[CIVIS-2863] Git repository unshallowing logic

lib/datadog/ci/git/local_repository.rb

@@ -159,17 +159,34 @@ def self.git_generate_packfiles(included_commits:, excluded_commits:, path:)
           nil
         end
 
+        def self.git_shallow_clone?
+          exec_git_command("git rev-parse --is-shallow-repository") == "true"
+        rescue => e
+          log_failure(e, "git shallow clone")
+          false
+        end
+
+        def self.git_unshallow
+          exec_git_command(
+            "git fetch " \
+            "--shallow-since=\"1 month ago\" " \
+            "--update-shallow " \
+            "--filter=\"blob:none\" " \
+            "--recurse-submodules=no " \
+            "$(git config --default origin --get clone.defaultRemoteName) $(git rev-parse HEAD)"
+          )
+        rescue => e
+          log_failure(e, "git unshallow")
+          nil
+        end
+
         # makes .exec_git_command private to make sure that this method
         # is not called from outside of this module with insecure parameters
         class << self
           private
 
           def filter_invalid_commits(commits)
-            commits.filter_map do |commit|
-              next unless Utils::Git.valid_commit_sha?(commit)
-
-              commit
-            end
+            commits.filter { |commit| Utils::Git.valid_commit_sha?(commit) }
           end
 
           def exec_git_command(cmd, stdin: nil)

lib/datadog/ci/git/tree_uploader.rb

@@ -26,8 +26,6 @@ def call(repository_url)
 
           Datadog.logger.debug { "Uploading git tree for repository #{repository_url}" }
 
-          # 2. Check if the repository clone is shallow and unshallow if appropriate
-          # TO BE ADDED IN CIVIS-2863
           latest_commits = LocalRepository.git_commits
           head_commit = latest_commits&.first
           if head_commit.nil?
@@ -36,23 +34,36 @@ def call(repository_url)
           end
 
           begin
-            excluded_commits, included_commits = split_known_commits(repository_url, latest_commits)
-            if included_commits.empty?
+            # ask the backend for the list of commits it already has
+            known_commits, new_commits = fetch_known_commits_and_split(repository_url, latest_commits)
+            # if all commits are present in the backend, we don't need to upload anything
+            if new_commits.empty?
               Datadog.logger.debug("No new commits to upload")
               return
             end
+
+            # quite often we deal with shallow clones in CI environment
+            if LocalRepository.git_shallow_clone? && LocalRepository.git_unshallow
+              Datadog.logger.debug("Detected shallow clone and unshallowed the repository, repeating commits search")
+
+              # re-run the search with the updated commit list after unshallowing
+              known_commits, new_commits = fetch_known_commits_and_split(
+                repository_url,
+                LocalRepository.git_commits
+              )
+            end
           rescue SearchCommits::ApiError => e
             Datadog.logger.debug("SearchCommits failed with #{e}, aborting git upload")
             return
           end
 
-          Datadog.logger.debug { "Uploading packfiles for commits: #{included_commits}" }
+          Datadog.logger.debug { "Uploading packfiles for commits: #{new_commits}" }
           uploader = UploadPackfile.new(
             api: api,
             head_commit_sha: head_commit,
             repository_url: repository_url
           )
-          Packfiles.generate(included_commits: included_commits, excluded_commits: excluded_commits) do |filepath|
+          Packfiles.generate(included_commits: new_commits, excluded_commits: known_commits) do |filepath|
             uploader.call(filepath: filepath)
           rescue UploadPackfile::ApiError => e
             Datadog.logger.debug("Packfile upload failed with #{e}")
@@ -62,7 +73,9 @@ def call(repository_url)
 
         private
 
-        def split_known_commits(repository_url, latest_commits)
+        # Split the latest commits list into known and new commits
+        # based on the backend response provided by /search_commits endpoint
+        def fetch_known_commits_and_split(repository_url, latest_commits)
           Datadog.logger.debug { "Checking the latest commits list with backend: #{latest_commits}" }
           backend_commits = SearchCommits.new(api: api).call(repository_url, latest_commits)
           latest_commits.partition do |commit|

sig/datadog/ci/git/local_repository.rbs

@@ -33,6 +33,10 @@ module Datadog
 
         def self.git_generate_packfiles: (included_commits: Enumerable[String], excluded_commits: Enumerable[String], path: String) -> String?
 
+        def self.git_shallow_clone?: () -> bool
+
+        def self.git_unshallow: () -> String?
+
         private
 
         def self.filter_invalid_commits: (Enumerable[String] commits) -> Array[String]

sig/datadog/ci/git/tree_uploader.rbs

@@ -11,7 +11,7 @@ module Datadog
 
         private
 
-        def split_known_commits: (String repository_url, Array[String] latest_commits) -> [Array[String], Array[String]]
+        def fetch_known_commits_and_split: (String repository_url, Array[String] latest_commits) -> [Array[String], Array[String]]
       end
     end
   end

spec/datadog/ci/git/local_repository_spec.rb

@@ -293,4 +293,82 @@ def with_custom_git_environment
       it { is_expected.to eq("first-tag") }
     end
   end
+
+  context "with shallow clone" do
+    let(:tmpdir) { Dir.mktmpdir }
+    after { FileUtils.remove_entry(tmpdir) }
+
+    before do
+      # shallow clone datadog-ci-rb repository
+      `cd #{tmpdir} && git clone --depth 1 https://github.com/DataDog/datadog-ci-rb`
+    end
+
+    def with_shallow_clone_git_dir
+      ClimateControl.modify("GIT_DIR" => File.join(tmpdir, "datadog-ci-rb/.git")) do
+        yield
+      end
+    end
+
+    describe ".git_shallow_clone?" do
+      subject do
+        with_shallow_clone_git_dir { described_class.git_shallow_clone? }
+      end
+
+      it { is_expected.to be_truthy }
+    end
+
+    describe ".git_commits" do
+      subject do
+        with_shallow_clone_git_dir { described_class.git_commits }
+      end
+
+      it "returns a list of single git commit sha" do
+        expect(subject).to be_kind_of(Array)
+        expect(subject).not_to be_empty
+        expect(subject).to have(1).item
+        expect(subject.first).to match(/^\h{40}$/)
+      end
+    end
+
+    describe ".git_unshallow" do
+      # skip for jruby for now - old git version DD docker image
+      before { skip if PlatformHelpers.jruby? }
+
+      subject do
+        with_shallow_clone_git_dir { described_class.git_unshallow }
+      end
+      let(:commits) do
+        with_shallow_clone_git_dir { described_class.git_commits }
+      end
+
+      it "unshallows the repository" do
+        expect(subject).to be_truthy
+        expect(commits.size).to be > 1
+      end
+    end
+  end
+
+  context "with full clone" do
+    let(:tmpdir) { Dir.mktmpdir }
+    after { FileUtils.remove_entry(tmpdir) }
+
+    before do
+      # shallow clone datadog-ci-rb repository
+      `cd #{tmpdir} && git clone https://github.com/DataDog/datadog-ci-rb`
+    end
+
+    def with_full_clone_git_dir
+      ClimateControl.modify("GIT_DIR" => File.join(tmpdir, "datadog-ci-rb/.git")) do
+        yield
+      end
+    end
+
+    describe ".git_shallow_clone?" do
+      subject do
+        with_full_clone_git_dir { described_class.git_shallow_clone? }
+      end
+
+      it { is_expected.to be_falsey }
+    end
+  end
 end

spec/datadog/ci/git/tree_uploader_spec.rb

@@ -19,7 +19,6 @@
     let(:search_commits) { double("search_commits", call: backend_commits) }
 
     before do
-      allow(Datadog::CI::Git::LocalRepository).to receive(:git_commits).and_return(latest_commits)
       allow(Datadog::CI::Git::SearchCommits).to receive(:new).with(api: api).and_return(search_commits)
     end
 
@@ -33,73 +32,122 @@
       end
     end
 
-    context "when the latest commits list is empty" do
-      let(:latest_commits) { [] }
-
-      it "logs a debug message and aborts the git upload" do
-        expect(Datadog.logger).to receive(:debug).with("Got empty latest commits list, aborting git upload")
-
-        tree_uploader.call(repository_url)
-      end
-    end
-
-    context "when the backend commits search fails" do
+    context "when API is configured" do
       before do
-        expect(search_commits).to receive(:call).and_raise(Datadog::CI::Git::SearchCommits::ApiError, "test error")
+        expect(Datadog::CI::Git::LocalRepository).to receive(:git_commits).and_return(latest_commits)
       end
 
-      it "logs a debug message and aborts the git upload" do
-        expect(Datadog.logger).to receive(:debug).with("SearchCommits failed with test error, aborting git upload")
-
-        tree_uploader.call(repository_url)
-      end
-    end
-
-    context "when all commits are known to the backend" do
-      let(:backend_commits) { latest_commits }
-
-      it "logs a debug message and aborts the git upload" do
-        expect(Datadog.logger).to receive(:debug).with("No new commits to upload")
-
-        tree_uploader.call(repository_url)
-      end
-    end
+      context "when the latest commits list is empty" do
+        let(:latest_commits) { [] }
 
-    context "when some commits are new" do
-      let(:upload_packfile) { double("upload_packfile", call: nil) }
+        it "logs a debug message and aborts the git upload" do
+          expect(Datadog.logger).to receive(:debug).with("Got empty latest commits list, aborting git upload")
 
-      before do
-        expect(Datadog::CI::Git::Packfiles).to receive(:generate).with(
-          included_commits: latest_commits - backend_commits.to_a,
-          excluded_commits: backend_commits
-        ).and_yield("packfile_path")
-
-        expect(Datadog::CI::Git::UploadPackfile).to receive(:new).with(
-          api: api,
-          head_commit_sha: head_commit,
-          repository_url: repository_url
-        ).and_return(upload_packfile)
+          tree_uploader.call(repository_url)
+        end
       end
 
-      context "when the packfile upload fails" do
+      context "when the backend commits search fails" do
         before do
-          expect(upload_packfile).to receive(:call).and_raise(Datadog::CI::Git::UploadPackfile::ApiError, "test error")
+          expect(search_commits).to receive(:call).and_raise(Datadog::CI::Git::SearchCommits::ApiError, "test error")
         end
 
         it "logs a debug message and aborts the git upload" do
-          expect(Datadog.logger).to receive(:debug).with("Packfile upload failed with test error")
+          expect(Datadog.logger).to receive(:debug).with("SearchCommits failed with test error, aborting git upload")
 
           tree_uploader.call(repository_url)
         end
       end
 
-      context "when the packfile upload succeeds" do
-        it "uploads the new commits" do
-          expect(upload_packfile).to receive(:call).with(filepath: "packfile_path").and_return(nil)
+      context "when all commits are known to the backend" do
+        let(:backend_commits) { latest_commits }
+
+        it "logs a debug message and aborts the git upload" do
+          expect(Datadog.logger).to receive(:debug).with("No new commits to upload")
 
           tree_uploader.call(repository_url)
         end
       end
+
+      context "when some commits are new" do
+        let(:upload_packfile) { double("upload_packfile", call: nil) }
+
+        context "when the repository is shallow cloned" do
+          before do
+            expect(Datadog::CI::Git::LocalRepository).to receive(:git_shallow_clone?).and_return(true)
+          end
+
+          context "when the unshallowing fails" do
+            before do
+              expect(Datadog::CI::Git::LocalRepository).to receive(:git_unshallow).and_return(nil)
+            end
+
+            it "uploads what we can upload" do
+              expect(Datadog::CI::Git::Packfiles).to receive(:generate).with(
+                included_commits: %w[13c988d4f15e06bcdd0b0af290086a3079cdadb0],
+                excluded_commits: backend_commits
+              ).and_yield("packfile_path")
+
+              tree_uploader.call(repository_url)
+            end
+          end
+
+          context "when the unshallowing succeeds" do
+            before do
+              expect(Datadog::CI::Git::LocalRepository).to receive(:git_unshallow).and_return("unshallow_result")
+              expect(Datadog::CI::Git::LocalRepository).to receive(:git_commits).and_return(
+                latest_commits + %w[782d09e3fbfd8cf1b5c13f3eb9621362f9089ed5]
+              )
+            end
+
+            it "uploads the new commits" do
+              expect(Datadog::CI::Git::Packfiles).to receive(:generate).with(
+                included_commits: %w[13c988d4f15e06bcdd0b0af290086a3079cdadb0 782d09e3fbfd8cf1b5c13f3eb9621362f9089ed5],
+                excluded_commits: backend_commits
+              ).and_yield("packfile_path")
+
+              tree_uploader.call(repository_url)
+            end
+          end
+        end
+
+        context "when the repository is not shallow cloned" do
+          before do
+            expect(Datadog::CI::Git::LocalRepository).to receive(:git_shallow_clone?).and_return(false)
+
+            expect(Datadog::CI::Git::Packfiles).to receive(:generate).with(
+              included_commits: latest_commits - backend_commits.to_a,
+              excluded_commits: backend_commits
+            ).and_yield("packfile_path")
+
+            expect(Datadog::CI::Git::UploadPackfile).to receive(:new).with(
+              api: api,
+              head_commit_sha: head_commit,
+              repository_url: repository_url
+            ).and_return(upload_packfile)
+          end
+
+          context "when the packfile upload fails" do
+            before do
+              expect(upload_packfile).to receive(:call).and_raise(Datadog::CI::Git::UploadPackfile::ApiError, "test error")
+            end
+
+            it "logs a debug message and aborts the git upload" do
+              expect(Datadog.logger).to receive(:debug).with("Packfile upload failed with test error")
+
+              tree_uploader.call(repository_url)
+            end
+          end
+
+          context "when the packfile upload succeeds" do
+            it "uploads the new commits" do
+              expect(upload_packfile).to receive(:call).with(filepath: "packfile_path").and_return(nil)
+
+              tree_uploader.call(repository_url)
+            end
+          end
+        end
+      end
     end
   end
 end