From 4bb609f350948869f9895a7e3dc7e4a61c5c417c Mon Sep 17 00:00:00 2001
From: Nick Thomas <nick@gitlab.com>
Date: Tue, 6 Apr 2021 17:02:14 +0100
Subject: [PATCH] Make blobs directly accessible through the graphql repository

Going through the tree is not always easy, especially when you want
a single blob in a large subdirectory
---
 app/graphql/resolvers/blobs_resolver.rb       | 37 ++++++++++
 app/graphql/types/repository_type.rb          |  2 +
 app/models/blob.rb                            |  1 +
 ...ke-blob-info-available-through-graphql.yml |  5 ++
 doc/api/graphql/reference/index.md            |  1 +
 spec/graphql/resolvers/blobs_resolver_spec.rb | 74 +++++++++++++++++++
 spec/graphql/types/repository_type_spec.rb    |  2 +
 7 files changed, 122 insertions(+)
 create mode 100644 app/graphql/resolvers/blobs_resolver.rb
 create mode 100644 changelogs/unreleased/323195-make-blob-info-available-through-graphql.yml
 create mode 100644 spec/graphql/resolvers/blobs_resolver_spec.rb

diff --git a/app/graphql/resolvers/blobs_resolver.rb b/app/graphql/resolvers/blobs_resolver.rb
new file mode 100644
index 00000000000000..521e04827591fe
--- /dev/null
+++ b/app/graphql/resolvers/blobs_resolver.rb
@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+module Resolvers
+  class BlobsResolver < BaseResolver
+    include Gitlab::Graphql::Authorize::AuthorizeResource
+
+    type Types::Tree::BlobType.connection_type, null: true
+    authorize :download_code
+    calls_gitaly!
+
+    alias_method :repository, :object
+
+    argument :paths, [GraphQL::STRING_TYPE],
+             required: true,
+             description: 'Array of desired blob paths.'
+    argument :ref, GraphQL::STRING_TYPE,
+             required: false,
+             default_value: nil,
+             description: 'The commit ref to get the blobs from. Default value is HEAD.'
+
+    # We fetch blobs from Gitaly efficiently but it still scales O(N) with the
+    # number of paths being fetched, so apply a scaling limit to that.
+    def self.resolver_complexity(args, child_complexity:)
+      super + args.fetch(:paths, []).size
+    end
+
+    def resolve(paths:, ref:)
+      authorize!(repository.container)
+
+      return [] if repository.empty?
+
+      ref ||= repository.root_ref
+
+      repository.blobs_at(paths.map { |path| [ref, path] })
+    end
+  end
+end
diff --git a/app/graphql/types/repository_type.rb b/app/graphql/types/repository_type.rb
index e319a5f312414a..fc835cdf642ab0 100644
--- a/app/graphql/types/repository_type.rb
+++ b/app/graphql/types/repository_type.rb
@@ -14,5 +14,7 @@ class RepositoryType < BaseObject
           description: 'Indicates a corresponding Git repository exists on disk.'
     field :tree, Types::Tree::TreeType, null: true, resolver: Resolvers::TreeResolver, calls_gitaly: true,
           description: 'Tree of the repository.'
+    field :blobs, Types::Tree::BlobType.connection_type, null: true, resolver: Resolvers::BlobsResolver, calls_gitaly: true,
+          description: 'Blobs contained within the repository'
   end
 end
diff --git a/app/models/blob.rb b/app/models/blob.rb
index 8a9db8b45ea220..2185233a1ac55e 100644
--- a/app/models/blob.rb
+++ b/app/models/blob.rb
@@ -2,6 +2,7 @@
 
 # Blob is a Rails-specific wrapper around Gitlab::Git::Blob, SnippetBlob and Ci::ArtifactBlob
 class Blob < SimpleDelegator
+  include GlobalID::Identification
   include Presentable
   include BlobLanguageFromGitAttributes
   include BlobActiveModel
diff --git a/changelogs/unreleased/323195-make-blob-info-available-through-graphql.yml b/changelogs/unreleased/323195-make-blob-info-available-through-graphql.yml
new file mode 100644
index 00000000000000..e4b6e55bb11946
--- /dev/null
+++ b/changelogs/unreleased/323195-make-blob-info-available-through-graphql.yml
@@ -0,0 +1,5 @@
+---
+title: Make blobs directly accessible through the graphql repository
+merge_request: 58677
+author:
+type: added
diff --git a/doc/api/graphql/reference/index.md b/doc/api/graphql/reference/index.md
index 56b5bcce3f1ab0..6da5ec2df2c432 100644
--- a/doc/api/graphql/reference/index.md
+++ b/doc/api/graphql/reference/index.md
@@ -5388,6 +5388,7 @@ Autogenerated return type of RepositionImageDiffNote.
 
 | Field | Type | Description |
 | ----- | ---- | ----------- |
+| `blobs` | [`BlobConnection`](#blobconnection) | Blobs contained within the repository. |
 | `empty` | [`Boolean!`](#boolean) | Indicates repository has no visible content. |
 | `exists` | [`Boolean!`](#boolean) | Indicates a corresponding Git repository exists on disk. |
 | `rootRef` | [`String`](#string) | Default branch of the repository. |
diff --git a/spec/graphql/resolvers/blobs_resolver_spec.rb b/spec/graphql/resolvers/blobs_resolver_spec.rb
new file mode 100644
index 00000000000000..bc0344796ee3b3
--- /dev/null
+++ b/spec/graphql/resolvers/blobs_resolver_spec.rb
@@ -0,0 +1,74 @@
+# frozen_string_literal: true
+
+require 'spec_helper'
+
+RSpec.describe Resolvers::BlobsResolver do
+  include GraphqlHelpers
+
+  describe '.resolver_complexity' do
+    it 'adds one per path being resolved' do
+      control = described_class.resolver_complexity({}, child_complexity: 1)
+
+      expect(described_class.resolver_complexity({ paths: %w[a b c] }, child_complexity: 1))
+        .to eq(control + 3)
+    end
+  end
+
+  describe '#resolve' do
+    let_it_be(:user) { create(:user) }
+    let_it_be(:project) { create(:project, :repository) }
+
+    let(:repository) { project.repository }
+    let(:args) { { paths: paths, ref: ref } }
+    let(:paths) { [] }
+    let(:ref) { nil }
+
+    subject(:resolve_blobs) { resolve(described_class, obj: repository, args: args, ctx: { current_user: user }) }
+
+    context 'when unauthorized' do
+      it 'raises an exception' do
+        expect { resolve_blobs }.to raise_error(Gitlab::Graphql::Errors::ResourceNotAvailable)
+      end
+    end
+
+    context 'when authorized' do
+      before do
+        project.add_developer(user)
+      end
+
+      context 'using no filter' do
+        it 'returns nothing' do
+          is_expected.to be_empty
+        end
+      end
+
+      context 'using paths filter' do
+        let(:paths) { ['README.md'] }
+
+        it 'returns the specified blobs for HEAD' do
+          is_expected.to contain_exactly(have_attributes(path: 'README.md'))
+        end
+
+        context 'specifying a non-existent blob' do
+          let(:paths) { ['non-existent'] }
+
+          it 'returns nothing' do
+            is_expected.to be_empty
+          end
+        end
+
+        context 'specifying a different ref' do
+          let(:ref) { 'add-pdf-file' }
+          let(:paths) { ['files/pdf/test.pdf', 'README.md'] }
+
+          it 'returns the specified blobs for that ref' do
+            is_expected.to contain_exactly(
+              have_attributes(path: 'files/pdf/test.pdf'),
+              have_attributes(path: 'README.md')
+            )
+          end
+        end
+      end
+    end
+  end
+end
diff --git a/spec/graphql/types/repository_type_spec.rb b/spec/graphql/types/repository_type_spec.rb
index e9199bd286eb2f..a3bb7e502f203a 100644
--- a/spec/graphql/types/repository_type_spec.rb
+++ b/spec/graphql/types/repository_type_spec.rb
@@ -12,4 +12,6 @@
   specify { expect(described_class).to have_graphql_field(:tree) }
 
   specify { expect(described_class).to have_graphql_field(:exists, calls_gitaly?: true, complexity: 2) }
+
+  specify { expect(described_class).to have_graphql_field(:blobs) }
 end
-- 
GitLab