From d4564c7afc4dfc1372891a5fed1b7bdffd23f4a3 Mon Sep 17 00:00:00 2001
From: Andrey Marchenko <andrey.marchenko@datadoghq.com>
Date: Wed, 31 Jan 2024 14:34:10 +0100
Subject: [PATCH 1/6] Add Datadog::Core::Utils::Url module to filter sensitive
 info from urls Add Datadog::Core::Environment::Git module that provides
 git_repository_url and git_commit_sha from environment variables

---
 lib/datadog/core/environment/git.rb       | 27 +++++++++
 lib/datadog/core/git/ext.rb               | 25 +-------
 lib/datadog/core/utils/url.rb             | 17 ++++++
 sig/datadog/core/environment/git.rbs      | 15 +++++
 sig/datadog/core/git/ext.rbs              | 42 +-------------
 sig/datadog/core/utils/url.rbs            |  9 +++
 spec/datadog/core/environment/git_spec.rb | 29 ++++++++++
 spec/datadog/core/utils/url_spec.rb       | 69 +++++++++++++++++++++++
 8 files changed, 170 insertions(+), 63 deletions(-)
 create mode 100644 lib/datadog/core/environment/git.rb
 create mode 100644 lib/datadog/core/utils/url.rb
 create mode 100644 sig/datadog/core/environment/git.rbs
 create mode 100644 sig/datadog/core/utils/url.rbs
 create mode 100644 spec/datadog/core/environment/git_spec.rb
 create mode 100644 spec/datadog/core/utils/url_spec.rb

diff --git a/lib/datadog/core/environment/git.rb b/lib/datadog/core/environment/git.rb
new file mode 100644
index 0000000000..52cecaa0af
--- /dev/null
+++ b/lib/datadog/core/environment/git.rb
@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+require_relative '../git/ext'
+require_relative '../utils/url'
+
+module Datadog
+  module Core
+    module Environment
+      # Retrieves garbage collection statistics
+      module Git
+        module_function
+
+        def git_repository_url
+          return @git_repository_url if defined?(@git_repository_url)
+
+          @git_repository_url = Utils::Url.filter_sensitive_info(ENV[Datadog::Core::Git::Ext::ENV_REPOSITORY_URL])
+        end
+
+        def git_commit_sha
+          return @git_commit_sha if defined?(@git_commit_sha)
+
+          @git_commit_sha = ENV[Datadog::Core::Git::Ext::ENV_COMMIT_SHA]
+        end
+      end
+    end
+  end
+end
diff --git a/lib/datadog/core/git/ext.rb b/lib/datadog/core/git/ext.rb
index 636f4edb93..e2014841ee 100644
--- a/lib/datadog/core/git/ext.rb
+++ b/lib/datadog/core/git/ext.rb
@@ -5,32 +5,11 @@ module Core
     module Git
       # Defines constants for Git tags
       module Ext
-        GIT_SHA_LENGTH = 40
-
-        TAG_BRANCH = 'git.branch'
-        TAG_REPOSITORY_URL = 'git.repository_url'
-        TAG_TAG = 'git.tag'
-
-        TAG_COMMIT_AUTHOR_DATE = 'git.commit.author.date'
-        TAG_COMMIT_AUTHOR_EMAIL = 'git.commit.author.email'
-        TAG_COMMIT_AUTHOR_NAME = 'git.commit.author.name'
-        TAG_COMMIT_COMMITTER_DATE = 'git.commit.committer.date'
-        TAG_COMMIT_COMMITTER_EMAIL = 'git.commit.committer.email'
-        TAG_COMMIT_COMMITTER_NAME = 'git.commit.committer.name'
-        TAG_COMMIT_MESSAGE = 'git.commit.message'
-        TAG_COMMIT_SHA = 'git.commit.sha'
+        TAG_REPOSITORY_URL = '_dd.git.repository_url'
+        TAG_COMMIT_SHA = '_dd.git.commit.sha'
 
         ENV_REPOSITORY_URL = 'DD_GIT_REPOSITORY_URL'
         ENV_COMMIT_SHA = 'DD_GIT_COMMIT_SHA'
-        ENV_BRANCH = 'DD_GIT_BRANCH'
-        ENV_TAG = 'DD_GIT_TAG'
-        ENV_COMMIT_MESSAGE = 'DD_GIT_COMMIT_MESSAGE'
-        ENV_COMMIT_AUTHOR_NAME = 'DD_GIT_COMMIT_AUTHOR_NAME'
-        ENV_COMMIT_AUTHOR_EMAIL = 'DD_GIT_COMMIT_AUTHOR_EMAIL'
-        ENV_COMMIT_AUTHOR_DATE = 'DD_GIT_COMMIT_AUTHOR_DATE'
-        ENV_COMMIT_COMMITTER_NAME = 'DD_GIT_COMMIT_COMMITTER_NAME'
-        ENV_COMMIT_COMMITTER_EMAIL = 'DD_GIT_COMMIT_COMMITTER_EMAIL'
-        ENV_COMMIT_COMMITTER_DATE = 'DD_GIT_COMMIT_COMMITTER_DATE'
       end
     end
   end
diff --git a/lib/datadog/core/utils/url.rb b/lib/datadog/core/utils/url.rb
new file mode 100644
index 0000000000..18a069ecc3
--- /dev/null
+++ b/lib/datadog/core/utils/url.rb
@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+module Datadog
+  module Core
+    module Utils
+      # Helpers class that provides methods to proces URLs
+      # such as filtering sensitive information.
+      module Url
+        def self.filter_sensitive_info(url)
+          return nil if url.nil?
+
+          url.gsub(%r{((https?|ssh)://)[^/]*@}, '\1')
+        end
+      end
+    end
+  end
+end
diff --git a/sig/datadog/core/environment/git.rbs b/sig/datadog/core/environment/git.rbs
new file mode 100644
index 0000000000..e8ad16a915
--- /dev/null
+++ b/sig/datadog/core/environment/git.rbs
@@ -0,0 +1,15 @@
+module Datadog
+  module Core
+    module Environment
+      module Git
+        @git_repository_url: String?
+
+        @git_commit_sha: String?
+
+        def self?.git_repository_url: () -> String?
+
+        def self?.git_commit_sha: () -> String?
+      end
+    end
+  end
+end
diff --git a/sig/datadog/core/git/ext.rbs b/sig/datadog/core/git/ext.rbs
index 3d000c7f0c..239e669369 100644
--- a/sig/datadog/core/git/ext.rbs
+++ b/sig/datadog/core/git/ext.rbs
@@ -2,51 +2,13 @@ module Datadog
   module Core
     module Git
       module Ext
-        GIT_SHA_LENGTH: ::Regexp
+        TAG_REPOSITORY_URL: "_dd.git.repository_url"
 
-        TAG_BRANCH: "git.branch"
-
-        TAG_REPOSITORY_URL: "git.repository_url"
-
-        TAG_TAG: "git.tag"
-
-        TAG_COMMIT_AUTHOR_DATE: "git.commit.author.date"
-
-        TAG_COMMIT_AUTHOR_EMAIL: "git.commit.author.email"
-
-        TAG_COMMIT_AUTHOR_NAME: "git.commit.author.name"
-
-        TAG_COMMIT_COMMITTER_DATE: "git.commit.committer.date"
-
-        TAG_COMMIT_COMMITTER_EMAIL: "git.commit.committer.email"
-
-        TAG_COMMIT_COMMITTER_NAME: "git.commit.committer.name"
-
-        TAG_COMMIT_MESSAGE: "git.commit.message"
-
-        TAG_COMMIT_SHA: "git.commit.sha"
+        TAG_COMMIT_SHA: "_dd.git.commit.sha"
 
         ENV_REPOSITORY_URL: "DD_GIT_REPOSITORY_URL"
 
         ENV_COMMIT_SHA: "DD_GIT_COMMIT_SHA"
-
-        ENV_BRANCH: "DD_GIT_BRANCH"
-
-        ENV_TAG: "DD_GIT_TAG"
-
-        ENV_COMMIT_MESSAGE: "DD_GIT_COMMIT_MESSAGE"
-
-        ENV_COMMIT_AUTHOR_NAME: "DD_GIT_COMMIT_AUTHOR_NAME"
-
-        ENV_COMMIT_AUTHOR_EMAIL: "DD_GIT_COMMIT_AUTHOR_EMAIL"
-
-        ENV_COMMIT_AUTHOR_DATE: "DD_GIT_COMMIT_AUTHOR_DATE"
-
-        ENV_COMMIT_COMMITTER_NAME: "DD_GIT_COMMIT_COMMITTER_NAME"
-
-        ENV_COMMIT_COMMITTER_EMAIL: "DD_GIT_COMMIT_COMMITTER_EMAIL"
-
-        ENV_COMMIT_COMMITTER_DATE: "DD_GIT_COMMIT_COMMITTER_DATE"
       end
     end
   end
diff --git a/sig/datadog/core/utils/url.rbs b/sig/datadog/core/utils/url.rbs
new file mode 100644
index 0000000000..e0d46970d2
--- /dev/null
+++ b/sig/datadog/core/utils/url.rbs
@@ -0,0 +1,9 @@
+module Datadog
+  module Core
+    module Utils
+      module Url
+        def self?.filter_sensitive_info: (::String? url) -> ::String?
+      end
+    end
+  end
+end
diff --git a/spec/datadog/core/environment/git_spec.rb b/spec/datadog/core/environment/git_spec.rb
new file mode 100644
index 0000000000..5b8975f4fd
--- /dev/null
+++ b/spec/datadog/core/environment/git_spec.rb
@@ -0,0 +1,29 @@
+require 'spec_helper'
+
+require 'datadog/core/environment/git'
+
+RSpec.describe Datadog::Core::Environment::Git do
+  around do |example|
+    ClimateControl.modify(env_key => env_value) do
+      example.run
+    end
+  end
+
+  describe '.git_repository_url' do
+    subject { described_class.git_repository_url }
+
+    let(:env_key) { Datadog::Core::Git::Ext::ENV_REPOSITORY_URL }
+    let(:env_value) { 'https://gitlab-ci-token:AAA_bbb@gitlab.com/DataDog/systems-test.git' }
+
+    it { is_expected.to eq('https://gitlab.com/DataDog/systems-test.git') }
+  end
+
+  describe '.git_commit_sha' do
+    subject { described_class.git_commit_sha }
+
+    let(:env_key) { Datadog::Core::Git::Ext::ENV_COMMIT_SHA }
+    let(:env_value) { '1234567890abcdef' }
+
+    it { is_expected.to eq('1234567890abcdef') }
+  end
+end
diff --git a/spec/datadog/core/utils/url_spec.rb b/spec/datadog/core/utils/url_spec.rb
new file mode 100644
index 0000000000..867e587c72
--- /dev/null
+++ b/spec/datadog/core/utils/url_spec.rb
@@ -0,0 +1,69 @@
+require 'spec_helper'
+
+require 'datadog/core/utils/url'
+
+RSpec.describe Datadog::Core::Utils::Url do
+  describe '.filter_sensitive_info' do
+    subject(:filtered_url) { described_class.filter_sensitive_info(url) }
+
+    context 'with https' do
+      context 'with username and password' do
+        let(:url) { 'https://gitlab-ci-token:AAA_bbb@gitlab.com/DataDog/systems-test.git' }
+
+        it { is_expected.to eq('https://gitlab.com/DataDog/systems-test.git') }
+      end
+
+      context 'with username' do
+        let(:url) { 'https://token@gitlab.com/user/project.git' }
+
+        it { is_expected.to eq('https://gitlab.com/user/project.git') }
+      end
+
+      context 'without credentials' do
+        let(:url) { 'https://gitlab.com/DataDog/systems-test.git' }
+
+        it { is_expected.to eq('https://gitlab.com/DataDog/systems-test.git') }
+      end
+    end
+
+    context 'with ssh' do
+      context 'with username and password' do
+        let(:url) { 'ssh://gitlab-ci-token:AAA_bbb@gitlab.com/DataDog/systems-test.git' }
+
+        it { is_expected.to eq('ssh://gitlab.com/DataDog/systems-test.git') }
+      end
+
+      context 'with username' do
+        let(:url) { 'ssh://token@gitlab.com/user/project.git' }
+
+        it { is_expected.to eq('ssh://gitlab.com/user/project.git') }
+      end
+
+      context 'without credentials' do
+        let(:url) { 'ssh://gitlab.com/DataDog/systems-test.git' }
+
+        it { is_expected.to eq('ssh://gitlab.com/DataDog/systems-test.git') }
+      end
+    end
+
+    context 'without protocol' do
+      context 'without credentials' do
+        let(:url) { 'gitlab.com/DataDog/systems-test.git' }
+
+        it { is_expected.to eq('gitlab.com/DataDog/systems-test.git') }
+      end
+
+      context 'with credentials' do
+        let(:url) { 'git@github.com:user/project.git' }
+
+        it { is_expected.to eq('git@github.com:user/project.git') }
+      end
+    end
+
+    context 'with nil' do
+      let(:url) { nil }
+
+      it { is_expected.to be_nil }
+    end
+  end
+end

From 3c7eaf9641632e5e34e5f3ea19d34f1150d3c2e2 Mon Sep 17 00:00:00 2001
From: Andrey Marchenko <andrey.marchenko@datadoghq.com>
Date: Wed, 31 Jan 2024 16:34:50 +0100
Subject: [PATCH 2/6] [tracing] send source code integration tags in the first
 span of trace chunk

---
 lib/datadog/core/environment/git.rb           |  4 +-
 lib/datadog/core/utils/url.rb                 | 14 +++-
 .../tracing/transport/trace_formatter.rb      | 31 ++++++++
 sig/datadog/core/utils/url.rbs                |  2 +-
 .../tracing/transport/trace_formatter.rbs     | 10 +++
 spec/datadog/core/environment/git_spec.rb     | 18 +++++
 spec/datadog/core/utils/url_spec.rb           |  4 +-
 .../tracing/transport/trace_formatter_spec.rb | 76 ++++++++++++++++++-
 8 files changed, 150 insertions(+), 9 deletions(-)

diff --git a/lib/datadog/core/environment/git.rb b/lib/datadog/core/environment/git.rb
index 52cecaa0af..382633a201 100644
--- a/lib/datadog/core/environment/git.rb
+++ b/lib/datadog/core/environment/git.rb
@@ -6,14 +6,14 @@
 module Datadog
   module Core
     module Environment
-      # Retrieves garbage collection statistics
+      # Retrieves git repository information from environment variables
       module Git
         module_function
 
         def git_repository_url
           return @git_repository_url if defined?(@git_repository_url)
 
-          @git_repository_url = Utils::Url.filter_sensitive_info(ENV[Datadog::Core::Git::Ext::ENV_REPOSITORY_URL])
+          @git_repository_url = Utils::Url.filter_basic_auth(ENV[Datadog::Core::Git::Ext::ENV_REPOSITORY_URL])
         end
 
         def git_commit_sha
diff --git a/lib/datadog/core/utils/url.rb b/lib/datadog/core/utils/url.rb
index 18a069ecc3..d11c0c9688 100644
--- a/lib/datadog/core/utils/url.rb
+++ b/lib/datadog/core/utils/url.rb
@@ -1,15 +1,23 @@
 # frozen_string_literal: true
 
+require 'uri'
+
 module Datadog
   module Core
     module Utils
-      # Helpers class that provides methods to proces URLs
+      # Helpers class that provides methods to process URLs
       # such as filtering sensitive information.
       module Url
-        def self.filter_sensitive_info(url)
+        def self.filter_basic_auth(url)
           return nil if url.nil?
 
-          url.gsub(%r{((https?|ssh)://)[^/]*@}, '\1')
+          URI(url).tap do |u|
+            u.user = nil
+            u.password = nil
+          end.to_s
+        # Git scheme: git@github.com:DataDog/dd-trace-rb.git
+        rescue URI::InvalidURIError
+          url
         end
       end
     end
diff --git a/lib/datadog/tracing/transport/trace_formatter.rb b/lib/datadog/tracing/transport/trace_formatter.rb
index e03c66f78b..41a5e2a4c4 100644
--- a/lib/datadog/tracing/transport/trace_formatter.rb
+++ b/lib/datadog/tracing/transport/trace_formatter.rb
@@ -2,6 +2,8 @@
 
 require_relative '../../core/environment/identity'
 require_relative '../../core/environment/socket'
+require_relative '../../core/environment/git'
+require_relative '../../core/git/ext'
 require_relative '../../core/runtime/ext'
 require_relative '../metadata/ext'
 require_relative '../trace_segment'
@@ -13,6 +15,7 @@ module Transport
       class TraceFormatter
         attr_reader \
           :root_span,
+          :first_span,
           :trace
 
         def self.format!(trace)
@@ -22,6 +25,9 @@ def self.format!(trace)
         def initialize(trace)
           @trace = trace
           @root_span = find_root_span(trace)
+          # source code integration uses the "first span in trace chunk" concept instead of root span
+          # see: https://github.com/DataDog/dd-trace-rb/pull/3424
+          @first_span = trace.spans.first
         end
 
         # Modifies a trace so suitable for transport
@@ -54,6 +60,11 @@ def format!
           tag_sampling_priority!
           tag_profiling_enabled!
 
+          return trace unless first_span
+
+          tag_git_repository_url!
+          tag_git_commit_sha!
+
           trace
         end
 
@@ -185,6 +196,18 @@ def tag_profiling_enabled!
           )
         end
 
+        def tag_git_repository_url!
+          return if git_repository_url.nil?
+
+          first_span.set_tag(Core::Git::Ext::TAG_REPOSITORY_URL, git_repository_url)
+        end
+
+        def tag_git_commit_sha!
+          return if git_commit_sha.nil?
+
+          first_span.set_tag(Core::Git::Ext::TAG_COMMIT_SHA, git_commit_sha)
+        end
+
         private
 
         def partial?
@@ -203,6 +226,14 @@ def find_root_span(trace)
           # when root span is not found, fall back to last span (partial flush)
           root_span || trace.spans.last
         end
+
+        def git_repository_url
+          Core::Environment::Git.git_repository_url
+        end
+
+        def git_commit_sha
+          Core::Environment::Git.git_commit_sha
+        end
       end
     end
   end
diff --git a/sig/datadog/core/utils/url.rbs b/sig/datadog/core/utils/url.rbs
index e0d46970d2..51b651c6c8 100644
--- a/sig/datadog/core/utils/url.rbs
+++ b/sig/datadog/core/utils/url.rbs
@@ -2,7 +2,7 @@ module Datadog
   module Core
     module Utils
       module Url
-        def self?.filter_sensitive_info: (::String? url) -> ::String?
+        def self?.filter_basic_auth: (::String? url) -> ::String?
       end
     end
   end
diff --git a/sig/datadog/tracing/transport/trace_formatter.rbs b/sig/datadog/tracing/transport/trace_formatter.rbs
index 7f3604db7f..cb8e14c698 100644
--- a/sig/datadog/tracing/transport/trace_formatter.rbs
+++ b/sig/datadog/tracing/transport/trace_formatter.rbs
@@ -4,6 +4,8 @@ module Datadog
       class TraceFormatter
         attr_reader root_span: untyped
 
+        attr_reader first_span: untyped
+
         attr_reader trace: untyped
 
         def self.format!: (untyped trace) -> untyped
@@ -40,11 +42,19 @@ module Datadog
 
         def tag_high_order_trace_id!: () -> (nil | untyped)
 
+        def tag_git_repository_url!: () -> (nil | untyped)
+
+        def tag_git_commit_sha!: () -> (nil | untyped)
+
         private
 
         def partial?: () -> untyped
 
         def find_root_span: (untyped trace) -> untyped
+
+        def git_repository_url: () -> (nil | untyped)
+
+        def git_commit_sha: () -> (nil | untyped)
       end
     end
   end
diff --git a/spec/datadog/core/environment/git_spec.rb b/spec/datadog/core/environment/git_spec.rb
index 5b8975f4fd..ad5b7e2149 100644
--- a/spec/datadog/core/environment/git_spec.rb
+++ b/spec/datadog/core/environment/git_spec.rb
@@ -3,12 +3,30 @@
 require 'datadog/core/environment/git'
 
 RSpec.describe Datadog::Core::Environment::Git do
+  def remove_cached_variables
+    if described_class.instance_variable_defined?(:@git_repository_url)
+      described_class.remove_instance_variable(:@git_repository_url)
+    end
+
+    if described_class.instance_variable_defined?(:@git_commit_sha)
+      described_class.remove_instance_variable(:@git_commit_sha)
+    end
+  end
+
   around do |example|
     ClimateControl.modify(env_key => env_value) do
       example.run
     end
   end
 
+  before do
+    remove_cached_variables
+  end
+
+  after do
+    remove_cached_variables
+  end
+
   describe '.git_repository_url' do
     subject { described_class.git_repository_url }
 
diff --git a/spec/datadog/core/utils/url_spec.rb b/spec/datadog/core/utils/url_spec.rb
index 867e587c72..e09c667b20 100644
--- a/spec/datadog/core/utils/url_spec.rb
+++ b/spec/datadog/core/utils/url_spec.rb
@@ -3,8 +3,8 @@
 require 'datadog/core/utils/url'
 
 RSpec.describe Datadog::Core::Utils::Url do
-  describe '.filter_sensitive_info' do
-    subject(:filtered_url) { described_class.filter_sensitive_info(url) }
+  describe '.filter_basic_auth' do
+    subject(:filtered_url) { described_class.filter_basic_auth(url) }
 
     context 'with https' do
       context 'with username and password' do
diff --git a/spec/datadog/tracing/transport/trace_formatter_spec.rb b/spec/datadog/tracing/transport/trace_formatter_spec.rb
index 2a560d73d0..59eb2e51c9 100644
--- a/spec/datadog/tracing/transport/trace_formatter_spec.rb
+++ b/spec/datadog/tracing/transport/trace_formatter_spec.rb
@@ -66,10 +66,32 @@
     end
   end
 
+  shared_context 'git environment stub' do
+    before do
+      allow(Datadog::Core::Environment::Git).to receive(:git_repository_url).and_return(git_repository_url)
+      allow(Datadog::Core::Environment::Git).to receive(:git_commit_sha).and_return(git_commit_sha)
+    end
+  end
+
+  shared_context 'git metadata' do
+    let(:git_repository_url) { 'git_repository_url' }
+    let(:git_commit_sha) { 'git_commit_sha' }
+
+    include_context 'git environment stub'
+  end
+
+  shared_context 'no git metadata' do
+    let(:git_repository_url) { nil }
+    let(:git_commit_sha) { nil }
+
+    include_context 'git environment stub'
+  end
+
   shared_context 'no root span' do
     let(:trace) { Datadog::Tracing::TraceSegment.new(spans, **trace_options) }
     let(:spans) { Array.new(3) { Datadog::Tracing::Span.new('my.job') } }
     let(:root_span) { spans.last }
+    let(:first_span) { spans.first }
   end
 
   shared_context 'missing root span' do
@@ -82,12 +104,14 @@
     end
     let(:spans) { Array.new(3) { Datadog::Tracing::Span.new('my.job') } }
     let(:root_span) { spans.last }
+    let(:first_span) { spans.first }
   end
 
   shared_context 'available root span' do
     let(:trace) { Datadog::Tracing::TraceSegment.new(spans, root_span_id: root_span.id, **trace_options) }
     let(:spans) { Array.new(3) { Datadog::Tracing::Span.new('my.job') } }
     let(:root_span) { spans[1] }
+    let(:first_span) { spans.first }
   end
 
   describe '::new' do
@@ -96,7 +120,8 @@
         it do
           is_expected.to have_attributes(
             trace: trace,
-            root_span: root_span
+            root_span: root_span,
+            first_span: first_span
           )
         end
       end
@@ -194,6 +219,25 @@
         end
       end
 
+      shared_examples 'first span with no git metadata' do
+        it do
+          format!
+          expect(first_span.meta).to_not include('_dd.git.repository_url', '_dd.git.commit.sha')
+        end
+      end
+
+      shared_examples 'first span with git metadata' do
+        it do
+          format!
+          expect(first_span.meta).to include(
+            {
+              '_dd.git.repository_url' => git_repository_url,
+              '_dd.git.commit.sha' => git_commit_sha
+            }
+          )
+        end
+      end
+
       context 'with no root span' do
         include_context 'no root span'
 
@@ -231,6 +275,16 @@
           it_behaves_like 'root span with tags'
           it_behaves_like 'root span without generic tags'
         end
+
+        context 'with git metadata' do
+          include_context 'git metadata'
+          it_behaves_like 'first span with git metadata'
+        end
+
+        context 'without git metadata' do
+          include_context 'no git metadata'
+          it_behaves_like 'first span with no git metadata'
+        end
       end
 
       context 'with missing root span' do
@@ -270,6 +324,16 @@
           it_behaves_like 'root span with tags'
           it_behaves_like 'root span without generic tags'
         end
+
+        context 'with git metadata' do
+          include_context 'git metadata'
+          it_behaves_like 'first span with git metadata'
+        end
+
+        context 'without git metadata' do
+          include_context 'no git metadata'
+          it_behaves_like 'first span with no git metadata'
+        end
       end
 
       context 'with a root span' do
@@ -311,6 +375,16 @@
           it_behaves_like 'root span with tags'
           it_behaves_like 'root span with generic tags'
         end
+
+        context 'with git metadata' do
+          include_context 'git metadata'
+          it_behaves_like 'first span with git metadata'
+        end
+
+        context 'without git metadata' do
+          include_context 'no git metadata'
+          it_behaves_like 'first span with no git metadata'
+        end
       end
     end
   end

From d292844fc59fd705363acdb6616da21109f3fc7f Mon Sep 17 00:00:00 2001
From: Andrey Marchenko <andrey.marchenko@datadoghq.com>
Date: Thu, 15 Feb 2024 13:59:10 +0100
Subject: [PATCH 3/6] [profiling] add source code integration tags to
 TagBuilder

---
 lib/datadog/core/environment/git.rb        |  6 ++---
 lib/datadog/profiling/tag_builder.rb       |  7 ++++++
 spec/datadog/profiling/tag_builder_spec.rb | 28 ++++++++++++++++++++++
 3 files changed, 37 insertions(+), 4 deletions(-)

diff --git a/lib/datadog/core/environment/git.rb b/lib/datadog/core/environment/git.rb
index 382633a201..1b40a59983 100644
--- a/lib/datadog/core/environment/git.rb
+++ b/lib/datadog/core/environment/git.rb
@@ -8,15 +8,13 @@ module Core
     module Environment
       # Retrieves git repository information from environment variables
       module Git
-        module_function
-
-        def git_repository_url
+        def self.git_repository_url
           return @git_repository_url if defined?(@git_repository_url)
 
           @git_repository_url = Utils::Url.filter_basic_auth(ENV[Datadog::Core::Git::Ext::ENV_REPOSITORY_URL])
         end
 
-        def git_commit_sha
+        def self.git_commit_sha
           return @git_commit_sha if defined?(@git_commit_sha)
 
           @git_commit_sha = ENV[Datadog::Core::Git::Ext::ENV_COMMIT_SHA]
diff --git a/lib/datadog/profiling/tag_builder.rb b/lib/datadog/profiling/tag_builder.rb
index b90b382820..be10c2bddc 100644
--- a/lib/datadog/profiling/tag_builder.rb
+++ b/lib/datadog/profiling/tag_builder.rb
@@ -1,12 +1,15 @@
 # frozen_string_literal: true
 
 require_relative '../core/utils'
+require_relative '../core/environment/git'
+require_relative '../core/git/ext'
 
 module Datadog
   module Profiling
     # Builds a hash of default plus user tags to be included in a profile
     module TagBuilder
       include Datadog::Profiling::Ext::Transport::HTTP # Tag name constants
+      include Datadog::Core::Git::Ext # Git tags constants
 
       def self.call(
         settings:,
@@ -23,6 +26,8 @@ def self.call(
         runtime_id: Core::Environment::Identity.id,
         runtime_platform: Core::Environment::Identity.lang_platform,
         runtime_version: Core::Environment::Identity.lang_version,
+        git_repository_url: Core::Environment::Git.git_repository_url,
+        git_commit_sha: Core::Environment::Git.git_commit_sha,
         # User-provided tags
         user_tags: settings.tags
       )
@@ -42,6 +47,8 @@ def self.call(
         tags[FORM_FIELD_TAG_ENV] = env if env
         tags[FORM_FIELD_TAG_SERVICE] = service if service
         tags[FORM_FIELD_TAG_VERSION] = version if version
+        tags[TAG_REPOSITORY_URL] = git_repository_url if git_repository_url
+        tags[TAG_COMMIT_SHA] = git_commit_sha if git_commit_sha
 
         # Make sure everything is an utf-8 string, to avoid encoding issues in native code/libddprof/further downstream
         user_tags.merge(tags).map do |key, value|
diff --git a/spec/datadog/profiling/tag_builder_spec.rb b/spec/datadog/profiling/tag_builder_spec.rb
index 44dbfd491f..3b82a70d4f 100644
--- a/spec/datadog/profiling/tag_builder_spec.rb
+++ b/spec/datadog/profiling/tag_builder_spec.rb
@@ -78,5 +78,33 @@
         expect(result).to include('ascii-key' => 'ascii-value')
       end
     end
+
+    describe 'source code integration' do
+      context 'when git environment is available' do
+        before do
+          allow(Datadog::Core::Environment::Git).to receive(:git_repository_url).and_return(
+            'git_repository_url'
+          )
+          allow(Datadog::Core::Environment::Git).to receive(:git_commit_sha).and_return('git_commit_sha')
+        end
+
+        it 'includes the git repository URL and commit SHA' do
+          expect(call).to include(
+            '_dd.git.repository_url' => 'git_repository_url', '_dd.git.commit.sha' => 'git_commit_sha'
+          )
+        end
+      end
+
+      context 'when git environment is not available' do
+        before do
+          allow(Datadog::Core::Environment::Git).to receive(:git_repository_url).and_return(nil)
+          allow(Datadog::Core::Environment::Git).to receive(:git_commit_sha).and_return(nil)
+        end
+
+        it 'includes the git repository URL and commit SHA' do
+          expect(call).to_not include('_dd.git.repository_url', '_dd.git.commit.sha')
+        end
+      end
+    end
   end
 end

From 45fdf4b552da2f79173ff9bc6dd92a923e63f20d Mon Sep 17 00:00:00 2001
From: Andrey Marchenko <andrey.marchenko@datadoghq.com>
Date: Thu, 15 Feb 2024 17:41:37 +0100
Subject: [PATCH 4/6] [profiling] do not use _dd prefix for profiling tags

---
 lib/datadog/profiling/ext.rb               | 2 ++
 lib/datadog/profiling/tag_builder.rb       | 6 ++----
 sig/datadog/profiling/ext.rbs              | 2 ++
 spec/datadog/profiling/tag_builder_spec.rb | 4 ++--
 4 files changed, 8 insertions(+), 6 deletions(-)

diff --git a/lib/datadog/profiling/ext.rb b/lib/datadog/profiling/ext.rb
index 2122ec7a95..ca8380b0e8 100644
--- a/lib/datadog/profiling/ext.rb
+++ b/lib/datadog/profiling/ext.rb
@@ -23,6 +23,8 @@ module HTTP
           FORM_FIELD_TAG_RUNTIME_VERSION = 'runtime_version'
           FORM_FIELD_TAG_SERVICE = 'service'
           FORM_FIELD_TAG_VERSION = 'version'
+          FORM_FIELD_TAG_GIT_REPOSITORY_URL = 'git.repository_url'
+          FORM_FIELD_TAG_GIT_COMMIT_SHA = 'git.commit.sha'
 
           PPROF_DEFAULT_FILENAME = 'rubyprofile.pprof'
           CODE_PROVENANCE_FILENAME = 'code-provenance.json'
diff --git a/lib/datadog/profiling/tag_builder.rb b/lib/datadog/profiling/tag_builder.rb
index be10c2bddc..65ae026084 100644
--- a/lib/datadog/profiling/tag_builder.rb
+++ b/lib/datadog/profiling/tag_builder.rb
@@ -2,14 +2,12 @@
 
 require_relative '../core/utils'
 require_relative '../core/environment/git'
-require_relative '../core/git/ext'
 
 module Datadog
   module Profiling
     # Builds a hash of default plus user tags to be included in a profile
     module TagBuilder
       include Datadog::Profiling::Ext::Transport::HTTP # Tag name constants
-      include Datadog::Core::Git::Ext # Git tags constants
 
       def self.call(
         settings:,
@@ -47,8 +45,8 @@ def self.call(
         tags[FORM_FIELD_TAG_ENV] = env if env
         tags[FORM_FIELD_TAG_SERVICE] = service if service
         tags[FORM_FIELD_TAG_VERSION] = version if version
-        tags[TAG_REPOSITORY_URL] = git_repository_url if git_repository_url
-        tags[TAG_COMMIT_SHA] = git_commit_sha if git_commit_sha
+        tags[FORM_FIELD_TAG_GIT_REPOSITORY_URL] = git_repository_url if git_repository_url
+        tags[FORM_FIELD_TAG_GIT_COMMIT_SHA] = git_commit_sha if git_commit_sha
 
         # Make sure everything is an utf-8 string, to avoid encoding issues in native code/libddprof/further downstream
         user_tags.merge(tags).map do |key, value|
diff --git a/sig/datadog/profiling/ext.rbs b/sig/datadog/profiling/ext.rbs
index f68c2a377b..873d8e40a2 100644
--- a/sig/datadog/profiling/ext.rbs
+++ b/sig/datadog/profiling/ext.rbs
@@ -23,6 +23,8 @@ module Datadog
           FORM_FIELD_TAG_RUNTIME_VERSION: "runtime_version"
           FORM_FIELD_TAG_SERVICE: "service"
           FORM_FIELD_TAG_VERSION: "version"
+          FORM_FIELD_TAG_GIT_REPOSITORY_URL: 'git.repository_url'
+          FORM_FIELD_TAG_GIT_COMMIT_SHA: 'git.commit.sha'
           PPROF_DEFAULT_FILENAME: "rubyprofile.pprof"
           CODE_PROVENANCE_FILENAME: "code-provenance.json"
         end
diff --git a/spec/datadog/profiling/tag_builder_spec.rb b/spec/datadog/profiling/tag_builder_spec.rb
index 3b82a70d4f..148f0b26dc 100644
--- a/spec/datadog/profiling/tag_builder_spec.rb
+++ b/spec/datadog/profiling/tag_builder_spec.rb
@@ -90,7 +90,7 @@
 
         it 'includes the git repository URL and commit SHA' do
           expect(call).to include(
-            '_dd.git.repository_url' => 'git_repository_url', '_dd.git.commit.sha' => 'git_commit_sha'
+            'git.repository_url' => 'git_repository_url', 'git.commit.sha' => 'git_commit_sha'
           )
         end
       end
@@ -102,7 +102,7 @@
         end
 
         it 'includes the git repository URL and commit SHA' do
-          expect(call).to_not include('_dd.git.repository_url', '_dd.git.commit.sha')
+          expect(call).to_not include('git.repository_url', 'git.commit.sha')
         end
       end
     end

From 92a95d781fa55cbdb5fe10a307708390d99e2b7f Mon Sep 17 00:00:00 2001
From: Andrey Marchenko <andrey.marchenko@datadoghq.com>
Date: Fri, 16 Feb 2024 13:12:14 +0100
Subject: [PATCH 5/6] [profiling] remove FORM_FIELD_ prefix from git tags

---
 lib/datadog/profiling/ext.rb         | 4 ++--
 lib/datadog/profiling/tag_builder.rb | 4 ++--
 sig/datadog/profiling/ext.rbs        | 4 ++--
 3 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/lib/datadog/profiling/ext.rb b/lib/datadog/profiling/ext.rb
index ca8380b0e8..9dea808ddf 100644
--- a/lib/datadog/profiling/ext.rb
+++ b/lib/datadog/profiling/ext.rb
@@ -23,8 +23,8 @@ module HTTP
           FORM_FIELD_TAG_RUNTIME_VERSION = 'runtime_version'
           FORM_FIELD_TAG_SERVICE = 'service'
           FORM_FIELD_TAG_VERSION = 'version'
-          FORM_FIELD_TAG_GIT_REPOSITORY_URL = 'git.repository_url'
-          FORM_FIELD_TAG_GIT_COMMIT_SHA = 'git.commit.sha'
+          TAG_GIT_REPOSITORY_URL = 'git.repository_url'
+          TAG_GIT_COMMIT_SHA = 'git.commit.sha'
 
           PPROF_DEFAULT_FILENAME = 'rubyprofile.pprof'
           CODE_PROVENANCE_FILENAME = 'code-provenance.json'
diff --git a/lib/datadog/profiling/tag_builder.rb b/lib/datadog/profiling/tag_builder.rb
index 65ae026084..2325a7b93f 100644
--- a/lib/datadog/profiling/tag_builder.rb
+++ b/lib/datadog/profiling/tag_builder.rb
@@ -45,8 +45,8 @@ def self.call(
         tags[FORM_FIELD_TAG_ENV] = env if env
         tags[FORM_FIELD_TAG_SERVICE] = service if service
         tags[FORM_FIELD_TAG_VERSION] = version if version
-        tags[FORM_FIELD_TAG_GIT_REPOSITORY_URL] = git_repository_url if git_repository_url
-        tags[FORM_FIELD_TAG_GIT_COMMIT_SHA] = git_commit_sha if git_commit_sha
+        tags[TAG_GIT_REPOSITORY_URL] = git_repository_url if git_repository_url
+        tags[TAG_GIT_COMMIT_SHA] = git_commit_sha if git_commit_sha
 
         # Make sure everything is an utf-8 string, to avoid encoding issues in native code/libddprof/further downstream
         user_tags.merge(tags).map do |key, value|
diff --git a/sig/datadog/profiling/ext.rbs b/sig/datadog/profiling/ext.rbs
index 873d8e40a2..1ad36590f8 100644
--- a/sig/datadog/profiling/ext.rbs
+++ b/sig/datadog/profiling/ext.rbs
@@ -23,8 +23,8 @@ module Datadog
           FORM_FIELD_TAG_RUNTIME_VERSION: "runtime_version"
           FORM_FIELD_TAG_SERVICE: "service"
           FORM_FIELD_TAG_VERSION: "version"
-          FORM_FIELD_TAG_GIT_REPOSITORY_URL: 'git.repository_url'
-          FORM_FIELD_TAG_GIT_COMMIT_SHA: 'git.commit.sha'
+          TAG_GIT_REPOSITORY_URL: "git.repository_url"
+          TAG_GIT_COMMIT_SHA: "git.commit.sha"
           PPROF_DEFAULT_FILENAME: "rubyprofile.pprof"
           CODE_PROVENANCE_FILENAME: "code-provenance.json"
         end

From 79079ec11ef086ca1846fda67e9eb8f9716ddc8e Mon Sep 17 00:00:00 2001
From: Andrey Marchenko <andrey.marchenko@datadoghq.com>
Date: Fri, 16 Feb 2024 15:34:04 +0100
Subject: [PATCH 6/6] addressing minor code review comments

---
 lib/datadog/tracing/transport/trace_formatter.rb       | 8 ++++----
 spec/datadog/tracing/transport/trace_formatter_spec.rb | 3 +--
 2 files changed, 5 insertions(+), 6 deletions(-)

diff --git a/lib/datadog/tracing/transport/trace_formatter.rb b/lib/datadog/tracing/transport/trace_formatter.rb
index 41a5e2a4c4..2140ccebdd 100644
--- a/lib/datadog/tracing/transport/trace_formatter.rb
+++ b/lib/datadog/tracing/transport/trace_formatter.rb
@@ -60,10 +60,10 @@ def format!
           tag_sampling_priority!
           tag_profiling_enabled!
 
-          return trace unless first_span
-
-          tag_git_repository_url!
-          tag_git_commit_sha!
+          if first_span
+            tag_git_repository_url!
+            tag_git_commit_sha!
+          end
 
           trace
         end
diff --git a/spec/datadog/tracing/transport/trace_formatter_spec.rb b/spec/datadog/tracing/transport/trace_formatter_spec.rb
index 59eb2e51c9..f2bd23d69a 100644
--- a/spec/datadog/tracing/transport/trace_formatter_spec.rb
+++ b/spec/datadog/tracing/transport/trace_formatter_spec.rb
@@ -120,8 +120,7 @@
         it do
           is_expected.to have_attributes(
             trace: trace,
-            root_span: root_span,
-            first_span: first_span
+            root_span: root_span
           )
         end
       end