Merge branch '259619-md5-hexdigest-on-checksummable' into 'master'

Add md5_hexdigest to compute MD5 hashes for files

See merge request gitlab-org/gitlab!71178

app/models/concerns/checksummable.rb

@@ -8,8 +8,12 @@ module Checksummable
       Zlib.crc32(data)
     end
 
-    def hexdigest(path)
+    def sha256_hexdigest(path)
       ::Digest::SHA256.file(path).hexdigest
     end
+
+    def md5_hexdigest(path)
+      ::Digest::MD5.file(path).hexdigest
+    end
   end
 end

app/models/lfs_object.rb

@@ -49,7 +49,7 @@ class LfsObject < ApplicationRecord
   end
 
   def self.calculate_oid(path)
-    self.hexdigest(path)
+    self.sha256_hexdigest(path)
   end
 end
 

app/models/upload.rb

@@ -67,7 +67,7 @@ class Upload < ApplicationRecord
     self.checksum = nil
     return unless needs_checksum?
 
-    self.checksum = self.class.hexdigest(absolute_path)
+    self.checksum = self.class.sha256_hexdigest(absolute_path)
   end
 
   # Initialize the associated Uploader class with current model

ee/app/models/concerns/geo/blob_replicator_strategy.rb

@@ -84,7 +84,7 @@ module Geo
     def calculate_checksum
       raise 'File is not checksummable' unless checksummable?
 
-      model.hexdigest(blob_path)
+      model.sha256_hexdigest(blob_path)
     end
 
     # Returns whether the file exists on disk or in remote storage

ee/lib/gitlab/geo/replication/file_retriever.rb

@@ -36,7 +36,7 @@ module Gitlab
           # Remove this when we implement checksums for files on the Object Storage
           return true unless recorded_file.local?
 
-          extra_params[:checksum] == Upload.hexdigest(recorded_file.absolute_path)
+          extra_params[:checksum] == Upload.sha256_hexdigest(recorded_file.absolute_path)
         end
       end
     end

ee/spec/lib/gitlab/geo/replication/blob_retriever_spec.rb

@@ -3,7 +3,7 @@ require 'spec_helper'
 
 RSpec.describe Gitlab::Geo::Replication::BlobRetriever, :aggregate_failures do
   let(:package_file) { create(:package_file, :npm) }
-  let(:package_checksum) { package_file.class.hexdigest(package_file.file.path) }
+  let(:package_checksum) { package_file.class.sha256_hexdigest(package_file.file.path) }
   let(:replicator_class) { Geo::PackageFileReplicator }
   let(:replicator) { replicator_class.new(model_record_id: package_file.id) }
 

ee/spec/support/shared_examples/models/concerns/blob_replicator_strategy_shared_examples.rb

@@ -207,7 +207,7 @@ RSpec.shared_examples 'a blob replicator' do
     context 'when the file is locally stored' do
       context 'when the file exists' do
         it 'returns hexdigest of the file' do
-          expected = described_class.model.hexdigest(subject.blob_path)
+          expected = described_class.model.sha256_hexdigest(subject.blob_path)
 
           expect(subject.calculate_checksum).to eq(expected)
         end

lib/gitlab/ci/trace.rb

@@ -261,7 +261,7 @@ module Gitlab
             project: job.project,
             file_type: :trace,
             file: stream,
-            file_sha256: self.class.hexdigest(path))
+            file_sha256: self.class.sha256_hexdigest(path))
 
           trace_metadata.track_archival!(trace_artifact.id)
         end

lib/gitlab/verify/uploads.rb

@@ -28,7 +28,7 @@ module Gitlab
       end
 
       def actual_checksum(upload)
-        Upload.hexdigest(upload.absolute_path)
+        Upload.sha256_hexdigest(upload.absolute_path)
       end
 
       def remote_object_exists?(upload)

spec/models/concerns/checksummable_spec.rb

@@ -13,11 +13,19 @@ RSpec.describe Checksummable do
     end
   end
 
-  describe ".hexdigest" do
+  describe ".sha256_hexdigest" do
     it 'returns the SHA256 sum of the file' do
       expected = Digest::SHA256.file(__FILE__).hexdigest
 
-      expect(subject.hexdigest(__FILE__)).to eq(expected)
+      expect(subject.sha256_hexdigest(__FILE__)).to eq(expected)
+    end
+  end
+
+  describe ".md5_hexdigest" do
+    it 'returns the MD5 sum of the file' do
+      expected = Digest::MD5.file(__FILE__).hexdigest
+
+      expect(subject.md5_hexdigest(__FILE__)).to eq(expected)
     end
   end
 end

spec/support/shared_examples/lib/gitlab/ci/ci_trace_shared_examples.rb

@@ -497,7 +497,7 @@ RSpec.shared_examples 'trace with disabled live trace feature' do
         expect(build.job_artifacts_trace.file.filename).to eq('job.log')
         expect(File.exist?(src_path)).to be_falsy
         expect(src_checksum)
-          .to eq(described_class.hexdigest(build.job_artifacts_trace.file.path))
+          .to eq(described_class.sha256_hexdigest(build.job_artifacts_trace.file.path))
         expect(build.job_artifacts_trace.file_sha256).to eq(src_checksum)
       end
     end
@@ -523,7 +523,7 @@ RSpec.shared_examples 'trace with disabled live trace feature' do
         expect(build.job_artifacts_trace.file.filename).to eq('job.log')
         expect(build.old_trace).to be_nil
         expect(src_checksum)
-          .to eq(described_class.hexdigest(build.job_artifacts_trace.file.path))
+          .to eq(described_class.sha256_hexdigest(build.job_artifacts_trace.file.path))
         expect(build.job_artifacts_trace.file_sha256).to eq(src_checksum)
       end
     end
@@ -861,7 +861,7 @@ RSpec.shared_examples 'trace with enabled live trace feature' do
         expect(build.job_artifacts_trace.file.filename).to eq('job.log')
         expect(Ci::BuildTraceChunk.where(build: build)).not_to be_exist
         expect(src_checksum)
-          .to eq(described_class.hexdigest(build.job_artifacts_trace.file.path))
+          .to eq(described_class.sha256_hexdigest(build.job_artifacts_trace.file.path))
         expect(build.job_artifacts_trace.file_sha256).to eq(src_checksum)
       end
     end