scrappy script to migrate Gitlab->Gitea

gl2gt.rb

#!/usr/bin/env ruby

## NOTE: sometimes (probably because GL can't handle load) the repo is created but the import/migrate fails. 
##       go to <gitea>/admin/repos?sort=size to find 0-byte repos, delete them, and try again. Unfortunately,
##       deleting these repos (either because of that page or because the repo isn't fully initialised) doesn't
##       clean up on disk, so I had to manually `rm -rf` them too.

require 'json'
require 'uri'
require 'net/https'
require 'pry-byebug'

GITLAB_TOKEN = 'REPLACE ME WITH ADMIN + READ_REPO TOKEN'
GITEA_TOKEN = 'REPLACE ME'
GITLAB_HEADER = {'Authorization' => "Bearer #{GITLAB_TOKEN}"}
GITEA_HEADER = {
  'Accept' => 'application/json',
  'Authorization' => "token #{GITEA_TOKEN}",
  'Content-Type' => 'application/json',
}

$gitea = Net::HTTP.new('git.bjeanes.com', 443)
$gitea.use_ssl = true
$gitlab = Net::HTTP.new('gitlab.bjeanes.com', 443)
$gitlab.use_ssl = true

# hacky "cache" to skip migrated ones when re-runnign script, due to GitLab's incredibly slow API
$migrated_file = File.open('./migrated.txt', 'a+')
$migrated = File.readlines('./migrated.txt').map(&:chomp)

def each_gitlab_repo(&block)
  path = '/api/v4/projects?pagination=keyset&per_page=10&order_by=id&sort=asc'

  loop do
    resp = $gitlab.get(path, GITLAB_HEADER)
    data = JSON.parse(resp.body)
    data.each(&block)
    next_path = Array(resp.to_hash['link']).grep(/rel="next"/)[0].to_s.scan(/<(.*?)>/).flatten.first
    return unless next_path
    path = URI.parse(next_path).request_uri
  end
end

def create_gitea_org(name)
  $created_orgs ||= []
  return true if $created_orgs.include?(name)
  resp = $gitea.post("/api/v1/orgs", JSON.generate({visibility: :private, username: name}), GITEA_HEADER)
  if resp.code == "201" || resp.body.match("user already exists")
    puts "Created org #{name}" if resp.code == 201
    $created_orgs << name
  else
    $migrated_file.close
    raise resp.body
  end
end

def migrate(repo)
  return if $migrated.include?(repo["path_with_namespace"])
  owner, repo_name = repo["path_with_namespace"].split('/', 2)
  
  # Everything under this account on my GL is a stale mirror of GH account.
  # I will pull these into Gitea separately so they stay as up-to-date mirrors
  if owner == "bjeanes"
    return
  else
    create_gitea_org(owner)
  end

  # Gitea doesn't support nested namespaces, unlike GitLab :(
  # See https://github.com/go-gitea/gitea/issues/1872
  repo_name = repo_name.gsub('/','__')
  is_private = repo["visibility"] == "private"

  resp = $gitea.post("/api/v1/repos/migrate", JSON.generate({
    service: "gitlab",
    private: is_private,
    clone_addr: repo["http_url_to_repo"],
    repo_owner: owner,
    repo_name: repo_name,
    auth_token: GITLAB_TOKEN,
    mirror: false,
    issues: false,
    labels: false,
    milestones: false,
    releases: false,
    pull_requests: false,
    wiki: false,
  }), GITEA_HEADER)

  if resp.code == "201"
    puts "Repo #{owner}/#{repo_name} created and queued for import :)"
  elsif resp.code == "409" && resp.body.match("Files already exist for this repository")
    # I had to fix these manually by `rm -rf` the files under Gitea
    puts "Gitea bug? Repo doesn't exist but can't be created"
    return
  elsif resp.code == "409" && resp.body.match("repository with the same name already exists.")
    # Assume we already imported this so just skip
    puts "Repo #{owner}/#{repo_name} already exists"
  else
    $migrated_file.close
    raise resp.body
  end

  $migrated_file.puts repo["path_with_namespace"]
  $migrated_file.flush

  # Un-watch repo
  $gitea.delete("/api/v1/repos/#{owner}/#{repo_name}/subscription", GITEA_HEADER)
end

each_gitlab_repo { |r|
  next if r['path'] =~ /gitlab/
  migrate(r)
}