Lex various number literals differently. Fix a few bugs

README.md

@@ -20,8 +20,10 @@ well as any paths specified in /etc/dmd.conf.
 code file.
 
 # Dot Completion
+This is currently under development.
 
 # Paren Completion
+This is currently under development.
 
 # JSON output
 Generates a JSON summary of the input file.

autocomplete.d

@@ -1,4 +1,3 @@
-
 //          Copyright Brian Schott (Sir Alaran) 2012.
 // Distributed under the Boost Software License, Version 1.0.
 //    (See accompanying file LICENSE_1_0.txt or copy at
@@ -28,100 +27,19 @@ immutable string[] versions = ["AIX", "all", "Alpha", "ARM", "BigEndian", "BSD",
 	"Win64", "Windows", "X86", "X86_64"
 ];
 
-/+/**
+/**
  * Returns: indicies into the token array
  */
-Tuple!(size_t, size_t) findEndOfStatement(const Token[] tokens, size_t index, out size_t)
+size_t findEndOfExpression(const Token[] tokens, size_t index)
 {
-
+	return index;
-}+/
-
-string[] callChainBackwards(const Token[] tokens, size_t index)
-{
-	if (index == 0)
-		return [tokens[index].value];
-	string[] callChain;
-	string current;
-	loop: while(true)
-	{
-		switch(tokens[index].type)
-		{
-		case TokenType.tThis:
-		case TokenType.identifier:
-		case TokenType.TYPES_BEGIN: .. case TokenType.TYPES_END:
-			current = tokens[index].value ~ current;
-			callChain = current ~ callChain;
-			current = "";
-			if (index == 0)
-				break loop;
-			else
-				--index;
-			if (tokens[index] == TokenType.not)
-				callChain = callChain[1 .. $];
-			break;
-		case TokenType.rBracket:
-			tokens.skipBrackets(index);
-			current ~= "[]";
-			break;
-		case TokenType.rParen:
-			tokens.skipParens(index);
-			break;
-		case TokenType.not:
-		case TokenType.dot:
-			if (index == 0)
-				break loop;
-			else
-				--index;
-			break;
-		default:
-			break loop;
-		}
-	}
-	return callChain;
 }
 
-
+size_t findBeginningOfExpression(const Token[] tokens, size_t index)
-string[] callChainForwards(const Token[] tokens, size_t index)
 {
-	string[] callChain;
+	return index;
-	while (index < tokens.length)
-	{
-		switch(tokens[index].type)
-		{
-		case TokenType.tNew:
-			++index;
-			break;
-		case TokenType.tThis:
-		case TokenType.identifier:
-		case TokenType.TYPES_BEGIN: .. case TokenType.TYPES_END:
-			callChain ~= tokens[index++].value;
-			break;
-		case TokenType.lParen:
-			tokens.skipParens(index);
-			break;
-		case TokenType.lBracket:
-			tokens.skipBrackets(index);
-			callChain[$ - 1] ~= "[i]";
-			break;
-		case TokenType.not:
-			++index;
-			if (tokens.startsWith(TokenType.lParen))
-				tokens.skipParens(index);
-			else
-				++index;
-			break;
-		default:
-			break;
-		}
-		if (index >= tokens.length || tokens[index] != TokenType.dot)
-			break;
-		else
-			++index;
-	}
-	return callChain;
 }
 
-
 struct AutoComplete
 {
 	this(const (Token)[] tokens, CompletionContext context)
@@ -130,39 +48,42 @@ struct AutoComplete
 		this.context = context;
 	}
 
-	string getTypeOfExpression(string[] chain, const Token[] tokens, size_t cursor)
+	string getTypeOfExpression(const(Token)[] expression, const Token[] tokens, size_t cursor)
 	{
-		if (chain.length == 0)
+		return "void";
-			return "void";
-		auto type = typeOfVariable(chain[0], cursor);
-		if (type == "void")
-			return type;
-		chain = chain[1 .. $];
-		while (chain.length >= 1)
-		{
-			auto typeMap = context.getMembersOfType(type);
-			if (typeMap is null)
-				return "void";
-			auto memberType = typeMap[chain[0]][0];
-			if (memberType is null)
-				return "void";
-			type = memberType;
-			chain = chain[1 .. $];
-		}
-		return type;
 	}
 
 	/**
 	 * This is where the magic happens
 	 */
-	string typeOfVariable(string symbol, size_t cursor)
+	string typeOfVariable(Token symbol, size_t cursor)
 	{
 		// int is of type int, double of type double, and so on
-		if (symbol in typeProperties)
+		if (symbol.value in typeProperties)
-			return symbol;
+			return symbol.value;
 
-		if (context.getMembersOfType(symbol))
+		switch (symbol.type)
-			return symbol;
+		{
+			case TokenType.floatLiteral:
+				return "float";
+			case TokenType.doubleLiteral:
+				return "double";
+			case TokenType.realLiteral:
+				return "real";
+			case TokenType.intLiteral:
+				return "int";
+			case TokenType.unsignedIntLiteral:
+				return "uint";
+			case TokenType.longLiteral:
+				return "long";
+			case TokenType.unsignedLongLiteral:
+				return "ulong";
+			default:
+				break;
+		}
+
+		if (context.getMembersOfType(symbol.value))
+			return symbol.value;
 
 		// Arbitrarily define the depth of the cursor position as zero
 		// iterate backwards through the code to try to find the variable
@@ -183,14 +104,13 @@ struct AutoComplete
 					|| p == TokenType.tConst)
 					&& preceedingTokens[index + 1] == TokenType.assign)
 				{
-					auto chain = callChainForwards(tokens, index + 2);
+					return null;
-					return getTypeOfExpression(chain, tokens, cursor);
 				}
-				if (p == TokenType.identifier
+				else if (p == TokenType.identifier
 					|| (p.type > TokenType.TYPES_BEGIN
 					&& p.type < TokenType.TYPES_END))
 				{
-					return preceedingTokens[index - 1].value;
+					return p.value;
 				}
 			}
 			if (index == 0)
@@ -207,7 +127,7 @@ struct AutoComplete
 			return minCount!("a.bodyStart > b.bodyStart")(structs)[0].name;
 		foreach (s; structs)
 		{
-			auto t = s.getMemberType(symbol);
+			auto t = s.getMemberType(symbol.value);
 			if (t !is null)
 				return t;
 		}
@@ -225,14 +145,16 @@ struct AutoComplete
 
 	string parenComplete(size_t cursor)
 	{
+		stderr.writeln("parenComplete");
 		auto index = assumeSorted(tokens).lowerBound(cursor).length - 2;
 		Token t = tokens[index];
+		stderr.writeln(t);
 		if (t.startIndex + t.value.length + 1 != cursor)
 			return "";
 		switch (tokens[index].type)
 		{
 		case TokenType.tVersion:
-			return to!string(array(join(map!`a ~ "?1"`(versions), " ")));
+			return to!string(join(map!`a ~ "?1"`(versions), " ").array());
 		case TokenType.tIf:
 		case TokenType.tCast:
 		case TokenType.tWhile:
@@ -251,20 +173,7 @@ struct AutoComplete
 		Token t = tokens[index];
 		if (t.startIndex + t.value.length + 1 != cursor)
 			return "";
-		stderr.writeln(t);
+		auto type = typeOfVariable(t, cursor);
-		string[] chain = callChainBackwards(tokens, index);
-		auto type = getTypeOfExpression(chain, tokens, cursor);
-
-		if (type && type in typeProperties)
-		{
-			string r;
-			foreach (i, prop; typeProperties[type])
-				if (i == typeProperties.length)
-					r = r ~ prop;
-				else
-					r = r ~ prop ~ " ";
-			return r;
-		}
 
 		const Tuple!(string, string)[string] typeMap = context.getMembersOfType(type);
 		if (typeMap is null)
@@ -272,7 +181,7 @@ struct AutoComplete
 		auto app = appender!(string[])();
 		foreach (k, t; typeMap)
 			app.put(k ~ t[1]);
-		return to!string(array(join(sort(app.data), " ")));
+		return to!string(array(join(sort!"a.toLower() < b.toLower()"(app.data), " ")));
 	}
 
 	const(Token)[] tokens;

build.sh

@@ -1,2 +1,2 @@
 dmd *.d -release -noboundscheck -O -w -wi -m64 -property -ofdscanner 
-#dmd *.d -g -unittest -m64 -w -wi -property -oftokenizer
+#dmd *.d -g -unittest -m64 -w -wi -property -ofdscanner

highlighter.d

@@ -30,7 +30,7 @@ html { background-color: #111; color: #ccc; }
 .string { color: Tomato; font-style: italic; }
 .property { color: HotPink; font-weight: bold;}
 .operator { color: tan; font-weight: bold; }
-.type { color: cyan; }
+.type { color: cyan; font-weight: bold; }
 </style>
 <pre>]");
 
@@ -50,7 +50,7 @@ html { background-color: #111; color: #ccc; }
 		case TokenType.stringLiteral:
 			writeSpan("string", t.value);
 			break;
-		case TokenType.numberLiteral:
+		case TokenType.NUMBERS_BEGIN: .. case TokenType.NUMBERS_END:
 			writeSpan("number", t.value);
 			break;
 		case TokenType.OPERATORS_BEGIN: .. case TokenType.OPERATORS_END:

langutils.d

@@ -254,11 +254,19 @@ enum TokenType: uint
 // Misc
 	MISC_BEGIN,
 	comment, /// /** comment */ or // comment or ///comment
+	NUMBERS_BEGIN,
+	floatLiteral, /// 123.456f or 0x123_45p-af
+	doubleLiteral, /// 123.456
+	realLiteral, /// 123.456L
+	intLiteral, /// 123 or 0b1101010101
+	unsignedIntLiteral, /// 123u
+	longLiteral, /// 123L
+	unsignedLongLiteral, /// 123uL
+	NUMBERS_END,
 	stringLiteral, /// "a string"
-	numberLiteral, /// int, float, etc...
+	identifier, /// anything else
-	identifier,
 	whitespace, /// whitespace
-	blank,
+	blank, /// unknown token type
 	MISC_END,
 }
 

parser.d

@@ -218,7 +218,6 @@ string parseTypeDeclaration(const Token[] tokens, ref size_t index)
 			break buildingType;
 		}
 	}
-	stderr.writeln("type = ", type);
 	return type;
 }
 

tokenizer.d

@@ -141,11 +141,15 @@ body
 		++endIndex;
 	}
 	++endIndex;
+	if (endIndex < inputString.length && (inputString[endIndex] == 'w'
+		|| inputString[endIndex] == 'd' || inputString[endIndex] == 'c'))
+	{
+		++endIndex;
+	}
 	endIndex = min(endIndex, inputString.length);
 	return inputString[startIndex .. endIndex];
 }
 
-
 /**
  * Lexes the various crazy D string literals such as q{}, q"WTF is this? WTF",
  * and q"<>".
@@ -218,79 +222,230 @@ string lexTokenString(S)(ref S inputString, ref size_t endIndex, ref uint lineNu
 	return "";
 }
 
-/**
+pure nothrow Token lexNumber(S)(ref S inputString, ref size_t endIndex)
- *
+	if (isSomeString!S)
- */
-pure nothrow string lexNumber(S)(ref S inputString, ref size_t endIndex) if (isSomeString!S)
 {
-	auto startIndex = endIndex;
+	Token token;
-	bool foundDot = false;
+	size_t startIndex = endIndex;
-	bool foundX = false;
+	if (inputString[endIndex] == '0')
-	bool foundB = false;
+	{
-	bool foundE = false;
+		endIndex++;
-	numberLoop: while (endIndex < inputString.length)
+		if (endIndex >= inputString.length)
+		{
+			token.type = TokenType.intLiteral;
+			token.value = inputString[startIndex .. endIndex];
+			return token;
+		}
+		switch (inputString[endIndex])
+		{
+		case '0': .. case '9':
+			// The current language spec doesn't cover octal literals, so this
+			// is decimal.
+			lexDecimal(inputString, startIndex, endIndex, token);
+			return token;
+		case 'b':
+		case 'B':
+			lexBinary(inputString, startIndex, ++endIndex, token);
+			return token;
+		case 'x':
+		case 'X':
+			lexHex(inputString, startIndex, ++endIndex, token);
+			return token;
+		default:
+			token.type = TokenType.intLiteral;
+			token.value = inputString[startIndex .. endIndex];
+			return token;
+		}
+	}
+	else
+	{
+		lexDecimal(inputString, startIndex, endIndex, token);
+		return token;
+	}
+}
+
+pure nothrow void lexBinary(S)(ref S inputString, size_t startIndex,
+	ref size_t endIndex, ref Token token) if (isSomeString!S)
+{
+	bool lexingSuffix = false;
+	bool isLong = false;
+	bool isUnsigned = false;
+	token.type = TokenType.intLiteral;
+	binaryLoop: while (endIndex < inputString.length)
 	{
 		switch (inputString[endIndex])
 		{
 		case '0':
-			if (!foundX)
+		case '1':
-			{
+		case '_':
-				++endIndex;
+			++endIndex;
-				if (endIndex < inputString.length
+			if (lexingSuffix)
-					&& (inputString[endIndex] == 'x' || inputString[endIndex] == 'X'))
+				break binaryLoop;
-				{
+			break;
-					++endIndex;
+		case 'u':
-					foundX = true;
+		case 'U':
-				}
+			++endIndex;
-			}
+			lexingSuffix = true;
+			if (isLong)
+				token.type = TokenType.unsignedLongLiteral;
 			else
-				++endIndex;
+				token.type = TokenType.unsignedIntLiteral;
 			break;
-		case 'b':
+		case 'L':
-			if (foundB)
-				break numberLoop;
-			foundB = true;
 			++endIndex;
+			if (isLong)
+				break binaryLoop;
+			if (isUnsigned)
+				token.type = TokenType.unsignedLongLiteral;
+			else
+				token.type = TokenType.longLiteral;
+			isLong = true;
 			break;
-		case '.':
+		default:
-			if (foundDot || foundX || foundE)
+			break binaryLoop;
-				break numberLoop;
+		}
-			foundDot = true;
+	}
+
+	token.value = inputString[startIndex .. endIndex];
+}
+
+pure nothrow void lexDecimal(S)(ref S inputString, size_t startIndex,
+	ref size_t endIndex, ref Token token) if (isSomeString!S)
+{
+	bool lexingSuffix = false;
+	bool isLong = false;
+	bool isUnsigned = false;
+	bool isFloat = false;
+	bool isReal = false;
+	bool isDouble = false;
+	bool foundDot = false;
+	bool foundE = false;
+	bool foundPlusMinus = false;
+	token.type = TokenType.intLiteral;
+	decimalLoop: while (endIndex < inputString.length)
+	{
+		switch (inputString[endIndex])
+		{
+		case '0': .. case '9':
+		case '_':
 			++endIndex;
+			if (lexingSuffix)
+				break decimalLoop;
+			break;
+		case 'e':
+		case 'E':
+			if (foundE)
+				break decimalLoop;
+			++endIndex;
+			foundE = true;
 			break;
 		case '+':
 		case '-':
-			if (!foundE)
+			if (foundPlusMinus || !foundE)
-				break numberLoop;
+				break decimalLoop;
+			foundPlusMinus = true;
 			++endIndex;
 			break;
+		case '.':
+			if (foundDot)
+				break decimalLoop;
+			++endIndex;
+			foundDot = true;
+			token.type = TokenType.doubleLiteral;
+			isDouble = true;
+			break;
+		case 'u':
+		case 'U':
+			++endIndex;
+			lexingSuffix = true;
+			if (isLong)
+				token.type = TokenType.unsignedLongLiteral;
+			else
+				token.type = TokenType.unsignedIntLiteral;
+			isUnsigned = true;
+			break;
+		case 'L':
+			++endIndex;
+			lexingSuffix = true;
+			if (isLong || isReal)
+				break decimalLoop;
+			if (isDouble)
+				token.type = TokenType.realLiteral;
+			else if (isUnsigned)
+				token.type = TokenType.unsignedLongLiteral;
+			else
+				token.type = TokenType.longLiteral;
+			isLong = true;
+			break;
+		case 'f':
+		case 'F':
+			lexingSuffix = true;
+			if (isUnsigned || isLong)
+				break decimalLoop;
+			++endIndex;
+			token.type = TokenType.floatLiteral;
+			break decimalLoop;
+		default:
+			break decimalLoop;
+		}
+	}
+
+	token.value = inputString[startIndex .. endIndex];
+}
+
+nothrow void lexHex(S)(ref S inputString, ref size_t startIndex,
+	ref size_t endIndex, ref Token token) if (isSomeString!S)
+{
+	bool lexingSuffix = false;
+	bool isLong = false;
+	bool isUnsigned = false;
+	bool isFloat = false;
+	bool isReal = false;
+	bool isDouble = false;
+	bool foundDot = false;
+	bool foundE = false;
+	bool foundPlusMinus = false;
+	token.type = TokenType.intLiteral;
+	hexLoop: while (endIndex < inputString.length)
+	{
+		switch (inputString[endIndex])
+		{
+		case '0': .. case '9':
+		case 'a': .. case 'f':
+		case 'A': .. case 'F':
+		case '_':
+			++endIndex;
+			if (lexingSuffix)
+				break hexLoop;
+			break;
 		case 'p':
 		case 'P':
-			if (!foundX)
+			if (foundE)
-				break numberLoop;
+				break hexLoop;
+			++endIndex;
 			foundE = true;
-			goto case '_';
+			break;
-		case 'e':
+		case '+':
-		case 'E':
+		case '-':
-			if (foundE || foundX)
+			if (foundPlusMinus || !foundE)
-				break numberLoop;
+				break hexLoop;
-			foundE = true;
+			foundPlusMinus = true;
-			goto case '_';
-		case '1': .. case '9':
-		case '_':
 			++endIndex;
 			break;
-		case 'F':
+		case '.':
-		case 'f':
+			if (foundDot)
-		case 'L':
+				break hexLoop;
-		case 'i':
 			++endIndex;
-			break numberLoop;
+			foundDot = true;
+			token.type = TokenType.doubleLiteral;
+			isDouble = true;
+			break;
 		default:
-			break numberLoop;
+			break hexLoop;
 		}
 	}
-	return inputString[startIndex .. endIndex];
+
+	token.value = inputString[startIndex .. endIndex];
 }
 
 
@@ -337,8 +492,10 @@ Token[] tokenize(S)(S inputString, IterationStyle iterationStyle = IterationStyl
 
 	size_t endIndex = 0;
 	uint lineNumber = 1;
+
 	while (endIndex < inputString.length)
 	{
+		size_t prevIndex = endIndex;
 		Token currentToken;
 		auto startIndex = endIndex;
 		if (isWhite(inputString[endIndex]))
@@ -421,11 +578,8 @@ Token[] tokenize(S)(S inputString, IterationStyle iterationStyle = IterationStyl
 			"^",    "TokenType.xor",
 			"^=",   "TokenType.xorEquals",
 		));
-
 		case '0': .. case '9':
-			currentToken.value = lexNumber(inputString, endIndex);
+			currentToken = lexNumber(inputString, endIndex);
-			currentToken.type = TokenType.numberLiteral;
-			currentToken.lineNumber = lineNumber;
 			break;
 		case '/':
 			++endIndex;
@@ -528,8 +682,15 @@ Token[] tokenize(S)(S inputString, IterationStyle iterationStyle = IterationStyl
 			currentToken.lineNumber = lineNumber;
 			break;
 		}
-//		writeln(currentToken);
+		//stderr.writeln(currentToken);
 		tokenAppender.put(currentToken);
+
+		// This should never happen.
+		if (endIndex <= prevIndex)
+		{
+			stderr.writeln("FAIL");
+			return [];
+		}
 	}
 	return tokenAppender.data;
 }