feat: Add skipped domains (#1293)

* feat: Add skipped domains

* clippy

core/src/smtp/error.rs

@@ -48,6 +48,8 @@ pub enum SmtpError {
 	HotmailError(HotmailError),
 	/// Error when verifying a Microsoft 365 email via HTTP request.
 	Microsoft365Error(Microsoft365Error),
+	/// Email is in the `skipped_domains` parameter.
+	SkippedDomain(String),
 }
 
 impl From<SocksError> for SmtpError {

core/src/smtp/mod.rs

@@ -57,6 +57,17 @@ pub async fn check_smtp(
 	input: &CheckEmailInput,
 ) -> Result<SmtpDetails, SmtpError> {
 	let host_lowercase = host.to_lowercase().to_string();
+
+	if input
+		.skipped_domains
+		.iter()
+		.any(|d| host_lowercase.contains(d))
+	{
+		return Err(SmtpError::SkippedDomain(format!(
+			"Reacher currently cannot verify emails from @{domain}"
+		)));
+	}
+
 	// FIXME Is this `contains` too lenient?
 	if input.yahoo_use_api && host_lowercase.contains("yahoo") {
 		return yahoo::check_yahoo(to_email, input)
@@ -129,4 +140,19 @@ mod tests {
 			_ => panic!("check_smtp did not time out"),
 		}
 	}
+
+	#[test]
+	fn should_skip_domains() {
+		let runtime = Runtime::new().unwrap();
+
+		let to_email = EmailAddress::from_str("foo@icloud.com").unwrap();
+		let host = Name::from_str("mx01.mail.icloud.com.").unwrap();
+		let input = CheckEmailInput::default();
+
+		let res = runtime.block_on(check_smtp(&to_email, &host, 25, "icloud.com", &input));
+		match res {
+			Err(SmtpError::SkippedDomain(_)) => (),
+			r => panic!("{:?}", r),
+		}
+	}
 }

core/src/util/input_output.rs

@@ -137,6 +137,25 @@ pub struct CheckEmailInput {
 	///
 	/// Defaults to Opportunistic.
 	pub smtp_security: SmtpSecurity,
+	/// **IMPORTANT:** This is a beta feature, and might be completely removed,
+	/// or moved somewhere else, before the next release.
+	///
+	/// List of domains to skip when doing an SMTP connection, because we know
+	/// they return "unknown". For each string in this list, we check if the MX
+	/// record **contains** the string; if yes, we return an error saying the
+	/// SMTP verification is skipped.
+	///
+	/// Related issue: https://github.com/reacherhq/check-if-email-exists/issues/937
+	///
+	/// ## Example
+	///
+	/// If you want to skip Zoho emails, it's good enough to ".zoho.com." to
+	/// the list. This way it will skip all Zoho emails as their MX records
+	/// show "mx{N}.zoho.com.". Simply putting "zoho" might give false
+	/// positives if you have an email provider like "mycustomzoho.com".
+	///
+	/// Defaults to: [""]
+	pub skipped_domains: Vec<String>,
 }
 
 impl Default for CheckEmailInput {
@@ -157,6 +176,21 @@ impl Default for CheckEmailInput {
 			check_gravatar: false,
 			haveibeenpwned_api_key: None,
 			retries: 2,
+			skipped_domains: vec![
+				// on @bluewin.ch
+				// - mx-v02.bluewin.ch.
+				".bluewin.ch.".into(),
+				// on @bluewin.ch
+				// - mxbw-bluewin-ch.hdb-cs04.ellb.ch.
+				"bluewin-ch.".into(),
+				// on @gmx.de, @gmx.ch, @gmx.net
+				".gmx.net.".into(),
+				// on @icloud.com
+				".mail.icloud.com.".into(),
+				// on @web.de
+				".web.de.".into(),
+				".zoho.com.".into(),
+			],
 		}
 	}
 }
@@ -305,6 +339,16 @@ impl CheckEmailInput {
 		self.hotmail_use_headless = use_headless;
 		self
 	}
+
+	/// **IMPORTANT:** This is a beta feature, and might be completely removed,
+	/// or moved somewhere else, before the next release.
+	///
+	/// List of domains to skip when doing an SMTP connection, because we know
+	/// they return "unknown".
+	pub fn set_skipped_domains(&mut self, domains: Vec<String>) -> &mut CheckEmailInput {
+		self.skipped_domains = domains;
+		self
+	}
 }
 
 /// An enum to describe how confident we are that the recipient address is