Optimize KURL protocols

third_party/WebKit/Source/platform/weborigin/KURL.cpp

Index: third_party/WebKit/Source/platform/weborigin/KURL.cpp
diff --git a/third_party/WebKit/Source/platform/weborigin/KURL.cpp b/third_party/WebKit/Source/platform/weborigin/KURL.cpp
index 72cbe2122c47cf988e8c7ea64f7062690d158836..7f64c97f8a5761bba3985bc16cea0f910beef2bf 100644
--- a/third_party/WebKit/Source/platform/weborigin/KURL.cpp
+++ b/third_party/WebKit/Source/platform/weborigin/KURL.cpp
@@ -240,7 +240,7 @@ KURL::KURL(const AtomicString& canonicalString,
       m_protocolIsInHTTPFamily(false),
       m_parsed(parsed),
       m_string(canonicalString) {
-  initProtocolIsInHTTPFamily();
+  initProtocolMetadata();
   initInnerURL();
 }
 
@@ -252,6 +252,7 @@ KURL::KURL(WTF::HashTableDeletedValueType)
 KURL::KURL(const KURL& other)
     : m_isValid(other.m_isValid),
       m_protocolIsInHTTPFamily(other.m_protocolIsInHTTPFamily),
+      m_protocol(other.m_protocol),
       m_parsed(other.m_parsed),
       m_string(other.m_string) {
   if (other.m_innerURL.get())
@@ -263,6 +264,7 @@ KURL::~KURL() {}
 KURL& KURL::operator=(const KURL& other) {
   m_isValid = other.m_isValid;
   m_protocolIsInHTTPFamily = other.m_protocolIsInHTTPFamily;
+  m_protocol = other.m_protocol;
   m_parsed = other.m_parsed;
   m_string = other.m_string;
   if (other.m_innerURL)
@@ -276,6 +278,7 @@ KURL KURL::copy() const {
   KURL result;
   result.m_isValid = m_isValid;
   result.m_protocolIsInHTTPFamily = m_protocolIsInHTTPFamily;
+  result.m_protocol = m_protocol.isolatedCopy();
   result.m_parsed = m_parsed;
   result.m_string = m_string.isolatedCopy();
   if (m_innerURL)
@@ -335,7 +338,7 @@ String KURL::lastPathComponent() const {
 }
 
 String KURL::protocol() const {
-  return componentString(m_parsed.scheme);
+  return m_protocol;
 }
 
 String KURL::host() const {
@@ -432,6 +435,7 @@ bool KURL::setProtocol(const String& protocol) {
   replacements.SetScheme(charactersOrEmpty(newProtocolUTF8),
                          url::Component(0, newProtocolUTF8.length()));
   replaceComponents(replacements);
+  initProtocolMetadata();
 
   // isValid could be false but we still return true here. This is because
   // WebCore or JS scripts can build up a URL by setting individual
@@ -722,7 +726,7 @@ void KURL::init(const KURL& base,
     init(base, relativeUTF8.data(), relativeUTF8.length(), queryEncoding);
   } else
     init(base, relative.characters16(), relative.length(), queryEncoding);
-  initProtocolIsInHTTPFamily();
+  initProtocolMetadata();
   initInnerURL();
   DCHECK_EQ(protocol(), protocol().lower());
 }
@@ -772,50 +776,22 @@ void KURL::initInnerURL() {
     m_innerURL.reset();
 }
 
-template <typename CHAR>
-bool internalProtocolIs(const url::Component& scheme,
-                        const CHAR* spec,
-                        const char* protocol) {
-  const CHAR* begin = spec + scheme.begin;
-  const CHAR* end = begin + scheme.len;
-
-  while (begin != end && *protocol) {
-    ASSERT(toASCIILower(*protocol) == *protocol);
-    if (toASCIILower(*begin++) != *protocol++)
-      return false;
-  }
-
-  // Both strings are equal (ignoring case) if and only if all of the characters
-  // were equal, and the end of both has been reached.
-  return begin == end && !*protocol;
-}
-
-template <typename CHAR>
-bool checkIfProtocolIsInHTTPFamily(const url::Component& scheme,
-                                   const CHAR* spec) {
-  if (scheme.len == 4)
-    return internalProtocolIs(scheme, spec, "http");
-  if (scheme.len == 5)
-    return internalProtocolIs(scheme, spec, "https");
-  if (scheme.len == 7)
-    return internalProtocolIs(scheme, spec, "http-so");
-  if (scheme.len == 8)
-    return internalProtocolIs(scheme, spec, "https-so");
-  return false;
-}
-
-void KURL::initProtocolIsInHTTPFamily() {
+void KURL::initProtocolMetadata() {
   if (!m_isValid) {
     m_protocolIsInHTTPFamily = false;
     return;
   }
 
   ASSERT(!m_string.isNull());
-  m_protocolIsInHTTPFamily =
-      m_string.is8Bit() ? checkIfProtocolIsInHTTPFamily(m_parsed.scheme,
-                                                        m_string.characters8())
-                        : checkIfProtocolIsInHTTPFamily(
-                              m_parsed.scheme, m_string.characters16());
+  m_protocol = componentString(m_parsed.scheme);
+  if (!m_protocol.isEmpty())
+    m_protocol = (m_protocol);

[esprehn, 2016/11/04 23:48:32]

what does this do? assigning m_protocol to itself seems like it doesn't do
anything.

[Charlie Harrison, 2016/11/04 23:58:35]

Oops. That should be m_protocol = AtomicString(m_protocol). Sorry about that :/

+
+  // TODO(csharrison): Add "http" and "https" to the AtomicString table at init
+  // time so that the normal case here is just a pointer comparison.
+  m_protocolIsInHTTPFamily = m_protocol == "http" || m_protocol == "https" ||
+                             m_protocol == "http-so" ||
+                             m_protocol == "https-so";
 }
 
 bool KURL::protocolIs(const char* protocol) const {
@@ -827,14 +803,9 @@ bool KURL::protocolIs(const char* protocol) const {
   // FIXME: Chromium code needs to be fixed for this assert to be enabled.
   // ASSERT(strcmp(protocol, "javascript"));
 
-  if (m_string.isNull() || m_parsed.scheme.len <= 0)
+  if (m_string.isNull() || m_protocol.isNull() || m_parsed.scheme.len <= 0)
     return *protocol == '\0';
-
-  return m_string.is8Bit()
-             ? internalProtocolIs(m_parsed.scheme, m_string.characters8(),
-                                  protocol)
-             : internalProtocolIs(m_parsed.scheme, m_string.characters16(),
-                                  protocol);
+  return m_protocol == protocol;
 }
 
 String KURL::stringForInvalidComponent() const {