From dc1f6f33eb8a608f4b9e15003ef1d92a0be85977 Mon Sep 17 00:00:00 2001
From: Dmitry Marakasov <amdmi3@amdmi3.ru>
Date: Thu, 12 Dec 2024 20:45:54 +0300
Subject: [PATCH] webapp: Extract domain name for ssltest link

Difference from python impl is better handling of invalid urls.
---
 Cargo.lock                                    |  1 +
 repology-webapp/Cargo.toml                    |  1 +
 repology-webapp/src/lib.rs                    |  1 +
 repology-webapp/src/template_funcs.rs         | 70 +++++++++++++++++++
 repology-webapp/templates/_macros/links.html  |  6 +-
 .../tests/integration_tests/link.rs           |  1 -
 6 files changed, 77 insertions(+), 3 deletions(-)
 create mode 100644 repology-webapp/src/template_funcs.rs

diff --git a/Cargo.lock b/Cargo.lock
index 33205f7..7cf97a3 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -2082,6 +2082,7 @@ dependencies = [
  "tracing-appender",
  "tracing-subscriber",
  "ttf-parser",
+ "url",
  "url-escape",
 ]
 
diff --git a/repology-webapp/Cargo.toml b/repology-webapp/Cargo.toml
index 9b68175..faf550e 100644
--- a/repology-webapp/Cargo.toml
+++ b/repology-webapp/Cargo.toml
@@ -39,6 +39,7 @@ tracing = "0.1.41"
 tracing-appender = "0.2.3"
 tracing-subscriber = "0.3.19"
 ttf-parser = "0.25.1"
+url = "2.5.4"
 url-escape = "0.1.1"
 
 [dev-dependencies]
diff --git a/repology-webapp/src/lib.rs b/repology-webapp/src/lib.rs
index a915177..484d64b 100644
--- a/repology-webapp/src/lib.rs
+++ b/repology-webapp/src/lib.rs
@@ -22,6 +22,7 @@ mod result;
 mod state;
 mod static_files;
 mod template_context;
+mod template_funcs;
 mod url_for;
 mod views;
 mod xmlwriter;
diff --git a/repology-webapp/src/template_funcs.rs b/repology-webapp/src/template_funcs.rs
new file mode 100644
index 0000000..da49ea7
--- /dev/null
+++ b/repology-webapp/src/template_funcs.rs
@@ -0,0 +1,70 @@
+// SPDX-FileCopyrightText: Copyright 2024 Dmitry Marakasov <amdmi3@amdmi3.ru>
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+use url::Url;
+
+/// Given an url, try to extract domain for Qualys SSL Server Test
+///
+/// Qualys requires domain (not an IP address) and does not allow
+/// custom ports, so only return host part if
+pub fn extract_domain_for_ssltest(url: &str) -> Option<String> {
+    let url = Url::parse(url).ok()?;
+    if url.port_or_known_default().is_none_or(|port| port == 443) {
+        url.domain().map(|domain| domain.into())
+    } else {
+        None
+    }
+}
+
+#[cfg(test)]
+#[coverage(off)]
+mod tests {
+    mod test_extract_domain_for_ssltest {
+        use super::super::*;
+
+        #[test]
+        fn test_basic() {
+            assert_eq!(
+                extract_domain_for_ssltest("https://example.com/").as_deref(),
+                Some("example.com")
+            );
+        }
+
+        #[test]
+        fn test_default_port() {
+            assert_eq!(
+                extract_domain_for_ssltest("https://example.com:443/").as_deref(),
+                Some("example.com")
+            );
+        }
+
+        #[test]
+        fn test_custom_port() {
+            assert_eq!(extract_domain_for_ssltest("https://example.com:444/"), None);
+        }
+
+        #[test]
+        fn test_custom_schema() {
+            assert_eq!(
+                extract_domain_for_ssltest("git+https://example.com/").as_deref(),
+                Some("example.com")
+            );
+        }
+
+        #[test]
+        fn test_not_https() {
+            assert_eq!(extract_domain_for_ssltest("http://example.com/"), None);
+        }
+
+        #[test]
+        fn test_invalid_urls() {
+            assert_eq!(extract_domain_for_ssltest(""), None);
+            assert_eq!(extract_domain_for_ssltest("!&?"), None);
+            assert_eq!(extract_domain_for_ssltest(" "), None);
+            assert_eq!(extract_domain_for_ssltest("..."), None);
+            assert_eq!(extract_domain_for_ssltest("\n"), None);
+            assert_eq!(extract_domain_for_ssltest("example.com"), None);
+            assert_eq!(extract_domain_for_ssltest("file:///example.com"), None);
+        }
+    }
+}
diff --git a/repology-webapp/templates/_macros/links.html b/repology-webapp/templates/_macros/links.html
index eaaaac4..d98ef21 100644
--- a/repology-webapp/templates/_macros/links.html
+++ b/repology-webapp/templates/_macros/links.html
@@ -42,8 +42,10 @@
 	{%- if code == -205 -%}
 		(<a href="https://bb6-ie.blogspot.com/2017/04/do-not-put-ipv4-mapped-ipv6-addresses.html" rel="nofollow">related reading</a>)
 	{%- endif -%}
-	{%- if code >= -599 && code <= -500 -%}{# TODO: we only need netloc here #}
-		(<a href="https://www.ssllabs.com/ssltest/analyze.html?d={{ url }}" rel="nofollow">check on SSL Labs</a>)
+	{%- if code >= -599 && code <= -500 -%}
+		{%- if let Some(domain) = crate::template_funcs::extract_domain_for_ssltest(url) -%}
+			(<a href="https://www.ssllabs.com/ssltest/analyze.html?d={{ domain }}" rel="nofollow">check on SSL Labs</a>)
+		{%- endif -%}
 	{%- endif -%}
 {%- endmacro -%}
 
diff --git a/repology-webapp/tests/integration_tests/link.rs b/repology-webapp/tests/integration_tests/link.rs
index 601d567..de7cbcd 100644
--- a/repology-webapp/tests/integration_tests/link.rs
+++ b/repology-webapp/tests/integration_tests/link.rs
@@ -92,7 +92,6 @@ async fn test_ipv6_redirect(pool: PgPool) {
     );
 }
 
-#[ignore]
 #[sqlx::test(migrator = "repology_common::MIGRATOR", fixtures("link_data"))]
 async fn test_ssl_failure(pool: PgPool) {
     check_response!(