$where optimization

6 files changed, 196 insertions, 12 deletions

diff --git a/db/db.cpp b/db/db.cpp
index cefbff22e4f..d2d73d9f67a 100644
--- a/db/db.cpp
+++ b/db/db.cpp
@@ -273,7 +273,7 @@ public:
 };

 

 void listen(int port) { 

-	cout << "db version: 102 mar2008 where" << endl;

+	cout << "db version: 103 mar2008 minilex" << endl;

 	pdfileInit();

 	testTheDb();

 	cout << curTimeMillis() % 10000 << " waiting for connections...\n" << endl;

@@ -552,9 +552,6 @@ void msg(const char *m, int extras = 0) {
 

 int main(int argc, char* argv[], char *envp[] )

 {

-	JavaJS = new JavaJSImpl();

-	javajstest();

-

 	srand(curTimeMillis());

 

 	if( argc >= 2 ) {

@@ -583,6 +580,8 @@ int main(int argc, char* argv[], char *envp[] )
 			return 0;

 		}

 		if( strcmp(argv[1], "run") == 0 ) {

+			JavaJS = new JavaJSImpl();

+			javajstest();

 			listen(port);

 			goingAway = true;

 			return 0;

diff --git a/db/db.vcproj b/db/db.vcproj
index 9752e6544f9..f375ef80324 100644
--- a/db/db.vcproj
+++ b/db/db.vcproj
@@ -281,6 +281,10 @@
 				>

 			</File>

 			<File

+				RelativePath=".\minilex.h"

+				>

+			</File>

+			<File

 				RelativePath="..\util\mmap.h"

 				>

 			</File>

diff --git a/db/javajs.cpp b/db/javajs.cpp
index ec5e533a2c6..e7a6ce41fa7 100644
--- a/db/javajs.cpp
+++ b/db/javajs.cpp
@@ -44,7 +44,7 @@ using namespace std;
 		  longer needed and can be removed. 

 */

 extern "C" void tss_cleanup_implemented(void) {

-	cout << "tss_cleanup_implemented called" << endl;

+	//cout << "tss_cleanup_implemented called" << endl;

 }
 #endif
 
diff --git a/db/jsobj.cpp b/db/jsobj.cpp
index ded7aa5c9dd..b8ad2b7b266 100644
--- a/db/jsobj.cpp
+++ b/db/jsobj.cpp
@@ -5,13 +5,88 @@
 #include "../util/goodies.h"

 #include "javajs.h"

 

+#if defined(_WIN32)

+

+#include <hash_map>

+using namespace stdext;

+

+typedef const char * MyStr;

+struct less_str {

+   bool operator()(const MyStr & x, const MyStr & y) const {

+      if ( strcmp(x, y) > 0)

+         return true;

+

+      return false;

+   }

+};

+

+typedef hash_map<const char*, int, hash_compare<const char *, less_str> > strhashmap;

+

+#else

+

+#include <ext/hash_map>

+using namespace __gnu_cxx;

+

+typedef const char * MyStr;

+struct eq_str {

+   bool operator()(const MyStr & x, const MyStr & y) const {

+      if ( strcmp(x, y) == 0)

+         return true;

+

+      return false;

+   }

+};

+

+typedef hash_map<const char*, int, hash<const char *>, eq_str > strhashmap;

+

+#endif

+

+#include "minilex.h"

+

+MiniLex minilex;

+

 class Where { 

 public:

+	Where() { codeCopy = 0; }

 	~Where() {

 		JavaJS->scopeFree(scope);

-		scope = 0; func = 0;

+		delete codeCopy;

+		scope = 0; func = 0; codeCopy = 0;

 	}

 	jlong scope, func;

+	strhashmap fields;

+//	map<string,int> fields;

+	bool fullObject;

+	int nFields;

+	char *codeCopy;

+

+	void setFunc(const char *code) {

+		codeCopy = new char[strlen(code)+1];

+		strcpy(codeCopy,code);

+		func = JavaJS->functionCreate( code );

+		minilex.grabVariables(codeCopy, fields);

+		fullObject = fields.count("fullObject") > 0;

+		nFields = fields.size();

+	}

+

+	void buildSubset(JSObj& src, JSObjBuilder& dst) { 

+		JSElemIter it(src);

+		int n = 0;

+		if( !it.more() ) return;

+		while( 1 ) {

+			Element e = it.next();

+			if( e.eoo() )

+				break;

+			if( //n == 0 && 

+				fields.find(e.fieldName()) != fields.end()

+				//fields.count(e.fieldName())

+				) {

+				dst.append(e);

+				if( ++n >= nFields )

+					break;

+			}

+		}

+	}

 };

 

 JSMatcher::~JSMatcher() { 

@@ -243,13 +318,10 @@ JSMatcher::JSMatcher(JSObj &_jsobj) :
 			assert( where == 0 );

 			where = new Where();

 			const char *code = e.valuestr();

-                        if ( ! JavaJS ){

-                          JavaJS = new JavaJSImpl();

-                          javajstest();

-                        }

+			assert( JavaJS );

 			where->scope = JavaJS->scopeCreate();

 			JavaJS->scopeSetString(where->scope, "$client", client->name.c_str());

-			where->func = JavaJS->functionCreate( code );

+			where->setFunc(code);

 			continue;

 		}

 

@@ -398,7 +470,14 @@ ok:
 	if( where ) { 

 		if( where->func == 0 )

 			return false; // didn't compile

-		JavaJS->scopeSetObject(where->scope, "obj", &jsobj);

+		if( jsobj.objsize() < 200 || where->fullObject ) {

+			JavaJS->scopeSetObject(where->scope, "obj", &jsobj);

+		} else {

+			JSObjBuilder b;

+			where->buildSubset(jsobj, b);

+			JSObj temp = b.done();

+			JavaJS->scopeSetObject(where->scope, "obj", &temp);

+		}

 		if( JavaJS->invoke(where->scope, where->func) )

 			return false;

 		return JavaJS->scopeGetBoolean(where->scope, "return") != 0;

diff --git a/db/jsobj.h b/db/jsobj.h
index cb4fcd18941..46ba252f43f 100644
--- a/db/jsobj.h
+++ b/db/jsobj.h
@@ -313,6 +313,11 @@ public:
 		int l;

 		return JSObj(decouple(l), true);

 	}

+

+	/* this version, jsobjbuilder still frees the jsobj

+	when the builder goes out of scope.  use it this way

+	by default, that's simplest.

+	*/

 	JSObj done() { 

 		return JSObj(_done());

 	}

diff --git a/db/minilex.h b/db/minilex.h
new file mode 100644
index 00000000000..64c98fb9ff8
--- /dev/null
+++ b/db/minilex.h
@@ -0,0 +1,97 @@
+// minilex.h

+// mini js lexical analyzer.  idea is to be dumb and fast.

+

+struct MiniLex {

+	strhashmap reserved;

+	bool ic[256]; // ic=Identifier Character

+	bool starter[256];

+

+	// dm: very dumb about comments and escaped quotes -- but we are faster then at least, 

+	// albeit returning too much (which is ok for jsbobj current usage).

+	void grabVariables(char *code /*modified and must stay in scope*/, strhashmap& vars) { 

+		char *p = code;

+		char last = 0;

+		while( *p ) { 

+			if( starter[*p] ) { 

+				char *q = p+1;

+				while( *q && ic[*q] ) q++;

+				const char *identifier = p;

+				bool done = *q == 0;

+				*q = 0;

+				if( !reserved.count(identifier) ) {

+					// we try to be smart about 'obj' but have to be careful as obj.obj 

+					// can happen; this is so that nFields is right for simplistic where cases

+					// so we can stop scanning in jsobj when we find the field of interest.

+					if( strcmp(identifier,"obj")==0 && p>code && p[-1] != '.' )

+						;

+					else

+						vars[identifier] = 1;

+				}

+				if( done )

+					break;

+				p = q + 1;

+				continue;

+			}

+

+			if( *p == '\'' ) {

+				p++;

+				while( *p && *p != '\'' ) p++;

+			}

+			else if( *p == '"' ) {

+				p++;

+				while( *p && *p != '"' ) p++;

+			}

+			p++;

+		}

+	}

+

+	MiniLex() { 

+		strhashmap atest;

+		atest["foo"] = 3;

+		assert( atest.count("bar") == 0 );

+		assert( atest.count("foo") == 1 );

+		assert( atest["foo"] == 3 );

+

+		for( int i = 0; i < 256; i++ ) { 

+			ic[i] = starter[i] = false;

+		}

+		for( int i = 'a'; i <= 'z'; i++ )

+			ic[i] = starter[i] = true;

+		for( int i = 'A'; i <= 'Z'; i++ )

+			ic[i] = starter[i] = true;

+		for( int i = '0'; i <= '9'; i++ )

+			ic[i] = true;

+		for( int i = 128; i < 256; i++ )

+			ic[i] = starter[i] = true;

+		ic['$'] = starter['$'] = true;

+		ic['_'] = starter['_'] = true;

+

+		reserved["break"] = true;

+		reserved["case"] = true;

+		reserved["catch"] = true;

+		reserved["continue"] = true;

+		reserved["default"] = true;

+		reserved["delete"] = true;

+		reserved["do"] = true;

+		reserved["else"] = true;

+		reserved["finally"] = true;

+		reserved["for"] = true;

+		reserved["function"] = true;

+		reserved["if"] = true;

+		reserved["in"] = true;

+		reserved["instanceof"] = true;

+		reserved["new"] = true;

+		reserved["return"] = true;

+		reserved["switch"] = true;

+		reserved["this"] = true;

+		reserved["throw"] = true;

+		reserved["try"] = true;

+		reserved["typeof"] = true;

+		reserved["var"] = true;

+		reserved["void"] = true;

+		reserved["while"] = true;

+		reserved["with "] = true;

+	}

+};

+

+