Detect changes to LFS pointers for pruning and integrity check

Gitlab::Git::Blob.batch_lfs_metadata can be used to check for LFS pointers. It uses a lazy enumorator and filters by blob size

1 files changed, 55 insertions, 0 deletions

diff --git a/spec/lib/gitlab/git/blob_spec.rb b/spec/lib/gitlab/git/blob_spec.rb
index 412a0093d97..c04a9688503 100644
--- a/spec/lib/gitlab/git/blob_spec.rb
+++ b/spec/lib/gitlab/git/blob_spec.rb
@@ -143,6 +143,16 @@ describe Gitlab::Git::Blob, seed_helper: true do
         expect(blob.loaded_size).to eq(blob_size)
       end
     end
+
+    context 'when sha references a tree' do
+      it 'returns nil' do
+        tree = Gitlab::Git::Commit.find(repository, 'master').tree
+
+        blob = Gitlab::Git::Blob.raw(repository, tree.oid)
+
+        expect(blob).to be_nil
+      end
+    end
   end
 
   describe '.raw' do
@@ -226,6 +236,51 @@ describe Gitlab::Git::Blob, seed_helper: true do
     end
   end
 
+  describe '.batch_lfs_pointers' do
+    let(:tree_object) { Gitlab::Git::Commit.find(repository, 'master').tree }
+
+    let(:non_lfs_blob) do
+      Gitlab::Git::Blob.find(
+        repository,
+        'master',
+        'README.md'
+      )
+    end
+
+    let(:lfs_blob) do
+      Gitlab::Git::Blob.find(
+        repository,
+        '33bcff41c232a11727ac6d660bd4b0c2ba86d63d',
+        'files/lfs/image.jpg'
+      )
+    end
+
+    it 'returns a list of Gitlab::Git::Blob' do
+      blobs = described_class.batch_lfs_pointers(repository, [lfs_blob.id])
+
+      expect(blobs.count).to eq(1)
+      expect(blobs).to all( be_a(Gitlab::Git::Blob) )
+    end
+
+    it 'silently ignores tree objects' do
+      blobs = described_class.batch_lfs_pointers(repository, [tree_object.oid])
+
+      expect(blobs).to eq([])
+    end
+
+    it 'silently ignores non lfs objects' do
+      blobs = described_class.batch_lfs_pointers(repository, [non_lfs_blob.id])
+
+      expect(blobs).to eq([])
+    end
+
+    it 'avoids loading large blobs into memory' do
+      expect(repository).not_to receive(:lookup)
+
+      described_class.batch_lfs_pointers(repository, [non_lfs_blob.id])
+    end
+  end
+
   describe 'encoding' do
     context 'file with russian text' do
       let(:blob) { Gitlab::Git::Blob.find(repository, SeedRepo::Commit::ID, "encoding/russian.rb") }