private static int ReadTextNode(TokenizerArgument tokenizerArg)
{
if (tokenizerArg == null)
{
throw new ArgumentNullException(nameof(tokenizerArg));
}
var index = tokenizerArg.Index;
var startTagIdentifierLength = StartTagIdentifier.Length;
var startTagIdentifierVariation1Length = StartTagIdentifierVariation1.Length;
if (index + startTagIdentifierVariation1Length < tokenizerArg.Input.Length &&
tokenizerArg.Input.Substring(index, startTagIdentifierVariation1Length).Equals(StartTagIdentifierVariation1))
{
tokenizerArg.Output.Add(StartTagIdentifierVariation1);
tokenizerArg.Status = ParseStatus.ReadEndTag;
index += startTagIdentifierVariation1Length;
}
else if (tokenizerArg.Input.Substring(index, startTagIdentifierLength).Equals(StartTagIdentifier))
{
tokenizerArg.Output.Add(StartTagIdentifier);
tokenizerArg.Status = ParseStatus.ReadStartTag;
index += startTagIdentifierLength;
}
else
{
var nextIndex = tokenizerArg.Input.IndexOf(StartTagIdentifier, index, StringComparison.OrdinalIgnoreCase);
tokenizerArg.Output.Add(nextIndex == -1
? tokenizerArg.Input.Substring(index)
: tokenizerArg.Input.Substring(index, nextIndex - index));
index = nextIndex;
}
return(index);
}
private static int ReadEndTag(TokenizerArgument tokenizerArg)
{
if (tokenizerArg == null)
{
throw new ArgumentNullException(nameof(tokenizerArg));
}
var index = tokenizerArg.Index;
index = GetIndexOfNextNonWhitespaceChar(tokenizerArg.Input, index);
var tagNameStartIndex = index;
index = GetIndexOfNextTagIdentifier(tokenizerArg.Input, index, EndTagIdentifier);
tokenizerArg.Output.Add(tokenizerArg.Input.Substring(tagNameStartIndex, index - tagNameStartIndex));
index = GetIndexOfNextNonWhitespaceChar(tokenizerArg.Input, index);
var endTagIdentifierLength = EndTagIdentifier.Length;
if (index < tokenizerArg.Input.Length &&
tokenizerArg.Input.Substring(index, endTagIdentifierLength).Equals(EndTagIdentifier))
{
tokenizerArg.Output.Add(EndTagIdentifier);
tokenizerArg.Status = ParseStatus.ReadText;
index += endTagIdentifierLength;
}
return(index);
}
private static int ReadAttributeValue(TokenizerArgument tokenizerArg)
{
if (tokenizerArg == null)
{
throw new ArgumentNullException(nameof(tokenizerArg));
}
var index = tokenizerArg.Index;
index = GetIndexOfNextNonWhitespaceChar(tokenizerArg.Input, index);
var valueStartIndex = index;
var value = index < tokenizerArg.Input.Length
? tokenizerArg.Input.Substring(index, 1)
: string.Empty;
var isDoubleQuotes = value.Equals(DoubleQuotesIdentifier);
var isSingleQuotes = value.Equals(SingleQuotesIdentifier);
if (isDoubleQuotes || isSingleQuotes)
{
index++;
index = GetIndexOfNextDelimiter(
tokenizerArg.Input,
index,
isDoubleQuotes ? DoubleQuotesIdentifier : SingleQuotesIdentifier);
tokenizerArg.Output.Add(tokenizerArg.Input.Substring(valueStartIndex + 1, index - valueStartIndex - 2));
tokenizerArg.Status = ParseStatus.ReadAttributeName;
}
else
{
index = GetIndexOfNextTagIdentifier(tokenizerArg.Input, index, EndTagIdentifierVariation1);
tokenizerArg.Output.Add(tokenizerArg.Input.Substring(valueStartIndex, index - valueStartIndex));
index = GetIndexOfNextNonWhitespaceChar(tokenizerArg.Input, index);
tokenizerArg.Status = ParseStatus.ReadAttributeName;
}
var endTagIdentifierLength = EndTagIdentifier.Length;
var endTagIdentifierVariation1Length = EndTagIdentifierVariation1.Length;
if (index + 1 < tokenizerArg.Input.Length &&
tokenizerArg.Input.Substring(index, endTagIdentifierVariation1Length).Equals(EndTagIdentifierVariation1))
{
tokenizerArg.Output.Add(EndTagIdentifierVariation1);
tokenizerArg.Status = ParseStatus.ReadText;
index += endTagIdentifierVariation1Length;
}
else if (index < tokenizerArg.Input.Length &&
tokenizerArg.Input.Substring(index, endTagIdentifierLength).Equals(EndTagIdentifier))
{
tokenizerArg.Output.Add(EndTagIdentifier);
tokenizerArg.Status = ParseStatus.ReadText;
index += endTagIdentifierLength;
}
return(index);
}
private static int ReadAttributeName(TokenizerArgument tokenizerArg)
{
if (tokenizerArg == null)
{
throw new ArgumentNullException(nameof(tokenizerArg));
}
var index = tokenizerArg.Index;
index = GetIndexOfNextNonWhitespaceChar(tokenizerArg.Input, index);
var attributeNameStartIndex = index;
index =
GetIndexOfNextTagIdentifier(tokenizerArg.Input, index, EndTagIdentifierVariation1 + EqualityIdentifier);
tokenizerArg.Output.Add(
tokenizerArg.Input.Substring(attributeNameStartIndex, index - attributeNameStartIndex));
index = GetIndexOfNextNonWhitespaceChar(tokenizerArg.Input, index);
var endTagIdentifierLength = EndTagIdentifier.Length;
var endTagIdentifierVariation1Length = EndTagIdentifierVariation1.Length;
var equalityIdentifierLength = EqualityIdentifier.Length;
var slashIdentifierLength = SlashIdentifier.Length;
if (index + 1 < tokenizerArg.Input.Length &&
tokenizerArg.Input.Substring(index, endTagIdentifierVariation1Length).Equals(EndTagIdentifierVariation1))
{
tokenizerArg.Output.Add(EndTagIdentifierVariation1);
tokenizerArg.Status = ParseStatus.ReadText;
index += endTagIdentifierVariation1Length;
}
else if (index < tokenizerArg.Input.Length &&
tokenizerArg.Input.Substring(index, endTagIdentifierLength).Equals(EndTagIdentifier))
{
tokenizerArg.Output.Add(EndTagIdentifier);
tokenizerArg.Status = ParseStatus.ReadText;
index += endTagIdentifierLength;
}
else if (index < tokenizerArg.Input.Length &&
tokenizerArg.Input.Substring(index, equalityIdentifierLength).Equals(EqualityIdentifier))
{
tokenizerArg.Output.Add(EqualityIdentifier);
tokenizerArg.Status = ParseStatus.ReadAttributeValue;
index += equalityIdentifierLength;
}
else if (index < tokenizerArg.Input.Length &&
tokenizerArg.Input.Substring(index, slashIdentifierLength).Equals(SlashIdentifier))
{
index += slashIdentifierLength;
}
return(index);
}
/// <summary>
/// This will tokenise the HTML input string.
/// </summary>
/// <param name="input"></param>
/// <returns></returns>
private StringCollection GetTokens(string input)
{
var tokens = new StringCollection();
if (input.IsNullOrWhiteSpace())
{
return(tokens);
}
var nodeProcessorMappings = new Dictionary <ParseStatus, ProcessNode>
{
[ParseStatus.ReadText] = ReadTextNode,
[ParseStatus.ReadStartTag] = ReadStartTag,
[ParseStatus.ReadEndTag] = ReadEndTag,
[ParseStatus.ReadAttributeName] = ReadAttributeName,
[ParseStatus.ReadAttributeValue] = ReadAttributeValue
};
var index = 0;
var status = ParseStatus.ReadText;
while (index < input.Length)
{
if (nodeProcessorMappings.ContainsKey(status))
{
var tokenizerArg = new TokenizerArgument(input, index, status, tokens);
index = nodeProcessorMappings[status].Invoke(tokenizerArg);
status = tokenizerArg.Status;
if (index == -1)
{
break;
}
}
}
return(tokens);
}
