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.

1 files changed, 329 insertions, 0 deletions

diff --git a/lib/gitlab/legacy_github_import/importer.rb b/lib/gitlab/legacy_github_import/importer.rb
new file mode 100644
index 00000000000..12c968805f5
--- /dev/null
+++ b/lib/gitlab/legacy_github_import/importer.rb
@@ -0,0 +1,329 @@
+module Gitlab
+  module LegacyGithubImport
+    class Importer
+      include Gitlab::ShellAdapter
+
+      attr_reader :errors, :project, :repo, :repo_url
+
+      def initialize(project)
+        @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'
+          }
+        end
+
+        @client = Client.new(credentials[:user], opts)
+      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)
+        import_comments(:pull_requests)
+        import_wiki
+
+        # Gitea doesn't have a Release API yet
+        # See https://github.com/go-gitea/gitea/issues/330
+        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
+      end
+
+      def handle_errors
+        return unless errors.any?
+
+        project.update_column(:import_error, {
+          message: 'The remote data could not be fully imported.',
+          errors: errors
+        }.to_json)
+      end
+
+      def import_labels
+        fetch_resources(:labels, repo, per_page: 100) do |labels|
+          labels.each do |raw|
+            begin
+              gh_label = LabelFormatter.new(project, raw)
+              gh_label.create!
+            rescue => e
+              errors << { type: :label, url: Gitlab::UrlSanitizer.sanitize(gh_label.url), errors: e.message }
+            end
+          end
+        end
+
+        cache_labels!
+      end
+
+      def import_milestones
+        fetch_resources(:milestones, repo, state: :all, per_page: 100) do |milestones|
+          milestones.each do |raw|
+            begin
+              gh_milestone = MilestoneFormatter.new(project, raw)
+              gh_milestone.create!
+            rescue => e
+              errors << { type: :milestone, url: Gitlab::UrlSanitizer.sanitize(gh_milestone.url), errors: e.message }
+            end
+          end
+        end
+      end
+
+      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 => e
+              errors << { type: :issue, url: Gitlab::UrlSanitizer.sanitize(gh_issue.url), errors: e.message }
+            end
+          end
+        end
+      end
+
+      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 => 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
+
+      def import_comments(issuable_type)
+        resource_type = "#{issuable_type}_comments".to_sym
+
+        # 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
+
+        fetch_resources(resource_type, repo, per_page: 100) do |comments|
+          if last_note
+            discard_inserted_comments(comments, last_note)
+            last_note = nil
+          end
+
+          create_comments(comments)
+        end
+      end
+
+      def create_comments(comments)
+        ActiveRecord::Base.no_touching do
+          comments.each do |raw|
+            begin
+              comment = CommentFormatter.new(project, raw, client)
+
+              # 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('/')
+
+              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
+
+              next unless issuable
+
+              issuable.notes.create!(comment.attributes)
+            rescue => e
+              errors << { type: :comment, url: Gitlab::UrlSanitizer.sanitize(raw.url), errors: e.message }
+            end
+          end
+        end
+      end
+
+      def discard_inserted_comments(comments, last_note)
+        last_note_attrs = nil
+
+        cut_off_index = comments.find_index do |raw|
+          comment           = CommentFormatter.new(project, raw)
+          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
+
+        # Otherwise, remove the resources we've already inserted
+        comments.shift(cut_off_index + 1)
+      end
+
+      def import_wiki
+        unless project.wiki.repository_exists?
+          wiki = WikiFormatter.new(project)
+          gitlab_shell.import_repository(project.repository_storage_path, wiki.disk_path, wiki.import_url)
+        end
+      rescue Gitlab::Shell::Error => 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
+      end
+
+      def import_releases
+        fetch_resources(:releases, repo, per_page: 100) do |releases|
+          releases.each do |raw|
+            begin
+              gh_release = ReleaseFormatter.new(project, raw)
+              gh_release.create! if gh_release.valid?
+            rescue => e
+              errors << { type: :release, url: Gitlab::UrlSanitizer.sanitize(gh_release.url), errors: e.message }
+            end
+          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)
+
+        opts.last[:page] = current_page(resource_type)
+
+        client.public_send(resource_type, *opts) do |resources| # rubocop:disable GitlabSecurity/PublicSend
+          yield resources
+          increment_page(resource_type)
+        end
+
+        imported!(resource_type)
+      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
+    end
+  end
+end