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..d162d44a70eb45d071350f1088c44576af6aaee5 100644
--- a/runtime/lib/regexp.cc
+++ b/runtime/lib/regexp.cc
@@ -13,6 +13,10 @@
 
 namespace dart {
 
+DECLARE_FLAG(bool, trace_irregexp);
+DEFINE_FLAG(bool, use_jscre, false,
+            "Use JSCRE instead of the irregexp engine.");
+
 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 +26,22 @@ 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();
-  return Jscre::Compile(pattern, multi_line, ignore_case);
+
+  if (FLAG_use_jscre) {
+    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.
+    // Compilation is done upon regexp execution.
+    return RegExpEngine::New(pattern, multi_line, ignore_case);
+  }
 }
 
 
@@ -65,30 +84,90 @@ DEFINE_NATIVE_ENTRY(JSSyntaxRegExp_getGroupCount, 1) {
 
 
 DEFINE_NATIVE_ENTRY(JSSyntaxRegExp_ExecuteMatch, 3) {
-  const JSRegExp& regexp = JSRegExp::CheckedHandle(arguments->NativeArgAt(0));
+  JSRegExp& regexp = JSRegExp::CheckedHandle(arguments->NativeArgAt(0));
   ASSERT(!regexp.IsNull());
   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 (FLAG_use_jscre) {
+    return Jscre::Execute(regexp, str, start_index.Value());
+  } else {
+    const bool is_one_byte_string = str.IsOneByteString();

[Florian Schneider, 2014/09/16 11:11:48]

Could be also ExternalOneByteString.

[jgruber1, 2014/09/22 18:58:03]

The generated code is now specialized for all four string types.

+
+    // Retrieve the cached function.
+    Function& fn = Function::Handle(is_one_byte_string ?
+          regexp.one_byte_function() : regexp.two_byte_function());
+
+    // Compile if needed.
+    if (fn.IsNull()) {
+      // Parse the pattern.
+      const String& pattern = String::Handle(regexp.pattern());
+      RegExpCompileData compileData;
+      if (!RegExpParser::ParseRegExp(pattern, regexp.is_multi_line(),
+                                    &compileData)) {
+        // Parsing failures are handled in the factory constructor.
+        UNREACHABLE();
+      }
+
+      // Compile it.
+      RegExpEngine::CompilationResult result =
+          RegExpEngine::Compile(
+            &compileData,
+            regexp.is_ignore_case(),
+            regexp.is_global(),  // A Dart regexp is always global
+            regexp.is_multi_line(),
+            pattern,
+            str,
+            str.IsOneByteString());
+      if (result.function.IsNull()) {
+        // Should never fail.
+        UNREACHABLE();
+      }
+
+      // Debug output of the generated state machine.
+      if (FLAG_trace_irregexp) {
+#ifdef DEBUG
+        RegExpEngine::DotPrint(pattern.ToCString(),
+                               compileData.node,
+                               regexp.is_ignore_case());
+#endif
+      }
+
+      // Cache the result.
+      if (is_one_byte_string) {
+        regexp.set_one_byte_function(result.function);
+      } else {
+        regexp.set_two_byte_function(result.function);
+      }
+      regexp.set_num_bracket_expressions(compileData.capture_count);
+
+      fn = result.function.raw();
+    }
+
+    ASSERT(!fn.IsNull());
+
+    // Create the output array.

[Florian Schneider, 2014/09/16 11:11:47]

Creating the output array should be part of the generated code so that we can
return without going to C++ first.

[jgruber1, 2014/09/22 18:58:03]

Done.

+    const Smi& num_bracket_exprs =
+        Smi::Handle(regexp.num_bracket_expressions());
+    intptr_t num_bracket_expressions = num_bracket_exprs.Value();
+    const intptr_t kMatchPair = 2;
+    Array& matches = Array::Handle(
+          Array::New(kMatchPair * (num_bracket_expressions + 1)));
+
+    // And finally call the generated code.
+    IRRegExpMacroAssembler::Result result =
+        IRRegExpMacroAssembler::Execute(fn, str, start_index,
+                                        &matches, isolate);
+
+    if (result == IRRegExpMacroAssembler::SUCCESS) {
+      return matches.raw();
+    } else {
+      return Array::null();
+    }
   }
 
-  return Jscre::Execute(regexp, str, start_index.Value());
+  UNREACHABLE();
+  return Array::null();
 }
 
 }  // namespace dart