Added an internal flag to enable caching.

uhop · Jun 5, 2024 · 335aed2 · 335aed2
1 parent e5b8978
commit 335aed2
Show file tree

Hide file tree

Showing 13 changed files with 72 additions and 26 deletions.
diff --git a/binding.gyp b/binding.gyp
@@ -3,8 +3,10 @@
     {
       "target_name": "re2",
       "sources": [
-        "lib/str-val.cc",
         "lib/addon.cc",
+        "lib/accessors.cc",
+        "lib/str-val.cc",
+        "lib/util.cc",
         "lib/new.cc",
         "lib/exec.cc",
         "lib/test.cc",
@@ -13,9 +15,6 @@
         "lib/search.cc",
         "lib/split.cc",
         "lib/to_string.cc",
-        "lib/accessors.cc",
-        "lib/util.cc",
-        "lib/str-val.cc",
         "vendor/re2/re2/bitmap256.cc",
         "vendor/re2/re2/bitstate.cc",
         "vendor/re2/re2/compile.cc",

diff --git a/lib/accessors.cc b/lib/accessors.cc
@@ -28,6 +28,30 @@ NAN_GETTER(WrappedRE2::GetInternalSource)
 	info.GetReturnValue().Set(Nan::New(re2->regexp.pattern()).ToLocalChecked());
 }
 
+NAN_GETTER(WrappedRE2::GetEnabledCache)
+{
+	if (!WrappedRE2::HasInstance(info.This()))
+	{
+		info.GetReturnValue().SetUndefined();
+		return;
+	}
+
+	auto re2 = Nan::ObjectWrap::Unwrap<WrappedRE2>(info.This());
+	info.GetReturnValue().Set(re2->enabledCache);
+}
+
+NAN_GETTER(WrappedRE2::GetIsCached)
+{
+	if (!WrappedRE2::HasInstance(info.This()))
+	{
+		info.GetReturnValue().SetUndefined();
+		return;
+	}
+
+	auto re2 = Nan::ObjectWrap::Unwrap<WrappedRE2>(info.This());
+	info.GetReturnValue().Set(!!re2->lastStringValue);
+}
+
 NAN_GETTER(WrappedRE2::GetFlags)
 {
 	if (!WrappedRE2::HasInstance(info.This()))
@@ -39,9 +63,13 @@ NAN_GETTER(WrappedRE2::GetFlags)
 	auto re2 = Nan::ObjectWrap::Unwrap<WrappedRE2>(info.This());
 
 	std::string flags;
+	if (re2->enabledCache)
+	{
+		flags += "\b";
+	}
 	if (re2->hasIndices)
 	{
-		flags = "d";
+		flags += "d";
 	}
 	if (re2->global)
 	{

diff --git a/lib/addon.cc b/lib/addon.cc
@@ -74,6 +74,8 @@ v8::Local<v8::Function> WrappedRE2::Init()
 	Nan::SetAccessor(instanceTemplate, Nan::New("hasIndices").ToLocalChecked(), GetHasIndices);
 	Nan::SetAccessor(instanceTemplate, Nan::New("lastIndex").ToLocalChecked(), GetLastIndex, SetLastIndex);
 	Nan::SetAccessor(instanceTemplate, Nan::New("internalSource").ToLocalChecked(), GetInternalSource);
+	Nan::SetAccessor(instanceTemplate, Nan::New("enabledCache").ToLocalChecked(), GetEnabledCache);
+	Nan::SetAccessor(instanceTemplate, Nan::New("isCached").ToLocalChecked(), GetIsCached);
 
 	auto ctr = Nan::GetFunction(tpl).ToLocalChecked();
 
@@ -104,19 +106,6 @@ void WrappedRE2::dropLastString()
 	}
 }
 
-inline size_t countBytes(const char *data, size_t from, size_t n)
-{
-	for (; n > 0; --n)
-	{
-		size_t s = getUtf8CharSize(data[from]);
-		from += s;
-		if (s == 4 && n >= 2)
-			--n; // this utf8 character will take two utf16 characters
-					 // the decrement above is protected to avoid an overflow of an unsigned integer
-	}
-	return from;
-}
-
 void WrappedRE2::weakLastStringCallback(const Nan::WeakCallbackInfo<WrappedRE2> &data)
 {
 	WrappedRE2* re2 = data.GetParameter();

diff --git a/lib/exec.cc b/lib/exec.cc
@@ -15,7 +15,7 @@ NAN_METHOD(WrappedRE2::Exec)
 		return;
 	}
 
-	re2->prepareLastString(info[0]);
+	PrepareLastString prepare(re2, info[0]);
 	StrValBase &str = *re2->lastStringValue;
 	if (str.isBad) return; // throws an exception
 

diff --git a/lib/match.cc b/lib/match.cc
@@ -15,7 +15,7 @@ NAN_METHOD(WrappedRE2::Match)
 		return;
 	}
 
-	re2->prepareLastString(info[0], re2->global);
+	PrepareLastString prepare(re2, info[0], re2->global);
 	StrValBase &str = *re2->lastStringValue;
 	if (str.isBad) return; // throws an exception
 

diff --git a/lib/new.cc b/lib/new.cc
@@ -233,6 +233,7 @@ NAN_METHOD(WrappedRE2::New)
 	bool unicode = false;
 	bool sticky = false;
 	bool hasIndices = false;
+	bool enabledCache = false;
 
 	auto context = Nan::GetCurrentContext();
 	bool needFlags = true;
@@ -256,6 +257,9 @@ NAN_METHOD(WrappedRE2::New)
 		{
 			switch (data[i])
 			{
+			case '\b':
+				enabledCache = true;
+				break;
 			case 'g':
 				global = true;
 				break;
@@ -339,6 +343,7 @@ NAN_METHOD(WrappedRE2::New)
 
 			if (needFlags)
 			{
+				enabledCache = re2->enabledCache;
 				global = re2->global;
 				ignoreCase = re2->ignoreCase;
 				multiline = re2->multiline;
@@ -401,7 +406,7 @@ NAN_METHOD(WrappedRE2::New)
 	options.set_dot_nl(dotAll);
 	options.set_log_errors(false); // inappropriate when embedding
 
-	std::unique_ptr<WrappedRE2> re2(new WrappedRE2(re2::StringPiece(data, size), options, source, global, ignoreCase, multiline, dotAll, sticky, hasIndices));
+	std::unique_ptr<WrappedRE2> re2(new WrappedRE2(re2::StringPiece(data, size), options, source, enabledCache, global, ignoreCase, multiline, dotAll, sticky, hasIndices));
 	if (!re2->regexp.ok())
 	{
 		return Nan::ThrowSyntaxError(re2->regexp.error().c_str());

diff --git a/lib/replace.cc b/lib/replace.cc
@@ -492,7 +492,7 @@ NAN_METHOD(WrappedRE2::Replace)
 		return;
 	}
 
-	re2->prepareLastString(info[0]);
+	PrepareLastString prepare(re2, info[0]);
 	StrValBase &replacee = *re2->lastStringValue;
 	if (replacee.isBad) return; // throws an exception
 

diff --git a/lib/search.cc b/lib/search.cc
@@ -13,7 +13,7 @@ NAN_METHOD(WrappedRE2::Search)
 		return;
 	}
 
-	re2->prepareLastString(info[0], true);
+	PrepareLastString prepare(re2, info[0], true);
 	StrValBase &str = *re2->lastStringValue;
 	if (str.isBad) return; // throws an exception
 

diff --git a/lib/split.cc b/lib/split.cc
@@ -20,7 +20,7 @@ NAN_METHOD(WrappedRE2::Split)
 		return;
 	}
 
-	re2->prepareLastString(info[0], true);
+	PrepareLastString prepare(re2, info[0], true);
 	StrValBase &str = *re2->lastStringValue;
 	if (str.isBad) return; // throws an exception
 

diff --git a/lib/test.cc b/lib/test.cc
@@ -15,7 +15,7 @@ NAN_METHOD(WrappedRE2::Test)
 		return;
 	}
 
-	re2->prepareLastString(info[0]);
+	PrepareLastString prepare(re2, info[0]);
 	StrValBase &str = *re2->lastStringValue;
 	if (str.isBad) return; // throws an exception
 

diff --git a/lib/to_string.cc b/lib/to_string.cc
@@ -20,6 +20,10 @@ NAN_METHOD(WrappedRE2::ToString)
 	buffer += re2->source;
 	buffer += "/";
 
+	if (re2->enabledCache)
+	{
+		buffer += "\b";
+	}
 	if (re2->global)
 	{
 		buffer += "g";

diff --git a/lib/wrapped_re2.h b/lib/wrapped_re2.h
@@ -16,13 +16,15 @@ class WrappedRE2 : public Nan::ObjectWrap
 		const re2::StringPiece &pattern,
 		const re2::RE2::Options &options,
 		const std::string &src,
+		const bool &c,
 		const bool &g,
 		const bool &i,
 		const bool &m,
 		const bool &s,
 		const bool &y,
 		const bool &d) : regexp(pattern, options),
 						 source(src),
+						 enabledCache(c),
 						 global(g),
 						 ignoreCase(i),
 						 multiline(m),
@@ -47,6 +49,8 @@ class WrappedRE2 : public Nan::ObjectWrap
 	static NAN_GETTER(GetLastIndex);
 	static NAN_SETTER(SetLastIndex);
 	static NAN_GETTER(GetInternalSource);
+	static NAN_GETTER(GetEnabledCache);
+	static NAN_GETTER(GetIsCached);
 
 	// RegExp methods
 	static NAN_METHOD(Exec);
@@ -89,6 +93,7 @@ class WrappedRE2 : public Nan::ObjectWrap
 
 	re2::RE2 regexp;
 	std::string source;
+	bool enabledCache;
 	bool global;
 	bool ignoreCase;
 	bool multiline;
@@ -97,6 +102,8 @@ class WrappedRE2 : public Nan::ObjectWrap
 	bool hasIndices;
 	size_t lastIndex;
 
+	friend class PrepareLastString;
+
 private:
 	Nan::Persistent<v8::Value> lastString; // weak pointer
 	StrValBase *lastStringValue;
@@ -107,6 +114,20 @@ class WrappedRE2 : public Nan::ObjectWrap
 	void prepareLastString(const v8::Local<v8::Value> &arg, bool ignoreLastIndex = false);
 };
 
+struct PrepareLastString
+{
+	PrepareLastString(WrappedRE2 *re2, const v8::Local<v8::Value> &arg, bool ignoreLastIndex = false) : re2(re2) {
+		re2->prepareLastString(arg, ignoreLastIndex);
+	}
+
+	~PrepareLastString() {
+		if (!re2->enabledCache || !(re2->global || re2->sticky))
+			re2->dropLastString();
+	}
+
+	WrappedRE2 *re2;
+};
+
 // utilities
 
 inline size_t getUtf8Length(const uint16_t *from, const uint16_t *to)

diff --git a/re2.js b/re2.js
@@ -24,7 +24,7 @@ if (typeof Symbol != 'undefined') {
       if (!this.global) {
         throw TypeError('String.prototype.matchAll called with a non-global RE2 argument');
       }
-      const re = new RE2(this);
+      const re = new RE2(this, this.flags + '\b');
       re.lastIndex = this.lastIndex;
       for (;;) {
         const result = re.exec(str);