Do not DNSBL check invalid domains (#1107)

crates/common/src/config/spamfilter.rs

@@ -821,6 +821,18 @@ impl Element {
             Element::Any => map,
         }
     }
+
+    pub fn as_str(&self) -> &'static str {
+        match self {
+            Element::Url => "url",
+            Element::Domain => "domain",
+            Element::Email => "email",
+            Element::Ip => "ip",
+            Element::Header => "header",
+            Element::Body => "body",
+            Element::Any => "any",
+        }
+    }
 }
 
 pub struct IpResolver {

crates/spam-filter/src/analysis/domain.rs

@@ -165,14 +165,12 @@ impl SpamFilterAnalyzeDomain for Server {
                             attributes.iter().find_map(|(attr, value)| {
                                 if *attr == HREF {
                                     let value = value.as_deref()?.strip_prefix("mailto:")?;
-                                    if value.contains('@') {
+                                    let email =
+                                        Email::new(value.split_once('?').map_or(value, |(e, _)| e));
+
+                                    if email.is_valid() {
                                         return Some(ElementLocation::new(
-                                            Recipient {
-                                                email: Email::new(
-                                                    value.split_once('?').map_or(value, |(e, _)| e),
-                                                ),
-                                                name: None,
-                                            },
+                                            Recipient { email, name: None },
                                             if is_body {
                                                 Location::BodyHtml
                                             } else {
@@ -203,17 +201,19 @@ impl SpamFilterAnalyzeDomain for Server {
                         }
                     }
 
-                    emails.insert(ElementLocation::new(
-                        Recipient {
-                            email: email.clone(),
-                            name: None,
-                        },
-                        if is_body {
-                            Location::BodyText
-                        } else {
-                            Location::Attachment
-                        },
-                    ));
+                    if email.is_valid() {
+                        emails.insert(ElementLocation::new(
+                            Recipient {
+                                email: email.clone(),
+                                name: None,
+                            },
+                            if is_body {
+                                Location::BodyText
+                            } else {
+                                Location::Attachment
+                            },
+                        ));
+                    }
                 }
             }
         }

crates/spam-filter/src/analysis/url.rs

@@ -250,14 +250,16 @@ impl SpamFilterAnalyzeUrl for Server {
                 }
 
                 // Check Domain DNSBL
-                check_dnsbl(
-                    self,
-                    ctx,
-                    &StringResolver(host.sld_or_default()),
-                    Element::Domain,
-                    el.location,
-                )
-                .await;
+                if let Some(sld) = &host.sld {
+                    check_dnsbl(
+                        self,
+                        ctx,
+                        &StringResolver(sld),
+                        Element::Domain,
+                        el.location,
+                    )
+                    .await;
+                }
             } else {
                 // URL is an ip address
                 ctx.result.add_tag("SUSPICIOUS_URL");

crates/spam-filter/src/modules/dnsbl.rs

@@ -55,6 +55,7 @@ pub(crate) async fn check_dnsbl(
                 server,
                 dnsbl,
                 SpamFilterResolver::new(ctx, resolver, location),
+                scope,
                 &mut checks,
             )
             .await
@@ -77,6 +78,7 @@ async fn is_dnsbl(
     server: &Server,
     config: &DnsBlServer,
     resolver: SpamFilterResolver<'_, impl ResolveVariable>,
+    element: Element,
     checks: &mut usize,
 ) -> Option<String> {
     let time = Instant::now();
@@ -133,6 +135,7 @@ async fn is_dnsbl(
                             .iter()
                             .map(|ip| trc::Value::from(ip.to_string()))
                             .collect::<Vec<_>>(),
+                        Details = element.as_str(),
                         Elapsed = time.elapsed()
                     );
 
@@ -159,6 +162,7 @@ async fn is_dnsbl(
                         Spam(SpamEvent::Dnsbl),
                         Hostname = zone.clone(),
                         Result = trc::Value::None,
+                        Details = element.as_str(),
                         Elapsed = time.elapsed()
                     );
 
@@ -175,6 +179,7 @@ async fn is_dnsbl(
                         Spam(SpamEvent::DnsblError),
                         Hostname = zone,
                         Elapsed = time.elapsed(),
+                        Details = element.as_str(),
                         CausedBy = err.to_string()
                     );
 

crates/spam-filter/src/modules/sanitize.rs

@@ -10,7 +10,7 @@ use crate::{Email, Hostname};
 
 impl Hostname {
     pub fn new(host: &str) -> Self {
-        let mut fqdn = host.to_lowercase();
+        let mut fqdn = host.trim_end_matches('.').to_lowercase();
 
         // Decode punycode
         if fqdn.contains("xn--") {
@@ -42,12 +42,20 @@ impl Hostname {
             .ok();
 
         Hostname {
-            ip,
             sld: if ip.is_none() {
-                psl::domain_str(&fqdn).map(str::to_string)
+                psl::domain(fqdn.as_bytes()).and_then(|domain| {
+                    if domain.suffix().typ().is_some() {
+                        std::str::from_utf8(domain.as_bytes())
+                            .ok()
+                            .map(str::to_string)
+                    } else {
+                        None
+                    }
+                })
             } else {
                 None
             },
+            ip,
             fqdn,
         }
     }

crates/utils/src/lib.rs

@@ -380,7 +380,10 @@ pub fn sanitize_email(email: &str) -> Option<String> {
         }
     }
 
-    if found_domain && last_ch != '.' && psl::domain(result.as_bytes()).is_some() {
+    if found_domain
+        && last_ch != '.'
+        && psl::domain(result.as_bytes()).is_some_and(|d| d.suffix().typ().is_some())
+    {
         Some(result)
     } else {
         None

tests/src/smtp/inbound/antispam.rs

@@ -102,6 +102,54 @@ path = "{PATH}/test_antispam.db"
 #type = "redis"
 #url = "redis://127.0.0.1"
 
+[http-lookup.STWT_OPENPHISH]
+enable = true
+url = "https://openphish.com/feed.txt"
+format = "list"
+retry = "1h"
+refresh = "12h"
+timeout = "30s"
+limits.size = 104857600
+limits.entries = 900000
+limits.entry-size = 512
+
+[http-lookup.STWT_PHISHTANK]
+enable = true
+url = "http://data.phishtank.com/data/online-valid.csv.gz"
+format = "csv"
+separator = ","
+index.key = 1
+skip-first = true
+gzipped = true
+retry = "1h"
+refresh = "6h"
+timeout = "30s"
+limits.size = 104857600
+limits.entries = 900000
+limits.entry-size = 512
+
+[http-lookup.STWT_DISPOSABLE_DOMAINS]
+enable = true
+url = "https://disposable.github.io/disposable-email-domains/domains_mx.txt"
+format = "list"
+retry = "1h"
+refresh = "24h"
+timeout = "30s"
+limits.size = 104857600
+limits.entries = 900000
+limits.entry-size = 512
+
+[http-lookup.STWT_FREE_DOMAINS]
+enable = true
+url = "https://gist.githubusercontent.com/okutbay/5b4974b70673dfdcc21c517632c1f984/raw/993a35930a8d24a1faab1b988d19d38d92afbba4/free_email_provider_domains.txt"
+format = "list"
+retry = "1h"
+refresh = "720h"
+timeout = "30s"
+limits.size = 104857600
+limits.entries = 900000
+limits.entry-size = 512
+
 [enterprise.ai.dummy]
 url = "https://127.0.0.1:9090/v1/chat/completions"
 type = "chat"