feat(data-collection): Add base DataCollection configuration with defaults and backfill

sentry-ruby/lib/sentry/configuration.rb

@@ -16,6 +16,7 @@
 require "sentry/structured_logger"
 require "sentry/log_event_buffer"
 require "sentry/metric_event_buffer"
+require "sentry/data_collection"
 
 module Sentry
   class Configuration
@@ -132,6 +133,7 @@ class Configuration
     attr_accessor :max_breadcrumbs
 
     # Number of lines of code context to capture, or nil for none
+    # @deprecated Use {#data_collection} and `frame_context_lines` instead.
     # @return [Integer, nil]
     attr_accessor :context_lines
 
@@ -167,6 +169,7 @@ class Configuration
     alias inspect_exception_causes_for_exclusion? inspect_exception_causes_for_exclusion
 
     # Whether to capture local variables from the raised exception's frame. Default is false.
+    # @deprecated Use {#data_collection} and `stack_frame_variables` instead.
     # @return [Boolean]
     attr_accessor :include_local_variables
 
@@ -232,8 +235,14 @@ class Configuration
     # - request body
     # - query string
     # will not be sent to Sentry.
+    # @deprecated Use {#data_collection} instead.
     # @return [Boolean]
-    attr_accessor :send_default_pii
+    attr_reader :send_default_pii
+
+    # Controls which categories of data may be collected.
+    # Replacement for send_default_pii.
+    # @return [DataCollection]
+    attr_accessor :data_collection
 
     # Capture queue time from X-Request-Start header set by reverse proxies.
     # Works with any Rack app behind Nginx, HAProxy, Heroku router, etc.
@@ -536,6 +545,7 @@ def initialize
       self.breadcrumbs_logger = []
       self.context_lines = 3
       self.include_local_variables = false
+
       self.environment = environment_from_env
       self.enabled_environments = nil
       self.exclude_loggers = []
@@ -548,6 +558,7 @@ def initialize
 
       self.sample_rate = 1.0
       self.send_modules = true
+      self.data_collection = DataCollection.new
       self.send_default_pii = false
       self.skip_rake_integration = false
       self.send_client_reports = true
@@ -595,6 +606,8 @@ def initialize
 
       yield(self) if block_given?
 
+      log_deprecations
+
       run_callbacks(:after, :configured)
     end
 
@@ -620,6 +633,11 @@ def dsn=(value)
       @dsn = init_dsn(value)
     end
 
+    def send_default_pii=(value)
+      @send_default_pii = value
+      self.data_collection = DataCollection.backfill(self)
+    end
+
     alias server= dsn=
 
     def release=(value)
@@ -803,6 +821,13 @@ def csp_report_uri
       end
     end
 
+    # @api private
+    def log_deprecations
+      log_warn("`send_default_pii` is deprecated; use `data_collection` instead.") if self.send_default_pii
+      log_warn("`include_local_variables` is deprecated; use `data_collection.stack_frame_variables` instead.") if include_local_variables
+      log_warn("`context_lines` is deprecated; use `data_collection.frame_context_lines` instead.") if context_lines != 3
+    end
+
     # @api private
     def stacktrace_builder
       @stacktrace_builder ||= StacktraceBuilder.new(

sentry-ruby/lib/sentry/data_collection.rb

@@ -0,0 +1,161 @@
+# frozen_string_literal: true
+
+module Sentry
+  class DataCollection
+    # Configuration for the categories of data collected by the SDK.
+    # Replacement for send_default_pii.
+    # Spec: https://develop.sentry.dev/sdk/foundations/client/data-collection/
+    #
+    # @example Configure data collection
+    #   Sentry.init do |config|
+    #     data_collection = config.data_collection
+    #     data_collection.user_info = true
+    #     data_collection.cookies.mode = :deny_list
+    #     data_collection.cookies.terms = ["session", "token"]
+    #     data_collection.http_headers.request.mode = :deny_list
+    #     data_collection.http_headers.request.terms = nil
+    #     data_collection.http_headers.response.mode = :allow_list
+    #     data_collection.http_headers.response.terms = ["my_special_header"]
+    #     data_collection.http_bodies = [:incoming_request]
+    #     data_collection.url_query_params.mode = :allow_list
+    #     data_collection.url_query_params.terms = ["page", "limit"]
+    #     data_collection.graphql.document = true
+    #     data_collection.graphql.variables = true
+    #     data_collection.database_query_data = true
+    #     data_collection.queues = true
+    #     data_collection.stack_frame_variables = true
+    #     data_collection.frame_context_lines = 5
+    #   end
+    MODES = %i[off deny_list allow_list].freeze
+    BODY_TYPES = %i[
+    incoming_request
+    outgoing_request
+    incoming_response
+    outgoing_response
+    ].freeze
+
+    # Configuration for key-value data collection.
+    class KeyValueCollection
+      # `mode` controls whether values are collected:
+      # - `:off` disables collection.
+      # - `:deny_list` collects values except those matching `terms`.
+      # - `:allow_list` collects only values matching `terms`.
+      # @return [:off, :deny_list, :allow_list]
+      attr_accessor :mode
+
+      # `terms` contains the keys or patterns used by the selected mode.
+      # @return [Array<String>, nil]
+      attr_accessor :terms
+
+      def initialize(mode:, terms:)
+        @mode = mode
+        @terms = terms
+      end
+    end
+
+    class HttpHeaders
+      # @return [KeyValueCollection]
+      attr_accessor :request
+
+      # @return [KeyValueCollection]
+      attr_accessor :response
+
+      def initialize(request:, response:)
+        @request = request
+        @response = response
+      end
+    end
+
+    class GraphQL
+      # @return [Boolean]
+      attr_accessor :document
+
+      # @return [Boolean]
+      attr_accessor :variables
+
+      def initialize(document:, variables:)
+        @document = document
+        @variables = variables
+      end
+    end
+
+    # @return [Boolean]
+    # @default `true`
+    attr_accessor :user_info
+
+    # @return [KeyValueCollection]
+    # @default `mode: :deny_list, terms: nil`
+    attr_accessor :cookies
+
+    # @return [HttpHeaders]
+    # @default request and response use `mode: :deny_list, terms: nil`
+    attr_accessor :http_headers
+
+    # @return [Array<Symbol>] containing values from BODY_TYPES
+    # @default `nil` (all valid body types)
+    attr_accessor :http_bodies
+
+    # @return [KeyValueCollection]
+    # @default `mode: :deny_list, terms: nil`
+    attr_accessor :url_query_params
+
+    # @return [Boolean]
+    # @default `true`
+    attr_accessor :database_query_data
+
+    # @return [GraphQL]
+    # @default `document: true, variables: true`
+    attr_accessor :graphql
+
+    # @return [Boolean]
+    # @default `true`
+    attr_accessor :queues
+
+    # @return [Boolean]
+    # @default `false`
+    attr_accessor :stack_frame_variables
+
+    # @return [Integer]
+    # @default `3`
+    attr_accessor :frame_context_lines
+
+    # Builds data collection settings compatible with the legacy send_default_pii
+    # configuration.
+    def self.backfill(configuration)
+      # the new DataCollection defaults are already correct if pii is enabled
+      data_collection = new
+      return data_collection if configuration.send_default_pii
+
+      # TODO-neel-data map to exact ruby behaviour for backwards compat behavior
+      data_collection.user_info = false
+      data_collection.cookies.mode = :off
+      data_collection.http_headers.request.mode = :off
+      data_collection.http_headers.response.mode = :off
+      data_collection.http_bodies = []
+      data_collection.url_query_params.mode = :off
+      data_collection.graphql.document = false
+      data_collection.graphql.variables = false
+      data_collection.database_query_data = false
+      data_collection.queues = false
+      data_collection.stack_frame_variables = configuration.include_local_variables
+      data_collection.frame_context_lines = configuration.context_lines
+      data_collection
+    end
+
+    def initialize
+      @user_info = true
+      @cookies = KeyValueCollection.new(mode: :deny_list, terms: nil)
+      @http_headers = HttpHeaders.new(
+        request: KeyValueCollection.new(mode: :deny_list, terms: nil),
+        response: KeyValueCollection.new(mode: :deny_list, terms: nil)
+      )
+      @http_bodies = nil
+      @url_query_params = KeyValueCollection.new(mode: :deny_list, terms: nil)
+      @database_query_data = true
+      @graphql = GraphQL.new(document: true, variables: true)
+      @queues = true
+      @stack_frame_variables = false
+      @frame_context_lines = 3
+    end
+  end
+end

sentry-ruby/spec/sentry/configuration_spec.rb

@@ -1,6 +1,46 @@
 # frozen_string_literal: true
 
 RSpec.describe Sentry::Configuration do
+  describe "#data_collection" do
+    it "allows explicit data collection configuration after enabling send_default_pii" do
+      configuration = Sentry::Configuration.new do |config|
+        config.send_default_pii = true
+        config.data_collection.user_info = false
+      end
+
+      expect(configuration.data_collection.user_info).to eq(false)
+    end
+  end
+
+  describe "deprecated configuration warnings" do
+    it "warns when send_default_pii is enabled" do
+      allow(subject).to receive(:log_warn)
+      subject.send_default_pii = true
+
+      subject.send(:log_deprecations)
+
+      expect(subject).to have_received(:log_warn).with("`send_default_pii` is deprecated; use `data_collection` instead.")
+    end
+
+    it "warns when include_local_variables is enabled" do
+      allow(subject).to receive(:log_warn)
+      subject.include_local_variables = true
+
+      subject.send(:log_deprecations)
+
+      expect(subject).to have_received(:log_warn).with("`include_local_variables` is deprecated; use `data_collection.stack_frame_variables` instead.")
+    end
+
+    it "warns when context_lines is changed" do
+      allow(subject).to receive(:log_warn)
+      subject.context_lines = 4
+
+      subject.send(:log_deprecations)
+
+      expect(subject).to have_received(:log_warn).with("`context_lines` is deprecated; use `data_collection.frame_context_lines` instead.")
+    end
+  end
+
   describe "#background_worker_threads" do
     it "sets to have of the processors count" do
       allow_any_instance_of(Sentry::Configuration).to receive(:processor_count).and_return(8)

sentry-ruby/spec/sentry/data_collection_spec.rb

@@ -0,0 +1,92 @@
+# frozen_string_literal: true
+
+RSpec.describe Sentry::DataCollection do
+  subject(:data_collection) { described_class.new }
+
+  describe ".backfill" do
+    it "uses the send_default_pii=false defaults" do
+      configuration = Sentry::Configuration.new
+      data_collection = described_class.backfill(configuration)
+
+      expect(data_collection.user_info).to eq(false)
+      expect(data_collection.cookies.mode).to eq(:off)
+      expect(data_collection.http_headers.request.mode).to eq(:off)
+      expect(data_collection.http_headers.response.mode).to eq(:off)
+      expect(data_collection.http_bodies).to eq([])
+      expect(data_collection.url_query_params.mode).to eq(:off)
+      expect(data_collection.database_query_data).to eq(false)
+      expect(data_collection.graphql.document).to eq(false)
+      expect(data_collection.graphql.variables).to eq(false)
+      expect(data_collection.queues).to eq(false)
+      expect(data_collection.stack_frame_variables).to eq(false)
+      expect(data_collection.frame_context_lines).to eq(3)
+    end
+
+    it "uses the send_default_pii=true defaults when enabled later" do
+      configuration = Sentry::Configuration.new
+      configuration.send_default_pii = true
+
+      expect(configuration.data_collection.user_info).to eq(true)
+      expect(configuration.data_collection.cookies.mode).to eq(:deny_list)
+    end
+  end
+
+  describe "defaults" do
+    it "uses the defaults from the Data Collection specification" do
+      expect(data_collection.user_info).to eq(true)
+      expect(data_collection.cookies.mode).to eq(:deny_list)
+      expect(data_collection.cookies.terms).to be_nil
+      expect(data_collection.http_headers.request.mode).to eq(:deny_list)
+      expect(data_collection.http_headers.request.terms).to be_nil
+      expect(data_collection.http_headers.response.mode).to eq(:deny_list)
+      expect(data_collection.http_headers.response.terms).to be_nil
+      expect(data_collection.http_bodies).to be_nil
+      expect(data_collection.url_query_params.mode).to eq(:deny_list)
+      expect(data_collection.url_query_params.terms).to be_nil
+      expect(data_collection.database_query_data).to eq(true)
+      expect(data_collection.graphql.document).to eq(true)
+      expect(data_collection.graphql.variables).to eq(true)
+      expect(data_collection.queues).to eq(true)
+      expect(data_collection.stack_frame_variables).to eq(false)
+      expect(data_collection.frame_context_lines).to eq(3)
+    end
+  end
+
+  describe "constants" do
+    it "defines the supported modes" do
+      expect(described_class::MODES).to eq(%i[off deny_list allow_list])
+    end
+
+    it "defines the supported body types" do
+      expect(described_class::BODY_TYPES).to eq(
+        %i[incoming_request outgoing_request incoming_response outgoing_response]
+      )
+    end
+  end
+
+  describe "nested configuration objects" do
+    it "supports configuring key-value collection modes and terms" do
+      data_collection.cookies.mode = :allow_list
+      data_collection.cookies.terms = ["page"]
+
+      expect(data_collection.cookies.mode).to eq(:allow_list)
+      expect(data_collection.cookies.terms).to eq(["page"])
+    end
+
+    it "supports configuring request and response headers independently" do
+      data_collection.http_headers.request.mode = :off
+      data_collection.http_headers.response.terms = ["x-request-id"]
+
+      expect(data_collection.http_headers.request.mode).to eq(:off)
+      expect(data_collection.http_headers.response.terms).to eq(["x-request-id"])
+    end
+
+    it "supports configuring GraphQL fields independently" do
+      data_collection.graphql.document = false
+      data_collection.graphql.variables = false
+
+      expect(data_collection.graphql.document).to eq(false)
+      expect(data_collection.graphql.variables).to eq(false)
+    end
+  end
+end