Ast.ComplexNumberNode ParseComplex(SeekableStringReader sr)
{
//complex = complextuple | imaginary .
//imaginary = ['+' | '-' ] ( float | int ) 'j' .
//complextuple = '(' ( float | int ) imaginary ')' .
if (sr.Peek() == '(')
{
// complextuple
sr.Read(); // (
string numberstr;
if (sr.Peek() == '-' || sr.Peek() == '+')
{
// starts with a sign, read that first otherwise the readuntil will return immediately
numberstr = sr.Read(1) + sr.ReadUntil(new char[] { '+', '-' });
}
else
{
numberstr = sr.ReadUntil(new char[] { '+', '-' });
}
double realpart;
try {
realpart = double.Parse(numberstr, CultureInfo.InvariantCulture);
} catch (FormatException x) {
throw new ParseException("invalid float format", x);
}
sr.Rewind(1); // rewind the +/-
double imaginarypart = ParseImaginaryPart(sr);
if (sr.Read() != ')')
{
throw new ParseException("expected ) to end a complex number");
}
return(new Ast.ComplexNumberNode()
{
Real = realpart,
Imaginary = imaginarypart
});
}
else
{
// imaginary
double imag = ParseImaginaryPart(sr);
return(new Ast.ComplexNumberNode()
{
Real = 0,
Imaginary = imag
});
}
}
Ast.TupleNode ParseTuple(SeekableStringReader sr)
{
//tuple = tuple_empty | tuple_one | tuple_more
//tuple_empty = '()' .
//tuple_one = '(' expr ',' <whitespace> ')' .
//tuple_more = '(' expr_list ')' .
sr.Read(); // (
sr.SkipWhitespace();
Ast.TupleNode tuple = new Ast.TupleNode();
if (sr.Peek() == ')')
{
sr.Read();
return(tuple); // empty tuple
}
Ast.INode firstelement = ParseExpr(sr);
if (sr.Peek() == ',')
{
sr.Read();
sr.SkipWhitespace();
if (sr.Read() == ')')
{
// tuple with just a single element
tuple.Elements.Add(firstelement);
return(tuple);
}
sr.Rewind(1); // undo the thing that wasn't a )
}
tuple.Elements = ParseExprList(sr);
tuple.Elements.Insert(0, firstelement);
if (!sr.HasMore())
{
throw new ParseException("missing ')'");
}
char closechar = sr.Read();
if (closechar == ',')
{
closechar = sr.Read();
}
if (closechar != ')')
{
throw new ParseException("expected ')'");
}
return(tuple);
}
public Ast.INode ParseSingle(SeekableStringReader sr)
{
// single = int | float | complex | string | bool | none .
sr.SkipWhitespace();
switch (sr.Peek())
{
case 'N':
return(ParseNone(sr));
case 'T':
case 'F':
return(ParseBool(sr));
case '\'':
case '"':
return(ParseString(sr));
}
// int or float or complex.
int bookmark = sr.Bookmark();
try {
return(ParseComplex(sr));
} catch (ParseException) {
sr.FlipBack(bookmark);
try {
return(ParseFloat(sr));
} catch (ParseException) {
sr.FlipBack(bookmark);
return(ParseInt(sr));
}
}
}
Ast.ListNode ParseList(SeekableStringReader sr)
{
// list = list_empty | list_nonempty .
// list_empty = '[]' .
// list_nonempty = '[' expr_list ']' .
sr.Read(); // [
sr.SkipWhitespace();
Ast.ListNode list = new Ast.ListNode();
if (sr.Peek() == ']')
{
sr.Read();
return(list); // empty list
}
list.Elements = ParseExprList(sr);
if (!sr.HasMore())
{
throw new ParseException("missing ']'");
}
char closechar = sr.Read();
if (closechar != ']')
{
throw new ParseException("expected ']'");
}
return(list);
}
/// <summary>
/// Parse from a string with the Python literal expression
/// </summary>
public Ast Parse(string expression)
{
Ast ast = new Ast();
if (string.IsNullOrEmpty(expression))
{
return(ast);
}
SeekableStringReader sr = new SeekableStringReader(expression);
if (sr.Peek() == '#')
{
sr.ReadUntil('\n'); // skip comment line
}
try {
ast.Root = ParseExpr(sr);
sr.SkipWhitespace();
if (sr.HasMore())
{
throw new ParseException("garbage at end of expression");
}
return(ast);
} catch (ParseException x) {
string faultLocation = ExtractFaultLocation(sr);
throw new ParseException(x.Message + " (at position " + sr.Bookmark() + "; '" + faultLocation + "')", x);
}
}
private Ast.INode ParseCompound(SeekableStringReader sr)
{
// compound = tuple | dict | list | set .
sr.SkipWhitespace();
switch (sr.Peek())
{
case '[':
return(ParseList(sr));
case '{':
{
int bm = sr.Bookmark();
try {
return(ParseSet(sr));
} catch (ParseException) {
sr.FlipBack(bm);
return(ParseDict(sr));
}
}
case '(':
// tricky case here, it can be a tuple but also a complex number:
// if the last character before the closing parenthesis is a 'j', it is a complex number
{
int bm = sr.Bookmark();
string betweenparens = sr.ReadUntil(')', '\n').TrimEnd();
sr.FlipBack(bm);
return(betweenparens.EndsWith("j") ? (Ast.INode)ParseComplex(sr) : ParseTuple(sr));
}
default:
throw new ParseException("invalid sequencetype char");
}
}
List <Ast.INode> ParseKeyValueList(SeekableStringReader sr)
{
//keyvalue_list = keyvalue { ',' keyvalue } .
List <Ast.INode> kvs = new List <Ast.INode>();
kvs.Add(ParseKeyValue(sr));
while (sr.HasMore() && sr.Peek() == ',')
{
sr.Read();
kvs.Add(ParseKeyValue(sr));
}
return(kvs);
}
List <Ast.INode> ParseExprList(SeekableStringReader sr)
{
//expr_list = expr { ',' expr } .
List <Ast.INode> exprList = new List <Ast.INode>();
exprList.Add(ParseExpr(sr));
while (sr.HasMore() && sr.Peek() == ',')
{
sr.Read();
exprList.Add(ParseExpr(sr));
}
return(exprList);
}
private Ast.KeyValueNode ParseKeyValue(SeekableStringReader sr)
{
//keyvalue = expr ':' expr .
Ast.INode key = ParseExpr(sr);
if (!sr.HasMore() || sr.Peek() != ':')
{
throw new ParseException("expected ':'");
}
sr.Read(); // :
Ast.INode value = ParseExpr(sr);
return(new Ast.KeyValueNode
{
Key = key,
Value = value
});
}
Ast.DictNode ParseDict(SeekableStringReader sr)
{
//dict = '{' keyvalue_list '}' .
//keyvalue_list = keyvalue { ',' keyvalue } .
//keyvalue = expr ':' expr .
sr.Read(); // {
sr.SkipWhitespace();
Ast.DictNode dict = new Ast.DictNode();
if (sr.Peek() == '}')
{
sr.Read();
return(dict); // empty dict
}
List <Ast.INode> elts = ParseKeyValueList(sr);
if (!sr.HasMore())
{
throw new ParseException("missing '}'");
}
char closechar = sr.Read();
if (closechar != '}')
{
throw new ParseException("expected '}'");
}
// make sure it has dict semantics (remove duplicate keys)
Dictionary <Ast.INode, Ast.INode> fixedDict = new Dictionary <Ast.INode, Ast.INode>(elts.Count);
foreach (Ast.KeyValueNode kv in elts)
{
fixedDict[kv.Key] = kv.Value;
}
foreach (var kv in fixedDict)
{
dict.Elements.Add(new Ast.KeyValueNode()
{
Key = kv.Key,
Value = kv.Value
});
}
return(dict);
}
List <Ast.INode> ParseKeyValueList(SeekableStringReader sr)
{
//keyvalue_list = keyvalue { ',' keyvalue } .
List <Ast.INode> kvs = new List <Ast.INode>();
kvs.Add(ParseKeyValue(sr));
while (sr.HasMore() && sr.Peek() == ',')
{
sr.Read();
try {
kvs.Add(ParseKeyValue(sr));
} catch (ParseException) {
sr.Rewind(1);
break;
}
}
return(kvs);
}
Ast.INode ParseExpr(SeekableStringReader sr)
{
// expr = [ <whitespace> ] single | compound [ <whitespace> ] .
sr.SkipWhitespace();
char c = sr.Peek();
Ast.INode node;
if (c == '{' || c == '[' || c == '(')
{
node = ParseCompound(sr);
}
else
{
node = ParseSingle(sr);
}
sr.SkipWhitespace();
return(node);
}
List <Ast.INode> ParseExprList(SeekableStringReader sr)
{
//expr_list = expr { ',' expr } .
List <Ast.INode> exprList = new List <Ast.INode>();
exprList.Add(ParseExpr(sr));
while (sr.HasMore() && sr.Peek() == ',')
{
sr.Read();
try {
exprList.Add(ParseExpr(sr));
} catch (ParseException) {
sr.Rewind(1);
break;
}
}
return(exprList);
}
private Ast.INode ParseExpr(SeekableStringReader sr)
{
// expr = [ <whitespace> ] single | compound [ <whitespace> ] .
sr.SkipWhitespace();
if (!sr.HasMore())
{
throw new ParseException("unexpected end of line, missing expression or close/open character");
}
char c = sr.Peek();
Ast.INode node;
if (c == '{' || c == '[' || c == '(')
{
node = ParseCompound(sr);
}
else
{
node = ParseSingle(sr);
}
sr.SkipWhitespace();
return(node);
}
Ast.INode ParseCompound(SeekableStringReader sr)
{
// compound = tuple | dict | list | set .
sr.SkipWhitespace();
switch (sr.Peek())
{
case '[':
return(ParseList(sr));
case '{':
{
int bm = sr.Bookmark();
try {
return(ParseSet(sr));
} catch (ParseException) {
sr.FlipBack(bm);
return(ParseDict(sr));
}
}
case '(':
// tricky case here, it can be a tuple but also a complex number.
// try complex number first
{
int bm = sr.Bookmark();
try {
return(ParseComplex(sr));
} catch (ParseException) {
sr.FlipBack(bm);
return(ParseTuple(sr));
}
}
default:
throw new ParseException("invalid sequencetype char");
}
}
private Ast.SetNode ParseSet(SeekableStringReader sr)
{
// set = '{' expr_list trailing_comma '}' .
// trailing_comma = '' | ',' .
sr.Read(); // {
sr.SkipWhitespace();
Ast.SetNode setnode = new Ast.SetNode();
var elts = ParseExprList(sr);
// handle trailing comma if present
sr.SkipWhitespace();
if (!sr.HasMore())
{
throw new ParseException("missing '}'");
}
if (sr.Peek() == ',')
{
sr.Read();
}
if (!sr.HasMore())
{
throw new ParseException("missing '}'");
}
char closechar = sr.Read();
if (closechar != '}')
{
throw new ParseException("expected '}'");
}
// make sure it has set semantics (remove duplicate elements)
var h = new HashSet <Ast.INode>(elts);
setnode.Elements = new List <Ast.INode>(h);
return(setnode);
}
private double ParseImaginaryPart(SeekableStringReader sr)
{
//imaginary = ['+' | '-' ] ( float | int ) 'j' .
// string numberstr = sr.ReadUntil('j');
// try {
// return this.ParseDouble(numberstr);
// } catch(FormatException x) {
// throw new ParseException("invalid float format", x);
// }
if (!sr.HasMore())
{
throw new ParseException("unexpected end of input string");
}
char signOrDigit = sr.Peek();
if (signOrDigit == '+')
{
sr.Read(); // skip the '+'
}
// now an int or float follows.
double doubleValue;
int bookmark = sr.Bookmark();
try {
doubleValue = ParseFloat(sr).Value;
} catch (ParseException) {
sr.FlipBack(bookmark);
var integerPart = ParseInt(sr);
var integerNode = integerPart as Ast.IntegerNode;
if (integerNode != null)
{
doubleValue = integerNode.Value;
}
else
{
var longNode = integerPart as Ast.LongNode;
if (longNode != null)
{
doubleValue = longNode.Value;
}
else
{
var decimalNode = integerPart as Ast.DecimalNode;
if (decimalNode != null)
{
doubleValue = Convert.ToDouble(decimalNode.Value);
}
else
{
throw new ParseException("not an integer for the imaginary part");
}
}
}
}
// now a 'j' must follow!
sr.SkipWhitespace();
try {
if (sr.Read() != 'j')
{
throw new ParseException("not an imaginary part");
}
} catch (IndexOutOfRangeException) {
throw new ParseException("not an imaginary part");
}
return(doubleValue);
}
private Ast.ComplexNumberNode ParseComplex(SeekableStringReader sr)
{
//complex = complextuple | imaginary .
//imaginary = ['+' | '-' ] ( float | int ) 'j' .
//complextuple = '(' ( float | int ) imaginary ')' .
if (sr.Peek() == '(')
{
// complextuple
sr.Read(); // (
string numberstr;
if (sr.Peek() == '-' || sr.Peek() == '+')
{
// starts with a sign, read that first otherwise the readuntil will return immediately
numberstr = sr.Read(1) + sr.ReadUntil('+', '-');
}
else
{
numberstr = sr.ReadUntil('+', '-');
}
sr.Rewind(1); // rewind the +/-
// because we're a bit more cautious here with reading chars than in the float parser,
// it can be that the parser now stopped directly after the 'e' in a number like "3.14e+20".
// ("3.14e20" is fine) So, check if the last char is 'e' and if so, continue reading 0..9.
if (numberstr.EndsWith("e", StringComparison.InvariantCultureIgnoreCase))
{
// if the next symbol is + or -, accept it, then read the exponent integer
if (sr.Peek() == '-' || sr.Peek() == '+')
{
numberstr += sr.Read(1);
}
numberstr += sr.ReadWhile("0123456789");
}
sr.SkipWhitespace();
double realpart;
try {
realpart = ParseDouble(numberstr);
} catch (FormatException x) {
throw new ParseException("invalid float format", x);
}
double imaginarypart = ParseImaginaryPart(sr);
if (sr.Read() != ')')
{
throw new ParseException("expected ) to end a complex number");
}
return(new Ast.ComplexNumberNode
{
Real = realpart,
Imaginary = imaginarypart
});
}
// imaginary
double imag = ParseImaginaryPart(sr);
return(new Ast.ComplexNumberNode
{
Real = 0,
Imaginary = imag
});
}
private Ast.INode ParseDict(SeekableStringReader sr)
{
//dict = '{' keyvalue_list trailing_comma '}' .
//keyvalue_list = keyvalue { ',' keyvalue } .
//keyvalue = expr ':' expr .
// trailing_comma = '' | ',' .
sr.Read(); // {
sr.SkipWhitespace();
Ast.DictNode dict = new Ast.DictNode();
if (sr.Peek() == '}')
{
sr.Read();
return(dict); // empty dict
}
var elts = ParseKeyValueList(sr);
// handle trailing comma if present
sr.SkipWhitespace();
if (!sr.HasMore())
{
throw new ParseException("missing '}'");
}
if (sr.Peek() == ',')
{
sr.Read();
}
if (!sr.HasMore())
{
throw new ParseException("missing '}'");
}
char closechar = sr.Read();
if (closechar != '}')
{
throw new ParseException("expected '}'");
}
// make sure it has dict semantics (remove duplicate keys)
var fixedDict = new Dictionary <Ast.INode, Ast.INode>(elts.Count);
foreach (var node in elts)
{
var kv = (Ast.KeyValueNode)node;
fixedDict[kv.Key] = kv.Value;
}
foreach (var kv in fixedDict)
{
dict.Elements.Add(new Ast.KeyValueNode
{
Key = kv.Key,
Value = kv.Value
});
}
// SPECIAL CASE: {'__class__':'float','value':'nan'} ---> Double.NaN
if (dict.Elements.Count != 2)
{
return(dict);
}
if (!dict.Elements.Contains(new Ast.KeyValueNode(new Ast.StringNode("__class__"),
new Ast.StringNode("float"))))
{
return(dict);
}
if (dict.Elements.Contains(new Ast.KeyValueNode(new Ast.StringNode("value"), new Ast.StringNode("nan"))))
{
return(new Ast.DoubleNode(double.NaN));
}
return(dict);
}