// Read a MalSequence, checking that it starts and terminates correctly.
// Named read_list to follow the ref, but has been genericized to handle vectors as well.
static public MalSeqBase read_list(Reader reader, MalSeqBase sequence, char start, char end)
{
// Check that we are in fact at the start of a list.
string token = reader.Next();
if (token[0] != start)
{
// Parse error - probably internal if the list code is correct.
throw new MalInternalError("Sequence expected '" + start + "' but got: " + token);
}
// Use read_form to get the list's contents, accumulating them into the list.
while (true)
{
token = reader.Peek();
if (token != null)
{
// We are in the list or at the end.
if (token[0] == end)
{
// Reached valid end of list. Consume the end char.
reader.Next();
// And we are done.
break;
}
// Mutually recurse to read the next list element.
MalVal newVal = read_form(reader);
sequence.Add(newVal);
}
else
{
// The input has finished but the list hasn't. Try to get more input.
reader.LoadMoreTokens(start, end);
}
}
return(sequence);
}
// Read a MalVal form - which is either an atom or a sequence.
static public MalVal read_form(Reader reader)
{
if (reader.Peek() == null)
{
// Reader is empty - caused by a comment line in the input.
return(null);
}
else if (reader.Peek().StartsWith('\''))
{
// Create a list containing the quote symbol and the quoted form.
// Skip the quote symbol.
reader.Next();
// Now read the quoted thing, and build a quote form.
return(new MalList(new MalQuote(), read_form(reader)));
// TODO handle quasiquotes and splices in the same way.
}
else if (reader.Peek().StartsWith('('))
{
// Create a new List and read it's body.
return(read_list(reader, new MalList(), '(', ')'));
}
else if (reader.Peek().StartsWith('['))
{
// Create a new Vector and read it's body.
return(read_list(reader, new MalVector(), '[', ']'));
}
else if (reader.Peek().StartsWith('{'))
{
// Create a new HashMap and read it's body.
// TODO - check that hashmap contains a list of keywords and values.
return(read_list(reader, new MalHashMap(), '{', '}'));
}
else if (reader.Peek().StartsWith(')') || reader.Peek().StartsWith(']') || reader.Peek().StartsWith('}'))
{
// A sequence close character that doesn't match a start.
// This correctly handles a case like [1 ( 2 ] 3).
throw new MalParseError("Expecting sequence or atom but got '" + reader.Peek() + "'");
}
else
{
// This isn't a list try so parse it as an atom.
return(read_atom(reader));
}
}
static public MalVal read_token(Reader reader)
{
// "If you have a problem and you think regex is the answer, now you have two problems!"
// Unlike the referenceC#-Mal, read_token handles floats and badly-formed symbols.
// In the Mal Guide this is called read_atom but I renamed it to avoid confusion
// with Mal Atoms.
string tokenToRead = reader.Next();
if (tokenToRead.Length <= 0)
{
throw new MalInternalError("Reader has returned empty string");
}
switch (tokenToRead[0])
{
case '+':
if (tokenToRead.Length == 1)
{
// Token is a solo '+', not the beginning of a number.
return(new MalSym(tokenToRead));
}
// Skip the sign and extract a positive number;
return(ParseNumber(tokenToRead.Substring(1), true));
case '-':
if (tokenToRead.Length == 1)
{
// Token is a solo '-', not the beginning of a number.
return(new MalSym(tokenToRead));
}
// Skip the sign and extract a negative number;
return(ParseNumber(tokenToRead.Substring(1), false));
case '.':
// An initial '.' is only allowed at the start of a number, as in '.2'.
return(ParseNumber(tokenToRead, true));
case '\"':
if (tokenToRead.EndsWith("\""))
{
// Get rid of the quotes before storing the string. Seems right although
// I haven't confirmed by checking the reference version.
char[] charsToTrim = { '"' };
tokenToRead = tokenToRead.Trim(charsToTrim);
return(new MalString(tokenToRead));
}
// TODO - never reaches this point. The reader regex seems to throw away '"' chars if there is only 1.
throw new MalParseError("String '" + tokenToRead + "' lacks a closing thingy");
case ':':
// Handle a keyword
if (tokenToRead.Length == 1)
{
// Can't have a solo colon.
throw new MalParseError("':' must be followed by a keyword");
}
return(new MalKeyword(tokenToRead));
default:
if (IsDigit(tokenToRead[0].ToString()))
{
// Token seems to be an unsigned number.
return(ParseNumber(tokenToRead, true));
}
else if (tokenToRead == "nil")
{
return(malNil);
}
else if (tokenToRead == "true")
{
return(malTrue);
}
else if (tokenToRead == "false")
{
return(malFalse);
}
else
{
// If here it is 'just' a symbol.
return(new MalSym(tokenToRead));
}
}
throw new MalInternalError("Can't process '" + tokenToRead + "'");
}
// Read a MalVal form - which is either an atom or a sequence.
static public MalVal read_form(Reader reader)
{
if (reader.Peek() == null)
{
// Reader is empty - caused by a comment line in the input.
return(null);
}
else if (reader.Peek().StartsWith('('))
{
// Create a new List and read it's body.
return(read_list(reader, new MalList(), '(', ')'));
}
else if (reader.Peek().StartsWith('['))
{
// Create a new Vector and read it's body.
return(read_list(reader, new MalVector(), '[', ']'));
}
else if (reader.Peek().StartsWith('{'))
{
// Create a new HashMap and read it's body. EVAL checks it has valid key val pairs.
return(read_list(reader, new MalHashMap(), '{', '}'));
}
else if (reader.Peek().StartsWith(')') || reader.Peek().StartsWith(']') || reader.Peek().StartsWith('}'))
{
// A sequence close character that doesn't match a start.
// This correctly handles a case like [1 ( 2 ] 3).
throw new MalParseError("Expecting sequence or atom but got '" + reader.Peek() + "'");
}
else if (reader.Peek().StartsWith('&'))
{
// Reader macro. We have '&atomName'. Convert this into (deref atomName);
string atomToDeref = reader.Peek();
if (atomToDeref.Length == 1)
{
// Treat a solo '&' as a varargs symbol,
reader.Next();
return(malVarArgsChar);
}
// Build a deref form, extracting the atom name from the token.
MalList derefForm = new MalList();
derefForm.Add(new MalSym("deref"));
derefForm.Add(new MalSym(atomToDeref.Substring(1)));
// TODO - ??? handle non-explicit atom names? E.g. where '&' is followed by a form that creates an atom name.
// Advance past the deref symbol and return the new form.
reader.Next();
return(derefForm);
}
else if (reader.Peek().StartsWith('\''))
{
// Return a list containing a quote symbol and the quoted form.
reader.Next();
MalList quoteForm = new MalList();
quoteForm.Add(new MalSym("quote"));
quoteForm.Add(read_form(reader));
return(quoteForm);
}
else if (reader.Peek().StartsWith('`'))
{
// Return a list containing a quasiquote symbol and the quasiquoted form.
reader.Next();
MalList quasiquoteForm = new MalList();
quasiquoteForm.Add(new MalSym("quasiquote"));
quasiquoteForm.Add(read_form(reader));
return(quasiquoteForm);
}
else if (reader.Peek().StartsWith("~@"))
{
// Return a list containing a splice-unquote symbol and the next form.
// Dammit! I'd missed the '~' here and spent several days wondering why (or ...) didn't work.
reader.Next();
MalList quasiquoteForm = new MalList();
quasiquoteForm.Add(new MalSym("splice-unquote"));
quasiquoteForm.Add(read_form(reader));
return(quasiquoteForm);
}
else if (reader.Peek().StartsWith('~'))
{
// Return a list containing an unquote symbol and the next form.
reader.Next();
MalList quasiquoteForm = new MalList();
quasiquoteForm.Add(new MalSym("unquote"));
quasiquoteForm.Add(read_form(reader));
return(quasiquoteForm);
}
else
{
// This isn't a list so parse it as an atom.
return(read_token(reader));
}
}
// Read a MalVal form - which is either an atom or a sequence.
static public MalVal read_form(Reader reader)
{
if (reader.Peek() == null)
{
// Reader is empty - caused by a comment line in the input.
return(null);
}
else if (reader.Peek().StartsWith('('))
{
// Create a new List and read it's body.
return(read_list(reader, new MalList(), '(', ')'));
}
else if (reader.Peek().StartsWith('['))
{
// Create a new Vector and read it's body.
return(read_list(reader, new MalVector(), '[', ']'));
}
else if (reader.Peek().StartsWith('{'))
{
// Create a new HashMap and read it's body. EVAL checks it has valid key val pairs.
return(read_list(reader, new MalHashMap(), '{', '}'));
}
else if (reader.Peek().StartsWith(')') || reader.Peek().StartsWith(']') || reader.Peek().StartsWith('}'))
{
// A sequence close character that doesn't match a start.
// This correctly handles a case like [1 ( 2 ] 3).
throw new MalParseError("Expecting sequence or atom but got '" + reader.Peek() + "'");
}
else if (reader.Peek().StartsWith('&'))
{
// Reader macro. We have '&atomName'. Convert this into (deref atomName);
string atomToDeref = reader.Peek();
if (atomToDeref.Length == 1)
{
throw new MalParseError("'&' lacks atom name");
}
// Build a deref form, extracting the atom name from the token.
MalList derefForm = new MalList();
derefForm.Add(new MalSym("deref"));
derefForm.Add(new MalSym(atomToDeref.Substring(1)));
// TODO - ??? handle non-explicit atom names? E.g. where '&' is followed by a form that creates an atom name.
// Advance past the deref symbol and return the new form.
reader.Next();
return(derefForm);
}
else if (reader.Peek().StartsWith('\''))
{
// Create a list containing the quote symbol and the quoted form.
// Skip the quote symbol.
reader.Next();
MalList quoteForm = new MalList();
quoteForm.Add(new MalQuote());
quoteForm.Add(read_form(reader));
// Now read the quoted thing, and build a quote form.
return(quoteForm);
// TODO handle quasiquotes and splices in the same way.
}
else
{
// This isn't a list so parse it as an atom.
return(read_token(reader));
}
}
static public MalVal read_atom(Reader reader)
{
// "If you have a problem and you think regex is the answer, now you have two problems!"
// Unlike the referenceC#-Mal, read_atom handles floats and badly-formed symbols.
string tokenToRead = reader.Next();
if (tokenToRead.Length <= 0)
{
throw new MalParseError("INTERNAL - reader has returned empty string");
}
switch (tokenToRead[0])
{
case '+':
if (tokenToRead.Length == 1)
{
// Token is a solo '+', not the beginning of a number.
return(new MalSym(tokenToRead));
}
// Skip the sign and extract a positive number;
return(ParseNumber(tokenToRead.Substring(1), true));
case '-':
if (tokenToRead.Length == 1)
{
// Token is a solo '-', not the beginning of a number.
return(new MalSym(tokenToRead));
}
// Skip the sign and extract a negative number;
return(ParseNumber(tokenToRead.Substring(1), false));
case '.':
// An initial '.' is only allowed at the start of a number, as in '.2'.
return(ParseNumber(tokenToRead, true));
case '\"':
if (tokenToRead.EndsWith("\""))
{
return(new MalString(tokenToRead));
}
// TODO - never reaches this point. The reader regex seems to throw away '"' chars if there is only 1.
throw new MalParseError("String '" + tokenToRead + "' lacks a closing thingy");
case ':':
// Handle a keyword
if (tokenToRead.Length == 1)
{
// Can't have a solo colon.
throw new MalParseError("':' must be followed by a keyword");
}
return(new MalKeyword(tokenToRead));
default:
if (IsDigit(tokenToRead[0].ToString()))
{
// Token seems to be an unsigned number.
return(ParseNumber(tokenToRead, true));
}
// If here it is 'just' a symbol.
return(new MalSym(tokenToRead));
}
throw new MalParseError("INTERNAL - can't process '" + tokenToRead + "'");
}
