private void TryExtractValue(List <CodeLexem> res, ref SourcePart text)
{
if (text[0] == '{')
{
res.Add(new CodeLexem(LexemType.Symbol, CutString(ref text, 1)));
}
else if (text[0] == '[')
{
res.Add(new CodeLexem(LexemType.Symbol, CutString(ref text, 1)));
}
else if (text[0] == '"')
{
res.Add(new CodeLexem(LexemType.Quotes, CutString(ref text, 1)));
// var end = text.IndexOf('"');
var end = FindEndQuoteForValue(text);
res.Add(new CodeLexem(LexemType.Value, CutString(ref text, end)));
res.Add(new CodeLexem(LexemType.Quotes, CutString(ref text, 1)));
}
else
{
var end = text.IndexOfAny(new[] { ',', '}' });
res.Add(new CodeLexem(LexemType.Value, CutString(ref text, end)));
res.Add(new CodeLexem(LexemType.Symbol, CutString(ref text, 1)));
}
}
protected override List<CodeLexem> Parse(SourcePart text)
{
var list = new List<CodeLexem>();
while (text.Length > 0)
{
var lenght = text.Length;
TryExtract(list, ref text, ByteOrderMark);
TryExtract(list, ref text, "[", LexemType.Symbol);
TryExtract(list, ref text, "{", LexemType.Symbol);
if (TryExtract(list, ref text, "\"", LexemType.Quotes))
{
ParseJsonPropertyName(list, ref text); // Extract Name
TryExtract(list, ref text, "\"", LexemType.Quotes);
TryExtract(list, ref text, ":", LexemType.Symbol);
TrySpace(list, ref text);
TryExtractValue(list, ref text); // Extract Value
}
ParseSymbol(list, ref text); // Parse extras
TrySpace(list, ref text);
TryExtract(list, ref text, "\r\n", LexemType.LineBreak);
TryExtract(list, ref text, "\n", LexemType.LineBreak);
TryExtract(list, ref text, "}", LexemType.Symbol);
TryExtract(list, ref text, "]", LexemType.Symbol);
if (lenght == text.Length)
break;
}
return list;
}
private int FindEndQuoteForValue(SourcePart text)
{
if (text == null || text.Length < 2)
{
return(-1);
}
var start = 1;
var end = -1;
var length = text.Length;
var indexMatch = false;
while (!indexMatch && start > 0 && start < length)
{
end = text.Substring(start, length - start).IndexOf("\"") + start;
if (end == -1)
{
return(-1);
}
indexMatch = text.Substring(end - 1, 1).IndexOf("\\") == -1;
start = end + 1;
}
return(indexMatch ? end : -1);
}
protected virtual List<CodeLexem> Parse(SourcePart text)
{
var list = new List<CodeLexem>();
var sourceString = text;
LineBreaks(list, ref sourceString, sourceString.Length, LexemType.PlainText);
return list;
}
private void ParseJsonPropertyName(ICollection<CodeLexem> res, ref SourcePart text)
{
var index = text.IndexOf("\":");
if (index <= 0)
return;
res.Add(new CodeLexem(LexemType.Property, CutString(ref text, index)));
}
protected virtual List <CodeLexem> Parse(SourcePart text)
{
var list = new List <CodeLexem>();
var sourceString = text;
LineBreaks(list, ref sourceString, sourceString.Length, LexemType.PlainText);
return(list);
}
protected bool TryExtract(List<CodeLexem> res, ref SourcePart text, string lex)
{
if (text.StartsWith(lex))
{
CutString(ref text, lex.Length);
return true;
}
return false;
}
protected string CutString(ref SourcePart text, int count)
{
if (count == 0)
return string.Empty;
previousSimbol = text[count - 1];
var result = text.Substring(0, count);
text = text.Substring(count);
return result;
}
protected bool TryExtract(List <CodeLexem> res, ref SourcePart text, string lex)
{
if (text.StartsWith(lex))
{
CutString(ref text, lex.Length);
return(true);
}
return(false);
}
protected bool TryExtract(List <CodeLexem> res, ref SourcePart text, string lex, LexemType type)
{
if (text.StartsWith(lex))
{
res.Add(new CodeLexem(type, CutString(ref text, lex.Length)));
return(true);
}
return(false);
}
private void ParseJsonPropertyName(ICollection <CodeLexem> res, ref SourcePart text)
{
var index = text.IndexOf("\":");
if (index <= 0)
{
return;
}
res.Add(new CodeLexem(LexemType.Property, CutString(ref text, index)));
}
protected void ParseValue(List <CodeLexem> res, ref SourcePart text)
{
const string lex = "\"";
var index = text.IndexOf(lex);
if (index < 0)
{
return;
}
LineBreaks(res, ref text, index + lex.Length - 1, LexemType.Value);
}
private void ParseSymbol(ICollection <CodeLexem> res, ref SourcePart text)
{
int index = text.IndexOfAny(XmlSymbol);
if (index != 0)
{
return;
}
res.Add(new CodeLexem(LexemType.Symbol, text.Substring(0, 1)));
text = text.Substring(1);
}
protected string CutString(ref SourcePart text, int count)
{
if (count == 0)
{
return(string.Empty);
}
previousSimbol = text[count - 1];
var result = text.Substring(0, count);
text = text.Substring(count);
return(result);
}
protected void TrySpace(List <CodeLexem> res, ref SourcePart text)
{
var stringBuilder = new StringBuilder();
while (SpaceChars.Contains(text[0]))
{
stringBuilder.Append(CutString(ref text, 1));
}
if (stringBuilder.Length > 0)
{
res.Add(new CodeLexem(LexemType.Space, stringBuilder.ToString()));
}
}
protected override List <CodeLexem> Parse(SourcePart text)
{
var list = new List <CodeLexem>();
while (text.Length > 0)
{
if (TryExtract(list, ref text, "<!--", LexemType.Comment))
{
TryExtractTo(list, ref text, "-->", LexemType.Comment);
}
if (text.StartsWith("<", StringComparison.Ordinal))
{
IsInsideBlock = false;
}
if (TryExtract(list, ref text, "\"{}", LexemType.Value))
{
TryExtractTo(list, ref text, "\"", LexemType.Value);
}
if (TryExtract(list, ref text, "</", LexemType.Symbol) ||
TryExtract(list, ref text, "<", LexemType.Symbol) ||
TryExtract(list, ref text, "{", LexemType.Symbol))
{
ParseXmlKeyWord(list, ref text, LexemType.Object);
}
if (TryExtract(list, ref text, "\"", LexemType.Quotes))
{
ParseValue(list, ref text);
}
ParseXmlKeyWord(list, ref text, IsInsideBlock ? LexemType.PlainText : LexemType.Property);
TryExtract(list, ref text, "\"", LexemType.Quotes);
TryExtract(list, ref text, "}", LexemType.Symbol);
if (text.StartsWith(">", StringComparison.Ordinal))
{
IsInsideBlock = true;
}
ParseSymbol(list, ref text);
TrySpace(list, ref text);
TryExtract(list, ref text, "\n", LexemType.LineBreak);
}
return(list);
}
protected void TryExtractTo(List <CodeLexem> res, ref SourcePart text, string lex, LexemType type, string except = null)
{
var index = text.IndexOf(lex);
if (except != null)
{
while (index >= 0 && text.Substring(0, index + 1).EndsWith(except))
{
index = text.IndexOf(lex, index + 1);
}
}
if (index < 0)
{
return;
}
LineBreaks(res, ref text, index + lex.Length, type);
}
protected void LineBreaks(List<CodeLexem> res, ref SourcePart text, int to, LexemType type)
{
while (text.Length > 0 && to > 0)
{
var index = text.IndexOf("\n");
if (index >= to)
{
res.Add(new CodeLexem(type, CutString(ref text, to)));
return;
}
if (index != 0)
{
res.Add(new CodeLexem(type, CutString(ref text, index)));
}
res.Add(new CodeLexem(LexemType.LineBreak, CutString(ref text, 1)));
to -= index + 1;
}
}
private void ParseXmlKeyWord(ICollection <CodeLexem> res, ref SourcePart text, LexemType type)
{
var index = text.IndexOfAny(XmlEndOfTerm);
if (index <= 0)
{
return;
}
var delimiterIndex = text.IndexOf(XmlNamespaceDelimeter);
if (delimiterIndex > 0 && delimiterIndex < index)
{
res.Add(new CodeLexem(type, CutString(ref text, delimiterIndex)));
res.Add(new CodeLexem(LexemType.Symbol, CutString(ref text, 1)));
res.Add(new CodeLexem(type, CutString(ref text, index - delimiterIndex - 1)));
return;
}
res.Add(new CodeLexem(type, CutString(ref text, index)));
}
protected void LineBreaks(List <CodeLexem> res, ref SourcePart text, int to, LexemType type)
{
while (text.Length > 0 && to > 0)
{
var index = text.IndexOf("\n");
if (index >= to)
{
res.Add(new CodeLexem(type, CutString(ref text, to)));
return;
}
if (index != 0)
{
res.Add(new CodeLexem(type, CutString(ref text, index)));
}
res.Add(new CodeLexem(LexemType.LineBreak, CutString(ref text, 1)));
to -= index + 1;
}
}
private int FindEndQuoteForValue(SourcePart text)
{
if (text == null || text.Length < 2)
return -1;
var start = 1;
var end = -1;
var length = text.Length;
var indexMatch = false;
while (!indexMatch && start > 0 && start < length)
{
end = text.Substring(start, length - start).IndexOf("\"") + start;
if (end == -1)
return -1;
indexMatch = text.Substring(end - 1, 1).IndexOf("\\") == -1;
start = end + 1;
}
return indexMatch ? end : -1;
}
protected override List<CodeLexem> Parse(SourcePart text)
{
var list = new List<CodeLexem>();
while (text.Length > 0)
{
if (TryExtract(list, ref text, "<!--", LexemType.Comment))
TryExtractTo(list, ref text, "-->", LexemType.Comment);
if (text.StartsWith("<", StringComparison.Ordinal))
IsInsideBlock = false;
if (TryExtract(list, ref text, "\"{}", LexemType.Value))
TryExtractTo(list, ref text, "\"", LexemType.Value);
if (TryExtract(list, ref text, "</", LexemType.Symbol) ||
TryExtract(list, ref text, "<", LexemType.Symbol) ||
TryExtract(list, ref text, "{", LexemType.Symbol))
{
ParseXmlKeyWord(list, ref text, LexemType.Object);
}
if (TryExtract(list, ref text, "\"", LexemType.Quotes))
{
ParseValue(list, ref text);
}
ParseXmlKeyWord(list, ref text, IsInsideBlock ? LexemType.PlainText : LexemType.Property);
TryExtract(list, ref text, "\"", LexemType.Quotes);
TryExtract(list, ref text, "}", LexemType.Symbol);
if (text.StartsWith(">", StringComparison.Ordinal))
IsInsideBlock = true;
ParseSymbol(list, ref text);
TrySpace(list, ref text);
TryExtract(list, ref text, "\n", LexemType.LineBreak);
}
return list;
}
protected override List <CodeLexem> Parse(SourcePart text)
{
var list = new List <CodeLexem>();
while (text.Length > 0)
{
var lenght = text.Length;
TryExtract(list, ref text, ByteOrderMark);
TryExtract(list, ref text, "[", LexemType.Symbol);
TryExtract(list, ref text, "{", LexemType.Symbol);
if (TryExtract(list, ref text, "\"", LexemType.Quotes))
{
ParseJsonPropertyName(list, ref text); // Extract Name
TryExtract(list, ref text, "\"", LexemType.Quotes);
TryExtract(list, ref text, ":", LexemType.Symbol);
TrySpace(list, ref text);
TryExtractValue(list, ref text); // Extract Value
}
ParseSymbol(list, ref text); // Parse extras
TrySpace(list, ref text);
TryExtract(list, ref text, "\r\n", LexemType.LineBreak);
TryExtract(list, ref text, "\n", LexemType.LineBreak);
TryExtract(list, ref text, "}", LexemType.Symbol);
TryExtract(list, ref text, "]", LexemType.Symbol);
if (lenght == text.Length)
{
break;
}
}
return(list);
}
protected void TryExtractTo(List<CodeLexem> res, ref SourcePart text, string lex, LexemType type, string except = null)
{
var index = text.IndexOf(lex);
if (except != null)
{
while (index >= 0 && text.Substring(0, index + 1).EndsWith(except))
{
index = text.IndexOf(lex, index + 1);
}
}
if (index < 0)
return;
LineBreaks(res, ref text, index + lex.Length, type);
}
private void ParseXmlKeyWord(ICollection<CodeLexem> res, ref SourcePart text, LexemType type)
{
var index = text.IndexOfAny(XmlEndOfTerm);
if (index <= 0)
return;
var delimiterIndex = text.IndexOf(XmlNamespaceDelimeter);
if (delimiterIndex > 0 && delimiterIndex < index)
{
res.Add(new CodeLexem(type, CutString(ref text, delimiterIndex)));
res.Add(new CodeLexem(LexemType.Symbol, CutString(ref text, 1)));
res.Add(new CodeLexem(type, CutString(ref text, index - delimiterIndex - 1)));
return;
}
res.Add(new CodeLexem(type, CutString(ref text, index)));
}
private void TryExtractValue(List<CodeLexem> res, ref SourcePart text)
{
if (text[0] == '{')
{
res.Add(new CodeLexem(LexemType.Symbol, CutString(ref text, 1)));
}
else if (text[0] == '[')
{
res.Add(new CodeLexem(LexemType.Symbol, CutString(ref text, 1)));
}
else if (text[0] == '"')
{
res.Add(new CodeLexem(LexemType.Quotes, CutString(ref text, 1)));
// var end = text.IndexOf('"');
var end = FindEndQuoteForValue(text);
res.Add(new CodeLexem(LexemType.Value, CutString(ref text, end)));
res.Add(new CodeLexem(LexemType.Quotes, CutString(ref text, 1)));
}
else
{
var end = text.IndexOfAny(new[] { ',', '}' });
res.Add(new CodeLexem(LexemType.Value, CutString(ref text, end)));
res.Add(new CodeLexem(LexemType.Symbol, CutString(ref text, 1)));
}
}
private void ParseSymbol(ICollection<CodeLexem> res, ref SourcePart text)
{
var index = text.IndexOfAny(JsonSymbol);
if (index != 0)
return;
res.Add(new CodeLexem(LexemType.Symbol, text.Substring(0, 1)));
text = text.Substring(1);
}
protected void ParseValue(List<CodeLexem> res, ref SourcePart text)
{
const string lex = "\"";
var index = text.IndexOf(lex);
if (index < 0)
return;
LineBreaks(res, ref text, index + lex.Length - 1, LexemType.Value);
}
protected void TrySpace(List<CodeLexem> res, ref SourcePart text)
{
var stringBuilder = new StringBuilder();
while (SpaceChars.Contains(text[0]))
{
stringBuilder.Append(CutString(ref text, 1));
}
if (stringBuilder.Length > 0)
{
res.Add(new CodeLexem(LexemType.Space, stringBuilder.ToString()));
}
}
protected bool TryExtract(List<CodeLexem> res, ref SourcePart text, string lex, LexemType type)
{
if (text.StartsWith(lex))
{
res.Add(new CodeLexem(type, CutString(ref text, lex.Length)));
return true;
}
return false;
}
