Improved the lexer to be functionally equlivant to the C++ version and ported the tests, commented out the dot generation from Import as it was slowing things down significantly.

src/Importer.cpp

@@ -130,7 +130,7 @@ void Importer::import(std::string fileName) {
 			return;
 		}
 		if (i.ast) {
-			outFileAST << i.ast->DOTGraphString() << std::endl;
+			//outFileAST << i.ast->DOTGraphString() << std::endl;
 		} else {
 			std::cout << "Tree returned from ASTTransformation for " << fileName << " is NULL!" << std::endl;
 		}
@@ -187,7 +187,7 @@ NodeTree<Symbol>* Importer::parseAndTrim(std::string fileName) {
 
 	if (parseTree) {
         //std::cout << parseTree->DOTGraphString() << std::endl;
-		outFile << parseTree->DOTGraphString() << std::endl;
+		//outFile << parseTree->DOTGraphString() << std::endl;
 	} else {
 		std::cout << "ParseTree returned from parser for " << fileName << " is NULL!" << std::endl;
 	    outFile.close(); outFileTransformed.close();
@@ -207,7 +207,7 @@ NodeTree<Symbol>* Importer::parseAndTrim(std::string fileName) {
 		parseTree = CollapseTransformation<Symbol>(collapseSymbols[i]).transform(parseTree);
 
 	if (parseTree) {
-		outFileTransformed << parseTree->DOTGraphString() << std::endl;
+		//outFileTransformed << parseTree->DOTGraphString() << std::endl;
 	} else {
 		std::cout << "Tree returned from transformation is NULL!" << std::endl;
 	}

stdlib/lexer.krak

@@ -30,16 +30,23 @@ obj lexer {
     fun add_regex(newOne: regex::regex) {
         regs.add(newOne)
     }
+    fun add_regex(newOne: char*) {
+        regs.add(regex::regex(newOne))
+    }
     fun set_input(in: string::string) {
         input = in
     }
     fun next(): symbol::symbol {
+        if (position >= input.length())
+            return symbol::symbol("$EOF$", true)
         var max = regs.map(fun(reg: regex::regex): util::pair<int, string::string> {
                 return util::make_pair(reg.long_match(input.slice(position, -1)), reg.regexString); })
                 .max(fun(first: util::pair<int, string::string>, second: util::pair<int, string::string>): bool
                         { return first.first < second.first; })
+        if (max.first < 0)
+            return symbol::symbol("$INVALID$", true)
         position += max.first
-        return symbol::symbol(input.slice(position-max.first, position), max.second, true)
+        return symbol::symbol(max.second, true, input.slice(position-max.first, position))
     }
 }
 

stdlib/symbol.krak

@@ -1,12 +1,22 @@
 import string
 
-fun symbol(dataIn: char*, nameIn: char*, terminalIn: bool): symbol {
+fun symbol(nameIn: char*, terminalIn: bool): symbol {
-    var toRet.construct(string::string(dataIn), string::string(nameIn), terminalIn): symbol
+    var toRet.construct(string::string(nameIn), terminalIn, string::string("no_value")): symbol
     return toRet
 }
 
-fun symbol(dataIn: string::string, nameIn: string::string, terminalIn: bool): symbol {
+fun symbol(nameIn: string::string, terminalIn: bool): symbol {
-    var toRet.construct(dataIn, nameIn, terminalIn): symbol
+    var toRet.construct(nameIn, terminalIn, string::string("no_value")): symbol
+    return toRet
+}
+
+fun symbol(nameIn: char*, terminalIn: bool, dataIn: char*): symbol {
+    var toRet.construct(string::string(nameIn), terminalIn, string::string(dataIn)): symbol
+    return toRet
+}
+
+fun symbol(nameIn: string::string, terminalIn: bool, dataIn: string::string): symbol {
+    var toRet.construct(nameIn, terminalIn, dataIn): symbol
     return toRet
 }
 
@@ -20,10 +30,10 @@ obj symbol {
         name.construct()
         return this
     }
-    fun construct(dataIn: string::string, nameIn: string::string, terminalIn: bool): symbol* {
+    fun construct(nameIn: string::string, terminalIn: bool, dataIn: string::string): symbol* {
-        data.construct(dataIn)
         name.construct(nameIn)
         terminal = terminalIn
+        data.construct(dataIn)
         return this
     }
     fun destruct() {

tests/test_lexer.expected_results

@@ -1,2 +1,22 @@
 a+: aaaa true
 test: test true
+old contributed tests
+b: b true
+b: b true
+$EOF$: no_value true
+
+a*: aaa true
+b: b true
+a*: aa true
+b: b true
+b: b true
+$EOF$: no_value true
+
+a|b: b true
+$INVALID$: no_value true
+
+xyzzy: xyzzy true
+$EOF$: no_value true
+
+(i|n|t|e)+: intent true
+$EOF$: no_value true

tests/test_lexer.krak

@@ -3,14 +3,55 @@ import regex:*
 import string:*
 import symbol:*
 import io:*
+import util:*
 
 fun main(): int {
     var lex.construct(): lexer
     lex.set_input(string("aaaatesta"))
     lex.add_regex(regex("a+"))
-    lex.add_regex(regex("test"))
+    lex.add_regex("test")
     println(lex.next().to_string())
     println(lex.next().to_string())
+
+    println("old contributed tests")
+
+    {
+        var lex.construct(): lexer
+        lex.add_regex("b")
+        lex.set_input(string("bb"))
+        range(3).for_each(fun(i: int) { println(lex.next().to_string()); } )
+    }
+    println()
+    {
+        var lex.construct(): lexer
+        lex.add_regex("a*")
+        lex.add_regex("b")
+        lex.set_input(string("aaabaabb"))
+        range(6).for_each(fun(i: int) { println(lex.next().to_string()); } )
+    }
+    println()
+    {
+        var lex.construct(): lexer
+        lex.add_regex("a|b")
+        lex.set_input(string("blah"))
+        range(2).for_each(fun(i: int) { println(lex.next().to_string()); } )
+    }
+    println()
+    {
+        var lex.construct(): lexer
+        lex.add_regex("xyzzy")
+        lex.set_input(string("xyzzy"))
+        range(2).for_each(fun(i: int) { println(lex.next().to_string()); } )
+    }
+    println()
+    {
+        var lex.construct(): lexer
+        lex.add_regex("int")
+        lex.add_regex("(i|n|t|e)+")
+        lex.set_input(string("intent"))
+        range(2).for_each(fun(i: int) { println(lex.next().to_string()); } )
+    }
+
     return 0
 }