new: prune the cache when too large

micke · Oct 26, 2024 · 4bf848d · 4bf848d
1 parent 96b8368
commit 4bf848d
Show file tree

Hide file tree

Showing 2 changed files with 138 additions and 5 deletions.
diff --git a/lib/valid_email2/address.rb b/lib/valid_email2/address.rb
@@ -8,13 +8,14 @@ module ValidEmail2
   class Address
     attr_accessor :address
 
-    # Cache structure: { domain (String): { records: [], cached_at: Time, ttl: Integer } }
-    @@mx_servers_cache = {}
-    @@mx_or_a_servers_cache = {}
-
     PROHIBITED_DOMAIN_CHARACTERS_REGEX = /[+!_\/\s'`]/
     DEFAULT_RECIPIENT_DELIMITER = '+'
     DOT_DELIMITER = '.'
+    MAX_CACHE_SIZE = 1_000
+
+    # Cache structure: { domain (String): { records: [], cached_at: Time, ttl: Integer } }
+    @@mx_servers_cache = {}
+    @@mx_or_a_servers_cache = {}
 
     def self.prohibited_domain_characters_regex
       @prohibited_domain_characters_regex ||= PROHIBITED_DOMAIN_CHARACTERS_REGEX
@@ -141,6 +142,10 @@ def address_contain_emoticons?(email)
     end
 
     def mx_servers
+      if @@mx_servers_cache.size > MAX_CACHE_SIZE
+        prune_cache(@@mx_servers_cache)
+      end
+
       domain = address.domain.downcase
 
       if @@mx_servers_cache[domain]
@@ -170,6 +175,10 @@ def null_mx?
     end
 
     def mx_or_a_servers
+      if @@mx_or_a_servers_cache.size > MAX_CACHE_SIZE
+        prune_cache(@@mx_or_a_servers_cache)
+      end
+
       domain = address.domain.downcase
 
       if @@mx_or_a_servers_cache[domain]
@@ -194,5 +203,11 @@ def mx_or_a_servers
 
       records
     end
+
+    def prune_cache(cache)
+      entries_sorted_by_cached_at_asc = (cache.sort_by { |_domain, data| data[:cached_at] }).flatten
+      entries_to_remove = entries_sorted_by_cached_at_asc.first(cache.size - MAX_CACHE_SIZE)
+      entries_to_remove.each { |domain| cache.delete(domain) }
+    end
   end
 end
diff --git a/spec/address_spec.rb b/spec/address_spec.rb
@@ -97,6 +97,65 @@
 
         expect(Resolv::DNS).to have_received(:open).once
       end
+
+      it "does not prune the cache when the cache size is less than the max cache size" do
+        expect(email_instance).not_to receive(:prune_cache)
+
+        email_instance.valid_strict_mx?
+      end
+
+      it "prunes the cache when the cache size is greater than the max cache size" do
+        stub_const("#{described_class}::MAX_CACHE_SIZE", 0)
+
+        expect(email_instance).to receive(:prune_cache).with(described_class.class_variable_get(:@@mx_servers_cache)).once
+
+        email_instance.valid_strict_mx?
+        email_instance.valid_strict_mx?
+      end
+
+      it "does not call the MX or A servers lookup when there is a cached entry for the domain and the cache size is less than the max cache size" do
+        stub_const("#{described_class}::MAX_CACHE_SIZE", 1)
+        described_class.class_variable_set(:@@mx_servers_cache, { email_instance.address.domain => { records: mock_mx_records, cached_at: Time.now, ttl: ttl }})
+
+        email_instance.valid_strict_mx?
+
+        expect(Resolv::DNS).not_to have_received(:open)
+      end
+
+      it "calls the MX or A servers lookup when there is a cached entry for the domain but the cache size is greater than the max cache size" do
+        stub_const("#{described_class}::MAX_CACHE_SIZE", 0)
+        described_class.class_variable_set(:@@mx_servers_cache, { email_instance.address.domain => { records: mock_mx_records, cached_at: Time.now, ttl: ttl }})
+
+        email_instance.valid_strict_mx?
+
+        expect(Resolv::DNS).to have_received(:open).once
+      end
+
+      it "does not prune older entries when the cache size is less than the max size" do
+        stub_const("#{described_class}::MAX_CACHE_SIZE", 1)
+        described_class.class_variable_set(:@@mx_servers_cache, {
+          'another_domain.com' => {
+            records: mock_mx_records, cached_at: Time.now - 100, ttl: ttl
+          }
+        })
+
+        email_instance.valid_strict_mx?
+
+        expect(described_class.class_variable_get(:@@mx_servers_cache).keys).to match_array([email_instance.address.domain, 'another_domain.com'])
+      end
+
+      it "prunes older entries when the cache size is greater than the max size" do
+        stub_const("#{described_class}::MAX_CACHE_SIZE", 0)
+        described_class.class_variable_set(:@@mx_servers_cache, {
+          'another_domain.com' => {
+            records: mock_mx_records, cached_at: Time.now - 100, ttl: ttl
+          }
+        })
+
+        email_instance.valid_strict_mx?
+
+        expect(described_class.class_variable_get(:@@mx_servers_cache).keys).to match_array([email_instance.address.domain])
+      end
     end
 
     describe "#valid_mx?" do
@@ -148,11 +207,70 @@
 
       it "calls the MX or A servers lookup when the time since last lookup is greater than the cached ttl entry" do
         described_class.class_variable_set(:@@mx_or_a_servers_cache, { email_instance.address.domain => { records: mock_a_records, cached_at: Time.now - ttl, ttl: ttl }})
-      
+
         email_instance.valid_mx?
 
         expect(Resolv::DNS).to have_received(:open).once
       end
+
+      it "does not prune the cache when the cache size is less than the max cache size" do
+        expect(email_instance).not_to receive(:prune_cache)
+
+        email_instance.valid_mx?
+      end
+
+      it "prunes the cache when the cache size is greater than the max cache size" do
+        stub_const("#{described_class}::MAX_CACHE_SIZE", 0)
+
+        expect(email_instance).to receive(:prune_cache).with(described_class.class_variable_get(:@@mx_or_a_servers_cache)).once
+
+        email_instance.valid_mx?
+        email_instance.valid_mx?
+      end
+
+      it "does not call the MX or A servers lookup when there is a cached entry for the domain and the cache size is less than the max cache size" do
+        stub_const("#{described_class}::MAX_CACHE_SIZE", 1)
+        described_class.class_variable_set(:@@mx_or_a_servers_cache, { email_instance.address.domain => { records: mock_a_records, cached_at: Time.now, ttl: ttl }})
+
+        email_instance.valid_mx?
+
+        expect(Resolv::DNS).not_to have_received(:open)
+      end
+
+      it "calls the MX or A servers lookup when there is a cached entry for the domain but the cache size is greater than the max cache size" do
+        stub_const("#{described_class}::MAX_CACHE_SIZE", 0)
+        described_class.class_variable_set(:@@mx_or_a_servers_cache, { email_instance.address.domain => { records: mock_a_records, cached_at: Time.now, ttl: ttl }})
+
+        email_instance.valid_mx?
+
+        expect(Resolv::DNS).to have_received(:open).once
+      end
+
+      it "does not prune older entries when the cache size is less than the max size" do
+        stub_const("#{described_class}::MAX_CACHE_SIZE", 1)
+        described_class.class_variable_set(:@@mx_or_a_servers_cache, {
+          'another_domain.com' => {
+            records: mock_a_records, cached_at: Time.now - 100, ttl: ttl
+          }
+        })
+
+        email_instance.valid_mx?
+
+        expect(described_class.class_variable_get(:@@mx_or_a_servers_cache).keys).to match_array([email_instance.address.domain, 'another_domain.com'])
+      end
+
+      it "prunes older entries when the cache size is greater than the max size" do
+        stub_const("#{described_class}::MAX_CACHE_SIZE", 0)
+        described_class.class_variable_set(:@@mx_or_a_servers_cache, {
+          'another_domain.com' => {
+            records: mock_a_records, cached_at: Time.now - 100, ttl: ttl
+          }
+        })
+
+        email_instance.valid_mx?
+
+        expect(described_class.class_variable_get(:@@mx_or_a_servers_cache).keys).to match_array([email_instance.address.domain])
+      end
     end
   end
 end