Merge branch 'process-commit-worker-migration-encoding' into 'master'

Encode when migrating ProcessCommitWorker jobs

## What does this MR do?

This adds encoding logic to the migration for ProcessCommitWorker, ensuring it doesn't throw errors when the input can not be converted to UTF-8 without extra help.

## What are the relevant issue numbers?

https://gitlab.com/gitlab-org/gitlab-ce/issues/25489

See merge request !8064

3 files changed, 44 insertions, 7 deletions

diff --git a/changelogs/unreleased/process-commit-worker-migration-encoding.yml b/changelogs/unreleased/process-commit-worker-migration-encoding.yml
new file mode 100644
index 00000000000..26aabd9b647
--- /dev/null
+++ b/changelogs/unreleased/process-commit-worker-migration-encoding.yml
@@ -0,0 +1,4 @@
+---
+title: Encode input when migrating ProcessCommitWorker jobs to prevent migration errors
+merge_request: 
+author: 
diff --git a/db/migrate/20161124141322_migrate_process_commit_worker_jobs.rb b/db/migrate/20161124141322_migrate_process_commit_worker_jobs.rb
index 453a44e271a..77e0c40d850 100644
--- a/db/migrate/20161124141322_migrate_process_commit_worker_jobs.rb
+++ b/db/migrate/20161124141322_migrate_process_commit_worker_jobs.rb
@@ -47,14 +47,14 @@ class MigrateProcessCommitWorkerJobs < ActiveRecord::Migration
 
         hash = {
           id: commit.oid,
-          message: commit.message,
+          message: encode(commit.message),
           parent_ids: commit.parent_ids,
           authored_date: commit.author[:time],
-          author_name: commit.author[:name],
-          author_email: commit.author[:email],
+          author_name: encode(commit.author[:name]),
+          author_email: encode(commit.author[:email]),
           committed_date: commit.committer[:time],
-          committer_email: commit.committer[:email],
-          committer_name: commit.committer[:name]
+          committer_email: encode(commit.committer[:email]),
+          committer_name: encode(commit.committer[:name])
         }
 
         payload['args'][2] = hash
@@ -89,4 +89,14 @@ class MigrateProcessCommitWorkerJobs < ActiveRecord::Migration
       end
     end
   end
+
+  def encode(data)
+    encoding = Encoding::UTF_8
+
+    if data.encoding == encoding
+      data
+    else
+      data.encode(encoding, invalid: :replace, undef: :replace)
+    end
+  end
 end
diff --git a/spec/migrations/migrate_process_commit_worker_jobs_spec.rb b/spec/migrations/migrate_process_commit_worker_jobs_spec.rb
index 52428547a9f..6a93deb5412 100644
--- a/spec/migrations/migrate_process_commit_worker_jobs_spec.rb
+++ b/spec/migrations/migrate_process_commit_worker_jobs_spec.rb
@@ -1,3 +1,5 @@
+# encoding: utf-8
+
 require 'spec_helper'
 require Rails.root.join('db', 'migrate', '20161124141322_migrate_process_commit_worker_jobs.rb')
 
@@ -59,6 +61,10 @@ describe MigrateProcessCommitWorkerJobs do
       Sidekiq.redis { |r| r.llen('queue:process_commit') }
     end
 
+    def pop_job
+      JSON.load(Sidekiq.redis { |r| r.lpop('queue:process_commit') })
+    end
+
     before do
       Sidekiq.redis do |redis|
         job = JSON.dump(args: [project.id, user.id, commit.oid])
@@ -92,11 +98,28 @@ describe MigrateProcessCommitWorkerJobs do
       expect(job_count).to eq(1)
     end
 
+    it 'encodes data to UTF-8' do
+      allow_any_instance_of(Rugged::Repository).to receive(:lookup).
+        with(commit.oid).
+        and_return(commit)
+
+      allow(commit).to receive(:message).
+        and_return('김치'.force_encoding('BINARY'))
+
+      migration.up
+
+      job = pop_job
+
+      # We don't care so much about what is being stored, instead we just want
+      # to make sure the encoding is right so that JSON encoding the data
+      # doesn't produce any errors.
+      expect(job['args'][2]['message'].encoding).to eq(Encoding::UTF_8)
+    end
+
     context 'a migrated job' do
       let(:job) do
         migration.up
-
-        JSON.load(Sidekiq.redis { |r| r.lpop('queue:process_commit') })
+        pop_job
       end
 
       let(:commit_hash) do