Merge pull request #736 from readthedocs/davidfischer/publisher-allow…

…ed-domains

Publisher allowed domains

adserver/migrations/0082_publisher_allowed_domains.py

@@ -0,0 +1,23 @@
+# Generated by Django 3.2.18 on 2023-04-18 16:54
+from django.db import migrations
+from django.db import models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('adserver', '0081_rollout_ad_prioritization_pacing'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='historicalpublisher',
+            name='allowed_domains',
+            field=models.CharField(blank=True, default='', help_text="A space separated list of domains where the publisher's ads can appear", max_length=1024, verbose_name='Allowed domains'),
+        ),
+        migrations.AddField(
+            model_name='publisher',
+            name='allowed_domains',
+            field=models.CharField(blank=True, default='', help_text="A space separated list of domains where the publisher's ads can appear", max_length=1024, verbose_name='Allowed domains'),
+        ),
+    ]

adserver/models.py

@@ -282,6 +282,17 @@ class Publisher(TimeStampedModel, IndestructibleModel):
         blank=True,
     )
 
+    # If this is blank, all domains are allowed
+    allowed_domains = models.CharField(
+        _("Allowed domains"),
+        max_length=1024,
+        help_text=_(
+            "A space separated list of domains where the publisher's ads can appear"
+        ),
+        default="",
+        blank=True,
+    )
+
     unauthed_ad_decisions = models.BooleanField(
         default=True,
         help_text=_(

adserver/tests/test_utils.py

@@ -19,6 +19,7 @@
 from ..utils import get_client_user_agent
 from ..utils import get_geoipdb_geolocation
 from ..utils import get_geolocation
+from ..utils import is_allowed_domain
 from ..utils import is_blocklisted_ip
 from ..utils import is_blocklisted_referrer
 from ..utils import is_blocklisted_user_agent
@@ -194,3 +195,26 @@ def test_parse_date_string(self):
             parse_date_string("2020-01-01"),
             datetime.datetime(year=2020, month=1, day=1, tzinfo=pytz.utc),
         )
+
+    def test_is_allowed_domain(self):
+        # Check some nulls
+        self.assertTrue(is_allowed_domain("", None))
+        self.assertTrue(is_allowed_domain("http://example.com", None))
+        self.assertTrue(is_allowed_domain(None, ("not-example.com",)))
+
+        self.assertTrue(
+            is_allowed_domain("http://example.com", ("not-example.com", "example.com"))
+        )
+        self.assertTrue(
+            is_allowed_domain(
+                "https://example.com/path.html", ("not-example.com", "example.com")
+            )
+        )
+
+        self.assertFalse(is_allowed_domain("http://example.com", ("not-example.com",)))
+        self.assertFalse(
+            is_allowed_domain("https://example.com/path.html", ("not-example.com",))
+        )
+
+        # Subdomains aren't included by default
+        self.assertFalse(is_allowed_domain("http://www.example.com", ("example.com",)))

adserver/utils.py

@@ -293,6 +293,20 @@ def is_proxy_ip(ip):
     return False
 
 
+def is_allowed_domain(url, allowed_domains):
+    """Check if a domain is in the domain allowed list."""
+    url_domain = get_domain_from_url(url)
+
+    # If there's no domain restrictions, all URLs are allowed
+    if not allowed_domains or not url_domain:
+        return True
+
+    if url_domain in allowed_domains:
+        return True
+
+    return False
+
+
 def get_geolocation(request, force=False):
     """Gets the geolocation for this IP address."""
     if force:

adserver/views.py

@@ -104,6 +104,7 @@
 from .utils import get_client_ip
 from .utils import get_client_user_agent
 from .utils import get_geolocation
+from .utils import is_allowed_domain
 from .utils import is_blocklisted_ip
 from .utils import is_blocklisted_referrer
 from .utils import is_blocklisted_user_agent
@@ -785,6 +786,16 @@ def ignore_tracking_reason(self, request, advertisement, offer):
                 user_agent,
             )
             reason = "Mismatched browser"
+        elif offer.publisher.allowed_domains and not is_allowed_domain(
+            offer.url, offer.publisher.allowed_domains.split()
+        ):
+            log.log(
+                self.log_security_level,
+                "Offer URL is not on the allowed domain list. Publisher: [%s], Offer URL: [%s]",
+                offer.publisher,
+                offer.url,
+            )
+            # Note: this does not set a reason so it only logs mismatches
 
         # This is out of the elif block and will be run everytime
         if offer and offer.ip != anonymize_ip_address(ip_address):