[2 of 4] Add some test strings to registry_controlled_domain_unittest.cc

net/base/registry_controlled_domains/registry_controlled_domain_unittest.cc

Index: net/base/registry_controlled_domains/registry_controlled_domain_unittest.cc
diff --git a/net/base/registry_controlled_domains/registry_controlled_domain_unittest.cc b/net/base/registry_controlled_domains/registry_controlled_domain_unittest.cc
index 261a05e7fa66324baa29eecf869d306e1530b34c..fd19789b8c5ac2859d8623df5c653c8a104dcc10 100644
--- a/net/base/registry_controlled_domains/registry_controlled_domain_unittest.cc
+++ b/net/base/registry_controlled_domains/registry_controlled_domain_unittest.cc
@@ -91,6 +91,7 @@ class RegistryControlledDomainTest : public testing::Test {
  protected:
   template <typename Graph>
   void UseDomainData(const Graph& graph) {
+    // This is undone in TearDown.
     SetFindDomainGraph(graph, sizeof(Graph));
   }
 
@@ -172,39 +173,61 @@ TEST_F(RegistryControlledDomainTest, TestGetRegistryLength) {
 
   // Test GURL version of GetRegistryLength().
   EXPECT_EQ(2U, GetRegistryLengthFromURL("http://a.baz.jp/file.html",
-                                         EXCLUDE_UNKNOWN_REGISTRIES)); // 1
+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 1
   EXPECT_EQ(3U, GetRegistryLengthFromURL("http://a.baz.jp./file.html",
-                                         EXCLUDE_UNKNOWN_REGISTRIES)); // 1
+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 1
   EXPECT_EQ(0U, GetRegistryLengthFromURL("http://ac.jp",
-                                         EXCLUDE_UNKNOWN_REGISTRIES)); // 2
+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 2
   EXPECT_EQ(0U, GetRegistryLengthFromURL("http://a.bar.jp",
-                                         EXCLUDE_UNKNOWN_REGISTRIES)); // 3
+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 3
   EXPECT_EQ(0U, GetRegistryLengthFromURL("http://bar.jp",
-                                         EXCLUDE_UNKNOWN_REGISTRIES)); // 3
+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 3
+  EXPECT_EQ(2U, GetRegistryLengthFromURL("http://xbar.jp",
+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 1

[Ryan Sleevi, 2017/01/25 22:05:24]

What are you denoting with the trailing comments? That's not as clear - and
digging through the blame doesn't really shed light on it.

[ncarter (slow), 2017/02/16 20:46:26]

These numbers seemed to correspond to which rules (in the gperf file) are
potential matches for input strings. They are 1-indexed starting from the first
rule.

   EXPECT_EQ(0U, GetRegistryLengthFromURL("http://baz.bar.jp",
-                                         EXCLUDE_UNKNOWN_REGISTRIES)); // 3 4
+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 3 4
+  EXPECT_EQ(0U, GetRegistryLengthFromURL("http://.baz.bar.jp",

[Ryan Sleevi, 2017/01/25 22:05:24]

This is an invalid URL - so what's actually being tested here?

[ncarter (slow), 2017/02/16 20:46:26]

What's being tested is the code in GetRegistryLengthImpl that deals with with
edge cases around multiple leading and trailing dots. And, an empty label in the
middle of a hostname is a corner case that the new implementation will need to
handle.

To my surprise, GURL canonicalization actually doesn't handle any of these cases
-- the following unittest passes:

TEST(RCD, EmptyLabelsNotCanonicalizedByGURL) {
  EXPECT_EQ(".baz.bar.jp", GURL("http://.baz.bar.jp").host());
  EXPECT_EQ("..baz.bar.jp", GURL("http://..baz.bar.jp").host());
  EXPECT_EQ("baz..bar.jp", GURL("http://baz..bar.jp").host());
  EXPECT_EQ("baz.bar.jp..", GURL("http://baz.bar.jp..").host());
  EXPECT_EQ("baz.bar.jp.", GURL("http://baz.bar.jp.").host());
  EXPECT_EQ("...a..b..c..d...", GURL("http://...a..b..c..d...").host());

  EXPECT_TRUE(GURL("http://.baz.bar.jp").is_valid());
  EXPECT_TRUE(GURL("http://..baz.bar.jp").is_valid());
  EXPECT_TRUE(GURL("http://baz..bar.jp").is_valid());
  EXPECT_TRUE(GURL("http://baz.bar.jp..").is_valid());
  EXPECT_TRUE(GURL("http://baz.bar.jp.").is_valid());
  EXPECT_TRUE(GURL("http://...a..b..c..d...").is_valid());
}

+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 3 4
+  EXPECT_EQ(0U, GetRegistryLengthFromURL("http://..baz.bar.jp",

[Ryan Sleevi, 2017/01/25 22:05:24]

Ditto

[ncarter (slow), 2017/02/16 20:46:26]

See above.

+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 3 4
+  EXPECT_EQ(11U, GetRegistryLengthFromURL("http://foo..baz.bar.jp",

[Ryan Sleevi, 2017/01/25 22:05:24]

Ditto

[ncarter (slow), 2017/02/16 20:46:26]

See above.

+                                          EXCLUDE_UNKNOWN_REGISTRIES));  // 3 4
+  EXPECT_EQ(0U, GetRegistryLengthFromURL("http://xbaz.bar.jp",
+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 3
+  EXPECT_EQ(11U, GetRegistryLengthFromURL("http://x.xbaz.bar.jp",
+                                          EXCLUDE_UNKNOWN_REGISTRIES));  // 3
   EXPECT_EQ(12U, GetRegistryLengthFromURL("http://a.b.baz.bar.jp",
-                                         EXCLUDE_UNKNOWN_REGISTRIES)); // 4
+                                          EXCLUDE_UNKNOWN_REGISTRIES));  // 4
   EXPECT_EQ(6U, GetRegistryLengthFromURL("http://baz.pref.bar.jp",
-                                         EXCLUDE_UNKNOWN_REGISTRIES)); // 5
+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 5
+  EXPECT_EQ(6U, GetRegistryLengthFromURL("http://z.baz.pref.bar.jp",
+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 5
+  EXPECT_EQ(10U, GetRegistryLengthFromURL("http://p.ref.bar.jp",
+                                          EXCLUDE_UNKNOWN_REGISTRIES));  // 5
+  EXPECT_EQ(0U, GetRegistryLengthFromURL("http://xpref.bar.jp",
+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 5
+  EXPECT_EQ(12U, GetRegistryLengthFromURL("http://baz.xpref.bar.jp",
+                                          EXCLUDE_UNKNOWN_REGISTRIES));  // 5
+  EXPECT_EQ(6U, GetRegistryLengthFromURL("http://baz..pref.bar.jp",

[Ryan Sleevi, 2017/01/25 22:05:24]

Also invalid

[ncarter (slow), 2017/02/16 20:46:26]

See above.

+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 5
   EXPECT_EQ(11U, GetRegistryLengthFromURL("http://a.b.bar.baz.com",
-                                         EXCLUDE_UNKNOWN_REGISTRIES)); // 6
+                                          EXCLUDE_UNKNOWN_REGISTRIES));  // 6
   EXPECT_EQ(3U, GetRegistryLengthFromURL("http://a.d.c",
-                                         EXCLUDE_UNKNOWN_REGISTRIES)); // 7
+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 7
   EXPECT_EQ(3U, GetRegistryLengthFromURL("http://.a.d.c",
-                                         EXCLUDE_UNKNOWN_REGISTRIES)); // 7
+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 7
   EXPECT_EQ(3U, GetRegistryLengthFromURL("http://..a.d.c",
-                                         EXCLUDE_UNKNOWN_REGISTRIES)); // 7
+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 7
   EXPECT_EQ(1U, GetRegistryLengthFromURL("http://a.b.c",
-                                         EXCLUDE_UNKNOWN_REGISTRIES)); // 7 8
+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // 7 8
   EXPECT_EQ(0U, GetRegistryLengthFromURL("http://baz.com",
-                                         EXCLUDE_UNKNOWN_REGISTRIES)); // none
+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // none
   EXPECT_EQ(0U, GetRegistryLengthFromURL("http://baz.com.",
-                                         EXCLUDE_UNKNOWN_REGISTRIES)); // none
+                                         EXCLUDE_UNKNOWN_REGISTRIES));  // none
   EXPECT_EQ(3U, GetRegistryLengthFromURL("http://baz.com",
-                                         INCLUDE_UNKNOWN_REGISTRIES)); // none
+                                         INCLUDE_UNKNOWN_REGISTRIES));  // none
   EXPECT_EQ(4U, GetRegistryLengthFromURL("http://baz.com.",
-                                         INCLUDE_UNKNOWN_REGISTRIES)); // none
+                                         INCLUDE_UNKNOWN_REGISTRIES));  // none
 
   EXPECT_EQ(std::string::npos,
       GetRegistryLengthFromURL(std::string(), EXCLUDE_UNKNOWN_REGISTRIES));