datadog 2.0.0.beta1

data/lib/datadog/tracing/contrib/opensearch/patcher.rb

@@ -0,0 +1,135 @@
+# frozen_string_literal: true
+
+require_relative '../../metadata/ext'
+require_relative '../analytics'
+require_relative 'ext'
+require_relative '../ext'
+require_relative '../integration'
+require_relative '../patcher'
+
+module Datadog
+  module Tracing
+    module Contrib
+      module OpenSearch
+        # Patcher enables patching of 'opensearch' module.
+        module Patcher
+          include Contrib::Patcher
+
+          module_function
+
+          def patch
+            require 'uri'
+            require 'json'
+            require_relative 'quantize'
+
+            ::OpenSearch::Transport::Client.prepend(Client)
+          end
+
+          # Patches OpenSearch::Transport::Client module
+          module Client
+            # rubocop:disable Metrics/MethodLength
+            # rubocop:disable Metrics/AbcSize
+            def perform_request(method, path, params = {}, body = nil, headers = nil)
+              response = nil
+              # rubocop:disable Metrics/BlockLength
+              Tracing.trace('opensearch.query', service: datadog_configuration[:service_name]) do |span|
+                begin
+                  # Set generic tags
+                  span.set_tag(Tracing::Metadata::Ext::TAG_COMPONENT, Ext::TAG_COMPONENT)
+                  span.set_tag(Tracing::Metadata::Ext::TAG_KIND, Tracing::Metadata::Ext::SpanKind::TAG_CLIENT)
+                  span.set_tag(Contrib::Ext::DB::TAG_SYSTEM, Ext::TAG_SYSTEM)
+
+                  # Set argument tags
+                  span.set_tag(OpenSearch::Ext::TAG_METHOD, method)
+                  span.set_tag(OpenSearch::Ext::TAG_PATH, path)
+
+                  tag_params(params, span)
+                  tag_body(body, span)
+
+                  # Parse url
+                  original_url = transport.get_connection.full_url(path, {})
+                  url = URI.parse(original_url)
+                  host = url.host
+                  port = url.port
+                  scheme = url.scheme
+                  # Set url.user to nil to remove sensitive information (i.e. user's username and password)
+                  url.user = nil
+
+                  if datadog_configuration[:peer_service]
+                    span.set_tag(
+                      Tracing::Metadata::Ext::TAG_PEER_SERVICE,
+                      datadog_configuration[:peer_service]
+                    )
+                  end
+
+                  # Tag original global service name if not used
+                  if span.service != Datadog.configuration.service
+                    span.set_tag(Tracing::Contrib::Ext::Metadata::TAG_BASE_SERVICE, Datadog.configuration.service)
+                  end
+
+                  # Set url tags
+                  span.set_tag(OpenSearch::Ext::TAG_URL, url)
+                  span.set_tag(OpenSearch::Ext::TAG_HOST, host)
+                  span.set_tag(OpenSearch::Ext::TAG_PORT, port)
+                  span.set_tag(OpenSearch::Ext::TAG_SCHEME, scheme)
+
+                  span.set_tag(Tracing::Metadata::Ext::TAG_PEER_HOSTNAME, host) if host
+
+                  # Define span resource
+                  quantized_url = OpenSearch::Quantize.format_url(url)
+                  span.resource = "#{method} #{quantized_url}"
+                  Contrib::SpanAttributeSchema.set_peer_service!(span, Ext::PEER_SERVICE_SOURCES)
+                rescue StandardError => e
+                  Datadog.logger.error(e.message)
+                ensure
+                  begin
+                    response = super
+                  rescue => e
+                    status_code = ::OpenSearch::Transport::Transport::ERRORS.key(e.class)
+                    span.set_tag(Tracing::Metadata::Ext::HTTP::TAG_STATUS_CODE, status_code) if status_code
+                    raise
+                  end
+                  # Set post-response tags
+                  span.set_tag(Tracing::Metadata::Ext::HTTP::TAG_STATUS_CODE, response.status)
+                  if response.headers['content-length']
+                    span.set_tag(
+                      OpenSearch::Ext::TAG_RESPONSE_CONTENT_LENGTH,
+                      response.headers['content-length'].to_i
+                    )
+                  end
+                end
+              end
+              # rubocop:enable Metrics/BlockLength
+              # rubocop:enable Metrics/AbcSize
+              response
+            end
+
+            def tag_params(params, span)
+              return unless params
+
+              params = JSON.generate(params) unless params.is_a?(String)
+              span.set_tag(OpenSearch::Ext::TAG_PARAMS, params)
+            end
+
+            def tag_body(body, span)
+              return unless body
+
+              body = JSON.generate(body) unless body.is_a?(String)
+              quantize_options = datadog_configuration[:quantize]
+              quantized_body = OpenSearch::Quantize.format_body(
+                body,
+                quantize_options
+              )
+              span.set_tag(OpenSearch::Ext::TAG_BODY, quantized_body)
+            end
+
+            def datadog_configuration
+              Datadog.configuration.tracing[:opensearch]
+            end
+          end
+          # rubocop:enable Metrics/MethodLength
+        end
+      end
+    end
+  end
+end

data/lib/datadog/tracing/contrib/opensearch/quantize.rb

@@ -0,0 +1,81 @@
+# frozen_string_literal: true
+
+require_relative '../utils/quantization/hash'
+
+module Datadog
+  module Tracing
+    module Contrib
+      module OpenSearch
+        # Removes sensitive data from OpenSearch strings (i.e. url and body).
+        module Quantize
+          PLACEHOLDER = '/?'
+          DEFAULT_PLACEHOLDER = '?'
+          EXCLUDE_KEYS = [].freeze
+          SHOW_KEYS = [:_index, :_type, :_id].freeze
+          DEFAULT_OPTIONS = {
+            exclude: EXCLUDE_KEYS,
+            show: SHOW_KEYS,
+            placeholder: PLACEHOLDER
+          }.freeze
+
+          module_function
+
+          def format_url(url)
+            # Url does not contain path, so just replace last digit(s) with '?' if '/' precedes the digit(s)
+            url.to_s.gsub(%r{/\d+$}, PLACEHOLDER)
+          end
+
+          # Use Elasticsearch implementation
+          def format_body(body, options = {})
+            format_body!(body, options)
+          rescue StandardError
+            options[:placeholder] || DEFAULT_PLACEHOLDER
+          end
+
+          def format_body!(body, options = {})
+            options = merge_options(DEFAULT_OPTIONS, options)
+
+            # Determine if bulk query or not, based on content
+            statements = body.end_with?("\n") ? body.split("\n") : [body]
+
+            # Parse each statement and quantize them.
+            statements.collect do |string|
+              reserialize_json(string, options[:placeholder]) do |obj|
+                Contrib::Utils::Quantization::Hash.format(obj, options)
+              end
+            end.join("\n")
+          end
+
+          def merge_options(original, additional)
+            {}.tap do |options|
+              # Show
+              # If either is :all, value becomes :all
+              options[:show] = if original[:show] == :all || additional[:show] == :all
+                                 :all
+                               else
+                                 (original[:show] || []).dup.concat(additional[:show] || []).uniq
+                               end
+
+              # Exclude
+              options[:exclude] = (original[:exclude] || []).dup.concat(additional[:exclude] || []).uniq
+            end
+          end
+
+          # Parses a JSON object from a string, passes its value
+          # to the block provided, and dumps its result back to JSON.
+          # If JSON parsing fails, it prints fail_value.
+          def reserialize_json(string, fail_value = DEFAULT_PLACEHOLDER)
+            return string unless block_given?
+
+            begin
+              JSON.dump(yield(JSON.parse(string)))
+            rescue JSON::ParserError
+              # If it can't parse/dump, don't raise an error.
+              fail_value
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/tracing/contrib/patchable.rb

@@ -0,0 +1,109 @@
+# frozen_string_literal: true
+
+module Datadog
+  module Tracing
+    module Contrib
+      # Base provides features that are shared across all integrations
+      module Patchable
+        def self.included(base)
+          base.extend(ClassMethods)
+          base.include(InstanceMethods)
+        end
+
+        # Class methods for integrations
+        # @public_api
+        module ClassMethods
+          # Version of the integration target code in the environment.
+          #
+          # This is the gem version, when the instrumentation target is a Ruby gem.
+          #
+          # If the target for instrumentation has concept of versioning, override {.version},
+          # otherwise override {.available?} and implement a custom target presence check.
+          # @return [Object] the target version
+          def version
+            nil
+          end
+
+          # Is the target available to be instrumented? (e.g. gem installed?)
+          #
+          # The target doesn't have to be loaded (e.g. `require`) yet, but needs to be able
+          # to be loaded before instrumentation can commence.
+          #
+          # By default, {.available?} checks if {.version} returned a non-nil object.
+          #
+          # If the target for instrumentation has concept of versioning, override {.version},
+          # otherwise override {.available?} and implement a custom target presence check.
+          # @return [Boolean] is the target available for instrumentation in this Ruby environment?
+          def available?
+            !version.nil?
+          end
+
+          # Is the target loaded into the application? (e.g. gem required? Constant defined?)
+          #
+          # The target's objects should be ready to be referenced by the instrumented when {.loaded}
+          # returns `true`.
+          #
+          # @return [Boolean] is the target ready to be referenced during instrumentation?
+          def loaded?
+            true
+          end
+
+          # Is this instrumentation compatible with the available target? (e.g. minimum version met?)
+          # @return [Boolean] is the available target compatible with this instrumentation?
+          def compatible?
+            available?
+          end
+
+          # Can the patch for this integration be applied?
+          #
+          # By default, this is equivalent to {#available?}, {#loaded?}, and {#compatible?}
+          # all being truthy.
+          def patchable?
+            available? && loaded? && compatible?
+          end
+        end
+
+        # Instance methods for integrations
+        # @public_api
+        module InstanceMethods
+          # The patcher module to inject instrumented objects into the instrumentation target.
+          #
+          # {Contrib::Patcher} includes the basic functionality of a patcher. `include`ing
+          # {Contrib::Patcher} into a new module is the recommend way to create a custom patcher.
+          #
+          # @return [Contrib::Patcher] a module that `include`s {Contrib::Patcher}
+          def patcher
+            nil
+          end
+
+          # @!visibility private
+          def patch
+            if !self.class.patchable? || patcher.nil?
+              return {
+                name: self.class.name,
+                available: self.class.available?,
+                loaded: self.class.loaded?,
+                compatible: self.class.compatible?,
+                patchable: self.class.patchable?
+              }
+            end
+
+            patcher.patch
+            true
+          end
+
+          # Can the patch for this integration be applied automatically?
+          # For example: test integrations should only be applied
+          # by the user explicitly setting `c.ci.instrument :rspec`
+          # and rails sub-modules are auto-instrumented by enabling rails
+          # so auto-instrumenting them on their own will cause changes in
+          # service naming behavior
+          # @return [Boolean] can the tracer activate this instrumentation without explicit user input?
+          def auto_instrument?
+            true
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/tracing/contrib/patcher.rb

@@ -0,0 +1,85 @@
+# frozen_string_literal: true
+
+require_relative '../../core/utils/only_once'
+
+module Datadog
+  module Tracing
+    module Contrib
+      # Common behavior for patcher modules.
+      #
+      # `include`ing {Contrib::Patcher} into a new module is the recommend way to create a custom patcher.
+      # The patcher can then be provided to a custom {Datadog::Tracing::Contrib::Integration} for instrumentation.
+      module Patcher
+        def self.included(base)
+          base.singleton_class.prepend(CommonMethods)
+          base.prepend(CommonMethods) if base.instance_of?(Class)
+        end
+
+        # Prepended instance methods for all patchers
+        # @public_api
+        module CommonMethods
+          attr_accessor \
+            :patch_error_result,
+            :patch_successful
+
+          def patch_name
+            self.class != Class && self.class != Module ? self.class.name : name
+          end
+
+          def patched?
+            patch_only_once.ran?
+          end
+
+          def patch
+            return unless defined?(super)
+
+            patch_only_once.run do
+              begin
+                super.tap do
+                  # Emit a metric
+                  Datadog.health_metrics.instrumentation_patched(1, tags: default_tags)
+                  @patch_successful = true
+                end
+              rescue StandardError => e
+                on_patch_error(e)
+              end
+            end
+          end
+
+          # Processes patching errors. This default implementation logs the error and reports relevant metrics.
+          # @param e [Exception]
+          def on_patch_error(e)
+            # Log the error
+            Datadog.logger.error("Failed to apply #{patch_name} patch. Cause: #{e} Location: #{Array(e.backtrace).first}")
+
+            @patch_error_result = {
+              type: e.class.name,
+              message: e.message,
+              line: Array(e.backtrace).first
+            }
+
+            # Emit a metric
+            tags = default_tags
+            tags << "error:#{e.class.name}"
+
+            Datadog.health_metrics.error_instrumentation_patch(1, tags: tags)
+          end
+
+          def default_tags
+            ["patcher:#{patch_name}"].tap do |tags|
+              tags << "target_version:#{target_version}" if respond_to?(:target_version) && !target_version.nil?
+              super.each { |t| tags << t } if defined?(super)
+            end
+          end
+
+          private
+
+          def patch_only_once
+            # NOTE: This is not thread-safe
+            @patch_only_once ||= Core::Utils::OnlyOnce.new
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/tracing/contrib/pg/configuration/settings.rb

@@ -0,0 +1,64 @@
+# frozen_string_literal: true
+
+require_relative '../../configuration/settings'
+require_relative '../ext'
+
+require_relative '../../propagation/sql_comment/ext'
+
+module Datadog
+  module Tracing
+    module Contrib
+      module Pg
+        module Configuration
+          # Custom settings for the Pg integration
+          # @public_api
+          class Settings < Contrib::Configuration::Settings
+            option :enabled do |o|
+              o.type :bool
+              o.env Ext::ENV_ENABLED
+              o.default true
+            end
+
+            # @!visibility private
+            option :analytics_enabled do |o|
+              o.type :bool
+              o.env Ext::ENV_ANALYTICS_ENABLED
+              o.default false
+            end
+
+            option :on_error do |o|
+              o.type :proc, nilable: true
+            end
+
+            option :analytics_sample_rate do |o|
+              o.type :float
+              o.env Ext::ENV_ANALYTICS_SAMPLE_RATE
+              o.default 1.0
+            end
+
+            option :service_name do |o|
+              o.type :string, nilable: true
+              o.default do
+                Contrib::SpanAttributeSchema.fetch_service_name(
+                  Ext::ENV_SERVICE_NAME,
+                  Ext::DEFAULT_PEER_SERVICE_NAME
+                )
+              end
+            end
+
+            option :comment_propagation do |o|
+              o.type :string
+              o.env Contrib::Propagation::SqlComment::Ext::ENV_DBM_PROPAGATION_MODE
+              o.default Contrib::Propagation::SqlComment::Ext::DISABLED
+            end
+
+            option :peer_service do |o|
+              o.type :string, nilable: true
+              o.env Ext::ENV_PEER_SERVICE
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/tracing/contrib/pg/ext.rb

@@ -0,0 +1,35 @@
+# frozen_string_literal: true
+
+module Datadog
+  module Tracing
+    module Contrib
+      module Pg
+        # pg integration constants
+        # @public_api Changing resource names, tag names, or environment variables creates breaking changes.
+        module Ext
+          ENV_ENABLED = 'DD_TRACE_PG_ENABLED'
+          ENV_SERVICE_NAME = 'DD_TRACE_PG_SERVICE_NAME'
+          ENV_PEER_SERVICE = 'DD_TRACE_PG_PEER_SERVICE'
+          # @!visibility private
+          ENV_ANALYTICS_ENABLED = 'DD_TRACE_PG_ANALYTICS_ENABLED'
+          ENV_ANALYTICS_SAMPLE_RATE = 'DD_TRACE_PG_ANALYTICS_SAMPLE_RATE'
+          DEFAULT_PEER_SERVICE_NAME = 'pg'
+          SPAN_EXEC = 'pg.exec'
+          SPAN_EXEC_PARAMS = 'pg.exec.params'
+          SPAN_EXEC_PREPARED = 'pg.exec.prepared'
+          SPAN_ASYNC_EXEC = 'pg.async.exec'
+          SPAN_ASYNC_EXEC_PARAMS = 'pg.async.exec.params'
+          SPAN_ASYNC_EXEC_PREPARED = 'pg.async.exec.prepared'
+          SPAN_SYNC_EXEC = 'pg.sync.exec'
+          SPAN_SYNC_EXEC_PARAMS = 'pg.sync.exec.params'
+          SPAN_SYNC_EXEC_PREPARED = 'pg.sync.exec.prepared'
+          TAG_DB_NAME = 'pg.db.name'
+          TAG_COMPONENT = 'pg'
+          TAG_OPERATION_QUERY = 'query'
+          TAG_SYSTEM = 'postgresql'
+          PEER_SERVICE_SOURCES = Contrib::Ext::DB::PEER_SERVICE_SOURCES.freeze
+        end
+      end
+    end
+  end
+end