datadog 2.38.0 → 2.47.0

data/lib/datadog/core/utils/network.rb

@@ -1,6 +1,6 @@
 # frozen_string_literal: true
 
-require 'ipaddr'
+require "ipaddr"
 
 module Datadog
   module Core
@@ -21,7 +21,7 @@ module Datadog
           cf-connecting-ipv6
         ].freeze
 
-        CGNAT_IP_RANGE = IPAddr.new('100.64.0.0/10')
+        CGNAT_IP_RANGE = IPAddr.new("100.64.0.0/10")
 
         class << self
           # Returns a client IP associated with the request if it was
@@ -75,8 +75,8 @@ module Datadog
 
               next unless value
 
-              ips = value.split(',')
-              ips = process_forwarded_header_values(ips) if name == 'forwarded'
+              ips = value.split(",")
+              ips = process_forwarded_header_values(ips) if name == "forwarded"
 
               ips.each do |ip|
                 parsed_ip = ip_to_ipaddr(ip.strip)
@@ -92,11 +92,11 @@ module Datadog
             values.each_with_object([]) do |value, acc|
               value.downcase!
 
-              value.split(';').each do |tuple_str|
+              value.split(";").each do |tuple_str|
                 tuple_str.strip!
-                next unless tuple_str.start_with?('for=')
+                next unless tuple_str.start_with?("for=")
 
-                tuple_str.delete_prefix!('for=')
+                tuple_str.delete_prefix!("for=")
                 tuple_str.delete!('"')
 
                 acc << tuple_str
@@ -110,14 +110,14 @@ module Datadog
           # The rationale is that in valid IPv6 addresses, colons will always preced dots,
           #   and in valid IPv4 addresses dots will always preced colons.
           def likely_ipv4?(value)
-            dot_index = value.index('.') || value.size
-            colon_index = value.index(':') || value.size
+            dot_index = value.index(".") || value.size
+            colon_index = value.index(":") || value.size
 
             dot_index < colon_index
           end
 
           def strip_zone_specifier(ipv6)
-            ipv6.gsub(/%.*/, '')
+            ipv6.gsub(/%.*/, "")
           end
 
           def strip_ipv6_port(ip)
@@ -129,7 +129,7 @@ module Datadog
           end
 
           def strip_ipv4_port(ip)
-            ip.gsub(/:\d+\z/, '')
+            ip.gsub(/:\d+\z/, "")
           end
 
           def global_ip?(parsed_ip)

data/lib/datadog/core/utils/only_once_successful.rb

@@ -1,6 +1,6 @@
 # frozen_string_literal: true
 
-require_relative 'only_once'
+require_relative "only_once"
 
 module Datadog
   module Core

data/lib/datadog/core/utils/safe_dup.rb

@@ -22,7 +22,8 @@ module Datadog
           end
         end
 
-        def self.frozen_dup(v)
+        # Steep: https://github.com/soutaro/steep/issues/2001
+        def self.frozen_dup(v) # steep:ignore MethodBodyTypeMismatch
           # For the case of a String we use the methods -@
           # That method are only for String objects
           # they are faster and chepaer on the memory side.

data/lib/datadog/core/utils/sequence.rb

@@ -5,18 +5,15 @@ module Datadog
     module Utils
       # Generates values from a consistent sequence
       class Sequence
-        def initialize(seed = 0, &block)
+        def initialize(seed = 0)
           @seed = seed
           @current = seed
-          @next_item = block
         end
 
         def next
-          # Steep: https://github.com/soutaro/steep/issues/477
-          # @type ivar @next_item: ^(::Integer) -> Integer
-          next_item = @next_item ? @next_item.call(@current) : @current
+          current = @current
           @current += 1
-          next_item
+          current
         end
 
         def reset!

data/lib/datadog/core/utils/spawn_monkey_patch.rb

@@ -1,6 +1,6 @@
 # frozen_string_literal: true
 
-require_relative '../../ruby_version'
+require_relative "../../ruby_version"
 
 module Datadog
   module Core
@@ -21,7 +21,7 @@ module Datadog
         # Prepends `Process.spawn` to merge `env_provider` output into the child's environment hash.
         module ProcessSpawnPatch
           # The One and Only Correct Delegation Pattern
-          if RubyVersion.is?('>= 3')
+          if RubyVersion.is?(">= 3")
             def spawn(*args, **kwargs) # steep:ignore DifferentMethodParameterKind
               super(*SpawnMonkeyPatch.inject_envs(args), **kwargs)
             end

data/lib/datadog/core/utils/time.rb

@@ -5,7 +5,16 @@ module Datadog
     module Utils
       # Common database-related utility functions.
       module Time
-        module_function
+        # timecop monkey-patches Time.now, Process.clock_gettime, etc:
+        # https://github.com/travisjeffery/timecop/blob/v0.9.11/lib/timecop/time_extensions.rb
+        # It keeps the original methods as aliases.
+        # We want the real not-mocked time here, so we use those aliases to the original method if present.
+        # If not present we assume these methods have not been monkey-patched and are the original.
+
+        original_clock_gettime_name = Process.respond_to?(:clock_gettime_without_mock) ? :clock_gettime_without_mock : :clock_gettime
+        define_singleton_method(:original_clock_gettime, &Process.method(original_clock_gettime_name))
+
+        MONOTONIC_CLOCK_ID = RUBY_PLATFORM.include?("darwin") ? Process::CLOCK_MONOTONIC_RAW : Process::CLOCK_MONOTONIC
 
         # Current monotonic time
         # On macOS, CLOCK_MONOTONIC only has microsecond precision,
@@ -13,71 +22,24 @@ module Datadog
         #
         # @param unit [Symbol] unit for the resulting value, same as ::Process#clock_gettime, defaults to :float_second
         # @return [Float|Integer] timestamp in the requested unit, since some unspecified starting point
-        MONOTONIC_CLOCK_ID = RUBY_PLATFORM.include?("darwin") ? Process::CLOCK_MONOTONIC_RAW : Process::CLOCK_MONOTONIC
-
-        def get_time(unit = :float_second)
-          Process.clock_gettime(MONOTONIC_CLOCK_ID, unit)
+        def self.get_time(unit = :float_second)
+          original_clock_gettime(MONOTONIC_CLOCK_ID, unit)
         end
 
         # Current wall time.
         #
         # @return [Time] current time object
-        def now
-          ::Time.now
-        end
-
-        # Overrides the implementation of `#now
-        # with the provided callable.
-        #
-        # Overriding the method `#now` instead of
-        # indirectly calling `block` removes
-        # one level of method call overhead.
-        #
-        # @param block [Proc] block that returns a `Time` object representing the current wall time
-        def now_provider=(block)
-          class << self
-            # Avoid method redefinition warning.
-            # `rescue nil` is added in case customers remove the method
-            # themselves to squelch the warning.
-            begin
-              remove_method(:now)
-            rescue
-              nil
-            end
-          end
-          define_singleton_method(:now, &block)
-        end
-
-        # Overrides the implementation of `#get_time
-        # with the provided callable.
-        #
-        # Overriding the method `#get_time` instead of
-        # indirectly calling `block` removes
-        # one level of method call overhead.
-        #
-        # @param block [Proc] block that accepts unit and returns timestamp in the requested unit
-        def get_time_provider=(block)
-          class << self
-            # Avoid method redefinition warning
-            # `rescue nil` is added in case customers remove the method
-            # themselves to squelch the warning.
-            begin
-              remove_method(:get_time)
-            rescue
-              nil
-            end
-          end
-          define_singleton_method(:get_time, &block)
-        end
+        original_time_now_name = ::Time.respond_to?(:now_without_mock_time) ? :now_without_mock_time : :now
+        define_singleton_method(:now, &::Time.method(original_time_now_name))
 
-        def measure(unit = :float_second)
+        def self.measure(unit = :float_second)
           before = get_time(unit)
           yield
           after = get_time(unit)
           after - before
         end
 
-        def as_utc_epoch_ns(time)
+        def self.as_utc_epoch_ns(time)
           # we use #to_r instead of #to_f because Float doesn't have enough precision to represent exact nanoseconds, see
           # https://rubyapi.org/3.0/o/time#method-i-to_f
           (time.to_r * 1_000_000_000).to_i

data/lib/datadog/core/utils/url.rb

@@ -1,6 +1,6 @@
 # frozen_string_literal: true
 
-require 'uri'
+require "uri"
 
 module Datadog
   module Core

data/lib/datadog/core/utils.rb

@@ -1,6 +1,6 @@
 # frozen_string_literal: true
 
-require_relative 'utils/forking'
+require_relative "utils/forking"
 
 module Datadog
   module Core
@@ -9,7 +9,7 @@ module Datadog
     module Utils
       extend Forking
 
-      EMPTY_STRING = ''.encode(::Encoding::UTF_8).freeze
+      EMPTY_STRING = "".encode(::Encoding::UTF_8).freeze
 
       # Stringifies `value` and ensures the outcome is
       # string is no longer than `size`.
@@ -20,7 +20,7 @@ module Datadog
       # If `omission.size` is larger than `size`, the output
       # will still be `omission.size` in length.
       # @!visibility private
-      def self.truncate(value, size, omission = '...')
+      def self.truncate(value, size, omission = "...")
         string = value.to_s
 
         return string if string.size <= size
@@ -52,7 +52,7 @@ module Datadog
         if binary
           # This option is useful for "gracefully" displaying binary data that
           # often contains text such as marshalled objects
-          str.encode('UTF-8', 'binary', invalid: :replace, undef: :replace, replace: '')
+          str.encode("UTF-8", "binary", invalid: :replace, undef: :replace, replace: "")
         elsif str.encoding == ::Encoding::UTF_8
           str
         elsif str.empty?

data/lib/datadog/core/workers/async.rb

@@ -1,6 +1,6 @@
 # frozen_string_literal: true
 
-require_relative '../logger'
+require_relative "../logger"
 
 module Datadog
   module Core

data/lib/datadog/core/workers/polling.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
-require_relative 'async'
-require_relative 'interval_loop'
+require_relative "async"
+require_relative "interval_loop"
 
 module Datadog
   module Core

data/lib/datadog/core/workers/runtime_metrics.rb

@@ -1,10 +1,10 @@
 # frozen_string_literal: true
 
-require_relative '../runtime/metrics'
+require_relative "../runtime/metrics"
 
-require_relative '../worker'
-require_relative 'async'
-require_relative 'polling'
+require_relative "../worker"
+require_relative "async"
+require_relative "polling"
 
 module Datadog
   module Core
@@ -32,7 +32,7 @@ module Datadog
           end
 
           # Workers::Async::Thread settings
-          self.fork_policy = options.fetch(:fork_policy, Workers::Async::Thread::FORK_POLICY_STOP)
+          self.fork_policy = options.fetch(:fork_policy, Workers::Async::Thread::FORK_POLICY_RESTART)
 
           # Workers::IntervalLoop settings
           self.loop_base_interval = options.fetch(:interval, DEFAULT_FLUSH_INTERVAL)

data/lib/datadog/core.rb

@@ -1,11 +1,11 @@
 # frozen_string_literal: true
 
 # Keep this at the top, this is needed at require-time by some files
-require_relative 'ruby_version'
+require_relative "ruby_version"
 
-require_relative 'core/deprecations'
-require_relative 'core/configuration/config_helper'
-require_relative 'core/extensions'
+require_relative "core/deprecations"
+require_relative "core/configuration/config_helper"
+require_relative "core/extensions"
 
 # We must load core extensions to make certain global APIs
 # accessible: both for Datadog features and the core itself.

data/lib/datadog/data_streams/configuration.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
-require_relative '../core/environment/variable_helpers'
-require_relative 'ext'
+require_relative "../core/environment/variable_helpers"
+require_relative "ext"
 
 module Datadog
   module DataStreams
@@ -38,7 +38,7 @@ module Datadog
               # @!visibility private
               option :interval do |o|
                 o.type :float
-                o.env '_DD_TRACE_STATS_WRITER_INTERVAL'
+                o.env "_DD_TRACE_STATS_WRITER_INTERVAL"
                 o.default 10.0
               end
             end

data/lib/datadog/data_streams/ext.rb

@@ -5,7 +5,7 @@ module Datadog
     # Constants for Data Streams Monitoring configuration
     # @public_api
     module Ext
-      ENV_ENABLED = 'DD_DATA_STREAMS_ENABLED'
+      ENV_ENABLED = "DD_DATA_STREAMS_ENABLED"
     end
   end
 end

data/lib/datadog/data_streams/extensions.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
-require_relative 'configuration'
-require_relative '../core/configuration'
+require_relative "configuration"
+require_relative "../core/configuration"
 
 module Datadog
   module DataStreams

data/lib/datadog/data_streams/pathway_context.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
-require 'stringio'
-require_relative '../core/utils/base64_codec'
+require "stringio"
+require_relative "../core/utils/base64_codec"
 
 module Datadog
   module DataStreams
@@ -58,7 +58,7 @@ module Datadog
         # - 8 bytes: hash value (little-endian)
         # - VarInt: pathway start time (milliseconds)
         # - VarInt: current edge start time (milliseconds)
-        [@hash].pack('Q') <<
+        [@hash].pack("Q") <<
           encode_var_int_64(time_to_ms(@pathway_start)) <<
           encode_var_int_64(time_to_ms(@current_edge_start))
       end
@@ -73,7 +73,7 @@ module Datadog
         hash_bytes = reader.read(8)
         return nil unless hash_bytes
 
-        hash_value = hash_bytes.unpack1('Q') # : Integer
+        hash_value = hash_bytes.unpack1("Q") # : Integer
 
         # Extract pathway start time (VarInt milliseconds)
         pathway_start_ms = decode_varint(reader)
@@ -98,40 +98,38 @@ module Datadog
       end
       private_class_method :decode
 
-      # Encode an unsigned 64-bit integer using LEB128 variable-length encoding.
+      # Encode a signed 64-bit integer using ZigZag + LEB128 variable-length encoding.
       #
-      # This implements unsigned LEB128 (Little Endian Base 128) encoding as specified
-      # in DWARF5 standard section 7.6:
-      # https://dwarfstd.org/doc/DWARF5.pdf#page=301
+      # dd-trace-go's DSM pathway encoder (internal/datastreams/propagator.go, via
+      # sketches-go's ddsketch/encoding package) ZigZag-maps the signed value before
+      # applying plain unsigned LEB128 (DWARF5 section 7.6:
+      # https://dwarfstd.org/doc/DWARF5.pdf#page=301). This implementation previously
+      # skipped the ZigZag step, which silently decoded to the wrong timestamp on any
+      # dd-trace-go consumer -- see https://github.com/DataDog/dd-trace-go/issues/5163.
       #
-      # Each byte uses 7 bits for data and 1 bit to indicate continuation.
-      # The high bit is set if more bytes follow, clear for the final byte.
+      # ZigZag maps signed values to unsigned ones so small-magnitude negative numbers
+      # still encode to few bytes: 0, -1, 1, -2, 2, ... -> 0, 1, 2, 3, 4, ...
       #
-      # @param value [Integer] Unsigned integer value to encode
+      # @param value [Integer] Signed integer value to encode
       # @return [String] Binary string of encoded bytes
       def encode_var_int_64(value)
+        zigzag = (value << 1) ^ (value >> 63)
         bytes = []
-        while value >= 0x80
-          bytes << ((value & 0x7F) | 0x80)
-          value >>= 7
+        while zigzag >= 0x80
+          bytes << ((zigzag & 0x7F) | 0x80)
+          zigzag >>= 7
         end
-        bytes << value
-        bytes.pack('C*')
+        bytes << zigzag
+        bytes.pack("C*")
       end
 
-      # Decode an unsigned LEB128 variable-length integer from IO stream.
+      # Decode a signed 64-bit integer encoded with ZigZag + LEB128 from an IO stream.
       #
-      # This implements unsigned LEB128 (Little Endian Base 128) decoding as specified
-      # in DWARF5 standard section 7.6:
-      # https://dwarfstd.org/doc/DWARF5.pdf#page=301
-      #
-      # VarInt format: Each byte uses 7 bits for data, 1 bit for continuation
-      # - High bit set = more bytes follow
-      # - High bit clear = final byte
-      # - Data bits accumulated in little-endian order
+      # See #encode_var_int_64 for why the ZigZag un-map is required to match
+      # dd-trace-go's wire format.
       #
       # @param io [StringIO] IO stream to read from
-      # @return [Integer, nil] Decoded unsigned integer, or nil if malformed
+      # @return [Integer, nil] Decoded signed integer, or nil if malformed
       def self.decode_varint(io)
         value = 0
         shift = 0
@@ -143,7 +141,7 @@ module Datadog
           value |= (byte & 0x7F) << shift
 
           # If high bit is clear, we're done
-          return value unless (byte & 0x80).nonzero?
+          return (value >> 1) ^ -(value & 1) unless (byte & 0x80).nonzero?
 
           shift += 7