entry') { |entry|
          issue = Issue.new(entry)
          issues.push(issue)
        }
        next_link = REXML::XPath.first(doc, "/feed/link[@rel = 'next']")
        if ! next_link
          return issues
        end
        url = next_link.attribute('href').value
      }
    end
  end

  def id
    @id
  end

  def title
    @title
  end

  def author
    @author
  end

  def owner
    @owner
  end

  def labels
    @labels
  end

  def comments_url
    @comments_url
  end

  def content_html
    @content_html
  end

  def content
    Nokogiri::HTML(self.content_html).text
  end

  def type
    in_label('Type')
  end

  def priority
    in_label('Priority')
  end

  def component
    in_label('Component')
  end

  def milestone
    in_label('Milestone')
  end

  def state
    @state
  end

  def status
    @status
  end

  def published
    @published
  end

  def updated
    @updated
  end

  private
  def in_label(prefix)
    @labels.each { |label|
      if label =~ /#{prefix}-(.+)/
        return Regexp.last_match[1]
      end
    }
    return nil
  end
end

class Comment
  @@id_counter = 1

  def initialize(issue, entry)
    @id = @@id_counter
    @@id_counter += 1
    @issue = issue
    content_text = REXML::XPath.first(entry, 'content/text()')
    @content_html = (content_text ? content_text.value : nil)
    @author = REXML::XPath.first(entry, 'author/name/text()').value
    @published = REXML::XPath.first(entry, 'published/text()').value
    @updated = REXML::XPath.first(entry, 'updated/text()').value
    @labels = REXML::XPath.each(entry, 'issues:updates/issues:label/text()').map {|t| t.value} .to_a
    status_text = REXML::XPath.first(entry, 'issues:updates/issues:status/text()')
    @status = (status_text ? status_text.value : nil)
  end

  def to_comment_obj(mapping)
    {
      'content' => self.content.gsub(/^/, '    '),
      'created_on' => self.published,
      'id' => self.id,
      'issue' => self.issue.id,
      'updated_on' => self.updated,
      'user' => mapping.user(self.author),
    }
  end

  def to_status_log_obj(mapping)
    {
      'changed_from' => nil,
      'changed_to' => mapping.status(self.status),
      'comment' => self.id,
      'created_on' => self.published,
      'field' => 'status',
      'issue' => self.issue.id,
      'user' => mapping.user(self.author),
    }
  end

  def self.read_comments(issues)
    comments = []
    issues.each { |issue|
      url = issue.comments_url
      while url
        open(url) { |atom|
          doc = REXML::Document.new(atom)
          REXML::XPath.each(doc, '/feed/entry') { |entry|
            comment = Comment.new(issue, entry)
            comments.push(comment)
          }
          next_link = REXML::XPath.first(doc, "/feed/link[@rel = 'next']")
          url = (next_link ? next_link.attribute('href').value : nil)
        }
      end
    }
    comments
  end

  def id
    @id
  end

  def issue
    @issue
  end

  def content_html
    @content_html
  end

  def content
    Nokogiri::HTML(self.content_html).text
  end

  def author
    @author
  end

  def published
    @published
  end

  def updated
    @updated
  end

  def labels
    @labels
  end

  def state
    @state
  end

  def status
    @status
  end

  def type
    in_label('Type')
  end

  def priority
    in_label('Priority')
  end

  def component
    in_label('Component')
  end

  def milestone
    in_label('Milestone')
  end

  private
  def in_label(prefix)
    @labels.each { |label|
      if label =~ /#{prefix}-(.+)/
        return Regexp.last_match[1]
      end
    }
    return nil
  end
end

class Mapping
  def initialize
    @user_mapping = {}
    @kind_mapping = {}
    @priority_mapping = {}
  end

  def map_user(google_user, bitbucket_user)
    @user_mapping[google_user] = bitbucket_user
    self
  end

  def user(google_user)
    @user_mapping[google_user]
  end

  def map_kind(google_type, bitbucket_kind)
    @kind_mapping[google_type] = bitbucket_kind
    self
  end

  def kind(google_type)
    @kind_mapping.fetch(google_type, 'bug')
  end

  def map_priority(google_priority, bitbucket_priority)
    @priority_mapping[google_priority] = bitbucket_priority
    self
  end

  def priority(google_priority)
    @priority_mapping.fetch(google_priority, 'major')
  end

  def status(google_status)
    case google_status
    when 'New', 'Accepted', 'Started'
      'new'
    when 'Fixed', 'Verified', 'Done'
      'resolved'
    when 'Invalid'
      'invalid'
    when 'Duplicate'
      'duplicate'
    when 'WontFix'
      'wontfix'
    else
      'new'
    end
  end
end

def extract_logs(comments, mapping)
  comments.select { |c|
    c.status
  } .map { |c|
    c.to_status_log_obj(mapping)
  }
end

def extract_components(issues, comments)
  (issues + comments).select { |e|
    e.component
  } .map { |e|
    e.component
  } .uniq.map { |component|
    {'name' => component}
  }
end

def extract_milestones(issues, comments)
  (issues + comments).select { |e|
    e.milestone
  } .map { |e|
    e.milestone
  } .uniq.map { |milestone|
    {'name' => milestone}
  }
end

issues = Issue.read_issues('kink-lang')
comments = Comment.read_comments(issues)

mapping = Mapping.new
  .map_user('miyakawa...@gmail.com', 'miyakawataku')
  .map_kind('Defect', 'bug')
  .map_kind('Enhancement', 'enhancement')
  .map_kind('Task', 'task')
  .map_kind('Review', 'task')
  .map_kind('Other', 'proposal')
  .map_priority('Critical', 'blocker')
  .map_priority('High', 'critical')
  .map_priority('Medium', 'major')
  .map_priority('Low', 'minor')

db = {
  'issues' => issues.map { |i| i.to_issue_obj(mapping) },
  'comments' => comments.map { |c| c.to_comment_obj(mapping) },
  'attachments' => [],
  'logs' => extract_logs(comments, mapping),
  'components' => extract_components(issues, comments),
  'milestones' => extract_milestones(issues, comments),
  'versions' => [],
  'meta' => {
    'default_assignee' => nil,
    'default_component' => nil,
    'default_kind' => 'bug',
    'default_milestone' => nil,
    'default_version' => nil,
  },
}

puts db.to_json