private string PrintValueNode(Node node)
{
if (node is Node <UriOrBlank> refNode)
{
return(PrintLinkNode(refNode));
}
if (node is Node <Text> textNode)
{
var val = textNode.Value;
return($"\"{ValueEncoder.LiteralEscape(val.Value)}\"" +
$"@{val.Language}");
}
//only write a node with
var dataType = node.ValueType.DataType == XMLSchema.String?
"" : "^^<" + ValueEncoder.UriEscape(node.ValueType.DataType) + ">";
if (node.ValueString is null)
{
throw new Exception("cannot print node without value string");
}
return($"\"{ValueEncoder.LiteralEscape(node.ValueString)}\"{dataType}");
}
private string PrintLinkNode(Node <UriOrBlank> node)
{
if (node is BlankNode b)
{
return(b.Scope == _g.Id ?
"_:" + b.Label :
throw new BlankNodeScopedToInvalidGraph());
}
if (node is UriNode u)
{
return($"<{ValueEncoder.UriEscape(u.Uri)}>");
}
throw new Exception($"Unknown {nameof(UriOrBlank)} node");
}
/// <summary>
/// Create token by scanning source data.
/// </summary>
public static Token NextToken(ReadOnlySpan <char> data)
{
if (data.IsEmpty)
{
return(Token.Invalid(data));
}
if (char.IsWhiteSpace(data[0]))
{
return(new Token(
NTripleTokenType.Whitespace, data.Slice(0, 1), data.Slice(1)));
}
if (data[0] == '.')
{
return(new Token(
NTripleTokenType.Dot, data.Slice(0, 1), data.Slice(1)));
}
int end;
if (data[0] == '"')
{
end = -1;
var escape = false;
for (var i = 1; i < data.Length; i++)
{
if (data[i] == '"' && !escape)
{
end = i; break;
}
//skip \\
if (data[i] == '\\' &&
i + 1 < data.Length && data[i + 1] == '\\')
{
i++; continue;
}
escape = data[i] == '\\';
}
if (end < 0)
{
return(Token.Invalid(data));
}
return(new Token(
NTripleTokenType.Literal,
data.Slice(1, end - 1),
data.Slice(end + 1)
));
}
if (data[0] == '<')
{
end = data.IndexOf(">");
if (end < 0)
{
return(Token.Invalid(data));
}
var val = ValueEncoder.LiteralUnEscape(data.Slice(1, end - 1));
return(new Token(
NTripleTokenType.Uri, val, data.Slice(end + 1)));
}
if (data[0] == '#')
{
//safe for EOL \r\n or \n
end = data.IndexOf('\n');
if (end < 0)
{
end = data.Length;
}
var remainder =
end == data.Length ?
ReadOnlySpan <char> .Empty :
data.Slice(end + 1);
if (data[end - 1] == '\r')
{
end--;
}
var value = end == data.Length ?
data.Slice(1) : data.Slice(1, end);
return(new Token(NTripleTokenType.Comment,
value, remainder));
}
if (data.StartsWith("_:"))
{
end = 2;
if (end == data.Length || IsInvalidBlankStartChar(data[end]))
{
throw new Exception(
"Blank label cannot start with a " +
(end == data.Length ? "no-char" : data[end].ToString()));
}
for (end = 3; end < data.Length; end++)
{
var ch = data[end];
if (!IsValidBlankChar(ch))
{
end--; break;
}
}
if (end == data.Length)
{
end--;
}
//cannot end on a .
if (data[end] == '.')
{
end--;
}
var rest = end + 1 < data.Length ? data.Slice(end + 1)
: ReadOnlySpan <char> .Empty;
return(new Token(
NTripleTokenType.Blank,
data.Slice(2, end - 1), rest));
}
if (data[0] == '@')
{
end = IndexOfWhitespace(data);
if (end < 0)
{
return(Token.Invalid(data));
}
return(new Token(
NTripleTokenType.Language,
data.Slice(1, end - 1),
data.Slice(end)
));
}
if (data.StartsWith("^^<"))
{
end = data.IndexOf('>');
if (end < 0)
{
return(Token.Invalid(data));
}
return(new Token(NTripleTokenType.DataType,
data.Slice(3, end - 3),
data.Slice(end + 1)));
}
return(Token.Invalid(data));
}
/// <summary>
/// Handle the next <see cref="NTripleTokenizer.Token"/>
/// </summary>
public void Next(NTripleTokenizer.Token token)
{
if (token.Type == NTripleTokenType.Invalid)
{
throw new Exception("Invalid token");
}
if (token.Type == NTripleTokenType.Whitespace)
{
return;
}
if (token.Type == NTripleTokenType.Comment)
{
return;
}
if (_subject is null)
{
if (token.Type == NTripleTokenType.Uri)
{
IsValidUri(token.Value);
_subject = _graph.Uri(ValueEncoder.UriUnEscape(token.Value));
return;
}
if (token.Type == NTripleTokenType.Blank)
{
_subject = _graph.Blank(token.Value);
return;
}
throw new Exception("Expected uri for subject");
}
if (_predicate is null)
{
if (token.Type == NTripleTokenType.Uri)
{
IsValidUri(token.Value);
_predicate = _graph.Uri(
ValueEncoder.UriUnEscape(token.Value));
return;
}
throw new Exception("Expected uri for predicate");
}
if (_object is null)
{
if (token.Type == NTripleTokenType.Uri)
{
IsValidUri(token.Value);
_object = _graph.Uri(token.Value);
return;
}
if (token.Type == NTripleTokenType.Blank)
{
_object = _graph.Blank(token.Value);
return;
}
if (_objectParts is null)
{
if (token.Type != NTripleTokenType.Literal)
{
throw new Exception("expected literal for object value");
}
_objectParts = new ObjectParts
{
Literal = ValueEncoder.LiteralUnEscape(token.Value)
};
return;
}
if (token.Type == NTripleTokenType.Language)
{
IsValidLanguageTag(token.Value);
_objectParts.Lang = new string(token.Value);
return;
}
if (token.Type == NTripleTokenType.DataType)
{
IsValidUri(token.Value);
_objectParts.ValueType = new string(token.Value);
return;
}
}
if (token.Type != NTripleTokenType.Dot)
{
throw new Exception("Expected dot to end triple.");
}
if (_object is null && _objectParts != null)
{
_object = _graph.NewNode(_objectParts.Literal,
ValueEncoder.UriUnEscape(_objectParts.ValueType),
_objectParts.Lang);
}
if (_object is null)
{
throw new Exception("Expected object to be created");
}
_graph.Assert(_subject, _predicate, _object);
//reset for next triple.
_subject = null;
_predicate = null;
_object = null;
_objectParts = null;
}
