}//Run
private void flush()
{
if (
(!isString) &&
(!isCommentBlock) &&
(!isCommentLine) &&
(buffer.Length == 0)
)
{
return;
}
string text = buffer.ToString();
object value = null;
buffer.Length = 0;
var type = JSONTokenType.tUnknown;
if (isString)
{
type = JSONTokenType.tStringLiteral;
if (!isVerbatim)
{
try //expand escapes
{
text = JSONStrings.UnescapeString(text);
}
catch (StringEscapeErrorException err)
{
lexer.EmitMessage(MessageType.Error, (int)JSONMsgCode.eInvalidStringEscape, tagStartPos, null, err.ErroredEscape);
return;
}
}
}
else if (isCommentLine && isDirective)//directives treated similar to line comments
{
type = JSONTokenType.tDirective;
}
else if (isCommentBlock || isCommentLine)
{
type = JSONTokenType.tComment;
}
else
{
try
{
value = JSONNumbers.Convert(text, out type);
}
catch (ArgumentException err)
{
lexer.EmitMessage(MessageType.Error, (int)JSONMsgCode.eValueTooBig, tagStartPos, null, err.Message);
return;
}
if (value == null) //not number
{
type = JSONKeywords.Resolve(text);
if (type == JSONTokenType.tIdentifier)
{
if (text.StartsWith("$"))
{
text = text.Remove(0, 1); //take care of verbatim names like: $class, $method, $var etc..
tagStartPos = new SourcePosition(tagStartPos.LineNumber, tagStartPos.ColNumber + 1, tagStartPos.CharNumber + 1);
}
if (!JSONIdentifiers.Validate(text))
{
lexer.EmitMessage(MessageType.Error, (int)JSONMsgCode.eInvalidIdentifier, tagStartPos, null, text);
return;
}
}
} //not number
} //not comment
if (type == JSONTokenType.tStringLiteral)
{
value = text;
}
tokens.Add(new JSONToken(lexer, type, tagStartPos, tagEndPos, text, value));
}
public IEnumerable <bool> Run()
{
const int YIELD_BATCH = 5;
var prevTokenCount = 0;
tokens.Add(new JSONToken(
lexer,
JSONTokenType.tBOF,
srcPos(),
srcPos(),
String.Empty));
#region Main walk
//=======================================================================================================================
while (!source.EOF)
{
moveNext();
#region CRLF
if ((chr == '\n') || (chr == '\r'))
{
if ((isString) && (!isVerbatim))
{
lexer.EmitMessage(MessageType.Error, (int)JSONMsgCode.eUnterminatedString, srcPos());
yield break;//no further parsing
}
if ((isString && isVerbatim) || (isCommentBlock))
{
bufferAdd(chr);
}
if (chr == '\n')
{
if ((!isString) && (!isCommentBlock))
{
flush();
if (isCommentLine)
{
isCommentLine = false;
isDirective = false;
}
freshLine = true;
}
posLine++;
}
posCol = 0;
continue;
}
#endregion
if (isString)
{
#region Inside String
if (isVerbatim || (chr != '\\') || (nchr != '\\'))//take care of 'c:\\dir\\';
{
//turn off strings
if (
((isVerbatim) && (chr == stringEnding) && (nchr == stringEnding)) ||
((!isVerbatim) && (chr == '\\') && (nchr == stringEnding))
)
{
//Verbatim: eat one extra: $"string ""test"" syntax" == string "test" syntax
//Regular: eat "\" escape: "string \"test\" syntax" == string "test" syntax
moveNext();
if (source.EOF)
{
lexer.EmitMessage(MessageType.Error, (int)JSONMsgCode.eUnterminatedString, srcPos());
yield break;//stop further processing, as string did not terminate but EOF reached
}
}
else if (chr == stringEnding)
{
flush();
isString = false;
continue; // eat terminating string char
}
}
else//take care of 'c:\\dir\\'
{
bufferAdd(chr); //preserve \
moveNext();
}
#endregion
}//in string
else
{
#region Not Inside String
if (!isCommentLine)
{
if (!isCommentBlock)
{
#region Not inside comments
#region Turn On Comments
//turn on comment block
if (((chr == '/') || (chr == '|')) && (nchr == '*'))
{
flush();
isCommentBlock = true;
commentBlockEnding = chr;
moveNext();
continue;
}
//turn on comment line
if ((chr == '/') && (nchr == '/'))
{
flush();
isCommentLine = true;
moveNext();
continue;
}
//turn on comment line mode for directive
//directives MUST be the first non-white char on the line
if (freshLine && chr == '#')
{
flush();
isCommentLine = true;
isDirective = true;
continue;
}
#endregion
#region Turn On Strings
if ((chr == '$') && ((nchr == '"') || (nchr == '\'')))
{
flush();
isString = true;
isVerbatim = true;
stringEnding = nchr;
moveNext();
continue;
}
if ((chr == '"') || (chr == '\''))
{
flush();
isString = true;
isVerbatim = false;
stringEnding = chr;
continue;
}
#endregion
#region Syntactic Separators - Space, colons and Symbols
if ((chr == ' ') || (chr == '\t')) //space or TAB
{
flush();
continue; //eat it
}
if (
(chr == ';') ||
(chr == '{') ||
(chr == '}') ||
(chr == '(') ||
(chr == ')') ||
(chr == '[') ||
(chr == ']') ||
(chr == ',') ||
(chr == ':') ||
((chr == '.') && (!JSONIdentifiers.ValidateDigit(nchr)))
)
{
flush();
bufferAdd(chr);
flush();
continue;
}
//Scientific numbers like: 2e+30, 45E-10
if (buffer.Length > 0 &&
JSONIdentifiers.ValidateDigit(buffer[0]) &&
(chr == 'e' || chr == 'E') &&
(nchr == '+' || nchr == '-')
)
{
bufferAdd(chr); //e
moveNext();
bufferAdd(chr); //+ or -
moveNext();
bufferAdd(chr); // add digit after + or -
continue;
}
//for operators like -- /= += etc...
if ((buffer.Length > 0) && (isSymbol(chr) != isSymbol(buffer[0])))
{
flush();
}
#endregion
#endregion
}
else
{
#region Turn Off Comment Block
if ((chr == '*') && (nchr == commentBlockEnding))
{
flush();
isCommentBlock = false;
moveNext();
continue;
}
#endregion
} //block comments off
} //NOT CommentLine
#endregion
}//not in string
bufferAdd(chr);
freshLine = false;
//yield the batch of new tokens
if (tokens.Count > prevTokenCount + YIELD_BATCH)
{
prevTokenCount = tokens.Count;
yield return(true);
}
}//while
//=======================================================================================================================
#endregion
flush(); //flush any remains
#region Post-walk check
if (tokens.Count < 2)
{
lexer.EmitMessage(MessageType.Error, (int)JSONMsgCode.ePrematureEOF, srcPos());
}
if (isCommentBlock)
{
lexer.EmitMessage(MessageType.Error, (int)JSONMsgCode.eUnterminatedComment, srcPos());
}
if (isString)
{
lexer.EmitMessage(MessageType.Error, (int)JSONMsgCode.eUnterminatedString, srcPos());
}
#endregion
tokens.Add(new JSONToken(lexer,
JSONTokenType.tEOF,
new SourcePosition(posLine, posCol, posChar),
new SourcePosition(posLine, posCol, posChar),
String.Empty));
yield return(true);
yield break;
}//Run