Hello guys! I have found that url_matcher::URLMatcher and URLPattern works
differently when matching domains that ends with a dot. Simple test shows
this behavior:

TEST(TrailingDotDomainMatch, Test) {
  const GURL normal_domain("http://example.com/");
  const GURL trailing_dot_domain("http://example.com./");

  // Test URLMatcher.
  using namespace url_matcher;
  URLMatcher matcher;
  URLMatcherConditionFactory* matcher_factory = matcher.condition_factory();
  URLMatcherConditionSet::Conditions conditions;
  conditions.insert(matcher_factory->CreateHostEqualsCondition("example.com"));
  const int kConditionSetId = 1;
  URLMatcherConditionSet::Vector conditions_vector;
  conditions_vector.push_back(make_scoped_refptr(
      new URLMatcherConditionSet(kConditionSetId, conditions)));
  matcher.AddConditionSets(conditions_vector);
  EXPECT_EQ(1u, matcher.MatchURL(normal_domain).size());
  EXPECT_EQ(1u, matcher.MatchURL(trailing_dot_domain).size());  // This match succeedes.

  // Test URLPattern.
  const URLPattern pattern(URLPattern::SCHEME_HTTP, "*://example.com/*");
  EXPECT_TRUE(pattern.MatchesURL(GURL(normal_domain)));
  EXPECT_TRUE(pattern.MatchesURL(GURL(trailing_dot_domain)));   // This match fails.
}

I think that in some cases this behavior of URLPattern can lead to some security
problems, like enabled extensions/userscripts on URLs where it is forbidden.

Is this a bug or a feature? Should we fix this behavior of URLPattern to properly
match both variations of domain?