Reduce buggy usage of the registry controlled domain service.

url/url_canon_unittest.cc

Index: url/url_canon_unittest.cc
diff --git a/url/url_canon_unittest.cc b/url/url_canon_unittest.cc
index 8f71a911188f4bd2031a3686f09d3b2d91447b4f..0d6056eeedd213885f7b5c27bcdd4a79e1a37818 100644
--- a/url/url_canon_unittest.cc
+++ b/url/url_canon_unittest.cc
@@ -887,6 +887,50 @@ TEST(URLCanonTest, IPEmpty) {
   EXPECT_FALSE(host_info.IsIPAddress());
 }
 
+// Verifies that CanonicalizeHostSubstring produces the expected output and
+// does not "fix" IP addresses. Because this code is a subset of
+// CanonicalizeHost, the shared functionality is not tested.
+TEST(URLCanonTest, CanonicalizeHostSubstring) {
+  // Basic sanity check.
+  {
+    std::string out_str;
+    StdStringCanonOutput output(&out_str);
+    EXPECT_TRUE(CanonicalizeHostSubstring(
+        "M\xc3\x9cNCHEN.com", Component(0, 12), &output));
+    output.Complete();
+    EXPECT_EQ("xn--mnchen-3ya.com", out_str);
+  }
+
+  // Failure case.
+  {
+    std::string out_str;
+    StdStringCanonOutput output(&out_str);
+    EXPECT_FALSE(CanonicalizeHostSubstring(
+        WStringToUTF16(L"\xfdd0zyx.com").c_str(), Component(0, 8), &output));
+    output.Complete();
+    EXPECT_EQ("%EF%BF%BDzyx.com", out_str);
+  }
+
+  // Should return true for empty input strings.
+  {
+    std::string out_str;
+    StdStringCanonOutput output(&out_str);
+    EXPECT_TRUE(CanonicalizeHostSubstring("", Component(0, 0), &output));
+    output.Complete();
+    EXPECT_EQ("", out_str);

[Peter Kasting, 2016/10/22 05:04:20]

Nit: Since |out_str| is a std::string I tend to prefer EXPECT_EQ(std::string(),
out_str) (or EXPECT_TRUE(out_str.empty())), but whatever

+  }
+
+  // Numbers that look like IP addresses should not be changed.
+  {
+    std::string out_str;
+    StdStringCanonOutput output(&out_str);
+    EXPECT_TRUE(CanonicalizeHostSubstring(
+        "01.02.03.04", Component(0, 11), &output));
+    output.Complete();
+    EXPECT_EQ("01.02.03.04", out_str);
+  }
+}
+
 TEST(URLCanonTest, UserInfo) {
   // Note that the canonicalizer should escape and treat empty components as
   // not being there.