gitlab-org--gitlab-foss/lib/gitlab/legacy_github_import/importer.rb

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

344 lines
12 KiB
Ruby
Raw Normal View History

# frozen_string_literal: true
2014-12-31 13:07:48 +00:00
module Gitlab
Rewrite the GitHub importer from scratch Prior to this MR there were two GitHub related importers: * Github::Import: the main importer used for GitHub projects * Gitlab::GithubImport: importer that's somewhat confusingly used for importing Gitea projects (apparently they have a compatible API) This MR renames the Gitea importer to Gitlab::LegacyGithubImport and introduces a new GitHub importer in the Gitlab::GithubImport namespace. This new GitHub importer uses Sidekiq for importing multiple resources in parallel, though it also has the ability to import data sequentially should this be necessary. The new code is spread across the following directories: * lib/gitlab/github_import: this directory contains most of the importer code such as the classes used for importing resources. * app/workers/gitlab/github_import: this directory contains the Sidekiq workers, most of which simply use the code from the directory above. * app/workers/concerns/gitlab/github_import: this directory provides a few modules that are included in every GitHub importer worker. == Stages The import work is divided into separate stages, with each stage importing a specific set of data. Stages will schedule the work that needs to be performed, followed by scheduling a job for the "AdvanceStageWorker" worker. This worker will periodically check if all work is completed and schedule the next stage if this is the case. If work is not yet completed this worker will reschedule itself. Using this approach we don't have to block threads by calling `sleep()`, as doing so for large projects could block the thread from doing any work for many hours. == Retrying Work Workers will reschedule themselves whenever necessary. For example, hitting the GitHub API's rate limit will result in jobs rescheduling themselves. These jobs are not processed until the rate limit has been reset. == User Lookups Part of the importing process involves looking up user details in the GitHub API so we can map them to GitLab users. The old importer used an in-memory cache, but this obviously doesn't work when the work is spread across different threads. The new importer uses a Redis cache and makes sure we only perform API/database calls if absolutely necessary. Frequently used keys are refreshed, and lookup misses are also cached; removing the need for performing API/database calls if we know we don't have the data we're looking for. == Performance & Models The new importer in various places uses raw INSERT statements (as generated by `Gitlab::Database.bulk_insert`) instead of using Rails models. This allows us to bypass any validations and callbacks, drastically reducing the number of SQL queries and Gitaly RPC calls necessary to import projects. To ensure the code produces valid data the corresponding tests check if the produced rows are valid according to the model validation rules.
2017-10-13 16:50:36 +00:00
module LegacyGithubImport
2014-12-31 13:07:48 +00:00
class Importer
2017-11-23 15:51:55 +00:00
def self.refmap
Gitlab::GithubImport.refmap
end
attr_reader :errors, :project, :repo, :repo_url
2014-12-31 13:07:48 +00:00
def initialize(project)
2016-05-09 22:45:37 +00:00
@project = project
@repo = project.import_source
@repo_url = project.import_url
@errors = []
@labels = {}
end
def client
return @client if defined?(@client)
unless credentials
raise Projects::ImportService::Error,
"Unable to find project import data credentials for project ID: #{@project.id}"
end
opts = {}
# Gitea plan to be GitHub compliant
if project.gitea_import?
uri = URI.parse(project.import_url)
host = "#{uri.scheme}://#{uri.host}:#{uri.port}#{uri.path}".sub(%r{/?[\w-]+/[\w-]+\.git\z}, '')
opts = {
host: host,
api_version: 'v1'
}
2016-03-21 14:11:05 +00:00
end
@client = Client.new(credentials[:user], **opts)
2014-12-31 13:07:48 +00:00
end
def execute
# The ordering of importing is important here due to the way GitHub structures their data
# 1. Labels are required by other items while not having a dependency on anything else
# so need to be first
# 2. Pull requests must come before issues. Every pull request is also an issue but not
# all issues are pull requests. Only the issue entity has labels defined in GitHub. GitLab
# doesn't structure data like this so we need to make sure that we've created the MRs
# before we attempt to add the labels defined in the GitHub issue for the related, already
# imported, pull request
import_labels
import_milestones
import_pull_requests
import_issues
import_comments(:issues)
# Gitea doesn't have an API endpoint for pull requests comments
unless project.gitea_import?
import_comments(:pull_requests)
end
import_wiki
# Gitea doesn't have a Release API yet
# See https://github.com/go-gitea/gitea/issues/330
# On re-enabling care should be taken to include releases `author_id` field and enable corresponding tests.
# See:
# 1) https://gitlab.com/gitlab-org/gitlab/-/issues/343448#note_985979730
# 2) https://gitlab.com/gitlab-org/gitlab/-/merge_requests/89694/diffs#dfc4a8141aa296465ea3c50b095a30292fb6ebc4_180_182
unless project.gitea_import?
import_releases
end
handle_errors
true
end
private
def credentials
return @credentials if defined?(@credentials)
@credentials = project.import_data ? project.import_data.credentials : nil
2016-03-21 14:11:05 +00:00
end
def handle_errors
return unless errors.any?
project.import_state.update_column(:last_error, {
message: 'The remote data could not be fully imported.',
errors: errors
}.to_json)
end
2016-04-18 02:52:34 +00:00
def import_labels
fetch_resources(:labels, repo, per_page: 100) do |labels|
labels.each do |raw|
2019-03-13 13:42:43 +00:00
gh_label = LabelFormatter.new(project, raw)
gh_label.create!
rescue StandardError => e
2019-03-13 13:42:43 +00:00
errors << { type: :label, url: Gitlab::UrlSanitizer.sanitize(gh_label.url), errors: e.message }
end
end
cache_labels!
2016-04-18 02:52:34 +00:00
end
2016-04-18 16:44:27 +00:00
def import_milestones
fetch_resources(:milestones, repo, state: :all, per_page: 100) do |milestones|
milestones.each do |raw|
2019-03-13 13:42:43 +00:00
gh_milestone = MilestoneFormatter.new(project, raw)
gh_milestone.create!
rescue StandardError => e
2019-03-13 13:42:43 +00:00
errors << { type: :milestone, url: Gitlab::UrlSanitizer.sanitize(gh_milestone.url), errors: e.message }
end
end
2016-04-18 16:44:27 +00:00
end
# rubocop: disable CodeReuse/ActiveRecord
def import_issues
fetch_resources(:issues, repo, state: :all, sort: :created, direction: :asc, per_page: 100) do |issues|
issues.each do |raw|
gh_issue = IssueFormatter.new(project, raw, client)
begin
issuable =
if gh_issue.pull_request?
MergeRequest.find_by(target_project_id: project.id, iid: gh_issue.number)
else
gh_issue.create!
end
apply_labels(issuable, raw)
rescue StandardError => e
errors << { type: :issue, url: Gitlab::UrlSanitizer.sanitize(gh_issue.url), errors: e.message }
end
2014-12-31 13:07:48 +00:00
end
end
end
# rubocop: enable CodeReuse/ActiveRecord
2014-12-31 13:07:48 +00:00
def import_pull_requests
fetch_resources(:pull_requests, repo, state: :all, sort: :created, direction: :asc, per_page: 100) do |pull_requests|
pull_requests.each do |raw|
gh_pull_request = PullRequestFormatter.new(project, raw, client)
next unless gh_pull_request.valid?
begin
restore_source_branch(gh_pull_request) unless gh_pull_request.source_branch_exists?
restore_target_branch(gh_pull_request) unless gh_pull_request.target_branch_exists?
merge_request = gh_pull_request.create!
# Gitea doesn't return PR in the Issue API endpoint, so labels must be assigned at this stage
if project.gitea_import?
apply_labels(merge_request, raw)
end
rescue StandardError => e
errors << { type: :pull_request, url: Gitlab::UrlSanitizer.sanitize(gh_pull_request.url), errors: e.message }
ensure
clean_up_restored_branches(gh_pull_request)
end
end
end
project.repository.after_remove_branch
end
def restore_source_branch(pull_request)
project.repository.create_branch(pull_request.source_branch_name, pull_request.source_branch_sha)
end
def restore_target_branch(pull_request)
project.repository.create_branch(pull_request.target_branch_name, pull_request.target_branch_sha)
end
def remove_branch(name)
project.repository.delete_branch(name)
rescue Gitlab::Git::Repository::DeleteBranchFailed
errors << { type: :remove_branch, name: name }
end
def clean_up_restored_branches(pull_request)
return if pull_request.opened?
remove_branch(pull_request.source_branch_name) unless pull_request.source_branch_exists?
remove_branch(pull_request.target_branch_name) unless pull_request.target_branch_exists?
end
def apply_labels(issuable, raw)
return unless raw.labels.count > 0
label_ids = raw.labels
.map { |attrs| @labels[attrs.name] }
.compact
issuable.update_attribute(:label_ids, label_ids)
end
# rubocop: disable CodeReuse/ActiveRecord
def import_comments(issuable_type)
resource_type = "#{issuable_type}_comments".to_sym
2016-10-19 18:42:31 +00:00
# Two notes here:
# 1. We don't have a distinctive attribute for comments (unlike issues iid), so we fetch the last inserted note,
# compare it against every comment in the current imported page until we find match, and that's where start importing
# 2. GH returns comments for _both_ issues and PRs through issues_comments API, while pull_requests_comments returns
# only comments on diffs, so select last note not based on noteable_type but on line_code
line_code_is = issuable_type == :pull_requests ? 'NOT NULL' : 'NULL'
last_note = project.notes.where("line_code IS #{line_code_is}").last
2016-10-19 18:42:31 +00:00
fetch_resources(resource_type, repo, per_page: 100) do |comments|
2016-10-19 18:42:31 +00:00
if last_note
discard_inserted_comments(comments, last_note)
last_note = nil
end
2016-10-27 13:08:37 +00:00
create_comments(comments)
end
end
# rubocop: enable CodeReuse/ActiveRecord
2014-12-31 13:07:48 +00:00
# rubocop: disable CodeReuse/ActiveRecord
2016-10-27 13:08:37 +00:00
def create_comments(comments)
ActiveRecord::Base.no_touching do
comments.each do |raw|
2019-03-13 13:42:43 +00:00
comment = CommentFormatter.new(project, raw, client)
2019-03-13 13:42:43 +00:00
# GH does not return info about comment's parent, so we guess it by checking its URL!
*_, parent, iid = URI(raw.html_url).path.split('/')
2019-03-13 13:42:43 +00:00
issuable = if parent == 'issues'
Issue.find_by(project_id: project.id, iid: iid)
else
MergeRequest.find_by(target_project_id: project.id, iid: iid)
end
2019-03-13 13:42:43 +00:00
next unless issuable
2019-03-13 13:42:43 +00:00
issuable.notes.create!(comment.attributes)
rescue StandardError => e
2019-03-13 13:42:43 +00:00
errors << { type: :comment, url: Gitlab::UrlSanitizer.sanitize(raw.url), errors: e.message }
end
end
2014-12-31 13:07:48 +00:00
end
# rubocop: enable CodeReuse/ActiveRecord
2016-01-06 23:15:37 +00:00
2016-10-19 18:42:31 +00:00
def discard_inserted_comments(comments, last_note)
last_note_attrs = nil
cut_off_index = comments.find_index do |raw|
comment = CommentFormatter.new(project, raw)
2016-10-19 18:42:31 +00:00
comment_attrs = comment.attributes
last_note_attrs ||= last_note.slice(*comment_attrs.keys)
comment_attrs.with_indifferent_access == last_note_attrs
end
# No matching resource in the collection, which means we got halted right on the end of the last page, so all good
return unless cut_off_index
2016-10-27 13:00:31 +00:00
# Otherwise, remove the resources we've already inserted
2016-10-19 18:42:31 +00:00
comments.shift(cut_off_index + 1)
end
2016-01-06 23:15:37 +00:00
def import_wiki
return if project.wiki.repository_exists?
wiki = WikiFormatter.new(project)
project.wiki.repository.import_repository(wiki.import_url)
rescue ::Gitlab::Git::CommandError => e
# GitHub error message when the wiki repo has not been created,
# this means that repo has wiki enabled, but have no pages. So,
# we can skip the import.
if e.message !~ /repository not exported/
errors << { type: :wiki, errors: e.message }
end
2016-01-06 23:15:37 +00:00
end
def import_releases
fetch_resources(:releases, repo, per_page: 100) do |releases|
releases.each do |raw|
2019-03-13 13:42:43 +00:00
gh_release = ReleaseFormatter.new(project, raw)
gh_release.create! if gh_release.valid?
rescue StandardError => e
2019-03-13 13:42:43 +00:00
errors << { type: :release, url: Gitlab::UrlSanitizer.sanitize(gh_release.url), errors: e.message }
end
end
end
def cache_labels!
project.labels.select(:id, :title).find_each do |label|
@labels[label.title] = label.id
end
end
def fetch_resources(resource_type, *opts)
return if imported?(resource_type)
2017-02-21 23:50:22 +00:00
opts.last[:page] = current_page(resource_type)
2016-10-19 18:42:31 +00:00
client.public_send(resource_type, *opts) do |resources| # rubocop:disable GitlabSecurity/PublicSend
yield resources
increment_page(resource_type)
end
2016-10-19 18:42:31 +00:00
imported!(resource_type)
rescue ::Octokit::NotFound => e
errors << { type: resource_type, errors: e.message }
2016-10-19 18:42:31 +00:00
end
def imported?(resource_type)
Rails.cache.read("#{cache_key_prefix}:#{resource_type}:imported")
end
def imported!(resource_type)
Rails.cache.write("#{cache_key_prefix}:#{resource_type}:imported", true, ex: 1.day)
end
def increment_page(resource_type)
key = "#{cache_key_prefix}:#{resource_type}:current-page"
# Rails.cache.increment calls INCRBY directly on the value stored under the key, which is
# a serialized ActiveSupport::Cache::Entry, so it will return an error by Redis, hence this ugly work-around
page = Rails.cache.read(key)
page += 1
Rails.cache.write(key, page)
page
end
def current_page(resource_type)
Rails.cache.fetch("#{cache_key_prefix}:#{resource_type}:current-page", ex: 1.day) { 1 }
end
def cache_key_prefix
@cache_key_prefix ||= "github-import:#{project.id}"
end
2014-12-31 13:07:48 +00:00
end
end
end