Initial scaffold for kestrove-core

Co-Authored-By: Claude Opus 5.5 <[email protected]>
2026-10-15 18:30:14 +09:00
commit 462fbbc523
34 changed files with 2063 additions and 0 deletions
@@ -0,0 +1,70 @@
+//! URL matching for autofill (docs/05 "Matching rules").
+
+use url::Url;
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord)]
+pub enum MatchLevel {
+    None = 0,
+    /// Same registrable domain (eTLD+1), different host.
+    Domain = 1,
+    /// Same host, different port.
+    Host = 2,
+    /// Same scheme, host and port.
+    Exact = 3,
+}
+
+/// Compares a saved item URL with the page URL. Never matches across schemes,
+/// except that an `http` item may match an `https` page (an upgrade) but never the reverse.
+pub fn match_url(item_url: &str, page_url: &str) -> MatchLevel {
+    let (Some(item), Some(page)) = (parse(item_url), parse(page_url)) else {
+        return MatchLevel::None;
+    };
+    let scheme_ok = item.scheme() == page.scheme() || (item.scheme() == "http" && page.scheme() == "https");
+    if !scheme_ok {
+        return MatchLevel::None;
+    }
+    let (Some(ih), Some(ph)) = (item.host_str(), page.host_str()) else {
+        return MatchLevel::None;
+    };
+    let (ih, ph) = (ih.to_ascii_lowercase(), ph.to_ascii_lowercase());
+    if ih == ph {
+        return if item.port_or_known_default() == page.port_or_known_default() && item.scheme() == page.scheme() {
+            MatchLevel::Exact
+        } else {
+            MatchLevel::Host
+        };
+    }
+    match (psl::domain_str(&ih), psl::domain_str(&ph)) {
+        (Some(a), Some(b)) if a == b => MatchLevel::Domain,
+        _ => MatchLevel::None,
+    }
+}
+
+fn parse(s: &str) -> Option<Url> {
+    let s = s.trim();
+    // Saved URLs are often stored without a scheme ("example.com/login").
+    if s.contains("://") {
+        Url::parse(s).ok()
+    } else {
+        Url::parse(&format!("https://{s}")).ok()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use MatchLevel::*;
+
+    #[test]
+    fn levels() {
+        assert_eq!(match_url("https://example.com", "https://example.com/login"), Exact);
+        assert_eq!(match_url("example.com", "https://example.com/"), Exact);
+        assert_eq!(match_url("https://example.com:8443", "https://example.com"), Host);
+        assert_eq!(match_url("https://example.com", "https://login.example.com"), Domain);
+        assert_eq!(match_url("https://a.github.io", "https://b.github.io"), None); // public suffix
+        assert_eq!(match_url("https://example.com", "https://example.com.evil.net"), None);
+        assert_eq!(match_url("https://example.com", "http://example.com"), None); // no downgrade
+        assert_eq!(match_url("http://example.com", "https://example.com"), Host); // upgrade allowed
+        assert_eq!(match_url("https://example.co.uk", "https://other.co.uk"), None);
+    }
+}