/// <summary>
/// Gets a string from the tokenized expression.
/// </summary>
/// <param name="iterator">The iterator to the tokenized expression.</param>
/// <param name="services">The shared assembly services.</param>
/// <returns></returns>
public static string GetString(RandomAccessIterator <Token> iterator, AssemblyServices services)
{
if (iterator.Current == null && !iterator.MoveNext())
{
return(string.Empty);
}
var token = iterator.Current;
if (IsStringLiteral(iterator))
{
iterator.MoveNext();
return(Regex.Unescape(token.Name.ToString()).TrimOnce('"'));
}
else if (token.Type == TokenType.Function && token.Name.Equals("format", services.StringComparison))
{
var str = GetFormatted(iterator, services);
if (!string.IsNullOrEmpty(str) && Token.IsEnd(iterator.Current))
{
return(str);
}
}
else if (token.Type == TokenType.Function && token.Name.Equals("char", services.StringComparison))
{
var code = (int)services.Evaluator.Evaluate(iterator, 0, 0x10FFFF);
return(char.ConvertFromUtf32(services.Encoding.GetCodePoint(code)));
}
else if (token.Type == TokenType.Operand &&
(char.IsLetter(token.Name[0]) || token.Name[0] == '_') &&
!services.Evaluator.IsReserved(token.Name))
{
var sym = services.SymbolManager.GetSymbol(token, services.CurrentPass > 0);
if (sym == null)
{
return(string.Empty);
}
if (sym.DataType == DataType.String)
{
if ((!iterator.MoveNext() || Token.IsEnd(iterator.Current)) && sym.StorageType == StorageType.Scalar)
{
return(sym.StringValue.TrimOnce('"').ToString());
}
else if (sym.StorageType == StorageType.Vector && iterator.Current.Name.Equals("["))
{
var current = iterator.Current;
var subscript = (int)services.Evaluator.Evaluate(iterator);
if (Token.IsEnd(iterator.Current))
{
if (subscript >= 0 && subscript < sym.StringVector.Count)
{
return(sym.StringVector[subscript].ToString());
}
throw new SyntaxException(current, "Index out of range.");
}
}
}
}
throw new SyntaxException(token, "Type mismatch.");
}
/// <summary>
/// Gets the formatted string from the tokenized expression.
/// </summary>
/// <param name="iterator">The iterator to the tokenized expression.</param>
/// <param name="services">The shared assembly services.</param>
/// <returns></returns>
public static string GetFormatted(RandomAccessIterator <Token> iterator, AssemblyServices services)
{
iterator.MoveNext();
var format = iterator.GetNext();
if (Token.IsEnd(format))
{
return(null);
}
string fmt;
if (!format.IsDoubleQuote())
{
if (format.Type != TokenType.Function && !format.Name.Equals("format", services.StringComparison))
{
return(null);
}
fmt = GetFormatted(iterator, services);
}
else
{
fmt = Regex.Unescape(format.Name.TrimOnce('"').ToString());
}
var parms = new List <object>();
if (iterator.MoveNext())
{
while (!Token.IsEnd(iterator.GetNext()))
{
if (ExpressionIsAString(iterator, services))
{
parms.Add(GetString(iterator, services));
}
else
{
var parmVal = services.Evaluator.Evaluate(iterator, false);
if (Regex.IsMatch(fmt, $"\\{{{parms.Count}(,-?\\d+)?:(d|D|x|X)\\d*\\}}"))
{
parms.Add((int)parmVal);
}
else
{
parms.Add(parmVal);
}
}
}
}
if (parms.Count == 0)
{
return(fmt);
}
return(string.Format(fmt, parms.ToArray()));
}
void ScanBlock(RandomAccessIterator <SourceLine> lines)
{
var ix = lines.Index;
var line = lines.Current;
var closures = new Stack <Token>();
closures.Push(line.Instruction);
while (lines.MoveNext() && closures.Count > 0)
{
if (lines.Current.Instruction != null)
{
if (lines.Current.Instruction.Name.Equals(_openClosures[closures.Peek().Name], Services.StringViewComparer))
{
closures.Pop();
}
else if (_openClosures.ContainsKey(lines.Current.Instruction.Name))
{
closures.Push(lines.Current.Instruction);
}
}
}
if (closures.Count > 0)
{
throw new SyntaxException(closures.Peek(),
$"Missing closure \"{_openClosures[closures.Peek().Name]}\" for directive \"{closures.Peek().Name}\".");
}
lines.SetIndex(ix);
}
/// <summary>
/// Determines whether the tokenized expression is a string.
/// </summary>
/// <param name="iterator">The iterator to the tokenized expression.</param>
/// <param name="services">The shared assembly services.</param>
/// <returns></returns>
public static bool ExpressionIsAString(RandomAccessIterator <Token> iterator, AssemblyServices services)
{
var token = iterator.Current;
if (token.IsDoubleQuote())
{
return(token.Name.Length > 2 && Token.IsEnd(iterator.PeekNext()));
}
var ix = iterator.Index;
var result = false;
if (token.Type == TokenType.Function &&
(token.Name.Equals("format", services.StringComparison) || token.Name.Equals("char", services.StringComparison)))
{
iterator.MoveNext();
var parms = Token.GetGroup(iterator);
var last = iterator.Current;
result = Token.IsEnd(last);
if (token.Name.Equals("char", services.StringComparison))
{
result &= services.Evaluator.Evaluate(parms.GetIterator(), 0, 0x10FFFF).IsInteger();
}
}
else if (token.Type == TokenType.Operand &&
(char.IsLetter(token.Name[0]) || token.Name[0] == '_') &&
!services.Evaluator.IsReserved(token.Name))
{
var sym = services.SymbolManager.GetSymbol(token, false);
if (sym != null)
{
if (iterator.MoveNext() && iterator.Current.Name.Equals("["))
{
var subscript = (int)services.Evaluator.Evaluate(iterator);
result = Token.IsEnd(iterator.Current) &&
subscript >= 0 && subscript < sym.StringVector.Count;
}
else
{
result = Token.IsEnd(iterator.Current) &&
sym.StorageType == StorageType.Scalar &&
sym.DataType == DataType.String;
}
}
}
iterator.SetIndex(ix);
return(result);
}
public double EvaluateFunction(RandomAccessIterator <Token> tokens)
{
tokens.MoveNext();
var param = tokens.GetNext();
if (param.Equals(")"))
{
throw new SyntaxException(param.Position, "Expected argument not provided.");
}
var symbolLookup = Services.SymbolManager.GetSymbol(param, false);
if (symbolLookup == null)
{
if (param.Type != TokenType.Operand || !char.IsLetter(param.Name[0]) || param.Name[0] != '_')
{
throw new SyntaxException(param.Position, "Function \"len\" expects a symbol.");
}
if (Services.CurrentPass > 0)
{
throw new SymbolException(param, SymbolException.ExceptionReason.NotDefined);
}
Services.PassNeeded = true;
return(0);
}
param = tokens.GetNext();
if (!param.Name.Equals(")"))
{
param = tokens.GetNext();
int subscript = -1;
if (param.Name.Equals("["))
{
subscript = (int)Services.Evaluator.Evaluate(tokens, 0, int.MaxValue);
}
if (subscript < 0 || !tokens.PeekNext().Equals(")"))
{
throw new SyntaxException(param.Position, "Unexpected argument.");
}
if (symbolLookup.StorageType != StorageType.Vector)
{
throw new SyntaxException(param.Position, "Type mismatch.");
}
if (symbolLookup.DataType == DataType.String)
{
if (subscript >= symbolLookup.StringVector.Count)
{
throw new SyntaxException(param.Position, "Index out of range.");
}
return(symbolLookup.StringVector[subscript].Length);
}
if (subscript >= symbolLookup.NumericVector.Count)
{
throw new SyntaxException(param.Position, "Index out of range.");
}
return(symbolLookup.NumericVector[subscript].Size());
}
return(symbolLookup.Length);
}
double CallFunction(RandomAccessIterator <Token> tokens, bool returnValueExpected)
{
var functionToken = tokens.Current;
var functionName = functionToken.Name;
tokens.MoveNext();
var evalParms = new List <object>();
Token token = tokens.GetNext();
while (!token.Name.Equals(")"))
{
if (token.IsSeparator())
{
tokens.MoveNext();
}
if (StringHelper.ExpressionIsAString(tokens, Services))
{
evalParms.Add(StringHelper.GetString(tokens, Services));
}
else
{
evalParms.Add(Services.Evaluator.Evaluate(tokens, false));
}
token = tokens.Current;
}
Services.SymbolManager.PushScopeEphemeral();
var value = _functionDefs[functionName].Invoke(evalParms);
Services.SymbolManager.PopScopeEphemeral();
if (double.IsNaN(value) && returnValueExpected)
{
throw new ReturnException(functionToken.Position,
$"Function name \"{functionName}\" did not return a value.");
}
return(value);
}
/// <summary>
/// Gets a grouping of tokens.
/// </summary>
/// <param name="tokens">The iterator to the full token expression.</param>
/// <returns>The grouped tokens</returns>
public static IEnumerable <Token> GetGroup(RandomAccessIterator <Token> tokens)
{
var list = new List <Token> {
tokens.Current
};
var open = tokens.Current.Name;
var closed = OpenClose[open];
var opens = 1;
while (tokens.MoveNext() && opens > 0)
{
list.Add(tokens.Current);
if (tokens.Current.Name.Equals(open))
{
opens++;
}
else if (tokens.Current.Name.Equals(closed))
{
opens--;
}
}
return(list);
}
/// <summary>
/// Parses the source string into a tokenized <see cref="SourceLine"/> collection.
/// </summary>
/// <param name="fileName">The source file's path/name.</param>
/// <param name="source">The source string.</param>
/// <returns>A collection of <see cref="SourceLine"/>s whose components are
/// properly tokenized for further evaluation and assembly.</returns>
/// <exception cref="ExpressionException"/>
public static IEnumerable <SourceLine> Parse(string fileName, string source)
{
var iterator = new RandomAccessIterator <char>(source.ToCharArray());
Token rootParent, currentParent;
Token token = null;
Reset();
Token currentOpen = null;
int currentLine = 1, lineNumber = currentLine;
// lineIndex is the iterator index at the start of each line for purposes of calculating token
// positions. sourceLindeIndex is the iterator index at the start of each new line
// of source. Usually lineIndex and sourceLindeIndex are the same, but for those source lines
// whose source code span multiple lines, they will be different.
int lineIndex = -1, opens = 0, sourceLineIndex = lineIndex;
var lines = new List <SourceLine>();
char previousChar = iterator.Current;
while (iterator.GetNext() != EOF)
{
if (iterator.Current != NewLine && iterator.Current != ':' && iterator.Current != ';')
{
try
{
token = ParseToken(previousChar, token, iterator);
if (token != null)
{
previousChar = iterator.Current;
token.Parent = currentParent;
token.Position = iterator.Index - lineIndex - token.Name.Length + 1;
if (token.OperatorType == OperatorType.Open || token.OperatorType == OperatorType.Closed || token.OperatorType == OperatorType.Separator)
{
if (token.OperatorType == OperatorType.Open)
{
opens++;
currentParent.AddChild(token);
currentOpen =
currentParent = token;
AddBlankSeparator();
}
else if (token.OperatorType == OperatorType.Closed)
{
if (currentOpen == null)
{
throw new ExpressionException(token, $"Missing opening for closure \"{token.Name}\"");
}
// check if matching ( to )
if (!Groups[currentOpen.Name].Equals(token.Name))
{
throw new ExpressionException(token, $"Mismatch between \"{currentOpen.Name}\" in column {currentOpen.Position} and \"{token.Name}\"");
}
// go up the ladder
currentOpen = currentParent = token.Parent = currentOpen.Parent;
while (currentOpen != null && currentOpen.OperatorType != OperatorType.Open)
{
currentOpen = currentOpen.Parent;
}
opens--;
}
else
{
currentParent = currentParent.Parent;
currentParent.AddChild(token);
currentParent = token;
}
}
else if (token.Type == TokenType.Instruction)
{
while (currentParent.Parent != rootParent)
{
currentParent = currentParent.Parent;
}
currentParent.AddChild(token);
AddBlankSeparator();
AddBlankSeparator();
}
else
{
currentParent.AddChild(token);
}
}
}
catch (ExpressionException ex)
{
Assembler.Log.LogEntry(fileName, lineNumber, ex.Position, ex.Message);
}
if (iterator.PeekNext() == NewLine)
{
iterator.MoveNext();
}
}
if (iterator.Current == ';')
{
_ = iterator.Skip(c => c != NewLine && (c != ':' || Assembler.Options.IgnoreColons) && c != EOF);
}
if (iterator.Current == NewLine || iterator.Current == ':' || iterator.Current == EOF)
{
previousChar = iterator.Current;
/* A new source line is when:
* 1. A line termination character (New Line, colon, EOF) is encountered
* 2. And either there are no more characters left or the most recent token created
* 3. Is not a binary operator nor it is a comma separator.
*/
var newLine = iterator.Current == EOF ||
(opens == 0 &&
(token == null ||
(token.OperatorType != OperatorType.Binary &&
token.OperatorType != OperatorType.Open &&
!token.Name.Equals(",")
)
)
);
if (iterator.Current == NewLine)
{
currentLine++;
}
if (newLine)
{
var newSourceLine = new SourceLine(fileName, lineNumber, GetSourceLineSource(), rootParent.Children[0]);
lines.Add(newSourceLine);
if (Assembler.Options.WarnLeft && newSourceLine.Label != null && newSourceLine.Label.Position != 1)
{
Assembler.Log.LogEntry(newSourceLine, newSourceLine.Label, "Label is not at the beginning of the line.", false);
}
Reset();
lineNumber = currentLine;
}
else
{
token = null;
}
lineIndex = iterator.Index;
if (newLine)
{
sourceLineIndex = iterator.Index;
}
}
}
if (currentOpen != null && currentOpen.OperatorType == OperatorType.Open)
{
Assembler.Log.LogEntry(fileName, 1, currentOpen.LastChild.Position, $"End of source reached without finding closing \"{Groups[currentOpen.Name]}\".");
}
if (token != null)
{
lines.Add(new SourceLine(fileName, lineNumber, GetSourceLineSource(), rootParent.Children[0]));
}
return(lines);
void AddBlankSeparator()
{
var sepToken = new Token()
{
Type = TokenType.Operator,
OperatorType = OperatorType.Separator,
Name = string.Empty,
Position = token == null ? 1 : token.Position,
Children = new List <Token>()
};
currentParent.AddChild(sepToken);
currentParent = sepToken;
}
string GetSourceLineSource()
{
if (iterator.Index > sourceLineIndex + 1)
{
return(source.Substring(sourceLineIndex + 1, iterator.Index - sourceLineIndex - 1));
}
return(string.Empty);
}
void Reset()
{
currentParent =
rootParent = new Token();
currentParent.Children = new List <Token>();
AddBlankSeparator();
AddBlankSeparator();
token = null;
}
}
static Token ParseToken(char previousChar, Token previousToken, RandomAccessIterator <char> iterator, bool parsingAssembly = true)
{
char c = iterator.Current;
while (char.IsWhiteSpace(c))
{
if (c == NewLine && parsingAssembly)
{
iterator.Rewind(iterator.Index - 1);
return(null);
}
c = iterator.GetNext();
}
if ((c == ';' && parsingAssembly) || c == EOF)
{
return(null);
}
var token = new Token();
//first case, simplest
var nextChar = iterator.PeekNext();
if (char.IsDigit(c) ||
char.IsLetter(c) ||
c == '_' ||
c == '?' ||
(c == '.' && char.IsLetterOrDigit(nextChar)) ||
(c == '\\' && char.IsLetterOrDigit(nextChar)))
{
token.Type = TokenType.Operand;
if (char.IsDigit(c) || (c == '.' && char.IsDigit(nextChar)))
{
if (char.IsDigit(c) && previousChar == '$')
{
token.Name = ScanTo(previousChar, iterator, FirstNonHex);
}
else if (c == '0' && (nextChar == 'b' ||
nextChar == 'B' ||
nextChar == 'o' ||
nextChar == 'O' ||
nextChar == 'x' ||
nextChar == 'X'))
{
token.Name = ScanTo(previousChar, iterator, FirstNonNonBase10);
}
else
{
token.Name = ScanTo(previousChar, iterator, FirstNonNumeric);
}
}
else if (c == '\\')
{
iterator.MoveNext();
token.Name = c + ScanTo(previousChar, iterator, FirstNonLetterOrDigit);
}
else if (c == '?')
{
token.UnparsedName =
token.Name = "?";
return(token);
}
else
{
token.UnparsedName =
token.Name = ScanTo(previousChar, iterator, FirstNonSymbol);
if (parsingAssembly && !Assembler.Options.CaseSensitive)
{
token.Name = token.Name.ToLower();
}
if (parsingAssembly && Assembler.InstructionLookupRules.Any(rule => rule(token.Name)))
{
token.Type = TokenType.Instruction;
}
else if (iterator.Current == '(' ||
(iterator.Current != NewLine && char.IsWhiteSpace(iterator.Current) &&
iterator.PeekNextSkipping(NonNewLineWhiteSpace) == '('))
{
token.Type = TokenType.Operator;
token.OperatorType = OperatorType.Function;
}
else
{
token.Type = TokenType.Operand;
}
}
}
else if (previousToken != null &&
previousToken.Name.Equals("%") &&
previousToken.OperatorType == OperatorType.Unary &&
(c == '.' || c == '#'))
{
// alternative binary string parsing
token.Type = TokenType.Operand;
token.Name = ScanTo(previousChar, iterator, FirstNonAltBin).Replace('.', '0')
.Replace('#', '1');
}
else if (c == '"' || c == SingleQuote)
{
var open = c;
var quoteBuilder = new StringBuilder(c.ToString());
var escaped = false;
while ((c = iterator.GetNext()) != open && c != char.MinValue)
{
quoteBuilder.Append(c);
if (c == '\\')
{
escaped = true;
quoteBuilder.Append(iterator.GetNext());
}
}
if (c == char.MinValue)
{
throw new ExpressionException(iterator.Index, $"Quote string not enclosed.");
}
quoteBuilder.Append(c);
var unescaped = escaped ? Regex.Unescape(quoteBuilder.ToString()) : quoteBuilder.ToString();
if (c == '\'' && unescaped.Length > 3)
{
throw new ExpressionException(iterator.Index, "Too many characters in character literal.");
}
token.Name = unescaped;
token.Type = TokenType.Operand;
}
else
{
if (c == '+' || c == '-')
{
/*
* Scenarios for parsing '+' or '-', since they can function as different things
* in an expression.
* 1. The binary operator:
* a. OPERAND+3 / ...)+(... => single '+' sandwiched between two operands/groupings
* b. OPERAND++3 / ...)++(... => the first '+' is a binary operator since it is to the
* right of an operand/grouping. We need to split off the single '++' to two
* separate '+' tokens. What kind of token is the second '+'? We worry about that later.
* c. OPERAND+++3 / ...)+++(... => again, the first '+' is a binary operator. We need to split
* it off from the rest of the string of '+' characters, and we worry about later.
* 2. The unary operator:
* a. +3 / +(... => single '+' immediately preceding an operand/grouping.
* b. ++3 / ++(... => parser doesn't accept C-style prefix (or postfix) operators, so one of these is an
* anonymous label. Which one? Easy, the first. Split the '+' string.
* 3. A full expression mixing both:
* a. OPERAND+++3 / ...)+++(... => From scenario 1.c, we know the first '+' is a binary operator,
* which leaves us with => '++3' left, which from scenario 2.b. we know the first '+'
* has to be an operand. So we split the string again, so that the next scan leaves us with
* '+3', so the third and final plus is a unary operator.
* OPERAND => operand
* + => binary operator
* + => operand
* + => unary operator
* 3/( => operand/grouping
* 4. A line reference:
* a. + => Simplest scenario.
* b. ++, +++, ++++, etc. => Treat as one.
*/
// Get the full string
token.Name = ScanTo(previousChar, iterator, FirstNonPlusMinus);
if (previousToken != null && (previousToken.Type == TokenType.Operand || previousToken.Name.Equals(")")))
{
// looking backward at the previous token, if it's an operand or grouping then we
// know this is a binary
token.Type = TokenType.Operator;
token.OperatorType = OperatorType.Binary;
if (token.Name.Length > 1) // we need to split off the rest of the string so we have a single char '+'
{
token.Name = c.ToString();
iterator.Rewind(iterator.Index - token.Position - 1);
}
}
else if (!IsNotOperand(nextChar) || nextChar == '(')
{
// looking at the very next character in the input stream, if it's an operand or grouping
// then we know this is a unary
if (token.Name.Length > 1)
{
// If the string is greater than one character,
// then it's not a unary, it's an operand AND a unary. So we split off the
// rest of the string.
token.Name = c.ToString();
iterator.Rewind(iterator.Index - token.Position - 1);
token.Type = TokenType.Operand;
}
else
{
token.Type = TokenType.Operator;
token.OperatorType = OperatorType.Unary;
}
}
else
{
token.Type = TokenType.Operand;
}
}
else if (c == '*')
{
// Same as +/- scenario above, if the previous token is an operand or grouping,
// we need to treat the splat as a binary operator.
if (previousToken != null && (previousToken.Type == TokenType.Operand || previousToken.Name.Equals(")")))
{
token.Type = TokenType.Operator;
token.OperatorType = OperatorType.Binary;
}
else
{
// but since there is no unary version of this we will treat as an operand, and let the evaluator
// deal with any problems like *OPERAND /*(
token.Type = TokenType.Operand;
}
token.Name = c.ToString();
}
else
{
// not a number, symbol, string, or special (+, -, *) character. So we just treat as an operator
token.Type = TokenType.Operator;
if (c.IsSeparator() || c.IsOpenOperator() || c.IsClosedOperator())
{
token.Name = c.ToString();
if (c.IsSeparator())
{
token.OperatorType = OperatorType.Separator;
}
else if (c.IsOpenOperator())
{
token.OperatorType = OperatorType.Open;
}
else
{
token.OperatorType = OperatorType.Closed;
}
}
else
{
token.Name = ScanTo(previousChar, iterator, FirstNonMatchingOperator);
token.UnparsedName = token.Name;
/* The general strategy to determine whether an operator is unary or binary:
* 1. Is it actually one of the defined unary types?
* 2. Peek at the next character. Is it a group or operand, or not?
* 3. Look behind at the previous token. Is it also a group or operand, or not?
* 4. If the token does NOT follow an operand or group, AND it precedes a group character,
* or operand character, then it is a unary.
* 5. All other cases, binary.
*
*/
if (
(
(
c.IsUnaryOperator() &&
(
!IsNotOperand(nextChar) ||
nextChar == '(' ||
nextChar.IsRadixOperator() ||
nextChar.IsUnaryOperator()
)
) ||
(
c.IsRadixOperator() && char.IsLetterOrDigit(nextChar)
) ||
(
c == '%' && (nextChar == '.' || nextChar == '#')
)
) &&
(previousToken == null ||
(previousToken.Type != TokenType.Operand &&
!previousToken.Name.Equals(")")
)
)
)
{
token.OperatorType = OperatorType.Unary;
}
else
{
token.OperatorType = OperatorType.Binary;
}
}
}
}
if (string.IsNullOrEmpty(token.UnparsedName))
{
token.UnparsedName = token.Name;
}
if (iterator.Current != token.Name[^ 1])
{
iterator.Rewind(iterator.Index - 1);
}
return(token);
}