Classify more types of input as queries.

This adds an oracle that can determine if a hostname is RFC 1738 compliant; if we get a hostname that's not an IP address and not compliant here, we almost certainly have a query.

This also rejects "ports" that don't consist of a number between 0 and 65535 (e.g. port "abc" or port "123456").

BUG=18817
TEST=The string "foo: bar" (with quotes) should default to search, not navigate, in the address bar.
Review URL: http://codereview.chromium.org/271116

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@29317 0039d316-1c4b-4281-b951-d872f2087c98

3 files changed, 90 insertions, 0 deletions

diff --git a/net/base/net_util.cc b/net/base/net_util.cc
index 1d7d558..05f5841 100644
--- a/net/base/net_util.cc
+++ b/net/base/net_util.cc
@@ -939,6 +939,54 @@ std::string GetDirectoryListingHeader(const string16& title) {
   return result;
 }
 
+inline bool IsHostCharAlpha(char c) {
+  // We can just check lowercase because uppercase characters have already been
+  // normalized.
+  return (c >= 'a') && (c <= 'z');
+}
+
+inline bool IsHostCharDigit(char c) {
+  return (c >= '0') && (c <= '9');
+}
+
+bool IsCanonicalizedHostRFC1738Compliant(const std::string& host) {
+  if (host.empty())
+    return false;
+
+  enum State {
+    NOT_IN_COMPONENT,
+    IN_COMPONENT_STARTED_DIGIT,
+    IN_COMPONENT_STARTED_ALPHA
+  } state = NOT_IN_COMPONENT;
+  bool last_char_was_hyphen = false;
+
+  for (std::string::const_iterator i(host.begin()); i != host.end(); ++i) {
+    const char c = *i;
+    if (state == NOT_IN_COMPONENT) {
+      if (IsHostCharDigit(c))
+        state = IN_COMPONENT_STARTED_DIGIT;
+      else if (IsHostCharAlpha(c))
+        state = IN_COMPONENT_STARTED_ALPHA;
+      else
+        return false;
+    } else {
+      if (c == '.') {
+        if (last_char_was_hyphen)
+          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;
+      } else {
+        return false;
+      }
+    }
+  }
+
+  return state == IN_COMPONENT_STARTED_ALPHA;
+}
+
 std::string GetDirectoryListingEntry(const string16& name,
                                      const std::string& raw_bytes,
                                      bool is_dir,
diff --git a/net/base/net_util.h b/net/base/net_util.h
index 4d7e0aa..302a55f 100644
--- a/net/base/net_util.h
+++ b/net/base/net_util.h
@@ -158,6 +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:
+//   * One or more components separated by '.'
+//   * Each component begins and ends with an alphanumeric character
+//   * Each component contains only alphanumeric characters and '-'
+//   * 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);
+
 // Call these functions to get the html snippet for a directory listing.
 // The return values of both functions are in UTF-8.
 std::string GetDirectoryListingHeader(const string16& title);
diff --git a/net/base/net_util_unittest.cc b/net/base/net_util_unittest.cc
index 6a01ec9..22d467c 100644
--- a/net/base/net_util_unittest.cc
+++ b/net/base/net_util_unittest.cc
@@ -344,6 +344,11 @@ const IDNTestCase idn_cases[] = {
 #endif
 };
 
+struct RFC1738Case {
+  const char* host;
+  bool expected_output;
+};
+
 struct SuggestedFilenameCase {
   const char* url;
   const char* content_disp_header;
@@ -810,6 +815,32 @@ TEST(NetUtilTest, IDNToUnicodeSlow) {
   }
 }
 
+TEST(NetUtilTest, RFC1738) {
+  const RFC1738Case rfc1738_cases[] = {
+    {"", false},
+    {"a", true},
+    {"-", false},
+    {".", false},
+    {"a.", false},
+    {"a.a", true},
+    {"9.a", true},
+    {"a.9", false},
+    {"a.a9", true},
+    {"a.9a", false},
+    {"a+9a", false},
+    {"1-.a-b", false},
+    {"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));
+  }
+}
+
 TEST(NetUtilTest, StripWWW) {
   EXPECT_EQ(L"", net::StripWWW(L""));
   EXPECT_EQ(L"", net::StripWWW(L"www."));