datadog 2.8.0 → 2.39.0

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

@@ -6,43 +6,43 @@ module Datadog
       # Helper methods for parsing string values into Numeric
       module Duration
         def self.call(value, base: :s)
-          cast = if value.include?('.')
-                   method(:Float)
-                 else
-                   method(:Integer)
-                 end
+          cast = if value.include?(".")
+            method(:Float)
+          else
+            method(:Integer)
+          end
 
           scale = case base
-                  when :s
-                    1_000_000_000
-                  when :ms
-                    1_000_000
-                  when :us
-                    1000
-                  when :ns
-                    1
-                  else
-                    raise ArgumentError, "invalid base: #{base.inspect}"
-                  end
+          when :s
+            1_000_000_000
+          when :ms
+            1_000_000
+          when :us
+            1000
+          when :ns
+            1
+          else
+            raise ArgumentError, "invalid base: #{base.inspect}"
+          end
 
           result = case value
-                   when /^(\d+(?:\.\d+)?)h$/
-                     cast.call(Regexp.last_match(1)) * 1_000_000_000 * 60 * 60 / scale
-                   when /^(\d+(?:\.\d+)?)m$/
-                     cast.call(Regexp.last_match(1)) * 1_000_000_000 * 60 / scale
-                   when /^(\d+(?:\.\d+)?)s$/
-                     cast.call(Regexp.last_match(1)) * 1_000_000_000 / scale
-                   when /^(\d+(?:\.\d+)?)ms$/
-                     cast.call(Regexp.last_match(1)) * 1_000_000 / scale
-                   when /^(\d+(?:\.\d+)?)us$/
-                     cast.call(Regexp.last_match(1)) * 1_000 / scale
-                   when /^(\d+(?:\.\d+)?)ns$/
-                     cast.call(Regexp.last_match(1)) / scale
-                   when /^(\d+(?:\.\d+)?)$/
-                     cast.call(Regexp.last_match(1))
-                   else
-                     raise ArgumentError, "invalid duration: #{value.inspect}"
-                   end
+          when /^(\d+(?:\.\d+)?)h$/
+            cast.call(Regexp.last_match(1)) * 1_000_000_000 * 60 * 60 / scale
+          when /^(\d+(?:\.\d+)?)m$/
+            cast.call(Regexp.last_match(1)) * 1_000_000_000 * 60 / scale
+          when /^(\d+(?:\.\d+)?)s$/
+            cast.call(Regexp.last_match(1)) * 1_000_000_000 / scale
+          when /^(\d+(?:\.\d+)?)ms$/
+            cast.call(Regexp.last_match(1)) * 1_000_000 / scale
+          when /^(\d+(?:\.\d+)?)us$/
+            cast.call(Regexp.last_match(1)) * 1_000 / scale
+          when /^(\d+(?:\.\d+)?)ns$/
+            cast.call(Regexp.last_match(1)) / scale
+          when /^(\d+(?:\.\d+)?)$/
+            cast.call(Regexp.last_match(1))
+          else
+            raise ArgumentError, "invalid duration: #{value.inspect}"
+          end
           # @type var result: Numeric
           result.round
         end

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

@@ -0,0 +1,29 @@
+# frozen_string_literal: true
+
+module Datadog
+  module Core
+    module Utils
+      # Bring newer Enumerable methods to older Ruby versions.
+      module EnumerableCompat
+        def self.filter_map(array, &block)
+          if array.respond_to?(:filter_map)
+            # DEV Supported since Ruby 2.7, saves an intermediate object creation
+            array.filter_map(&block)
+          elsif array.is_a?(Enumerator::Lazy)
+            # You would think that .compact would work here, but it does not:
+            # the result of .map could be an Enumerator::Lazy instance which
+            # does not implement #compact on Ruby 2.5/2.6.
+            array.map(&block).reject do |item|
+              item.nil?
+            end
+          else
+            array.each_with_object([]) do |item, memo|
+              new_item = block.call(item)
+              memo.push(new_item) unless new_item.nil?
+            end
+          end
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+
+module Datadog
+  module Core
+    module Utils
+      module FNV
+        # FNV-1a 64-bit non-cryptographic hash function.
+        # @see https://en.wikipedia.org/wiki/Fowler%E2%80%93Noll%E2%80%93Vo_hash_function#FNV-1a_hash Algorithm
+        # @see https://en.wikipedia.org/wiki/Fowler%E2%80%93Noll%E2%80%93Vo_hash_function#FNV_hash_parameters Prime and Offset Basis
+
+        FNV_OFFSET_BASIS = 14695981039346656037
+        FNV_PRIME = 1099511628211
+        FNV_64_BIT_MASK = 0xFFFFFFFFFFFFFFFF
+
+        def self.fnv1_64(data)
+          hash_value = FNV_OFFSET_BASIS
+          data.each_byte do |byte|
+            hash_value ^= byte
+            hash_value = (hash_value * FNV_PRIME) & FNV_64_BIT_MASK
+          end
+          hash_value
+        end
+      end
+    end
+  end
+end

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

@@ -1,5 +1,7 @@
 # frozen_string_literal: true
 
+require_relative "../../ruby_version"
+
 module Datadog
   module Core
     module Utils
@@ -45,14 +47,14 @@ module Datadog
           # object will cause forking to not be detected in the fork when it should have.
           #
           # This wrapper prevents this by initializing the fork PID when the object is created.
-          if RUBY_VERSION >= '3'
+          if RubyVersion.is?(">= 3")
             def initialize(*args, **kwargs, &block)
-              super(*args, **kwargs, &block)
+              super
               update_fork_pid!
             end
           else
             def initialize(*args, &block)
-              super(*args, &block)
+              super
               update_fork_pid!
             end
           end

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

@@ -3,30 +3,7 @@
 module Datadog
   module Core
     module Utils
-      # Refinements for {Hash}.
       module Hash
-        # This refinement ensures modern rubies are allowed to use newer,
-        # simpler, and more performant APIs.
-        module Refinement
-          # Introduced in Ruby 2.4
-          unless ::Hash.method_defined?(:compact)
-            refine ::Hash do
-              def compact
-                reject { |_k, v| v.nil? }
-              end
-            end
-          end
-
-          # Introduced in Ruby 2.4
-          unless ::Hash.method_defined?(:compact!)
-            refine ::Hash do
-              def compact!
-                reject! { |_k, v| v.nil? }
-              end
-            end
-          end
-        end
-
         # A minimal {Hash} wrapper that provides case-insensitive access
         # to hash keys, without the overhead of copying the original hash.
         #

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

@@ -0,0 +1,45 @@
+# frozen_string_literal: true
+
+require "forwardable"
+
+module Datadog
+  module Core
+    module Utils
+      # An LRU (Least Recently Used) cache implementation that relies on the
+      # Ruby 1.9+ `Hash` implementation that guarantees insertion order.
+      #
+      # WARNING: This implementation is NOT thread-safe and should be used
+      #          in a single-threaded context or guarded by Mutex.
+      class LRUCache
+        extend Forwardable
+
+        def_delegators :@store, :clear, :empty?
+
+        def initialize(max_size)
+          raise ArgumentError, "max_size must be an Integer" unless max_size.is_a?(Integer)
+          raise ArgumentError, "max_size must be greater than 0" if max_size <= 0
+
+          @max_size = max_size
+          @store = {}
+        end
+
+        # NOTE: Accessing a key moves it to the end of the list.
+        def [](key)
+          if (entry = @store.delete(key))
+            @store[key] = entry
+          end
+        end
+
+        def []=(key, value)
+          if @store.delete(key)
+            @store[key] = value
+          else
+            # NOTE: evict the oldest entry if store reached the maximum allowed size
+            @store.shift if @store.size >= @max_size
+            @store[key] = value
+          end
+        end
+      end
+    end
+  end
+end

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

@@ -1,6 +1,6 @@
 # frozen_string_literal: true
 
-require 'ipaddr'
+require "ipaddr"
 
 module Datadog
   module Core
@@ -13,6 +13,7 @@ module Datadog
           true-client-ip
           x-client-ip
           x-forwarded
+          forwarded
           forwarded-for
           x-cluster-client-ip
           fastly-client-ip
@@ -20,6 +21,8 @@ module Datadog
           cf-connecting-ipv6
         ].freeze
 
+        CGNAT_IP_RANGE = IPAddr.new("100.64.0.0/10")
+
         class << self
           # Returns a client IP associated with the request if it was
           #   retrieved successfully.
@@ -32,7 +35,7 @@ module Datadog
           def stripped_ip_from_request_headers(headers, ip_headers_to_check: DEFAULT_IP_HEADERS_NAMES)
             ip = ip_header(headers, ip_headers_to_check)
 
-            ip ? ip.to_s : nil
+            ip&.to_s
           end
 
           # @param [String] IP value.
@@ -40,7 +43,7 @@ module Datadog
           # @return [nil] when no valid IP value found.
           def stripped_ip(ip)
             ip = ip_to_ipaddr(ip)
-            ip ? ip.to_s : nil
+            ip&.to_s
           end
 
           private
@@ -52,10 +55,10 @@ module Datadog
             return unless ip
 
             clean_ip = if likely_ipv4?(ip)
-                         strip_ipv4_port(ip)
-                       else
-                         strip_zone_specifier(strip_ipv6_port(ip))
-                       end
+              strip_ipv4_port(ip)
+            else
+              strip_zone_specifier(strip_ipv6_port(ip))
+            end
 
             begin
               IPAddr.new(clean_ip)
@@ -72,7 +75,9 @@ module Datadog
 
               next unless value
 
-              ips = value.split(',')
+              ips = value.split(",")
+              ips = process_forwarded_header_values(ips) if name == "forwarded"
+
               ips.each do |ip|
                 parsed_ip = ip_to_ipaddr(ip.strip)
 
@@ -83,20 +88,36 @@ module Datadog
             nil
           end
 
+          def process_forwarded_header_values(values)
+            values.each_with_object([]) do |value, acc|
+              value.downcase!
+
+              value.split(";").each do |tuple_str|
+                tuple_str.strip!
+                next unless tuple_str.start_with?("for=")
+
+                tuple_str.delete_prefix!("for=")
+                tuple_str.delete!('"')
+
+                acc << tuple_str
+              end
+            end
+          end
+
           # Returns whether the given value is more likely to be an IPv4 than an IPv6 address.
           #
           # This is done by checking if a dot (`'.'`) character appears before a colon (`':'`) in the value.
           # 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)
@@ -108,11 +129,11 @@ module Datadog
           end
 
           def strip_ipv4_port(ip)
-            ip.gsub(/:\d+\z/, '')
+            ip.gsub(/:\d+\z/, "")
           end
 
           def global_ip?(parsed_ip)
-            parsed_ip && !parsed_ip.private? && !parsed_ip.loopback? && !parsed_ip.link_local?
+            parsed_ip && !parsed_ip.private? && !parsed_ip.loopback? && !parsed_ip.link_local? && !CGNAT_IP_RANGE.include?(parsed_ip)
           end
         end
       end

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

@@ -3,7 +3,7 @@
 module Datadog
   module Core
     module Utils
-      # Helper class to execute something only once such as not repeating warning logs, and instrumenting classes
+      # Helper class to execute something only once, such as not repeating warning logs and instrumenting classes
       # only once.
       #
       # Thread-safe when used correctly (e.g. be careful of races when lazily initializing instances of this class).

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

@@ -1,24 +1,39 @@
 # frozen_string_literal: true
 
-require_relative 'only_once'
+require_relative "only_once"
 
 module Datadog
   module Core
     module Utils
-      # Helper class to execute something with only one success.
+      # Helper class to execute something with only one successful execution.
       #
-      # This is useful for cases where we want to ensure that a block of code is only executed once, and only if it
-      # succeeds. One such example is sending app-started telemetry event.
+      # If limit is not provided to the constructor, +run+ will execute the
+      # block an unlimited number of times until the block indicates that it
+      # executed successfully by returning a truthy value. After a block
+      # executes successfully, subsequent +run+ calls will not invoke the
+      # block.
       #
-      # Successful execution is determined by the return value of the block: any truthy value is considered success.
+      # If a non-zero limit is provided to the constructor, +run+ will
+      # execute the block up to that many times, and will mark the instance
+      # of OnlyOneSuccessful as failed if none of the executions succeeded.
       #
-      # Thread-safe when used correctly (e.g. be careful of races when lazily initializing instances of this class).
+      # One consumer of this class is sending the app-started telemetry event.
+      #
+      # Successful execution is determined by the return value of the block:
+      # any truthy value is considered success.
+      #
+      # This class is thread-safe (however, instances of it must also be
+      # created in a thread-safe manner).
       #
       # Note: In its current state, this class is not Ractor-safe.
       # In https://github.com/DataDog/dd-trace-rb/pull/1398#issuecomment-797378810 we have a discussion of alternatives,
       # including an alternative implementation that is Ractor-safe once spent.
       class OnlyOnceSuccessful < OnlyOnce
-        def initialize(limit = 0)
+        def initialize(limit = nil)
+          if limit && limit <= 0
+            raise ArgumentError, "Limit must be a positive integer if provided: #{limit}"
+          end
+
           super()
 
           @limit = limit
@@ -52,7 +67,9 @@ module Datadog
 
         private
 
+        # Use this method only after checking that limit is not nil.
         def check_limit!
+          # @type ivar @limit: Integer
           if @retries >= @limit
             @failed = true
             @ran_once = true
@@ -60,7 +77,7 @@ module Datadog
         end
 
         def limited?
-          !@limit.nil? && @limit.positive?
+          !@limit.nil?
         end
 
         def reset_ran_once_state_for_tests

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

@@ -5,8 +5,8 @@ module Datadog
     module Utils
       # Helper methods for safer dup
       module SafeDup
-        # String#+@ was introduced in Ruby 2.3
-        def self.frozen_or_dup(v)
+        # Steep: https://github.com/soutaro/steep/issues/2001
+        def self.frozen_or_dup(v) # steep:ignore MethodBodyTypeMismatch
           # For the case of a String we use the methods +@ and -@.
           # Those methods are only for String objects
           # they are faster and chepaer on the memory side.
@@ -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

@@ -12,6 +12,8 @@ module Datadog
         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 += 1
           next_item

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

@@ -0,0 +1,68 @@
+# frozen_string_literal: true
+
+require_relative "../../ruby_version"
+
+module Datadog
+  module Core
+    module Utils
+      # Applies the Process.spawn wrapper used to merge additional environment variables
+      # into child processes.
+      module SpawnMonkeyPatch
+        # @param env_provider [#call] returns a Hash of env vars to merge into the child process
+        def self.apply!(env_provider:)
+          @env_provider = env_provider
+
+          # Idempotent: tests, reloads, or repeated Components init must not stack prepends.
+          return if ::Process.singleton_class.ancestors.include?(ProcessSpawnPatch)
+
+          ::Process.singleton_class.prepend(ProcessSpawnPatch)
+        end
+
+        # 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")
+            def spawn(*args, **kwargs) # steep:ignore DifferentMethodParameterKind
+              super(*SpawnMonkeyPatch.inject_envs(args), **kwargs)
+            end
+          else
+            def spawn(*args)
+              super(*SpawnMonkeyPatch.inject_envs(args))
+            end
+            ruby2_keywords :spawn if respond_to?(:ruby2_keywords, true)
+          end
+        end
+
+        # Merge the env vars from `env_provider` with the optional env `Hash` from {Process.spawn}.
+        #
+        # `env` is the first argument when it is a {Hash}; see MRI `spawn([env, ] *args, options)`:
+        # https://docs.ruby-lang.org/en/master/Process.html#method-c-spawn
+        #
+        # When there is **no** leading env Hash, MRI inherits the parent's `ENV`; we prepend only the
+        # `env_provider` hash so spawned children see parent env plus injections.
+        #
+        # When callers pass `unsetenv_others: true`, MRI only forwards the explicitly passed env Hash;
+        # replacing a missing hash with DATADOG_ENV.to_h would wrongly carry over parent variables.
+        # Prepending only the provider hash preserves `unsetenv_others` semantics.
+        #
+        # See https://docs.ruby-lang.org/en/master/Process.html#module-Process-label-Environment+Variables+-28-3Aunsetenv_others-29
+        #
+        # NOTE: `::Hash` (not bare `Hash`) is required because this module is nested under
+        # `Datadog::Core::Utils`, and `Datadog::Core::Utils::Hash` exists.
+        # Bare `Hash` resolves to that module via Module.nesting, making `Hash === some_hash`
+        # silently return `false`. See https://github.com/DataDog/dd-trace-rb/issues/5621.
+        def self.inject_envs(args)
+          provided_env = @env_provider.call
+
+          if ::Hash === args.first
+            args[0] = args.first.merge(provided_env)
+          else
+            args.unshift(provided_env)
+          end
+
+          args
+        end
+      end
+    end
+  end
+end

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

@@ -8,11 +8,15 @@ module Datadog
         module_function
 
         # Current monotonic time
+        # On macOS, CLOCK_MONOTONIC only has microsecond precision,
+        # so we use CLOCK_MONOTONIC_RAW which has nanosecond precision instead.
         #
         # @param unit [Symbol] unit for the resulting value, same as ::Process#clock_gettime, defaults to :float_second
-        # @return [Numeric] timestamp in the requested unit, since some unspecified starting point
+        # @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(Process::CLOCK_MONOTONIC, unit)
+          Process.clock_gettime(MONOTONIC_CLOCK_ID, unit)
         end
 
         # Current wall time.
@@ -31,6 +35,16 @@ module Datadog
         #
         # @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
 
@@ -43,6 +57,16 @@ module Datadog
         #
         # @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
 

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

@@ -0,0 +1,21 @@
+# frozen_string_literal: true
+
+module Datadog
+  module Core
+    module Utils
+      # Helper methods for truncating data
+      module Truncation
+        module_function
+
+        def truncate_in_middle(string, max_prefix_length, max_suffix_length)
+          max_length = max_prefix_length + 3 + max_suffix_length
+          if string.length > max_length
+            "#{string[0...max_prefix_length]}...#{string[-max_suffix_length..-1]}"
+          else
+            string
+          end
+        end
+      end
+    end
+  end
+end

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

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