Bug 1915419 - Add nsIEffectiveTLDService getSchemelessSiteFromHost. r=bvandersloot,necko-reviewers,valentin

Differential Revision: https://phabricator.services.mozilla.com/D220604

netwerk/dns/nsEffectiveTLDService.cpp

@@ -206,6 +206,23 @@ nsEffectiveTLDService::GetSchemelessSite(nsIURI* aURI, nsACString& aSite) {
   return rv;
 }
 
+// Variant of GetSchemelessSite which accepts a host string instead of a URI.
+NS_IMETHODIMP
+nsEffectiveTLDService::GetSchemelessSiteFromHost(const nsACString& aHostname,
+                                                 nsACString& aSite) {
+  NS_ENSURE_TRUE(!aHostname.IsEmpty(), NS_ERROR_FAILURE);
+
+  nsresult rv = GetBaseDomainFromHost(aHostname, 0, aSite);
+  if (rv == NS_ERROR_HOST_IS_IP_ADDRESS ||
+      rv == NS_ERROR_INSUFFICIENT_DOMAIN_LEVELS) {
+    aSite.Assign(aHostname);
+    nsContentUtils::MaybeFixIPv6Host(aSite);
+
+    return NS_OK;
+  }
+  return rv;
+}
+
 // External function for dealing with URIs to get site correctly.
 // Calls through to GetSchemelessSite(), and serializes with the scheme and
 // "://" prepended.

netwerk/dns/nsIEffectiveTLDService.idl

@@ -119,6 +119,28 @@ interface nsIEffectiveTLDService : nsISupports
      */
     ACString getSchemelessSite(in nsIURI aURI);
 
+    /**
+     * Same as getSchemelessSite but takes a host string. Prefer the URI variant
+     * if possible.
+     *
+     * @param aHost
+     *        The host to analyze.
+     *
+     * @return the Site.
+     *
+     * @throws NS_ERROR_UNEXPECTED
+     *         or other error returned by nsIIDNService::normalize when
+     *         the hostname contains characters disallowed in URIs
+     *
+     * @see    getSchemelessSite()
+     * @see    getBaseDomain()
+     * @see    getSite()
+     *
+     * @warning This function should not be used without good reason. Please
+     * use getSite() or the Origin if you are not absolutely certain.
+     */
+    ACString getSchemelessSiteFromHost(in AUTF8String aHost);
+
     /**
      * Get the Site for the origin of aURI; e.g. for
      * "https://www.bbc.co.uk/index.html", this would be "https://bbc.co.uk".

netwerk/dns/tests/unit/test_nsEffectiveTLDService_getSite.js

@@ -64,7 +64,10 @@ add_task(() => {
     ["https://test.example.com", "example.com"],
     ["https://test1.test2.example.com", "example.com"],
     ["https://test1.test2.example.co.uk", "example.co.uk"],
-    ["https://test1.test2.example.公司.香港", "example.xn--55qx5d.xn--j6w193g"],
+    [
+      "https://test1.test2.example.\u516c\u53f8.\u9999\u6e2f",
+      "example.xn--55qx5d.xn--j6w193g",
+    ],
   ]) {
     let origin = Services.io.newURI(originString);
     let answer = Services.eTLD.getSchemelessSite(origin);
@@ -75,3 +78,29 @@ add_task(() => {
     );
   }
 });
+
+add_task(function test_getSchemelessSiteFromHost() {
+  for (let [host, result] of [
+    ["com", "com"],
+    ["test", "test"],
+    ["test.", "test."],
+    ["::1", "[::1]"],
+    ["localhost", "localhost"],
+    ["127.0.0.1", "127.0.0.1"],
+    ["example.com", "example.com"],
+    ["test.example.com", "example.com"],
+    ["test1.test2.example.com", "example.com"],
+    ["test1.test2.example.co.uk", "example.co.uk"],
+    [
+      "test1.test2.example.\u516c\u53f8.\u9999\u6e2f",
+      "example.xn--55qx5d.xn--j6w193g",
+    ],
+  ]) {
+    let answer = Services.eTLD.getSchemelessSiteFromHost(host);
+    Assert.equal(
+      answer,
+      result,
+      `"${host}" should have schemeless site ${result}, got ${answer}.`
+    );
+  }
+});