Loosen RFC 1738 compliance check to allow underscores where we already allowed...

Loosen RFC 1738 compliance check to allow underscores where we already allowed hyphens, to match real-world needs.

I don't believe further loosening will be required but that data will hopefully be coming soon.  In the meantime people are asking for this fix.

BUG=25714
TEST=Entring "a_b.com" in the omnibox should default to navigate, not search
Review URL: http://codereview.chromium.org/339017

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@30245 0039d316-1c4b-4281-b951-d872f2087c98

chrome/browser/autocomplete/autocomplete.cc

@@ -175,14 +175,14 @@ AutocompleteInput::Type AutocompleteInput::Parse(
   if (registry_length == std::wstring::npos)
     return QUERY;  // Could be a broken IP address, etc.
 
-  // See if the hostname is valid per RFC 1738.  While IE and GURL allow
-  // hostnames to contain many other characters (perhaps for weird intranet
-  // machines), it's extremely unlikely that a user would be trying to type
-  // those in for anything other than a search query.
+  // See if the hostname is valid.  While IE and GURL allow hostnames to contain
+  // many other characters (perhaps for weird intranet machines), it's extremely
+  // unlikely that a user would be trying to type those in for anything other
+  // than a search query.
   url_canon::CanonHostInfo host_info;
   const std::string canonicalized_host(net::CanonicalizeHost(host, &host_info));
   if ((host_info.family == url_canon::CanonHostInfo::NEUTRAL) &&
-      !net::IsCanonicalizedHostRFC1738Compliant(canonicalized_host))
+      !net::IsCanonicalizedHostCompliant(canonicalized_host))
     return QUERY;
 
   // Presence of a port means this is likely a URL, if the port is really a port

chrome/browser/autocomplete/autocomplete_unittest.cc

@@ -243,7 +243,9 @@ TEST(AutocompleteTest, InputType) {
     { L"http://foo", AutocompleteInput::URL },
     { L"http://foo.c", AutocompleteInput::URL },
     { L"http://foo.com", AutocompleteInput::URL },
+    { L"http://foo_bar.com", AutocompleteInput::URL },
     { L"http://-.com", AutocompleteInput::QUERY },
+    { L"http://_foo_.com", AutocompleteInput::QUERY },
     { L"http://foo.com:abc", AutocompleteInput::QUERY },
     { L"http://foo.com:123456", AutocompleteInput::QUERY },
     { L"http:user@foo.com", AutocompleteInput::URL },

net/base/net_util.cc

@@ -961,7 +961,7 @@ inline bool IsHostCharDigit(char c) {
   return (c >= '0') && (c <= '9');
 }
 
-bool IsCanonicalizedHostRFC1738Compliant(const std::string& host) {
+bool IsCanonicalizedHostCompliant(const std::string& host) {
   if (host.empty())
     return false;
 
@@ -970,7 +970,7 @@ bool IsCanonicalizedHostRFC1738Compliant(const std::string& host) {
     IN_COMPONENT_STARTED_DIGIT,
     IN_COMPONENT_STARTED_ALPHA
   } state = NOT_IN_COMPONENT;
-  bool last_char_was_hyphen = false;
+  bool last_char_was_hyphen_or_underscore = false;
 
   for (std::string::const_iterator i(host.begin()); i != host.end(); ++i) {
     const char c = *i;
@@ -983,13 +983,13 @@ bool IsCanonicalizedHostRFC1738Compliant(const std::string& host) {
         return false;
     } else {
       if (c == '.') {
-        if (last_char_was_hyphen)
+        if (last_char_was_hyphen_or_underscore)
           return false;
         state = NOT_IN_COMPONENT;
       } else if (IsHostCharAlpha(c) || IsHostCharDigit(c)) {
-        last_char_was_hyphen = false;
-      } else if (c == '-') {
-        last_char_was_hyphen = true;
+        last_char_was_hyphen_or_underscore = false;
+      } else if ((c == '-') || (c == '_')) {
+        last_char_was_hyphen_or_underscore = true;
       } else {
         return false;
       }

net/base/net_util.h

@@ -158,16 +158,17 @@ std::string CanonicalizeHost(const std::string& host,
 std::string CanonicalizeHost(const std::wstring& host,
                              url_canon::CanonHostInfo* host_info);
 
-// Returns true if |host| is RFC 1738-compliant (and not an IP address).  The
-// rules are:
+// Returns true if |host| is not an IP address and is compliant with a set of
+// rules based on RFC 1738 and tweaked to be compatible with the real world.
+// The rules are:
 //   * One or more components separated by '.'
 //   * Each component begins and ends with an alphanumeric character
-//   * Each component contains only alphanumeric characters and '-'
+//   * Each component contains only alphanumeric characters and '-' or '_'
 //   * The last component does not begin with a digit
 //
 // NOTE: You should only pass in hosts that have been returned from
 // CanonicalizeHost(), or you may not get accurate results.
-bool IsCanonicalizedHostRFC1738Compliant(const std::string& host);
+bool IsCanonicalizedHostCompliant(const std::string& host);
 
 // Call these functions to get the html snippet for a directory listing.
 // The return values of both functions are in UTF-8.

net/base/net_util_unittest.cc

@@ -344,7 +344,7 @@ const IDNTestCase idn_cases[] = {
 #endif
 };
 
-struct RFC1738Case {
+struct CompliantHostCase {
   const char* host;
   bool expected_output;
 };
@@ -815,8 +815,8 @@ TEST(NetUtilTest, IDNToUnicodeSlow) {
   }
 }
 
-TEST(NetUtilTest, RFC1738) {
-  const RFC1738Case rfc1738_cases[] = {
+TEST(NetUtilTest, CompliantHost) {
+  const CompliantHostCase compliant_host_cases[] = {
     {"", false},
     {"a", true},
     {"-", false},
@@ -825,19 +825,20 @@ TEST(NetUtilTest, RFC1738) {
     {"a.a", true},
     {"9.a", true},
     {"a.9", false},
+    {"_9a", false},
     {"a.a9", true},
     {"a.9a", false},
     {"a+9a", false},
     {"1-.a-b", false},
-    {"1-2.a-b", true},
+    {"1-2.a_b", true},
     {"a.b.c.d.e", true},
     {"1.2.3.4.e", true},
     {"a.b.c.d.5", false},
   };
 
-  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(rfc1738_cases); ++i) {
-    EXPECT_EQ(rfc1738_cases[i].expected_output,
-              net::IsCanonicalizedHostRFC1738Compliant(rfc1738_cases[i].host));
+  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(compliant_host_cases); ++i) {
+    EXPECT_EQ(compliant_host_cases[i].expected_output,
+              net::IsCanonicalizedHostCompliant(compliant_host_cases[i].host));
   }
 }