regex

db/db.vcproj.Dwight-PC.Dwight Merriman.user

@@ -11,7 +11,7 @@
 			<DebugSettings
 				Command="$(TargetPath)"
 				WorkingDirectory=""
-				CommandArguments=""
+				CommandArguments="run"
 				Attach="false"
 				DebuggerType="3"
 				Remote="1"
@@ -22,7 +22,7 @@
 				SQLDebugging=""
 				Environment=""
 				EnvironmentMerge="true"
-				DebuggerFlavor=""
+				DebuggerFlavor="0"
 				MPIRunCommand=""
 				MPIRunArguments=""
 				MPIRunWorkingDirectory=""

db/jsobj.cpp

@@ -36,7 +36,7 @@ int Element::size() {
 				const char *p = value();
 				int len1 = strlen(p);
 				p = p + len1 + 1;
-				x = 1 + len1 + strlen(p);
+				x = 1 + len1 + strlen(p) + 2;
 			}
 			break;
 		default:
@@ -48,13 +48,38 @@ int Element::size() {
 }
 
 JSMatcher::JSMatcher(JSObj &_jsobj) : 
-   jsobj(_jsobj) 
+   jsobj(_jsobj), nRegex(0)
 {
 	JSElemIter i(jsobj);
 	n = 0;
 	while( i.more() ) {
 		Element e = i.next();
-		if( !e.eoo() ) {
+		if( e.eoo() )
+			break;
+
+		if( e.type() == RegEx ) {
+			if( nRegex >= 4 ) {
+				cout << "ERROR: too many regexes in query" << endl;
+			}
+			else {
+				pcrecpp::RE_Options options;
+				options.set_utf8(true);
+				const char *flags = e.regexFlags();
+				while( flags && *flags ) { 
+					if( *flags == 'i' )
+						options.set_caseless(true);
+					else if( *flags == 'm' )
+						options.set_multiline(true);
+					else if( *flags == 'x' )
+						options.set_extended(true);
+					flags++;
+				}
+				regexs[nRegex].re = new pcrecpp::RE(e.regex(), options);
+				regexs[nRegex].fieldName = e.fieldName();
+				nRegex++;
+			}
+		}
+		else {
 			toMatch.push_back(e);
 			n++;
 		}
@@ -62,7 +87,6 @@ JSMatcher::JSMatcher(JSObj &_jsobj) :
 }
 
 //#include <boost/regex.hpp>
-#include <pcrecpp.h> 
 //#include <pcre.h>
 
 struct RXTest { 
@@ -85,10 +109,26 @@ struct RXTest {
 
 bool JSMatcher::matches(JSObj& jsobj) {
 
-
 	/* assuming there is usually only one thing to match.  if more this
 	could be slow sometimes. */
 
+	for( int r = 0; r < nRegex; r++ ) { 
+		RegexMatcher& rm = regexs[r];
+		JSElemIter k(jsobj);
+		while( 1 ) {
+			if( !k.more() )
+				return false;
+			Element e = k.next();
+			if( strcmp(e.fieldName(), rm.fieldName) == 0 ) {
+				if( e.type() != String )
+					return false;
+				if( !rm.re->PartialMatch(e.valuestr()) )
+					return false;
+				break;
+			}
+		}
+	}
+
 	for( int i = 0; i < n; i++ ) {
 		Element& m = toMatch[i];
 		JSElemIter k(jsobj);

db/jsobj.h

@@ -81,6 +81,8 @@ public:
 	bool eoo() { return type() == EOO; }
 	int size();
 
+	const char * fieldName() { return data + 1; }
+
 	// raw data be careful:
 	const char * value() { return (data + fieldNameSize + 1); }
 
@@ -88,10 +90,18 @@ public:
 	int valuestrsize() { 
 		return *((int *) value());
 	}
+
 	// for strings.  also gives you start of the data for an embedded object
 	const char * valuestr() { return value() + 4; }
+
 	void* embeddedObject() { valuestr(); }
 
+	const char *regex() { assert(type() == RegEx); return value(); }
+	const char *regexFlags() { 
+		const char *p = regex();
+		return p + strlen(p) + 1;
+	}
+
 	bool operator==(Element& r) { 
 		int sz = size();
 		return sz == r.size() && 
@@ -100,7 +110,7 @@ public:
 
 private:
 	Element(const char *d) : data(d) {
-		fieldNameSize = eoo() ? 0 : strlen(data+1) + 1;
+		fieldNameSize = eoo() ? 0 : strlen(fieldName()) + 1;
 		totalSize = -1;
 	}
 	const char *data;
@@ -189,6 +199,16 @@ private:
 	const char *theend;
 };
 
+#include <pcrecpp.h> 
+
+class RegexMatcher { 
+public:
+	const char *fieldName;
+	pcrecpp::RE *re;
+	RegexMatcher() { re = 0; }
+	~RegexMatcher() { delete re; }
+};
+
 /* For when a js object is a pattern... */
 class JSMatcher { 
 public:
@@ -199,6 +219,9 @@ private:
 	JSObj& jsobj;
 	vector<Element> toMatch;
 	int n;
+
+	RegexMatcher regexs[4];
+	int nRegex;
 };
 
 /*- just for testing -- */