datadog-ci 1.18.0 → 1.20.0

data/lib/datadog/ci/ext/environment/providers/gitlab.rb

@@ -104,14 +104,14 @@ module Datadog
               env["CI_MERGE_REQUEST_TARGET_BRANCH_NAME"]
             end
 
-            def git_pull_request_base_branch_sha
-              env["CI_MERGE_REQUEST_TARGET_BRANCH_SHA"]
-            end
-
             def git_commit_head_sha
               env["CI_MERGE_REQUEST_SOURCE_BRANCH_SHA"]
             end
 
+            def pr_number
+              env["CI_MERGE_REQUEST_IID"]
+            end
+
             private
 
             def extract_name_email

data/lib/datadog/ci/ext/environment/providers/jenkins.rb

@@ -65,6 +65,14 @@ module Datadog
               env["GIT_BRANCH"]
             end
 
+            def git_pull_request_base_branch
+              env["CHANGE_TARGET"]
+            end
+
+            def pr_number
+              env["CHANGE_ID"]
+            end
+
             def ci_env_vars
               {
                 "DD_CUSTOM_TRACE_ID" => env["DD_CUSTOM_TRACE_ID"]

data/lib/datadog/ci/ext/environment/providers/teamcity.rb

@@ -25,6 +25,14 @@ module Datadog
             def job_url
               env["BUILD_URL"]
             end
+
+            def git_pull_request_base_branch
+              env["TEAMCITY_PULLREQUEST_TARGET_BRANCH"]
+            end
+
+            def pr_number
+              env["TEAMCITY_PULLREQUEST_NUMBER"]
+            end
           end
         end
       end

data/lib/datadog/ci/ext/environment/providers/travis.rb

@@ -61,6 +61,28 @@ module Datadog
             def git_commit_message
               env["TRAVIS_COMMIT_MESSAGE"]
             end
+
+            def git_pull_request_base_branch
+              return nil unless pull_request?
+
+              env["TRAVIS_BRANCH"]
+            end
+
+            def git_commit_head_sha
+              return nil unless pull_request?
+
+              env["TRAVIS_PULL_REQUEST_SHA"]
+            end
+
+            def pr_number
+              return nil unless pull_request?
+
+              env["TRAVIS_PULL_REQUEST"]
+            end
+
+            def pull_request?
+              env["TRAVIS_EVENT_TYPE"] == "pull_request"
+            end
           end
         end
       end

data/lib/datadog/ci/ext/environment/providers.rb

@@ -10,6 +10,7 @@ require_relative "providers/buddy"
 require_relative "providers/buildkite"
 require_relative "providers/circleci"
 require_relative "providers/codefresh"
+require_relative "providers/drone"
 require_relative "providers/github_actions"
 require_relative "providers/gitlab"
 require_relative "providers/jenkins"
@@ -34,6 +35,7 @@ module Datadog
             Providers::Buildkite,
             Providers::Circleci,
             Providers::Codefresh,
+            Providers::Drone,
             Providers::GithubActions,
             Providers::Gitlab,
             Providers::Jenkins,

data/lib/datadog/ci/ext/environment.rb

@@ -3,6 +3,7 @@
 require "datadog/core/telemetry/logging"
 
 require_relative "git"
+require_relative "environment/configuration_discrepancy_checker"
 require_relative "environment/extractor"
 
 require_relative "../utils/git"
@@ -24,6 +25,7 @@ module Datadog
         TAG_NODE_LABELS = "ci.node.labels"
         TAG_NODE_NAME = "ci.node.name"
         TAG_CI_ENV_VARS = "_dd.ci.env_vars"
+        TAG_PR_NUMBER = "pr.number"
 
         module Provider
           APPVEYOR = "appveyor"
@@ -35,6 +37,7 @@ module Datadog
           BUILDKITE = "buildkite"
           CIRCLECI = "circleci"
           CODEFRESH = "codefresh"
+          DRONE = "drone"
           GITHUB = "github"
           GITLAB = "gitlab"
           JENKINS = "jenkins"
@@ -51,9 +54,8 @@ module Datadog
           tags = Environment::Extractor.new(env).tags
 
           # If user defined metadata is defined, overwrite
-          tags.merge!(
-            Environment::Extractor.new(env, provider_klass: Providers::UserDefinedTags).tags
-          )
+          user_provided_tags = Environment::Extractor.new(env, provider_klass: Providers::UserDefinedTags).tags
+          tags.merge!(user_provided_tags)
 
           # Fill out tags from local git as fallback
           local_git_tags = Environment::Extractor.new(env, provider_klass: Providers::LocalGit).tags
@@ -61,6 +63,10 @@ module Datadog
             tags[key] ||= value
           end
 
+          # send some telemetry for the cases where git commit sha is overriden
+          discrepancy_checker = ConfigurationDiscrepancyChecker.new(tags, local_git_tags, user_provided_tags)
+          discrepancy_checker.check_for_discrepancies
+
           ensure_post_conditions(tags)
 
           tags

data/lib/datadog/ci/ext/telemetry.rb

@@ -68,6 +68,9 @@ module Datadog
 
         METRIC_TEST_SESSION = "test_session"
 
+        METRIC_GIT_COMMIT_SHA_MATCH = "git.commit_sha_match"
+        METRIC_GIT_COMMIT_SHA_DISCREPANCY = "git.commit_sha_discrepancy"
+
         TAG_TEST_FRAMEWORK = "test_framework"
         TAG_EVENT_TYPE = "event_type"
         TAG_HAS_CODEOWNER = "has_codeowner"

data/lib/datadog/ci/ext/test.rb

@@ -159,6 +159,12 @@ module Datadog
           RETRY_FAILED = "auto_test_retries"
           RETRY_FLAKY_FIXED = "attempt_to_fix"
           RETRY_EXTERNAL = "external"
+
+          DATADOG_RETRY_REASONS = [
+            RETRY_FAILED,
+            RETRY_DETECT_FLAKY,
+            RETRY_FLAKY_FIXED
+          ]
         end
 
         # possible reasons why a test was skipped

data/lib/datadog/ci/git/changed_lines.rb

@@ -0,0 +1,109 @@
+# frozen_string_literal: true
+
+module Datadog
+  module CI
+    module Git
+      # Helper class to efficiently store and query changed line intervals for a single file
+      # Uses merged sorted intervals with binary search for O(log n) query performance
+      class ChangedLines
+        def initialize
+          @intervals = [] # Array of [start, end] pairs
+          @built = false
+          @mutex = Mutex.new
+        end
+
+        # Add an interval (defers merging until build! is called)
+        def add_interval(start_line, end_line)
+          return if start_line > end_line
+
+          @mutex.synchronize do
+            @intervals << [start_line, end_line]
+            @built = false
+          end
+        end
+
+        # Sort and merge all intervals
+        # Call this after all intervals have been added
+        def build!
+          @mutex.synchronize do
+            return false if @built
+
+            @built = true
+            return false if @intervals.empty?
+
+            # Sort intervals by start line
+            @intervals.sort_by!(&:first)
+
+            # Merge overlapping intervals
+            merged = []
+
+            # @type var current_start: Integer
+            # @type var current_end: Integer
+            current_start, current_end = @intervals.first
+
+            @intervals.each_with_index do |interval, index|
+              next if index == 0
+              # @type var start_line: Integer
+              # @type var end_line: Integer
+              start_line, end_line = interval
+
+              if start_line <= current_end + 1
+                # Overlapping or adjacent intervals, merge them
+                current_end = [current_end, end_line].max
+              else
+                # Non-overlapping interval, save current and start new
+                merged << [current_start, current_end]
+                current_start = start_line
+                current_end = end_line
+              end
+            end
+
+            merged << [current_start, current_end]
+
+            @intervals = merged
+            true
+          end
+        end
+
+        # Check if any line in the query interval overlaps with changed lines
+        # Uses binary search for O(log n) performance
+        def overlaps?(query_start, query_end)
+          build! unless @built
+
+          return false if @intervals.empty? || query_start > query_end
+
+          # Binary search for the first interval that might overlap
+          left = 0
+          right = @intervals.length - 1
+
+          while left <= right
+            mid = (left + right) / 2
+            # @type var interval_start: Integer
+            # @type var interval_end: Integer
+            interval_start, interval_end = @intervals[mid]
+
+            if interval_end < query_start
+              left = mid + 1
+            elsif interval_start > query_end
+              right = mid - 1
+            else
+              # Found overlap
+              return true
+            end
+          end
+
+          false
+        end
+
+        def empty?
+          @intervals.empty?
+        end
+
+        def intervals
+          build! unless @built
+          @intervals.dup
+        end
+      end
+    end
+  end
+end

data/lib/datadog/ci/git/diff.rb

@@ -0,0 +1,90 @@
+# frozen_string_literal: true
+
+require "set"
+require_relative "changed_lines"
+
+module Datadog
+  module CI
+    module Git
+      class Diff
+        FILE_CHANGE_REGEX = /^diff --git a\/(?<file>.+?) b\//.freeze
+        LINES_CHANGE_REGEX = /^@@ -\d+(?:,\d+)? \+(?<start>\d+)(?:,(?<count>\d+))? @@/.freeze
+
+        def initialize(changed_files: {})
+          @changed_files = changed_files # Hash of file_path => ChangedLines
+        end
+
+        # Check if any lines in the given range are changed for the specified file
+        def lines_changed?(file_path, start_line: nil, end_line: nil)
+          changed_lines = @changed_files[file_path]
+          unless changed_lines
+            Datadog.logger.debug { "No changes found for file: #{file_path}" }
+            return false
+          end
+
+          # If either start_line or end_line is nil, return true if file is present
+          return true if start_line.nil? || end_line.nil?
+
+          changed_lines.overlaps?(start_line, end_line)
+        end
+
+        def empty?
+          @changed_files.empty?
+        end
+
+        # for debug purposes
+        def size
+          @changed_files.size
+        end
+
+        # for debug purposes
+        def inspect
+          @changed_files.inspect
+        end
+
+        def self.parse_diff_output(output)
+          return new if output.nil? || output.empty?
+
+          changed_files = {}
+          current_file = nil
+
+          output.each_line do |line|
+            # Match lines like: diff --git a/foo/bar.rb b/foo/bar.rb
+            # This captures git changes on file level
+            match = FILE_CHANGE_REGEX.match(line)
+            if match && match[:file]
+              # this path here is already relative from the git root
+              changed_file = match[:file]
+
+              unless changed_file.nil? || changed_file.empty?
+                current_file = changed_file
+                changed_files[current_file] ||= ChangedLines.new
+              end
+
+              Datadog.logger.debug { "matched changed_file: #{changed_file} from git diff line: #{line}" }
+
+              next
+            end
+
+            # Match lines like: @@ -1,2 +3,4 @@
+            match = LINES_CHANGE_REGEX.match(line)
+            if match && match[:start] && current_file
+              start_line = match[:start].to_i
+
+              line_count = 1 # Default to 1 line if count not specified
+              line_count = match[:count].to_i if match[:count]
+
+              end_line = start_line + line_count - 1
+
+              changed_files[current_file].add_interval(start_line, end_line)
+
+              Datadog.logger.debug { "Added interval [#{start_line}, #{end_line}] for file: #{current_file}" }
+            end
+          end
+
+          new(changed_files: changed_files)
+        end
+      end
+    end
+  end
+end

data/lib/datadog/ci/git/local_repository.rb

@@ -8,6 +8,7 @@ require_relative "../ext/telemetry"
 require_relative "../utils/command"
 require_relative "base_branch_sha_detector"
 require_relative "cli"
+require_relative "diff"
 require_relative "telemetry"
 require_relative "user"
 
@@ -37,14 +38,22 @@ module Datadog
             # that root is a prefix of the path
             return "" if path.size < prefix_index
 
-            prefix_index += 1 if path[prefix_index] == File::SEPARATOR
-            res = path[prefix_index..]
+            # this means that the root is not a prefix of this path somehow
+            return "" if path[prefix_index] != File::SEPARATOR
+
+            res = path[prefix_index + 1..]
           else
             # prefix_to_root is a difference between the root path and the given path
-            if @prefix_to_root == ""
-              return path
-            elsif @prefix_to_root
-              return File.join(@prefix_to_root, path)
+            if defined?(@prefix_to_root)
+              # if path starts with ./ remove the dot before applying the optimization
+              # @type var path: String
+              path = path[1..] if path.start_with?("./")
+
+              if @prefix_to_root == ""
+                return path
+              elsif @prefix_to_root
+                return File.join(@prefix_to_root, path)
+              end
             end
 
             pathname = Pathname.new(File.expand_path(path))
@@ -308,10 +317,10 @@ module Datadog
           res
         end
 
-        # Returns a Set of normalized file paths changed since the given base_commit.
+        # Returns a Diff object with relative file paths for files that were changed since the given base_commit.
         # If base_commit is nil, returns nil. On error, returns nil.
-        def self.get_changed_files_from_diff(base_commit)
-          return nil if base_commit.nil?
+        def self.get_changes_since(base_commit)
+          return Diff.new if base_commit.nil?
 
           Datadog.logger.debug { "calculating git diff from base_commit: #{base_commit}" }
 
@@ -329,29 +338,14 @@ module Datadog
 
             Datadog.logger.debug { "git diff output: #{output}" }
 
-            return nil if output.nil?
-
-            # 2. Parse the output to extract which files changed
-            changed_files = Set.new
-            output.each_line do |line|
-              # Match lines like: diff --git a/foo/bar.rb b/foo/bar.rb
-              # This captures git changes on file level
-              match = /^diff --git a\/(?<file>.+?) b\//.match(line)
-              if match && match[:file]
-                changed_file = match[:file]
-                # Normalize to repo root
-                normalized_changed_file = relative_to_root(changed_file)
-                changed_files << normalized_changed_file unless normalized_changed_file.nil? || normalized_changed_file.empty?
-
-                Datadog.logger.debug { "matched changed_file: #{changed_file} from line: #{line}" }
-                Datadog.logger.debug { "normalized_changed_file: #{normalized_changed_file}" }
-              end
-            end
-            changed_files
+            return Diff.new if output.nil?
+
+            # 2. Parse the output using Git::Diff
+            Diff.parse_diff_output(output)
           rescue => e
             Telemetry.track_error(e, Ext::Telemetry::Command::DIFF)
             log_failure(e, "get changed files from diff")
-            nil
+            Diff.new
           end
         end
 

data/lib/datadog/ci/impacted_tests_detection/component.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require "set"
-
 require_relative "../ext/test"
 require_relative "../git/local_repository"
 
@@ -11,7 +9,7 @@ module Datadog
       class Component
         def initialize(enabled:)
           @enabled = enabled
-          @changed_files = Set.new
+          @git_diff = Git::Diff.new
         end
 
         def configure(library_settings, test_session)
@@ -29,21 +27,21 @@ module Datadog
             return
           end
 
-          changed_files = Git::LocalRepository.get_changed_files_from_diff(base_commit_sha)
-          if changed_files.nil?
+          git_diff = Git::LocalRepository.get_changes_since(base_commit_sha)
+          if git_diff.empty?
             Datadog.logger.debug { "Impacted tests detection disabled: could not get changed files" }
             @enabled = false
             return
           end
 
           Datadog.logger.debug do
-            "Impacted tests detection: found #{changed_files.size} changed files"
+            "Impacted tests detection: found #{git_diff.size} changed files"
           end
           Datadog.logger.debug do
-            "Impacted tests detection: changed files: #{changed_files.inspect}"
+            "Impacted tests detection: changed files: #{git_diff.inspect}"
           end
 
-          @changed_files = changed_files
+          @git_diff = git_diff
           @enabled = true
         end
 
@@ -57,7 +55,15 @@ module Datadog
           source_file = test_span.source_file
           return false if source_file.nil?
 
-          @changed_files.include?(source_file)
+          # convert to relative path without leading slash
+          # @type var source_file: String
+          source_file = source_file[1..] if source_file.start_with?("/")
+
+          result = @git_diff.lines_changed?(source_file, start_line: test_span.start_line, end_line: test_span.end_line)
+          Datadog.logger.debug do
+            "Impacted tests detection: test #{test_span.name} with source file #{source_file} is modified: #{result}"
+          end
+          result
         end
 
         def tag_modified_test(test_span)

data/lib/datadog/ci/remote/library_settings_client.rb

@@ -81,7 +81,7 @@ module Datadog
                 "service" => test_session.service,
                 "env" => @dd_env,
                 "repository_url" => test_session.git_repository_url,
-                "branch" => test_session.git_branch,
+                "branch" => test_session.git_branch || test_session.git_tag,
                 "sha" => test_session.git_commit_sha,
                 "test_level" => Ext::Test::ITR_TEST_SKIPPING_MODE,
                 "configurations" => {

data/lib/datadog/ci/span.rb

@@ -72,6 +72,12 @@ module Datadog
         tracer_span.get_tag(Ext::Test::TAG_STATUS) == Ext::Test::Status::SKIP
       end
 
+      # Returns the current status of the span.
+      # @return [String, nil] the status of the span ("pass", "fail", "skip"), or nil if not set
+      def status
+        tracer_span.get_tag(Ext::Test::TAG_STATUS)
+      end
+
       # Sets the status of the span to "pass".
       # @return [void]
       def passed!
@@ -165,6 +171,12 @@ module Datadog
         tracer_span.get_tag(Ext::Git::TAG_BRANCH)
       end
 
+      # Returns the git tag extracted from the environment.
+      # @return [String, nil] the tag or nil if not set.
+      def git_tag
+        tracer_span.get_tag(Ext::Git::TAG_TAG)
+      end
+
       # Returns the base commit SHA for the pull request, if available.
       # @return [String, nil] the base commit SHA or nil if not set.
       def base_commit_sha

data/lib/datadog/ci/test.rb

@@ -66,6 +66,22 @@ module Datadog
         get_tag(Ext::Test::TAG_TEST_SESSION_ID)
       end
 
+      # Returns the starting line number of the test in the source file.
+      # @return [Integer] the starting line number
+      # @return [nil] if the starting line is not available
+      def start_line
+        line = get_tag(Ext::Test::TAG_SOURCE_START)
+        line&.to_i
+      end
+
+      # Returns the ending line number of the test in the source file.
+      # @return [Integer] the ending line number
+      # @return [nil] if the ending line is not available
+      def end_line
+        line = get_tag(Ext::Test::TAG_SOURCE_END)
+        line&.to_i
+      end
+
       # Returns "true" if test span represents a retry.
       # @return [Boolean] true if this test is a retry, false otherwise.
       def is_retry?

data/lib/datadog/ci/test_retries/component.rb

@@ -86,6 +86,12 @@ module Datadog
           end
         end
 
+        def record_test_started(test_span)
+          # mark test as retry in the beginning
+          # if this is a first execution, the current_retry_driver is nil and this is noop
+          current_retry_driver&.mark_as_retry(test_span)
+        end
+
         def record_test_finished(test_span)
           if current_retry_driver.nil?
             # we always run test at least once and after the first pass create a correct retry driver

data/lib/datadog/ci/test_retries/driver/base.rb

@@ -11,17 +11,22 @@ module Datadog
             false
           end
 
-          def record_retry(test_span)
+          def mark_as_retry(test_span)
             test_span&.set_tag(Ext::Test::TAG_IS_RETRY, "true")
             test_span&.set_tag(Ext::Test::TAG_RETRY_REASON, retry_reason)
           end
 
+          def record_retry(test_span)
+          end
+
           # duration in float seconds
           def record_duration(duration)
           end
 
           def retry_reason
-            "unknown"
+            # we set retry reason to be external (ie retried outside of datadog)
+            # by default if we don't know why the test was retried
+            Ext::Test::RetryReason::RETRY_EXTERNAL
           end
         end
       end

data/lib/datadog/ci/test_visibility/component.rb

@@ -267,6 +267,8 @@ module Datadog
 
           test_optimisation.mark_if_skippable(test)
           test_optimisation.start_coverage(test)
+
+          test_retries.record_test_started(test)
         end
 
         def on_test_session_finished(test_session)
@@ -293,9 +295,8 @@ module Datadog
           test_optimisation.stop_coverage(test)
           test_optimisation.on_test_finished(test, maybe_remote_context)
 
-          Telemetry.event_finished(test)
-
           test_retries.record_test_finished(test)
+          Telemetry.event_finished(test)
         end
 
         def on_after_test_span_finished(tracer_span)

data/lib/datadog/ci/test_visibility/transport.rb

@@ -19,10 +19,6 @@ module Datadog
       class Transport < Datadog::CI::Transport::EventPlatformTransport
         attr_reader :serializers_factory, :dd_env
 
-        def self.log_once
-          @log_once ||= Datadog::Core::Utils::OnlyOnce.new
-        end
-
         def initialize(
           api:,
           dd_env:,
@@ -72,12 +68,17 @@ module Datadog
 
             encoded
           else
-            message = "Invalid event skipped: #{serializer} Errors: #{serializer.validation_errors}"
-            Datadog.logger.warn(message)
-            CI::Transport::Telemetry.endpoint_payload_dropped(1, endpoint: telemetry_endpoint_tag)
+            message = "Event with type #{serializer.event_type}(name=#{serializer.name}) is invalid: #{serializer.validation_errors}"
+
+            if serializer.event_type == "span"
+              # events of type span are often skipped because of missing resource field
+              # (because they are misconfigured in tests context)
+              Datadog.logger.debug(message)
+            else
+              Datadog.logger.warn(message)
+              CI::Transport::Telemetry.endpoint_payload_dropped(1, endpoint: telemetry_endpoint_tag)
 
-            # log invalid message once as error to internal telemetry
-            self.class.log_once.run do
+              # for CI events log all events to internal telemetry
               Core::Telemetry::Logger.error(message)
             end