Cache header normalization to reduce object allocation

lib/http/header_normalizer.rb

@@ -0,0 +1,87 @@
+# frozen_string_literal: true
+
+module HTTP
+  class HeaderNormalizer
+    # Matches HTTP header names when in "Canonical-Http-Format"
+    CANONICAL_NAME_RE = /\A[A-Z][a-z]*(?:-[A-Z][a-z]*)*\z/
+
+    # Matches valid header field name according to RFC.
+    # @see http://tools.ietf.org/html/rfc7230#section-3.2
+    COMPLIANT_NAME_RE = /\A[A-Za-z0-9!#$%&'*+\-.^_`|~]+\z/
+
+    MAX_CACHE_SIZE = 200
+
+    def initialize
+      @cache = LRUCache.new(MAX_CACHE_SIZE)
+    end
+
+    # Transforms `name` to canonical HTTP header capitalization
+    def normalize(name)
+      @cache[name] ||= normalize_header(name)
+    end
+
+    private
+
+    # Transforms `name` to canonical HTTP header capitalization
+    #
+    # @param [String] name
+    # @raise [HeaderError] if normalized name does not
+    #   match {COMPLIANT_NAME_RE}
+    # @return [String] canonical HTTP header name
+    def normalize_header(name)
+      return name if CANONICAL_NAME_RE.match?(name)
+
+      normalized = name.split(/[\-_]/).each(&:capitalize!).join("-")
+
+      return normalized if COMPLIANT_NAME_RE.match?(normalized)
+
+      raise HeaderError, "Invalid HTTP header field name: #{name.inspect}"
+    end
+
+    class LRUCache
+      def initialize(max_size)
+        @max_size = max_size
+        @cache = {}
+        @order = []
+      end
+
+      def get(key)
+        return unless @cache.key?(key)
+
+        # Move the accessed item to the end of the order array
+        @order.delete(key)
+        @order.push(key)
+        @cache[key]
+      end
+
+      def set(key, value)
+        @cache[key] = value
+        @order.push(key)
+
+        # Maintain cache size
+        return unless @order.size > @max_size
+
+        oldest = @order.shift
+        @cache.delete(oldest)
+      end
+
+      def size
+        @cache.size
+      end
+
+      def key?(key)
+        @cache.key?(key)
+      end
+
+      def [](key)
+        get(key)
+      end
+
+      def []=(key, value)
+        set(key, value)
+      end
+    end
+
+    private_constant :LRUCache
+  end
+end

lib/http/headers.rb

@@ -5,20 +5,14 @@
 require "http/errors"
 require "http/headers/mixin"
 require "http/headers/known"
+require "http/header_normalizer"
 
 module HTTP
   # HTTP Headers container.
   class Headers
     extend Forwardable
     include Enumerable
 
-    # Matches HTTP header names when in "Canonical-Http-Format"
-    CANONICAL_NAME_RE = /\A[A-Z][a-z]*(?:-[A-Z][a-z]*)*\z/
-
-    # Matches valid header field name according to RFC.
-    # @see http://tools.ietf.org/html/rfc7230#section-3.2
-    COMPLIANT_NAME_RE = /\A[A-Za-z0-9!#$%&'*+\-.^_`|~]+\z/
-
     # Class constructor.
     def initialize
       # The @pile stores each header value using a three element array:
@@ -219,20 +213,15 @@ def coerce(object)
 
     private
 
+    class << self
+      def header_normalizer
+        @header_normalizer ||= HeaderNormalizer.new
+      end
+    end
+
     # Transforms `name` to canonical HTTP header capitalization
-    #
-    # @param [String] name
-    # @raise [HeaderError] if normalized name does not
-    #   match {HEADER_NAME_RE}
-    # @return [String] canonical HTTP header name
     def normalize_header(name)
-      return name if CANONICAL_NAME_RE.match?(name)
-
-      normalized = name.split(/[\-_]/).each(&:capitalize!).join("-")
-
-      return normalized if COMPLIANT_NAME_RE.match?(normalized)
-
-      raise HeaderError, "Invalid HTTP header field name: #{name.inspect}"
+      self.class.header_normalizer.normalize(name)
     end
 
     # Ensures there is no new line character in the header value

spec/lib/http/header_normalizer_spec.rb

@@ -0,0 +1,24 @@
+# frozen_string_literal: true
+
+RSpec.describe HTTP::HeaderNormalizer do
+  subject(:normalizer) { described_class.new }
+
+  describe "#normalize" do
+    it "normalizes the header" do
+      expect(normalizer.normalize("content_type")).to eq "Content-Type"
+    end
+
+    it "caches normalized headers" do
+      object_id = normalizer.normalize("content_type").object_id
+      expect(object_id).to eq normalizer.normalize("content_type").object_id
+    end
+
+    it "only caches up to MAX_CACHE_SIZE headers" do
+      (1..described_class::MAX_CACHE_SIZE + 1).each do |i|
+        normalizer.normalize("header#{i}")
+      end
+
+      expect(normalizer.instance_variable_get(:@cache).size).to eq described_class::MAX_CACHE_SIZE
+    end
+  end
+end