feat(core): Skip catch-all for known domains (#1336)

* feat(core): Skip catch-all for known domains

* Only skip catch-all for hotmail

core/src/smtp/connect.rs

@@ -29,7 +29,7 @@ use std::time::Duration;
 
 use trust_dns_proto::rr::Name;
 
-use super::parser;
+use super::{gmail::is_gmail, outlook::is_hotmail, parser, yahoo::is_yahoo};
 use super::{SmtpDetails, SmtpError};
 use crate::util::{constants::LOG_TARGET, input_output::CheckEmailInput};
 
@@ -220,7 +220,14 @@ async fn email_deliverable(
 async fn smtp_is_catch_all(
 	smtp_transport: &mut SmtpTransport,
 	domain: &str,
+	host: &Name,
 ) -> Result<bool, SmtpError> {
+	// Skip catch-all check for known providers.
+	let host = host.to_string();
+	if is_gmail(&host) || is_hotmail(&host) || is_yahoo(&host) {
+		return Ok(false);
+	}
+
 	// Create a random 15-char alphanumerical string.
 	let mut rng = SmallRng::from_entropy();
 	let random_email: String = iter::repeat(())
@@ -249,7 +256,7 @@ async fn create_smtp_future(
 	// Ok(SmtpDetails { can_connect_smtp: false, ... }).
 	let mut smtp_transport = connect_to_host(host, port, input).await?;
 
-	let is_catch_all = smtp_is_catch_all(&mut smtp_transport, domain)
+	let is_catch_all = smtp_is_catch_all(&mut smtp_transport, domain, host)
 		.await
 		.unwrap_or(false);
 	let deliverability = if is_catch_all {

core/src/smtp/error.rs

@@ -16,8 +16,8 @@
 
 use super::gmail::GmailError;
 #[cfg(feature = "headless")]
-use super::microsoft::hotmail::HotmailError;
-use super::microsoft::microsoft365::Microsoft365Error;
+use super::outlook::hotmail::HotmailError;
+use super::outlook::microsoft365::Microsoft365Error;
 use super::parser;
 use super::yahoo::YahooError;
 use crate::util::ser_with_display::ser_with_display;

core/src/smtp/gmail.rs

@@ -76,6 +76,11 @@ pub async fn check_gmail(
 	})
 }
 
+/// Check if the MX host is from Gmail.
+pub fn is_gmail(host: &str) -> bool {
+	host.to_lowercase().ends_with(".google.com.")
+}
+
 #[cfg(test)]
 mod tests {
 	use std::str::FromStr;

core/src/smtp/mod.rs

@@ -18,7 +18,7 @@ mod connect;
 mod error;
 mod gmail;
 mod http_api;
-mod microsoft;
+mod outlook;
 mod parser;
 mod yahoo;
 
@@ -32,6 +32,12 @@ use crate::{util::input_output::CheckEmailInput, LOG_TARGET};
 use connect::check_smtp_with_retry;
 pub use error::*;
 
+use self::{
+	gmail::is_gmail,
+	outlook::{is_hotmail, is_outlook},
+	yahoo::is_yahoo,
+};
+
 /// Details that we gathered from connecting to this email via SMTP
 #[derive(Debug, Default, Deserialize, Serialize)]
 pub struct SmtpDetails {
@@ -68,19 +74,18 @@ pub async fn check_smtp(
 		)));
 	}
 
-	// FIXME Is this `contains` too lenient?
-	if input.yahoo_use_api && host_lowercase.contains("yahoo") {
+	if input.yahoo_use_api && is_yahoo(&host_lowercase) {
 		return yahoo::check_yahoo(to_email, input)
 			.await
 			.map_err(|err| err.into());
 	}
-	if input.gmail_use_api && host_lowercase.ends_with(".google.com.") {
+	if input.gmail_use_api && is_gmail(&host_lowercase) {
 		return gmail::check_gmail(to_email, input)
 			.await
 			.map_err(|err| err.into());
 	}
-	if input.microsoft365_use_api && host_lowercase.ends_with(".mail.protection.outlook.com.") {
-		match microsoft::microsoft365::check_microsoft365_api(to_email, input).await {
+	if input.microsoft365_use_api && is_outlook(&host_lowercase) {
+		match outlook::microsoft365::check_microsoft365_api(to_email, input).await {
 			Ok(Some(smtp_details)) => return Ok(smtp_details),
 			// Continue in the event of an error/ambiguous result.
 			Err(err) => {
@@ -99,16 +104,8 @@ pub async fn check_smtp(
 		// The password recovery page do not always work with Microsoft 365
 		// addresses. So we only test with @hotmail and @outlook addresses.
 		// ref: https://github.com/reacherhq/check-if-email-exists/issues/1185
-		//
-		// After some testing, I got:
-		// - *@outlook.com -> `outlook-com.olc.protection.outlook.com.`
-		// - *@outlook.fr -> `eur.olc.protection.outlook.com.`
-		// - *@hotmail.com -> `hotmail-com.olc.protection.outlook.com.`
-		// - *@hotmail.fr -> `eur.olc.protection.outlook.com.`
-		//
-		// So it seems that outlook/hotmail addresses end with `olc.protection.outlook.com.`
-		if host_lowercase.ends_with("olc.protection.outlook.com.") {
-			return microsoft::hotmail::check_password_recovery(to_email, webdriver)
+		if is_hotmail(&host_lowercase) {
+			return outlook::hotmail::check_password_recovery(to_email, webdriver)
 				.await
 				.map_err(|err| err.into());
 		}

core/src/smtp/outlook/mod.rs

@@ -0,0 +1,25 @@
+#[cfg(feature = "headless")]
+pub mod hotmail;
+pub mod microsoft365;
+
+/// Check if a MX host is from outlook (includes @hotmail.*, @outlook.* and
+/// all Microsoft 365 addresses).
+pub fn is_outlook(host: &str) -> bool {
+	host.to_lowercase()
+		.ends_with(".mail.protection.outlook.com.")
+}
+
+/// Check if a MX host is an @hotmail.* or @outlook.* email.
+///
+/// After some testing, I got:
+/// - *@outlook.com -> `outlook-com.olc.protection.outlook.com.`
+/// - *@outlook.fr -> `eur.olc.protection.outlook.com.`
+/// - *@hotmail.com -> `hotmail-com.olc.protection.outlook.com.`
+/// - *@hotmail.fr -> `eur.olc.protection.outlook.com.`
+/// - *@hotmail.nl -> `eur.olc.protection.outlook.com.`
+///
+/// So it seems that outlook/hotmail addresses end with `olc.protection.outlook.com.`
+pub fn is_hotmail(host: &str) -> bool {
+	host.to_lowercase()
+		.ends_with(".mail.protection.outlook.com.")
+}

core/src/smtp/yahoo.rs

@@ -224,3 +224,11 @@ pub async fn check_yahoo(
 		..Default::default()
 	})
 }
+
+/// Check if the MX host is from Yahoo.
+/// Examples:
+/// - mta7.am0.yahoodns.net.
+/// - mx-eu.mail.am0.yahoodns.net.
+pub fn is_yahoo(host: &str) -> bool {
+	host.to_lowercase().ends_with(".yahoodns.net.")
+}