sentry-ruby 5.10.0 → 5.26.0

data/lib/sentry/log_event.rb

@@ -0,0 +1,206 @@
+# frozen_string_literal: true
+
+module Sentry
+  # Event type that represents a log entry with its attributes
+  #
+  # @see https://develop.sentry.dev/sdk/telemetry/logs/#log-envelope-item-payload
+  class LogEvent
+    TYPE = "log"
+
+    DEFAULT_PARAMETERS = [].freeze
+    DEFAULT_ATTRIBUTES = {}.freeze
+
+    SERIALIZEABLE_ATTRIBUTES = %i[
+      level
+      body
+      timestamp
+      environment
+      release
+      server_name
+      trace_id
+      attributes
+      contexts
+    ]
+
+    SENTRY_ATTRIBUTES = {
+      "sentry.trace.parent_span_id" => :parent_span_id,
+      "sentry.environment" => :environment,
+      "sentry.release" => :release,
+      "sentry.address" => :server_name,
+      "sentry.sdk.name" => :sdk_name,
+      "sentry.sdk.version" => :sdk_version,
+      "sentry.message.template" => :template
+    }
+
+    USER_ATTRIBUTES = {
+      "user.id" => :user_id,
+      "user.name" => :user_username,
+      "user.email" => :user_email
+    }
+
+    LEVELS = %i[trace debug info warn error fatal].freeze
+
+    attr_accessor :level, :body, :template, :attributes, :user
+
+    attr_reader :configuration, *SERIALIZEABLE_ATTRIBUTES
+
+    SERIALIZERS = %i[
+      attributes
+      body
+      level
+      parent_span_id
+      sdk_name
+      sdk_version
+      timestamp
+      trace_id
+      user_id
+      user_username
+      user_email
+    ].map { |name| [name, :"serialize_#{name}"] }.to_h
+
+    VALUE_TYPES = Hash.new("string").merge!({
+      TrueClass => "boolean",
+      FalseClass => "boolean",
+      Integer => "integer",
+      Float => "double"
+    }).freeze
+
+    TOKEN_REGEXP = /%\{(\w+)\}/
+
+    def initialize(configuration: Sentry.configuration, **options)
+      @configuration = configuration
+      @type = TYPE
+      @server_name = configuration.server_name
+      @environment = configuration.environment
+      @release = configuration.release
+      @timestamp = Sentry.utc_now
+      @level = options.fetch(:level)
+      @body = options[:body]
+      @template = @body if is_template?
+      @attributes = options[:attributes] || DEFAULT_ATTRIBUTES
+      @user = options[:user] || {}
+      @contexts = {}
+    end
+
+    def to_hash
+      SERIALIZEABLE_ATTRIBUTES.each_with_object({}) do |name, memo|
+        memo[name] = serialize(name)
+      end
+    end
+
+    private
+
+    def serialize(name)
+      serializer = SERIALIZERS[name]
+
+      if serializer
+        __send__(serializer)
+      else
+        public_send(name)
+      end
+    end
+
+    def serialize_level
+      level.to_s
+    end
+
+    def serialize_sdk_name
+      Sentry.sdk_meta["name"]
+    end
+
+    def serialize_sdk_version
+      Sentry.sdk_meta["version"]
+    end
+
+    def serialize_timestamp
+      timestamp.to_f
+    end
+
+    def serialize_trace_id
+      contexts.dig(:trace, :trace_id)
+    end
+
+    def serialize_parent_span_id
+      contexts.dig(:trace, :parent_span_id)
+    end
+
+    def serialize_body
+      if parameters.empty?
+        body
+      elsif parameters.is_a?(Hash)
+        body % parameters
+      else
+        sprintf(body, *parameters)
+      end
+    end
+
+    def serialize_user_id
+      user[:id]
+    end
+
+    def serialize_user_username
+      user[:username]
+    end
+
+    def serialize_user_email
+      user[:email]
+    end
+
+    def serialize_attributes
+      hash = {}
+
+      attributes.each do |key, value|
+        hash[key] = attribute_hash(value)
+      end
+
+      SENTRY_ATTRIBUTES.each do |key, name|
+        if (value = serialize(name))
+          hash[key] = attribute_hash(value)
+        end
+      end
+
+      USER_ATTRIBUTES.each do |key, name|
+        if (value = serialize(name))
+          hash[key] = value
+        end
+      end
+
+      hash
+    end
+
+    def attribute_hash(value)
+      { value: value, type: value_type(value) }
+    end
+
+    def value_type(value)
+      VALUE_TYPES[value.class]
+    end
+
+    def parameters
+      @parameters ||= begin
+        return DEFAULT_PARAMETERS unless template
+
+        parameters = template_tokens.empty? ?
+          attributes.fetch(:parameters, DEFAULT_PARAMETERS) : attributes.slice(*template_tokens)
+
+        if parameters.is_a?(Hash)
+          parameters.each do |key, value|
+            attributes["sentry.message.parameter.#{key}"] = value
+          end
+        else
+          parameters.each_with_index do |param, index|
+            attributes["sentry.message.parameter.#{index}"] = param
+          end
+        end
+      end
+    end
+
+    def template_tokens
+      @template_tokens ||= body.scan(TOKEN_REGEXP).flatten.map(&:to_sym)
+    end
+
+    def is_template?
+      body.include?("%s") || TOKEN_REGEXP.match?(body)
+    end
+  end
+end

data/lib/sentry/log_event_buffer.rb

@@ -0,0 +1,75 @@
+# frozen_string_literal: true
+
+require "sentry/threaded_periodic_worker"
+
+module Sentry
+  # LogEventBuffer buffers log events and sends them to Sentry in a single envelope.
+  #
+  # This is used internally by the `Sentry::Client`.
+  #
+  # @!visibility private
+  class LogEventBuffer < ThreadedPeriodicWorker
+    FLUSH_INTERVAL = 5 # seconds
+    DEFAULT_MAX_EVENTS = 100
+
+    # @!visibility private
+    attr_reader :pending_events
+
+    def initialize(configuration, client)
+      super(configuration.sdk_logger, FLUSH_INTERVAL)
+
+      @client = client
+      @pending_events = []
+      @max_events = configuration.max_log_events || DEFAULT_MAX_EVENTS
+      @mutex = Mutex.new
+
+      log_debug("[Logging] Initialized buffer with max_events=#{@max_events}, flush_interval=#{FLUSH_INTERVAL}s")
+    end
+
+    def start
+      ensure_thread
+      self
+    end
+
+    def flush
+      @mutex.synchronize do
+        return if empty?
+
+        log_debug("[LogEventBuffer] flushing #{size} log events")
+
+        send_events
+      end
+
+      log_debug("[LogEventBuffer] flushed #{size} log events")
+
+      self
+    end
+    alias_method :run, :flush
+
+    def add_event(event)
+      raise ArgumentError, "expected a LogEvent, got #{event.class}" unless event.is_a?(LogEvent)
+
+      @mutex.synchronize do
+        @pending_events << event
+        send_events if size >= @max_events
+      end
+
+      self
+    end
+
+    def empty?
+      @pending_events.empty?
+    end
+
+    def size
+      @pending_events.size
+    end
+
+    private
+
+    def send_events
+      @client.send_logs(@pending_events)
+      @pending_events.clear
+    end
+  end
+end

data/lib/sentry/logger.rb

@@ -1,6 +1,6 @@
 # frozen_string_literal: true
 
-require 'logger'
+require "logger"
 
 module Sentry
   class Logger < ::Logger

data/lib/sentry/metrics/aggregator.rb

@@ -0,0 +1,248 @@
+# frozen_string_literal: true
+
+module Sentry
+  module Metrics
+    class Aggregator < ThreadedPeriodicWorker
+      FLUSH_INTERVAL = 5
+      ROLLUP_IN_SECONDS = 10
+
+      # this is how far removed from user code in the backtrace we are
+      # when we record code locations
+      DEFAULT_STACKLEVEL = 4
+
+      KEY_SANITIZATION_REGEX = /[^a-zA-Z0-9_\-.]+/
+      UNIT_SANITIZATION_REGEX = /[^a-zA-Z0-9_]+/
+      TAG_KEY_SANITIZATION_REGEX = /[^a-zA-Z0-9_\-.\/]+/
+
+      TAG_VALUE_SANITIZATION_MAP = {
+        "\n" => "\\n",
+        "\r" => "\\r",
+        "\t" => "\\t",
+        "\\" => "\\\\",
+        "|" => "\\u{7c}",
+        "," => "\\u{2c}"
+      }
+
+      METRIC_TYPES = {
+        c: CounterMetric,
+        d: DistributionMetric,
+        g: GaugeMetric,
+        s: SetMetric
+      }
+
+      # exposed only for testing
+      attr_reader :client, :thread, :buckets, :flush_shift, :code_locations
+
+      def initialize(configuration, client)
+        super(configuration.sdk_logger, FLUSH_INTERVAL)
+        @client = client
+        @before_emit = configuration.metrics.before_emit
+        @enable_code_locations = configuration.metrics.enable_code_locations
+        @stacktrace_builder = configuration.stacktrace_builder
+
+        @default_tags = {}
+        @default_tags["release"] = configuration.release if configuration.release
+        @default_tags["environment"] = configuration.environment if configuration.environment
+
+        @mutex = Mutex.new
+
+        # a nested hash of timestamp -> bucket keys -> Metric instance
+        @buckets = {}
+
+        # the flush interval needs to be shifted once per startup to create jittering
+        @flush_shift = Random.rand * ROLLUP_IN_SECONDS
+
+        # a nested hash of timestamp (start of day) -> meta keys -> frame
+        @code_locations = {}
+      end
+
+      def add(type,
+              key,
+              value,
+              unit: "none",
+              tags: {},
+              timestamp: nil,
+              stacklevel: nil)
+        return unless ensure_thread
+        return unless METRIC_TYPES.keys.include?(type)
+
+        updated_tags = get_updated_tags(tags)
+        return if @before_emit && !@before_emit.call(key, updated_tags)
+
+        timestamp ||= Sentry.utc_now
+
+        # this is integer division and thus takes the floor of the division
+        # and buckets into 10 second intervals
+        bucket_timestamp = (timestamp.to_i / ROLLUP_IN_SECONDS) * ROLLUP_IN_SECONDS
+
+        serialized_tags = serialize_tags(updated_tags)
+        bucket_key = [type, key, unit, serialized_tags]
+
+        added = @mutex.synchronize do
+          record_code_location(type, key, unit, timestamp, stacklevel: stacklevel) if @enable_code_locations
+          process_bucket(bucket_timestamp, bucket_key, type, value)
+        end
+
+        # for sets, we pass on if there was a new entry to the local gauge
+        local_value = type == :s ? added : value
+        process_span_aggregator(bucket_key, local_value)
+      end
+
+      def flush(force: false)
+        flushable_buckets = get_flushable_buckets!(force)
+        code_locations = get_code_locations!
+        return if flushable_buckets.empty? && code_locations.empty?
+
+        envelope = Envelope.new
+
+        unless flushable_buckets.empty?
+          payload = serialize_buckets(flushable_buckets)
+          envelope.add_item(
+            { type: "statsd", length: payload.bytesize },
+            payload
+          )
+        end
+
+        unless code_locations.empty?
+          code_locations.each do |timestamp, locations|
+            payload = serialize_locations(timestamp, locations)
+            envelope.add_item(
+              { type: "metric_meta", content_type: "application/json" },
+              payload
+            )
+          end
+        end
+
+        @client.capture_envelope(envelope)
+      end
+
+      alias_method :run, :flush
+
+      private
+
+      # important to sort for key consistency
+      def serialize_tags(tags)
+        tags.flat_map do |k, v|
+          if v.is_a?(Array)
+            v.map { |x| [k.to_s, x.to_s] }
+          else
+            [[k.to_s, v.to_s]]
+          end
+        end.sort
+      end
+
+      def get_flushable_buckets!(force)
+        @mutex.synchronize do
+          flushable_buckets = {}
+
+          if force
+            flushable_buckets = @buckets
+            @buckets = {}
+          else
+            cutoff = Sentry.utc_now.to_i - ROLLUP_IN_SECONDS - @flush_shift
+            flushable_buckets = @buckets.select { |k, _| k <= cutoff }
+            @buckets.reject! { |k, _| k <= cutoff }
+          end
+
+          flushable_buckets
+        end
+      end
+
+      def get_code_locations!
+        @mutex.synchronize do
+          code_locations = @code_locations
+          @code_locations = {}
+          code_locations
+        end
+      end
+
+      # serialize buckets to statsd format
+      def serialize_buckets(buckets)
+        buckets.map do |timestamp, timestamp_buckets|
+          timestamp_buckets.map do |metric_key, metric|
+            type, key, unit, tags = metric_key
+            values = metric.serialize.join(":")
+            sanitized_tags = tags.map { |k, v| "#{sanitize_tag_key(k)}:#{sanitize_tag_value(v)}" }.join(",")
+
+            "#{sanitize_key(key)}@#{sanitize_unit(unit)}:#{values}|#{type}|\##{sanitized_tags}|T#{timestamp}"
+          end
+        end.flatten.join("\n")
+      end
+
+      def serialize_locations(timestamp, locations)
+        mapping = locations.map do |meta_key, location|
+          type, key, unit = meta_key
+          mri = "#{type}:#{sanitize_key(key)}@#{sanitize_unit(unit)}"
+
+          # note this needs to be an array but it really doesn't serve a purpose right now
+          [mri, [location.merge(type: "location")]]
+        end.to_h
+
+        { timestamp: timestamp, mapping: mapping }
+      end
+
+      def sanitize_key(key)
+        key.gsub(KEY_SANITIZATION_REGEX, "_")
+      end
+
+      def sanitize_unit(unit)
+        unit.gsub(UNIT_SANITIZATION_REGEX, "")
+      end
+
+      def sanitize_tag_key(key)
+        key.gsub(TAG_KEY_SANITIZATION_REGEX, "")
+      end
+
+      def sanitize_tag_value(value)
+        value.chars.map { |c| TAG_VALUE_SANITIZATION_MAP[c] || c }.join
+      end
+
+      def get_transaction_name
+        scope = Sentry.get_current_scope
+        return nil unless scope && scope.transaction_name
+        return nil if scope.transaction_source_low_quality?
+
+        scope.transaction_name
+      end
+
+      def get_updated_tags(tags)
+        updated_tags = @default_tags.merge(tags)
+
+        transaction_name = get_transaction_name
+        updated_tags["transaction"] = transaction_name if transaction_name
+
+        updated_tags
+      end
+
+      def process_span_aggregator(key, value)
+        scope = Sentry.get_current_scope
+        return nil unless scope && scope.span
+        return nil if scope.transaction_source_low_quality?
+
+        scope.span.metrics_local_aggregator.add(key, value)
+      end
+
+      def process_bucket(timestamp, key, type, value)
+        @buckets[timestamp] ||= {}
+
+        if (metric = @buckets[timestamp][key])
+          old_weight = metric.weight
+          metric.add(value)
+          metric.weight - old_weight
+        else
+          metric = METRIC_TYPES[type].new(value)
+          @buckets[timestamp][key] = metric
+          metric.weight
+        end
+      end
+
+      def record_code_location(type, key, unit, timestamp, stacklevel: nil)
+        meta_key =  [type, key, unit]
+        start_of_day = Time.utc(timestamp.year, timestamp.month, timestamp.day).to_i
+
+        @code_locations[start_of_day] ||= {}
+        @code_locations[start_of_day][meta_key] ||= @stacktrace_builder.metrics_code_location(caller[stacklevel || DEFAULT_STACKLEVEL])
+      end
+    end
+  end
+end

data/lib/sentry/metrics/configuration.rb

@@ -0,0 +1,47 @@
+# frozen_string_literal: true
+
+module Sentry
+  module Metrics
+    class Configuration
+      include ArgumentCheckingHelper
+
+      # Enable metrics usage.
+      # Starts a new {Sentry::Metrics::Aggregator} instance to aggregate metrics
+      # and a thread to aggregate flush every 5 seconds.
+      # @return [Boolean]
+      attr_accessor :enabled
+
+      # Enable code location reporting.
+      # Will be sent once per day.
+      # True by default.
+      # @return [Boolean]
+      attr_accessor :enable_code_locations
+
+      # Optional Proc, called before emitting a metric to the aggregator.
+      # Use it to filter keys (return false/nil) or update tags.
+      # Make sure to return true at the end.
+      #
+      # @example
+      #   config.metrics.before_emit = lambda do |key, tags|
+      #     return nil if key == 'foo'
+      #     tags[:bar] = 42
+      #     tags.delete(:baz)
+      #     true
+      #   end
+      #
+      # @return [Proc, nil]
+      attr_reader :before_emit
+
+      def initialize
+        @enabled = false
+        @enable_code_locations = true
+      end
+
+      def before_emit=(value)
+        check_callable!("metrics.before_emit", value)
+
+        @before_emit = value
+      end
+    end
+  end
+end

data/lib/sentry/metrics/counter_metric.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+module Sentry
+  module Metrics
+    class CounterMetric < Metric
+      attr_reader :value
+
+      def initialize(value)
+        @value = value.to_f
+      end
+
+      def add(value)
+        @value += value.to_f
+      end
+
+      def serialize
+        [value]
+      end
+
+      def weight
+        1
+      end
+    end
+  end
+end

data/lib/sentry/metrics/distribution_metric.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+module Sentry
+  module Metrics
+    class DistributionMetric < Metric
+      attr_reader :value
+
+      def initialize(value)
+        @value = [value.to_f]
+      end
+
+      def add(value)
+        @value << value.to_f
+      end
+
+      def serialize
+        value
+      end
+
+      def weight
+        value.size
+      end
+    end
+  end
+end

data/lib/sentry/metrics/gauge_metric.rb

@@ -0,0 +1,35 @@
+# frozen_string_literal: true
+
+module Sentry
+  module Metrics
+    class GaugeMetric < Metric
+      attr_reader :last, :min, :max, :sum, :count
+
+      def initialize(value)
+        value = value.to_f
+        @last = value
+        @min = value
+        @max = value
+        @sum = value
+        @count = 1
+      end
+
+      def add(value)
+        value = value.to_f
+        @last = value
+        @min = [@min, value].min
+        @max = [@max, value].max
+        @sum += value
+        @count += 1
+      end
+
+      def serialize
+        [last, min, max, sum, count]
+      end
+
+      def weight
+        5
+      end
+    end
+  end
+end