Add possibility to cache dns lookups

Add optional argument dns_resolver_cache to validate_email. dns_resolver_cache can be either an instance of dns.resolver.Cache or dns.resolver.LRUCache. If passed lookups will be cached according to the specified class.

dns.resolver.Cache can take a cleaning_interval argument and dns.resolver.LRUCache can take a max_size argument.

This can lead to fewer dns lookups needed since many domains of large email providers will likely be in cache.

email_validator/__init__.py

@@ -186,6 +186,7 @@ def validate_email(
     allow_empty_local=False,
     check_deliverability=True,
     timeout=DEFAULT_TIMEOUT,
+    dns_resolver_cache=None
 ):
     """
     Validates an email address, raising an EmailNotValidError if the address is not valid or returning a dict of
@@ -273,7 +274,9 @@ def validate_email(
     if check_deliverability:
         # Validate the email address's deliverability and update the
         # return dict with metadata.
-        deliverability_info = validate_email_deliverability(ret["domain"], ret["domain_i18n"], timeout)
+        deliverability_info = validate_email_deliverability(
+            ret["domain"], ret["domain_i18n"], timeout, dns_resolver_cache
+        )
         if "mx" in deliverability_info:
             ret.mx = deliverability_info["mx"]
             ret.mx_fallback_type = deliverability_info["mx-fallback"]
@@ -443,7 +446,12 @@ def validate_email_domain_part(domain):
     }
 
 
-def validate_email_deliverability(domain, domain_i18n, timeout=DEFAULT_TIMEOUT):
+def validate_email_deliverability(
+    domain,
+    domain_i18n,
+    timeout=DEFAULT_TIMEOUT,
+    dns_resolver_cache=None
+):
     # Check that the domain resolves to an MX record. If there is no MX record,
     # try an A or AAAA record which is a deprecated fallback for deliverability.
 
@@ -470,6 +478,8 @@ def dns_resolver_resolve_shim(resolver, domain, record):
             raise dns.exception.Timeout()
 
         resolver = dns.resolver.get_default_resolver()
+        if resolver.cache is None and dns_resolver_cache is not None:
+            resolver.cache = dns_resolver_cache
 
         if timeout:
             resolver.lifetime = timeout

tests/test_main.py

@@ -1,3 +1,5 @@
+from unittest import mock
+import dns.resolver
 import pytest
 from email_validator import EmailSyntaxError, EmailUndeliverableError, \
                             validate_email, validate_email_deliverability, \
@@ -344,3 +346,17 @@ def test_main_output_shim(monkeypatch, capsys):
     # The \n is part of the print statement, not part of the string, which is what the b'...' is
     # Since we're mocking py 2.7 here instead of actually using 2.7, this was the closest I could get
     assert stdout == "b'An email address cannot have a period immediately after the @-sign.'\n"
+
+
+@mock.patch("dns.resolver.LRUCache.put")
+def test_validate_email__with_dns_resolver_cache(mocked_put):
+    validate_email("test@gmail.com", dns_resolver_cache=dns.resolver.LRUCache(max_size=10000))
+    assert mocked_put.called
+
+    # Check that global resolver has the cache class set
+    resolver = dns.resolver.get_default_resolver()
+    assert resolver.cache.__class__ == dns.resolver.LRUCache
+
+    with mock.patch("dns.resolver.LRUCache.get") as mocked_get:
+        validate_email("test@gmail.com", dns_resolver_cache=dns.resolver.LRUCache(max_size=10000))
+        assert mocked_get.called