public void IsAtEndReturnsTrueWhenCurrentIndexAtEnd()
{
//Arrange
var input = "123".AsMemory();
var scanState = new SourceScanState(input);
//Act
scanState.Advance();
scanState.Advance();
scanState.Advance();
//Assert
scanState.IsAtEnd().Should().BeTrue();
}
// ReSharper restore InconsistentNaming
public static Token ScanNextToken(ref SourceScanState state)
{
if (state.IsAtEnd())
{
return(EOF_TOKEN);
}
state.StartIndex = state.CurrentPos;
var current = state.Advance();
// A '&' in plain text
//Possible HTML entity
// hello > world
// ^
if (current == HTML_ENTITY_START)
{
//a valid entity ends on ';'
// the only spaces that are allowed are before the terminating ';'
// next should be either a letter or a '#'
var next = state.PeekNext();
if (char.IsLetter(next) || next == '#')
{
state.Advance();
next = state.PeekNext();
if (char.IsLetterOrDigit(next))
{
//&[a-z][a-z]  or &#x
// ^ ^ ^
current = state.Advance();
while (current != '\0' && char.IsLetterOrDigit(state.PeekNext()))
{
current = state.Advance();
if (current == ';')
{
//end of entity
return(CreateToken(TokenType.HtmlEntity, HtmlElementType.None, state));
}
}
//end of contiguous chars, but no ';' found
//next non-whitespace character should be a ';'
// ex: > ;
// ^
current = AdvanceWhiteSpaces(ref state, current);
if (state.PeekNext() == ';' /*current == ';'*/)
{
state.Advance();
//end of entity
return(CreateToken(TokenType.HtmlEntity, HtmlElementType.None, state));
}
if (!char.IsLetterOrDigit(current))
{
//Invalid entity, treat as text
return(CreateToken(TokenType.Text, HtmlElementType.None, state));
}
// else it is not a valid entity and we can ignore it
}
}
}
if (current == HTML_TAG_START)
{
current = state.Advance();
if (state.IsInTag)
{
//We found a < inside a tag, ex: "<u<p>>
// ^
//We found a < inside a tag, ex: "<<p>>
// ^
//Finish the current Tag and start a new one
return(CreateToken(TokenType.HtmlOpenTag, HtmlElementType.Invalid, state));
}
if (state.IsAtEnd())
{
//we ended on a '<' character
return(CreateToken(TokenType.Text, HtmlElementType.Invalid, state));
}
state.IsInTag = true;
}
if (state.IsInTag)
{
return(ScanTag(ref state, current));
}
// not in a tag and we found a '<'
if (current == HTML_TAG_END)
{
if (state.IsInTag)
{
state.IsInTag = false;
}
//hello > world
// ^
return(CreateToken(TokenType.Text, HtmlElementType.Invalid, state));
}
while (!state.IsAtEnd() &&
state.PeekNextUnsafe() != HTML_TAG_START &&
state.PeekNextUnsafe() != HTML_ENTITY_START)
{
state.Advance();
}
return(CreateToken(TokenType.Text, HtmlElementType.None, state));
}
