private void AttributeValue(char quote)
{
_attributeValueStart = Position();
while (true)
{
switch (Current())
{
case '\0':
_type = HtmlTokenType.Comment;
Consume();
return;
case char c when c == quote:
_attributeValueEnd = Position();
Consume();
_attributeEnd = Position();
AddAttribute();
BeforeAttributeName();
return;
default:
Consume();
break;
}
}
}
private void AddAttribute()
{
if (_attributeNameEnd.Index == _attributeNameStart.Index)
{
return;
}
_attributes ??= ArrayPool <HtmlAttribute> .Shared.Rent(4);
// Grow array as needed
if (_attributesLength == _attributes.Length)
{
var newAttributes = ArrayPool <HtmlAttribute> .Shared.Rent(_attributes.Length * 2);
Array.Copy(_attributes, 0, newAttributes, 0, _attributes.Length);
ArrayPool <HtmlAttribute> .Shared.Return(_attributes);
_attributes = newAttributes;
}
_attributes[_attributesLength++] = new HtmlAttribute(
_attributeType,
_html.AsMemory(_attributeNameStart.Index, _attributeNameEnd.Index - _attributeNameStart.Index),
_html.AsMemory(_attributeValueStart.Index, _attributeValueEnd.Index - _attributeValueStart.Index),
_html.AsMemory(_attributeStart.Index, _attributeEnd.Index - _attributeStart.Index),
new HtmlTextRange(_attributeStart, _attributeEnd),
new HtmlTextRange(_attributeNameStart, _attributeNameEnd),
new HtmlTextRange(_attributeValueStart, _attributeValueEnd));
_attributeNameStart = _attributeNameEnd = default;
}
private void EndTagOpen()
{
_type = HtmlTokenType.EndTag;
switch (Current())
{
case '>':
Consume();
_nameEnd = _nameStart;
break;
case '\0':
Consume();
_type = HtmlTokenType.Text;
break;
case char c when IsASCIIAlpha(c):
TagName(readAttributes: false);
break;
default:
Consume();
BogusComment();
break;
}
}
private void AfterAttributeName()
{
while (true)
{
switch (Current())
{
case '\0':
_type = HtmlTokenType.Comment;
Consume();
return;
case '>':
if (_attributeEnd.Index == _attributeStart.Index)
{
_attributeEnd = Position();
}
Consume();
AddAttribute();
return;
case '/':
if (_attributeEnd.Index == _attributeStart.Index)
{
_attributeEnd = Position();
}
Consume();
AddAttribute();
SelfClosingStartTag();
return;
case '=':
Consume();
BeforeAttributeValue();
return;
case '\t':
case '\r':
case '\n':
case '\f':
case ' ':
if (_attributeEnd.Index == _attributeStart.Index)
{
_attributeEnd = Position();
}
Consume();
break;
default:
AddAttribute();
AttributeName();
return;
}
}
}
private void TagName(bool readAttributes)
{
_nameStart = _nameEnd = Position();
while (true)
{
switch (Current())
{
case '\0':
_type = HtmlTokenType.Comment;
Consume();
return;
case '>':
if (_nameEnd.Index == _nameStart.Index)
{
_nameEnd = Position();
}
Consume();
return;
case '/':
if (_nameEnd.Index == _nameStart.Index)
{
_nameEnd = Position();
}
Consume();
SelfClosingStartTag();
return;
case '\t':
case '\r':
case '\n':
case '\f':
case ' ':
if (_nameEnd.Index == _nameStart.Index)
{
_nameEnd = Position();
}
Consume();
if (readAttributes)
{
BeforeAttributeName();
return;
}
break;
default:
Consume();
break;
}
}
}
public bool Read(out HtmlToken token)
{
// Simplified parsing algorithm based on https://html.spec.whatwg.org/multipage/parsing.html
_attributesLength = 0;
_tokenStart = _tokenEnd = Position();
_nameStart = _nameEnd = default;
_isSelfClosing = false;
if (_position >= _length)
{
if (_attributes != null)
{
ArrayPool <HtmlAttribute> .Shared.Return(_attributes);
_attributes = null;
}
token = default;
return(false);
}
switch (Current())
{
case '<':
Consume();
TagOpen();
break;
default:
Consume();
Data();
break;
}
_tokenEnd = Position();
token = new HtmlToken(
_type,
_isSelfClosing,
_html.AsMemory(_nameStart.Index, _nameEnd.Index - _nameStart.Index),
_html.AsMemory(_tokenStart.Index, _tokenEnd.Index - _tokenStart.Index),
_attributes.AsMemory(0, _attributesLength),
new HtmlTextRange(_tokenStart, _tokenEnd),
new HtmlTextRange(_nameStart, _nameEnd));
return(true);
}
private void BeforeAttributeValue()
{
while (true)
{
switch (Current())
{
case '\t':
case '\r':
case '\n':
case '\f':
case ' ':
Consume();
break;
case '\'':
_attributeType = HtmlAttributeType.SingleQuoted;
Consume();
AttributeValue('\'');
return;
case '"':
_attributeType = HtmlAttributeType.DoubleQuoted;
Consume();
AttributeValue('"');
return;
case '>':
_attributeValueEnd = _attributeValueStart;
_attributeEnd = Position();
Consume();
AddAttribute();
return;
default:
_attributeType = HtmlAttributeType.Unquoted;
AttributeValueUnquoted();
return;
}
}
}
private void AttributeName(bool consumeOnce = false)
{
_attributeType = HtmlAttributeType.NameOnly;
_attributeNameStart = _attributeNameEnd = _attributeStart = _attributeEnd = Position();
_attributeValueStart = _attributeValueEnd = default;
if (consumeOnce)
{
Consume();
}
while (true)
{
switch (Current())
{
case '\t':
case '\r':
case '\n':
case '\f':
case ' ':
case '/':
case '>':
case '\0':
_attributeNameEnd = Position();
AfterAttributeName();
return;
case '=':
_attributeNameEnd = Position();
Consume();
BeforeAttributeValue();
return;
default:
Consume();
break;
}
}
}
private void AttributeValueUnquoted()
{
_attributeValueStart = Position();
while (true)
{
switch (Current())
{
case '>':
_attributeValueEnd = Position();
_attributeEnd = Position();
Consume();
AddAttribute();
return;
case '\0':
_type = HtmlTokenType.Comment;
Consume();
return;
case '\t':
case '\r':
case '\n':
case '\f':
case ' ':
_attributeValueEnd = Position();
_attributeEnd = Position();
Consume();
AddAttribute();
BeforeAttributeName();
return;
default:
Consume();
break;
}
}
}