Пример #1
0
        /// <summary>
        /// Called once an &amp; character is being seen.
        /// </summary>
        /// <param name="c">The next character after the &amp; character.</param>
        /// <returns>The entity token.</returns>
        String CharacterReference(Char c)
        {
            var start = _stringBuffer.Length;
            var hex = false;
            var numeric = c == Symbols.Num;

            if (numeric)
            {
                c = GetNext();
                hex = c == 'x' || c == 'X';

                if (hex)
                {
                    c = GetNext();

                    while (c.IsHex())
                    {
                        _stringBuffer.Append(c);
                        c = GetNext();
                    }
                }
                else
                {
                    while (c.IsDigit())
                    {
                        _stringBuffer.Append(c);
                        c = GetNext();
                    }
                }
            }
            else if (c.IsXmlNameStart())
            {
                do
                {
                    _stringBuffer.Append(c);
                    c = GetNext();
                }
                while (c.IsXmlName());
            }

            if (c == Symbols.Semicolon && _stringBuffer.Length > start)
            {
                var length = _stringBuffer.Length - start;
                var content = _stringBuffer.ToString(start, length);
                _stringBuffer.Remove(start, length);

                if (numeric)
                {
                    var number = numeric ? content.FromHex() : content.FromDec();

                    if (!number.IsValidAsCharRef())
                        throw XmlParseError.CharacterReferenceInvalidNumber.At(_position);

                    return number.ConvertFromUtf32();
                }
                else
                {
                    var entity = _resolver.GetSymbol(content);

                    if (String.IsNullOrEmpty(entity))
                        throw XmlParseError.CharacterReferenceInvalidCode.At(_position);

                    return entity;
                }
            }

            throw XmlParseError.CharacterReferenceNotTerminated.At(GetCurrentPosition());
        }
Пример #2
0
        protected String EReference(Char c)
        {
            var buffer = Pool.NewStringBuilder();

            if (c.IsXmlNameStart())
            {
                do
                {
                    buffer.Append(c);
                    c = _stream.Next;
                }
                while (c.IsXmlName());

                var temp = buffer.ToPool();

                if (temp.Length > 0 && c == Specification.SC)
                {
                    var p = _container.GetEntity(temp);

                    if (p != null)
                        return p.NodeValue;
                }
            }
            else if (c == Specification.NUM)
            {
                c = _src.Next;
                var hex = c == 'x' || c == 'X';

                if (hex)
                    c = _stream.Next;

                while (hex ? c.IsHex() : c.IsDigit())
                {
                    buffer.Append(c);
                    c = _src.Next;
                }

                var temp = buffer.ToPool();

                if (temp.Length > 0 && c == Specification.SC)
                {
                    var num = hex ? temp.FromHex() : temp.FromDec();

                    if (num.IsValidAsCharRef())
                        return Char.ConvertFromUtf32(num);

                    throw Errors.Xml(ErrorCode.CharacterReferenceInvalidNumber);
                }
            }

            throw Errors.Xml(ErrorCode.CharacterReferenceNotTerminated);
        }
Пример #3
0
        /// <summary>
        /// More http://www.w3.org/TR/REC-xml/#sec-pi.
        /// </summary>
        /// <param name="c">The next input character.</param>
        XmlToken ProcessingStart(Char c)
        {
            if (c.IsXmlNameStart())
            {
                _stringBuffer.Append(c);
                return ProcessingTarget(GetNext(), NewProcessing());
            }

            throw XmlParseError.XmlInvalidPI.At(GetCurrentPosition());
        }
Пример #4
0
        /// <summary>
        /// Called once an &amp; character is being seen.
        /// </summary>
        /// <param name="c">The next character after the &amp; character.</param>
        /// <returns>The entity token.</returns>
        XmlEntityToken CharacterReference(Char c)
        {
            if (c == Symbols.Num)
            {
                c = GetNext();
                var isHex = c == 'x' || c == 'X';

                if (isHex)
                {
                    c = GetNext();

                    while (c.IsHex())
                    {
                        _stringBuffer.Append(c);
                        c = GetNext();
                    }
                }
                else
                {
                    while (c.IsDigit())
                    {
                        _stringBuffer.Append(c);
                        c = GetNext();
                    }
                }

                if (_stringBuffer.Length > 0 && c == Symbols.Semicolon)
                    return NewEntity(numeric: true, hex: isHex);
            }
            else if (c.IsXmlNameStart())
            {
                do
                {
                    _stringBuffer.Append(c);
                    c = GetNext();
                }
                while (c.IsXmlName());

                if (c == Symbols.Semicolon)
                    return NewEntity();
            }

            throw XmlParseError.CharacterReferenceNotTerminated.At(GetCurrentPosition());
        }
Пример #5
0
        protected void PEReference(Char c, Boolean use = true)
        {
            var buffer = Pool.NewStringBuilder();

            if (c.IsXmlNameStart())
            {
                do
                {
                    buffer.Append(c);
                    c = _stream.Next;
                }
                while (c.IsXmlName());

                var temp = buffer.ToPool();

                if (c == Specification.SC)
                {
                    var p = _container.GetParameter(temp);

                    if (p != null)
                    {
                        if (use)
                        {
                            _stream.Push(temp, p.NodeValue);
                            return;
                        }
                        else
                            throw Errors.Xml(ErrorCode.DtdPEReferenceInvalid);
                    }
                }
            }

            if (use)
                throw Errors.Xml(ErrorCode.DtdPEReferenceInvalid);

            _stringBuffer.Append(Specification.PERCENT).Append(buffer.ToString());
        }
Пример #6
0
        /// <summary>
        /// See 8.2.4.53 Before DOCTYPE name state
        /// </summary>
        /// <param name="c">The next input character.</param>
        XmlToken DoctypeNameBefore(Char c)
        {
            while (c.IsSpaceCharacter())
                c = GetNext();

            if (c.IsXmlNameStart())
            {
                _stringBuffer.Append(c);
                return DoctypeName(GetNext(), NewDoctype());
            }

            throw XmlParseError.DoctypeInvalid.At(GetCurrentPosition());
        }
Пример #7
0
        /// <summary>
        /// More http://www.w3.org/TR/REC-xml/#sec-starttags.
        /// </summary>
        /// <param name="c">The next input character.</param>
        XmlToken TagOpen(Char c)
        {
            if (c == Specification.EM)
                return MarkupDeclaration(_src.Next);

            if (c == Specification.QM)
            {
                c = _src.Next;

                if (_src.ContinuesWith(Tags.XML, false))
                {
                    _src.Advance(2);
                    return DeclarationStart(_src.Next);
                }

                return ProcessingStart(c);
            }

            if (c == Specification.SOLIDUS)
                return TagEnd(_src.Next);

            if (c.IsXmlNameStart())
            {
                _stringBuffer.Clear();
                _stringBuffer.Append(c);
                return TagName(_src.Next, XmlToken.OpenTag());
            }

            throw Errors.Xml(ErrorCode.XmlInvalidStartTag);
        }
Пример #8
0
        /// <summary>
        /// More http://www.w3.org/TR/REC-xml/#sec-pi.
        /// </summary>
        /// <param name="c">The next input character.</param>
        DtdToken ProcessingStart(Char c)
        {
            if (c.IsXmlNameStart())
            {
                _stringBuffer.Clear();
                _stringBuffer.Append(c);
                return ProcessingTarget(_stream.Next, new DtdPIToken());
            }

            throw Errors.Xml(ErrorCode.XmlInvalidPI);
        }
Пример #9
0
        Boolean DeclarationNameBefore(Char c, DtdToken decl)
        {
            while (c.IsSpaceCharacter())
                c = _stream.Next;

            if (c == Specification.EOF)
                throw Errors.Xml(ErrorCode.EOF);

            if (c == Specification.PERCENT)
            {
                PEReference(_stream.Next);
                return DeclarationNameBefore(_stream.Current, decl);
            }
            
            if (c.IsXmlNameStart())
            {
                _stringBuffer.Clear();
                _stringBuffer.Append(c);
                return DeclarationName(_stream.Next, decl);
            }

            return false;
        }
Пример #10
0
        ElementChildrenDeclarationEntry TypeDeclarationChildren(Char c)
        {
            var entries = new List<ElementQuantifiedDeclarationEntry>();
            var connection = Specification.NULL;

            while (true)
            {
                if (entries.Count > 0)
                {
                    if (c != Specification.PIPE && c != Specification.COMMA)
                        throw Errors.Xml(ErrorCode.DtdTypeContent);

                    if (entries.Count == 1)
                        connection = c;
                    else if (connection != c)
                        throw Errors.Xml(ErrorCode.DtdTypeContent);

                    c = _stream.Next;
                }

                while (c.IsSpaceCharacter())
                    c = _stream.Next;

                if (c.IsXmlNameStart())
                {
                    var name = TypeDeclarationName(c);
                    entries.Add(name);
                }
                else if (c == Specification.RBO)
                    entries.Add(TypeDeclarationChildren(_stream.Next));
                else
                    throw Errors.Xml(ErrorCode.DtdTypeContent);

                c = _stream.Current;

                while (c.IsSpaceCharacter())
                    c = _stream.Next;

                if (c == Specification.RBC)
                    break;
            }

            c = _stream.Next;

            if (entries.Count == 0)
                throw Errors.Xml(ErrorCode.DtdTypeInvalid);

            if (connection == Specification.COMMA)
            {
                var sequence = new ElementSequenceDeclarationEntry();
                sequence.Sequence.AddRange(entries);
                sequence.Quantifier = TypeDeclarationQuantifier(c);
                return sequence;
            }
            else
            {
                var choice = new ElementChoiceDeclarationEntry();
                choice.Choice.AddRange(entries);
                choice.Quantifier = TypeDeclarationQuantifier(c);
                return choice;
            }
        }
Пример #11
0
        ElementMixedDeclarationEntry TypeDeclarationMixed(Char c)
        {
            var entry = new ElementMixedDeclarationEntry();

            while (true)
            {
                while (c.IsSpaceCharacter())
                    c = _stream.Next;

                if (c == Specification.RBC)
                {
                    c = _stream.Next;

                    if (c == Specification.ASTERISK)
                    {
                        entry.Quantifier = ElementQuantifier.ZeroOrMore;
                        _stream.Advance();
                        return entry;
                    }

                    if (entry.Names.Count == 0)
                        break;
                }
                else if (c == Specification.PIPE)
                {
                    c = _stream.Next;

                    while (c.IsSpaceCharacter())
                        c = _stream.Next;

                    _stringBuffer.Clear();

                    if (c.IsXmlNameStart())
                    {
                        _stringBuffer.Append(c);

                        while ((c = _stream.Next).IsXmlName())
                            _stringBuffer.Append(c);

                        entry.Names.Add(_stringBuffer.ToString());
                        continue;
                    }
                }
                
                throw Errors.Xml(ErrorCode.DtdTypeContent);
            }

            return entry;
        }
Пример #12
0
        /// <summary>
        /// See 8.2.4.53 Before DOCTYPE name state
        /// </summary>
        /// <param name="c">The next input character.</param>
        XmlToken DoctypeNameBefore(Char c)
        {
            while (c.IsSpaceCharacter())
                c = _src.Next;

            if (c.IsXmlNameStart())
            {
                _stringBuffer.Clear();
                _stringBuffer.Append(c);
                return DoctypeName(_src.Next, XmlToken.Doctype());
            }

            throw Errors.Xml(ErrorCode.DoctypeInvalid);
        }
Пример #13
0
        /// <summary>
        /// Called once an &amp; character is being seen.
        /// </summary>
        /// <param name="c">The next character after the &amp; character.</param>
        /// <returns>The entity token.</returns>
        XmlEntityToken CharacterReference(Char c)
        {
            var buffer = Pool.NewStringBuilder();

            if (c == Specification.NUM)
            {
                c = _src.Next;
                var hex = c == 'x' || c == 'X';

                if (hex)
                {
                    c = _src.Next;

                    while (c.IsHex())
                    {
                        buffer.Append(c);
                        c = _src.Next;
                    }
                }
                else
                {
                    while (c.IsDigit())
                    {
                        buffer.Append(c);
                        c = _src.Next;
                    }
                }

                if (buffer.Length > 0 && c == Specification.SC)
                    return new XmlEntityToken { Value = buffer.ToPool(), IsNumeric = true, IsHex = hex };
            }
            else if (c.IsXmlNameStart())
            {
                do
                {
                    buffer.Append(c);
                    c = _src.Next;
                }
                while (c.IsXmlName());

                if (c == Specification.SC)
                    return new XmlEntityToken { Value = buffer.ToPool() };
            }

            buffer.ToPool();
            throw Errors.Xml(ErrorCode.CharacterReferenceNotTerminated);
        }
Пример #14
0
        /// <summary>
        /// More http://www.w3.org/TR/REC-xml/#NT-Attribute.
        /// </summary>
        /// <param name="c">The next input character.</param>
        /// <param name="tag">The current tag token.</param>
        XmlToken AttributeBeforeName(Char c, XmlTagToken tag)
        {
            while (c.IsSpaceCharacter())
                c = _src.Next;

            if (c == Specification.SOLIDUS)
                return TagSelfClosing(_src.Next, tag);
            else if (c == Specification.GT)
                return tag;
            else if (c == Specification.EOF)
                throw Errors.Xml(ErrorCode.EOF);

            if (c.IsXmlNameStart())
            {
                _stringBuffer.Clear();
                _stringBuffer.Append(c);
                return AttributeName(_src.Next, tag);
            }

            throw Errors.Xml(ErrorCode.XmlInvalidAttribute);
        }
Пример #15
0
        /// <summary>
        /// More http://www.w3.org/TR/REC-xml/#sec-starttags.
        /// </summary>
        /// <param name="c">The next input character.</param>
        XmlToken TagOpen(Char c)
        {
            if (c == Symbols.ExclamationMark)
                return MarkupDeclaration(GetNext());

            if (c == Symbols.QuestionMark)
            {
                c = GetNext();

                if (ContinuesWithSensitive(TagNames.Xml))
                {
                    Advance(2);
                    return DeclarationStart(GetNext());
                }

                return ProcessingStart(c);
            }

            if (c == Symbols.Solidus)
                return TagEnd(GetNext());
            
            if (c.IsXmlNameStart())
            {
                _stringBuffer.Append(c);
                return TagName(GetNext(), NewOpenTag());
            }

            throw XmlParseError.XmlInvalidStartTag.At(GetCurrentPosition());
        }
Пример #16
0
        DtdToken EntityDeclarationAfter(Char c, DtdEntityToken decl)
        {
            if (c.IsSpaceCharacter())
            {
                c = SkipSpaces(c);

                if (decl.IsExtern && !decl.IsParameter && String.IsNullOrEmpty(decl.ExternNotation) && _stream.ContinuesWith(NDATA))
                {
                    _stream.Advance(4);
                    c = _stream.Next;

                    while (c.IsSpaceCharacter())
                        c = _stream.Next;

                    if (c.IsXmlNameStart())
                    {
                        _stringBuffer.Clear();

                        do
                        {
                            _stringBuffer.Append(c);
                            c = _stream.Next;
                        }
                        while (c.IsXmlName());

                        decl.ExternNotation = _stringBuffer.ToString();
                        return EntityDeclarationAfter(c, decl);
                    }

                    throw Errors.Xml(ErrorCode.DtdEntityInvalid);
                }
            }

            if (c == Specification.EOF)
                throw Errors.Xml(ErrorCode.EOF);
            else if (c == Specification.GT)
                return decl;

            throw Errors.Xml(ErrorCode.DtdEntityInvalid);
        }
Пример #17
0
        /// <summary>
        /// More http://www.w3.org/TR/REC-xml/#dt-etag.
        /// </summary>
        /// <param name="c">The next input character.</param>
        XmlToken TagEnd(Char c)
        {
            if (c.IsXmlNameStart())
            {
                do
                {
                    _stringBuffer.Append(c);
                    c = GetNext();
                }
                while (c.IsXmlName());

                while (c.IsSpaceCharacter())
                    c = GetNext();

                if (c == Symbols.GreaterThan)
                {
                    var tag = NewCloseTag();
                    tag.Name = FlushBuffer();
                    return tag;
                }
            }
            
            if (c == Symbols.EndOfFile)
                throw XmlParseError.EOF.At(GetCurrentPosition());

            throw XmlParseError.XmlInvalidEndTag.At(GetCurrentPosition());
        }
Пример #18
0
        /// <summary>
        /// More http://www.w3.org/TR/REC-xml/#attdecls.
        /// </summary>
        /// <param name="c">The next input character.</param>
        DtdAttributeToken AttributeDeclaration(Char c)
        {
            var decl = new DtdAttributeToken();

            if (DeclarationNameBefore(_stream.Next, decl))
            {
                c = SkipSpaces(c);

                while (true)
                {
                    if (c == Specification.GT)
                        return AttributeDeclarationAfter(c, decl);
                    else if (!c.IsXmlNameStart())
                        break;

                    _stringBuffer.Clear();
                    decl.Attributes.Add(AttributeDeclarationName(c));
                    c = _stream.Current;

                    if(c.IsSpaceCharacter())
                        c = SkipSpaces(c);
                }
            }

            throw Errors.Xml(ErrorCode.DtdAttListInvalid);
        }
Пример #19
0
        /// <summary>
        /// More http://www.w3.org/TR/REC-xml/#NT-Attribute.
        /// </summary>
        /// <param name="c">The next input character.</param>
        /// <param name="tag">The current tag token.</param>
        XmlToken AttributeBeforeName(Char c, XmlTagToken tag)
        {
            while (c.IsSpaceCharacter())
                c = GetNext();

            if (c == Symbols.Solidus)
                return TagSelfClosing(GetNext(), tag);
            else if (c == Symbols.GreaterThan)
                return tag;
            else if (c == Symbols.EndOfFile)
                throw XmlParseError.EOF.At(GetCurrentPosition());

            if (c.IsXmlNameStart())
            {
                _stringBuffer.Append(c);
                return AttributeName(GetNext(), tag);
            }

            throw XmlParseError.XmlInvalidAttribute.At(GetCurrentPosition());
        }
Пример #20
0
        /// <summary>
        /// More http://www.w3.org/TR/REC-xml/#dt-etag.
        /// </summary>
        /// <param name="c">The next input character.</param>
        XmlToken TagEnd(Char c)
        {
            if (c.IsXmlNameStart())
            {
                _stringBuffer.Clear();

                do
                {
                    _stringBuffer.Append(c);
                    c = _src.Next;
                }
                while (c.IsXmlName());

                while (c.IsSpaceCharacter())
                    c = _src.Next;

                if (c == Specification.GT)
                {
                    var tag = XmlToken.CloseTag();
                    tag.Name = _stringBuffer.ToString();
                    return tag;
                }
            }

            if (c == Specification.EOF)
                throw Errors.Xml(ErrorCode.EOF);

            throw Errors.Xml(ErrorCode.XmlInvalidEndTag);
        }