Merge branch 'fix/optimize-github-importer-for-speed-and-memory' into 'master'

Optimize GitHub importer for speed and memory

See merge request !6552

CHANGELOG

@@ -7,6 +7,7 @@ v 8.13.0 (unreleased)
   - Fix robots.txt disallowing access to groups starting with "s" (Matt Harrison)
   - Revoke button in Applications Settings underlines on hover.
   - Add organization field to user profile
+  - Optimize GitHub importing for speed and memory
 
 v 8.12.2 (unreleased)
   - Fix Import/Export not recognising correctly the imported services.

lib/gitlab/github_import/client.rb

@@ -52,7 +52,7 @@ module Gitlab
 
       def method_missing(method, *args, &block)
         if api.respond_to?(method)
-          request { api.send(method, *args, &block) }
+          request(method, *args, &block)
         else
           super(method, *args, &block)
         end
@@ -99,20 +99,19 @@ module Gitlab
         rate_limit.resets_in + GITHUB_SAFE_SLEEP_TIME
       end
 
-      def request
+      def request(method, *args, &block)
         sleep rate_limit_sleep_time if rate_limit_exceed?
 
-        data = yield
+        data = api.send(method, *args, &block)
+        yield data
 
         last_response = api.last_response
 
         while last_response.rels[:next]
           sleep rate_limit_sleep_time if rate_limit_exceed?
           last_response = last_response.rels[:next].get
-          data.concat(last_response.data) if last_response.data.is_a?(Array)
+          yield last_response.data if last_response.data.is_a?(Array)
         end
-
-        data
       end
     end
   end

lib/gitlab/github_import/importer.rb

@@ -10,6 +10,7 @@ module Gitlab
         @repo     = project.import_source
         @repo_url = project.import_url
         @errors   = []
+        @labels   = {}
 
         if credentials
           @client = Client.new(credentials[:user])
@@ -23,6 +24,7 @@ module Gitlab
         import_milestones
         import_issues
         import_pull_requests
+        import_comments
         import_wiki
         import_releases
         handle_errors
@@ -46,20 +48,20 @@ module Gitlab
       end
 
       def import_labels
-        labels = client.labels(repo, per_page: 100)
-
+        client.labels(repo, per_page: 100) do |labels|
           labels.each do |raw|
             begin
-            LabelFormatter.new(project, raw).create!
+              label = LabelFormatter.new(project, raw).create!
+              @labels[label.title] = label.id
             rescue => e
               errors << { type: :label, url: Gitlab::UrlSanitizer.sanitize(raw.url), errors: e.message }
             end
           end
         end
+      end
 
       def import_milestones
-        milestones = client.milestones(repo, state: :all, per_page: 100)
-
+        client.milestones(repo, state: :all, per_page: 100) do |milestones|
           milestones.each do |raw|
             begin
               MilestoneFormatter.new(project, raw).create!
@@ -68,38 +70,37 @@ module Gitlab
             end
           end
         end
+      end
 
       def import_issues
-        issues = client.issues(repo, state: :all, sort: :created, direction: :asc, per_page: 100)
-
+        client.issues(repo, state: :all, sort: :created, direction: :asc, per_page: 100) do |issues|
           issues.each do |raw|
             gh_issue = IssueFormatter.new(project, raw)
 
             if gh_issue.valid?
               begin
                 issue = gh_issue.create!
-              apply_labels(issue)
-              import_comments(issue) if gh_issue.has_comments?
+                apply_labels(issue, raw)
               rescue => e
                 errors << { type: :issue, url: Gitlab::UrlSanitizer.sanitize(raw.url), errors: e.message }
               end
             end
           end
         end
+      end
 
       def import_pull_requests
-        pull_requests = client.pull_requests(repo, state: :all, sort: :created, direction: :asc, per_page: 100)
-        pull_requests = pull_requests.map { |raw| PullRequestFormatter.new(project, raw) }.select(&:valid?)
+        client.pull_requests(repo, state: :all, sort: :created, direction: :asc, per_page: 100) do |pull_requests|
+          pull_requests.each do |raw|
+            pull_request = PullRequestFormatter.new(project, raw)
+            next unless pull_request.valid?
 
-        pull_requests.each do |pull_request|
             begin
               restore_source_branch(pull_request) unless pull_request.source_branch_exists?
               restore_target_branch(pull_request) unless pull_request.target_branch_exists?
 
               merge_request = pull_request.create!
-            apply_labels(merge_request)
-            import_comments(merge_request)
-            import_comments_on_diff(merge_request)
+              apply_labels(merge_request, raw)
             rescue => e
               errors << { type: :pull_request, url: Gitlab::UrlSanitizer.sanitize(pull_request.url), errors: e.message }
             ensure
@@ -108,6 +109,9 @@ module Gitlab
           end
         end
 
+        project.repository.after_remove_branch
+      end
+
       def restore_source_branch(pull_request)
         project.repository.fetch_ref(repo_url, "pull/#{pull_request.number}/head", pull_request.source_branch_name)
       end
@@ -125,37 +129,38 @@ module Gitlab
       def clean_up_restored_branches(pull_request)
         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?
-
-        project.repository.after_remove_branch
       end
 
-      def apply_labels(issuable)
-        issue = client.issue(repo, issuable.iid)
-
-        if issue.labels.count > 0
-          label_ids = issue.labels
-            .map { |attrs| project.labels.find_by(title: attrs.name).try(:id) }
+      def apply_labels(issuable, raw_issuable)
+        if raw_issuable.labels.count > 0
+          label_ids = raw_issuable.labels
+            .map { |attrs| @labels[attrs.name] }
             .compact
 
           issuable.update_attribute(:label_ids, label_ids)
         end
       end
 
-      def import_comments(issuable)
-        comments = client.issue_comments(repo, issuable.iid, per_page: 100)
-        create_comments(issuable, comments)
+      def import_comments
+        client.issues_comments(repo, per_page: 100) do |comments|
+          create_comments(comments, :issue)
         end
 
-      def import_comments_on_diff(merge_request)
-        comments = client.pull_request_comments(repo, merge_request.iid, per_page: 100)
-        create_comments(merge_request, comments)
+        client.pull_requests_comments(repo, per_page: 100) do |comments|
+          create_comments(comments, :pull_request)
+        end
       end
 
-      def create_comments(issuable, comments)
+      def create_comments(comments, issuable_type)
         ActiveRecord::Base.no_touching do
           comments.each do |raw|
             begin
               comment        = CommentFormatter.new(project, raw)
+              issuable_class = issuable_type == :issue ? Issue : MergeRequest
+              iid            = raw.send("#{issuable_type}_url").split('/').last # GH doesn't return parent ID directly
+              issuable       = issuable_class.find_by_iid(iid)
+              next unless issuable
+
               issuable.notes.create!(comment.attributes)
             rescue => e
               errors << { type: :comment, url: Gitlab::UrlSanitizer.sanitize(raw.url), errors: e.message }
@@ -180,7 +185,7 @@ module Gitlab
       end
 
       def import_releases
-        releases = client.releases(repo, per_page: 100)
+        client.releases(repo, per_page: 100) do |releases|
           releases.each do |raw|
             begin
               gh_release = ReleaseFormatter.new(project, raw)
@@ -192,4 +197,5 @@ module Gitlab
         end
       end
     end
+  end
 end

spec/lib/gitlab/github_import/client_spec.rb

@@ -66,6 +66,6 @@ describe Gitlab::GithubImport::Client, lib: true do
     stub_request(:get, /api.github.com/)
     allow(client.api).to receive(:rate_limit!).and_raise(Octokit::NotFound)
 
-    expect { client.issues }.not_to raise_error
+    expect { client.issues {} }.not_to raise_error
   end
 end

spec/lib/gitlab/github_import/importer_spec.rb

@@ -57,7 +57,8 @@ describe Gitlab::GithubImport::Importer, lib: true do
           created_at: created_at,
           updated_at: updated_at,
           closed_at: nil,
-          url: 'https://api.github.com/repos/octocat/Hello-World/issues/1347'
+          url: 'https://api.github.com/repos/octocat/Hello-World/issues/1347',
+          labels: [double(name: 'Label #1')],
         )
       end
 
@@ -75,7 +76,8 @@ describe Gitlab::GithubImport::Importer, lib: true do
           created_at: created_at,
           updated_at: updated_at,
           closed_at: nil,
-          url: 'https://api.github.com/repos/octocat/Hello-World/issues/1348'
+          url: 'https://api.github.com/repos/octocat/Hello-World/issues/1348',
+          labels: [double(name: 'Label #2')],
         )
       end
 
@@ -94,7 +96,8 @@ describe Gitlab::GithubImport::Importer, lib: true do
           updated_at: updated_at,
           closed_at: nil,
           merged_at: nil,
-          url: 'https://api.github.com/repos/octocat/Hello-World/pulls/1347'
+          url: 'https://api.github.com/repos/octocat/Hello-World/pulls/1347',
+          labels: [double(name: 'Label #3')],
         )
       end
 
@@ -129,6 +132,8 @@ describe Gitlab::GithubImport::Importer, lib: true do
         allow_any_instance_of(Octokit::Client).to receive(:milestones).and_return([milestone, milestone])
         allow_any_instance_of(Octokit::Client).to receive(:issues).and_return([issue1, issue2])
         allow_any_instance_of(Octokit::Client).to receive(:pull_requests).and_return([pull_request, pull_request])
+        allow_any_instance_of(Octokit::Client).to receive(:issues_comments).and_return([])
+        allow_any_instance_of(Octokit::Client).to receive(:pull_requests_comments).and_return([])
         allow_any_instance_of(Octokit::Client).to receive(:last_response).and_return(double(rels: { next: nil }))
         allow_any_instance_of(Octokit::Client).to receive(:releases).and_return([release1, release2])
         allow_any_instance_of(Gitlab::Shell).to receive(:import_repository).and_raise(Gitlab::Shell::Error)
@@ -148,9 +153,7 @@ describe Gitlab::GithubImport::Importer, lib: true do
           errors: [
             { type: :label, url: "https://api.github.com/repos/octocat/Hello-World/labels/bug", errors: "Validation failed: Title can't be blank, Title is invalid" },
             { type: :milestone, url: "https://api.github.com/repos/octocat/Hello-World/milestones/1", errors: "Validation failed: Title has already been taken" },
-            { type: :issue, url: "https://api.github.com/repos/octocat/Hello-World/issues/1347", errors: "Invalid Repository. Use user/repo format." },
             { type: :issue, url: "https://api.github.com/repos/octocat/Hello-World/issues/1348", errors: "Validation failed: Title can't be blank, Title is too short (minimum is 0 characters)" },
-            { type: :pull_request, url: "https://api.github.com/repos/octocat/Hello-World/pulls/1347", errors: "Invalid Repository. Use user/repo format." },
             { type: :pull_request, url: "https://api.github.com/repos/octocat/Hello-World/pulls/1347", errors: "Validation failed: Validate branches Cannot Create: This merge request already exists: [\"New feature\"]" },
             { type: :wiki, errors: "Gitlab::Shell::Error" },
             { type: :release, url: 'https://api.github.com/repos/octocat/Hello-World/releases/2', errors: "Validation failed: Description can't be blank" }