Port and integrate the irregexp engine from V8

runtime/lib/regexp.cc

Index: runtime/lib/regexp.cc
diff --git a/runtime/lib/regexp.cc b/runtime/lib/regexp.cc
index 646261628738d6904f277787168a02655548bfbd..4d10f7a2cd450b80c4186b55d7e2b9aa00b01793 100644
--- a/runtime/lib/regexp.cc
+++ b/runtime/lib/regexp.cc
@@ -4,6 +4,7 @@
 
 #include "platform/assert.h"
 #include "vm/bootstrap_natives.h"
+#include "vm/compiler.h"
 #include "vm/exceptions.h"
 #include "vm/native_entry.h"
 #include "vm/object.h"
@@ -13,6 +14,9 @@
 
 namespace dart {
 
+DECLARE_FLAG(bool, trace_irregexp);
+
+
 DEFINE_NATIVE_ENTRY(JSSyntaxRegExp_factory, 4) {
   ASSERT(TypeArguments::CheckedHandle(arguments->NativeArgAt(0)).IsNull());
   GET_NON_NULL_NATIVE_ARGUMENT(String, pattern, arguments->NativeArgAt(1));
@@ -22,7 +26,25 @@ DEFINE_NATIVE_ENTRY(JSSyntaxRegExp_factory, 4) {
       Instance, handle_case_sensitive, arguments->NativeArgAt(3));
   bool ignore_case = handle_case_sensitive.raw() != Bool::True().raw();
   bool multi_line = handle_multi_line.raw() == Bool::True().raw();
+
+#if defined(USE_JSCRE) ||                                                      \
+    defined(TARGET_ARCH_MIPS)
   return Jscre::Compile(pattern, multi_line, ignore_case);
+#else
+  // Parse the pattern once in order to throw any format exceptions within
+  // the factory constructor. It is parsed again upon compilation.
+  RegExpCompileData compileData;
+  if (!RegExpParser::ParseRegExp(pattern, multi_line, &compileData)) {
+    // Parsing failures throw an exception.
+    UNREACHABLE();
+  }
+
+  // Create a JSRegExp object containing only the initial parameters.
+  JSRegExp& regexp = JSRegExp::Handle(
+        isolate, RegExpEngine::New(isolate, pattern, multi_line, ignore_case));
+
+  return regexp.raw();
+#endif
 }
 
 
@@ -70,25 +92,16 @@ DEFINE_NATIVE_ENTRY(JSSyntaxRegExp_ExecuteMatch, 3) {
   GET_NON_NULL_NATIVE_ARGUMENT(String, str, arguments->NativeArgAt(1));
   GET_NON_NULL_NATIVE_ARGUMENT(Smi, start_index, arguments->NativeArgAt(2));
 
-  // The irregexp parser runs alongside the jscre parser while the port is
-  // still in progress. When done, it will replace jscre completely.
-  const String& pattern = String::Handle(regexp.pattern());
-  RegExpCompileData compileData;
-  if (RegExpParser::ParseRegExp(pattern, regexp.is_multi_line(),
-                                &compileData)) {
-    RegExpEngine::Compile(&compileData,
-                          regexp.is_ignore_case(),
-                          regexp.is_global(),  // A Dart regexp is always global
-                          regexp.is_multi_line(),
-                          pattern,
-                          str,
-                          str.IsOneByteString());
-    RegExpEngine::DotPrint(pattern.ToCString(),
-                           compileData.node,
-                           regexp.is_ignore_case());
-  }
-
+#if defined(USE_JSCRE) ||                                                      \
+    defined(TARGET_ARCH_MIPS)
   return Jscre::Execute(regexp, str, start_index.Value());
+#else
+  // This function is intrinsified. See Intrinsifier::JSRegExp_ExecuteMatch.

[srdjan, 2014/09/22 22:34:07]

We occasionally run with --no-intrinsify. Will it work then?

[jgruber1, 2014/09/23 10:58:46]

Good point, I wasn't aware of that flag. Added back the code for the
non-intrinsified version of this function.

+  UNREACHABLE();
+  USE(&str);
+  USE(&start_index);
+  return Array::null();
+#endif
 }
 
 }  // namespace dart