Merge branch 'gitaly-get-blob' into 'master'

Add gitaly_git_blob_raw feature

See merge request !12712

2 files changed, 62 insertions, 13 deletions

diff --git a/lib/gitlab/git/blob.rb b/lib/gitlab/git/blob.rb
index ffe4f3ca95f..ea386c2ddcb 100644
--- a/lib/gitlab/git/blob.rb
+++ b/lib/gitlab/git/blob.rb
@@ -41,10 +41,6 @@ module Gitlab
               commit_id: sha
             )
           when :BLOB
-            # EncodingDetector checks the first 1024 * 1024 bytes for NUL byte, libgit2 checks
-            # only the first 8000 (https://github.com/libgit2/libgit2/blob/2ed855a9e8f9af211e7274021c2264e600c0f86b/src/filter.h#L15),
-            # which is what we use below to keep a consistent behavior.
-            detect = CharlockHolmes::EncodingDetector.new(8000).detect(entry.data)
             new(
               id: entry.oid,
               name: name,
@@ -53,7 +49,7 @@ module Gitlab
               mode: entry.mode.to_s(8),
               path: path,
               commit_id: sha,
-              binary: detect && detect[:type] == :binary
+              binary: binary?(entry.data)
             )
           end
         end
@@ -87,14 +83,28 @@ module Gitlab
         end
 
         def raw(repository, sha)
-          blob = repository.lookup(sha)
+          Gitlab::GitalyClient.migrate(:git_blob_raw) do |is_enabled|
+            if is_enabled
+              Gitlab::GitalyClient::Blob.new(repository).get_blob(oid: sha, limit: MAX_DATA_DISPLAY_SIZE)
+            else
+              blob = repository.lookup(sha)
+    
+              new(
+                id: blob.oid,
+                size: blob.size,
+                data: blob.content(MAX_DATA_DISPLAY_SIZE),
+                binary: blob.binary?
+              )
+            end
+          end
+        end
 
-          new(
-            id: blob.oid,
-            size: blob.size,
-            data: blob.content(MAX_DATA_DISPLAY_SIZE),
-            binary: blob.binary?
-          )
+        def binary?(data)
+          # EncodingDetector checks the first 1024 * 1024 bytes for NUL byte, libgit2 checks
+          # only the first 8000 (https://github.com/libgit2/libgit2/blob/2ed855a9e8f9af211e7274021c2264e600c0f86b/src/filter.h#L15),
+          # which is what we use below to keep a consistent behavior.
+          detect = CharlockHolmes::EncodingDetector.new(8000).detect(data)
+          detect && detect[:type] == :binary
         end
 
         # Recursive search of blob id by path
@@ -165,8 +175,17 @@ module Gitlab
         return if @data == '' # don't mess with submodule blobs
         return @data if @loaded_all_data
 
+        Gitlab::GitalyClient.migrate(:git_blob_load_all_data) do |is_enabled|
+          @data = begin
+            if is_enabled
+              Gitlab::GitalyClient::Blob.new(repository).get_blob(oid: id, limit: -1).data
+            else
+              repository.lookup(id).content
+            end
+          end
+        end
+
         @loaded_all_data = true
-        @data = repository.lookup(id).content
         @loaded_size = @data.bytesize
         @binary = nil
       end
diff --git a/lib/gitlab/gitaly_client/blob.rb b/lib/gitlab/gitaly_client/blob.rb
new file mode 100644
index 00000000000..0c398b46a08
--- /dev/null
+++ b/lib/gitlab/gitaly_client/blob.rb
@@ -0,0 +1,30 @@
+module Gitlab
+  module GitalyClient
+    class Blob
+      def initialize(repository)
+        @gitaly_repo = repository.gitaly_repository
+      end
+      
+      def get_blob(oid:, limit:)
+        request = Gitaly::GetBlobRequest.new(
+          repository: @gitaly_repo,
+          oid: oid,
+          limit: limit
+        )
+        response = GitalyClient.call(@gitaly_repo.storage_name, :blob_service, :get_blob, request)
+
+        blob = response.first
+        return unless blob.oid.present?
+
+        data = response.reduce(blob.data.dup) { |memo, msg| memo << msg.data.dup }
+
+        Gitlab::Git::Blob.new(
+          id: blob.oid,
+          size: blob.size,
+          data: data,
+          binary: Gitlab::Git::Blob.binary?(data)
+        )
+      end
+    end
+  end
+end