Fix for multipart checksums involving file chunks made entirely of li…

…ne tabulation (0x0B) elements

lib/atc/utils/aws_checksum_utils.rb

@@ -54,7 +54,7 @@ def self.multipart_checksum_for_file(file_path, calculate_whole_object: false)
   def self.digest_file(file_path, part_size, crc32c_accumulator, whole_object_digester)
     File.open(file_path, 'rb') do |file|
       buffer = String.new
-      while file.read(part_size, buffer).present?
+      while file.read(part_size, buffer) != nil
         crc32c_accumulator << Digest::CRC32c.digest(buffer)
         whole_object_digester&.update(buffer)
       end

spec/atc/utils/aws_checksum_utils_spec.rb

@@ -26,6 +26,38 @@
         expect(described_class.checksum_string_for_file(f.path, multipart_threshold)).to eq('g38HMg==')
       end
     end
+
+    # NOTE: We ran into this issue in the past, so that's why this test exists
+    it 'properly handles chunks that are made entirely of line tabulation (0x0B) elements' do
+      Tempfile.create('example-file-to-checksum') do |f|
+        f.write('A' * 5.megabytes)
+        f.write(Atc::Utils::HexUtils.hex_to_bin('0B') * 5.megabytes)
+        f.write(Atc::Utils::HexUtils.hex_to_bin('0B') * 5.megabytes)
+        f.write('A' * 5.megabytes)
+        expect(described_class.checksum_string_for_file(f.path, 5.megabytes)).to eq('YjA5PQ==-4')
+      end
+    end
+  end
+
+  describe '.digest_file' do
+    let(:crc32c_accumulator) { [] }
+    let(:whole_object_digester) { Digest::CRC32c.new }
+    let(:part_size) { 5.megabytes }
+
+    # NOTE: We ran into this issue in the past, so that's why this test exists
+    it 'properly handles chunks that are made entirely of line tabulation (0x0B) elements' do
+      Tempfile.create('example-file-to-checksum') do |f|
+        f.write('A' * part_size)
+        f.write(Atc::Utils::HexUtils.hex_to_bin('0B') * part_size)
+        f.write(Atc::Utils::HexUtils.hex_to_bin('0B') * part_size)
+        f.write('A' * part_size)
+        described_class.digest_file(f.path, part_size, crc32c_accumulator, whole_object_digester)
+      end
+      expect(
+        crc32c_accumulator.map { |checksum_bin_value| Base64.strict_encode64(checksum_bin_value) }
+      ).to eq(['MDaLrw==', '6TPWQg==', '6TPWQg==', 'MDaLrw=='])
+      expect(whole_object_digester.base64digest).to eq('vobXrQ==')
+    end
   end
 
   describe '.compute_default_part_size' do