public static File Parse(string filePath, string encoding, IXmlFlavor flavor)
{
Tracer.Trace($"Using {flavor.GetType().Name} flavor for '{filePath}'.");
var encodingToUse = Encoding.GetEncoding(encoding);
File file;
using (var finder = CharacterPositionFinder.CreateFrom(filePath, encodingToUse))
{
file = ParseCore(filePath, finder, flavor, encodingToUse);
Resorter.Resort(file);
GapFiller.Fill(file, finder);
}
CommentCleaner.Clean(file);
return(file);
}
public static File ParseCore(string filePath, CharacterPositionFinder finder, IXmlFlavor flavor, Encoding encoding)
{
using (var reader = new XmlTextReader(new StreamReader(SystemFile.OpenRead(filePath), encoding)))
{
var file = new File
{
Name = filePath,
FooterSpan = CharacterSpan.None, // there is no footer
};
var fileBegin = new LineInfo(reader.LineNumber + 1, reader.LinePosition);
try
{
var dummyRoot = new Container();
// Parse the XML and display the text content of each of the elements.
// as there are XMLs that have the declaration on same line as the root element, we just loop and parse until the end
while (!reader.EOF)
{
Parse(reader, dummyRoot, finder, flavor);
}
var root = dummyRoot.Children.OfType <Container>().Last();
var rootEnd = FixRoot(root, dummyRoot);
var fileEnd = new LineInfo(reader.LineNumber, reader.LinePosition - 1);
var positionAfterLastElement = new LineInfo(rootEnd.LineNumber, rootEnd.LinePosition + 1); // we calculate the next one (either a new line character or a regular one)
file.LocationSpan = new LocationSpan(fileBegin, fileEnd);
if (positionAfterLastElement < fileEnd)
{
file.FooterSpan = GetCharacterSpan(new LocationSpan(positionAfterLastElement, fileEnd), finder);
}
file.Children.Add(root);
}
catch (XmlException ex)
{
// try to adjust location span to include full file content
// but ignore empty files as parsing errors
var lines = SystemFile.ReadLines(filePath).Count();
if (lines == 0)
{
file.LocationSpan = new LocationSpan(LineInfo.None, LineInfo.None);
}
else
{
file.ParsingErrors.Add(new ParsingError
{
ErrorMessage = ex.Message,
Location = new LineInfo(ex.LineNumber, ex.LinePosition),
});
file.LocationSpan = new LocationSpan(new LineInfo(1, 0), new LineInfo(lines + 1, 0));
}
}
return(file);
}
}
//// ATTENTION !!!! SIDE EFFECT AS WE READ FURTHER !!!!
private static LocationSpan GetLocationSpanWithParseAttributes(XmlTextReader reader, Container container, CharacterPositionFinder finder, IXmlFlavor flavor)
{
return(GetLocationSpan(reader, _ =>
{
if (_.HasAttributes)
{
// we have to read the attributes
ParseAttributes(_, container, finder, flavor);
}
}));
}
private static void ParseTerminalNode(XmlTextReader reader, Container parent, CharacterPositionFinder finder, IXmlFlavor flavor)
{
var name = flavor.GetName(reader);
var type = flavor.GetType(reader);
var content = flavor.GetContent(reader);
var locationSpan = GetLocationSpan(reader);
var span = GetCharacterSpan(locationSpan, finder);
var child = AddTerminalNode(parent, type, name, content, locationSpan, span);
flavor.FinalAdjustAfterParsingComplete(child);
}
private static void ParseAttribute(XmlTextReader reader, Container parent, CharacterPositionFinder finder, IXmlFlavor flavor)
{
var attributeStartPos = new LineInfo(reader.LineNumber, reader.LinePosition);
var name = flavor.GetName(reader);
var type = flavor.GetType(reader);
// important call to be able to read the attribute value
reader.ReadAttributeValue();
var content = flavor.GetContent(reader);
var attributeEndPos = new LineInfo(reader.LineNumber, reader.LinePosition + content.Length);
var startPos = finder.GetCharacterPosition(attributeStartPos);
var endPos = finder.GetCharacterPosition(attributeEndPos);
var locationSpan = new LocationSpan(attributeStartPos, attributeEndPos);
var span = new CharacterSpan(startPos, endPos);
var child = AddTerminalNode(parent, type, name, content, locationSpan, span);
flavor.FinalAdjustAfterParsingComplete(child);
}
private static void ParseAttributes(XmlTextReader reader, Container parent, CharacterPositionFinder finder, IXmlFlavor flavor)
{
if (flavor.ParseAttributesEnabled)
{
reader.MoveToFirstAttribute();
// read all attributes
do
{
ParseAttribute(reader, parent, finder, flavor);
}while (reader.MoveToNextAttribute());
}
}
private static void ParseElement(XmlTextReader reader, Container parent, CharacterPositionFinder finder, IXmlFlavor flavor)
{
var name = flavor.GetName(reader);
var type = flavor.GetType(reader);
var content = flavor.GetContent(reader);
var container = new Container
{
Type = type,
Name = name,
Content = content,
};
var isEmpty = reader.IsEmptyElement;
var startingSpan = GetLocationSpanWithParseAttributes(reader, container, finder, flavor);
var headerSpan = GetCharacterSpan(startingSpan, finder);
if (isEmpty)
{
// there is no content, so we have to get away of the footer by just using the '/>' characters as footer
var headerSpanCorrected = new CharacterSpan(headerSpan.Start, Math.Max(headerSpan.Start, headerSpan.End - 2));
var footerSpanCorrected = new CharacterSpan(Math.Max(0, headerSpan.End - 1), headerSpan.End);
container.LocationSpan = startingSpan;
container.HeaderSpan = headerSpanCorrected;
container.FooterSpan = footerSpanCorrected;
}
else
{
while (reader.NodeType != XmlNodeType.EndElement)
{
Parse(reader, container, finder, flavor);
// we had a side effect (reading further on stream to get the location span), so we have to check whether we found already an element or end element
if (reader.NodeType == XmlNodeType.Element)
{
continue;
}
if (reader.NodeType == XmlNodeType.EndElement)
{
break;
}
}
var endingSpan = GetLocationSpan(reader);
var footerSpan = GetCharacterSpan(endingSpan, finder);
container.LocationSpan = new LocationSpan(startingSpan.Start, endingSpan.End);
container.HeaderSpan = headerSpan;
container.FooterSpan = footerSpan;
}
// check whether we can use a terminal node instead
var child = flavor.FinalAdjustAfterParsingComplete(container);
parent.Children.Add(child);
}
private static void Parse(XmlTextReader reader, Container parent, CharacterPositionFinder finder, IXmlFlavor flavor)
{
var nodeType = reader.NodeType;
switch (nodeType)
{
case XmlNodeType.Element:
{
ParseElement(reader, parent, finder, flavor);
break;
}
case XmlNodeType.ProcessingInstruction:
case XmlNodeType.Comment:
case XmlNodeType.XmlDeclaration:
case XmlNodeType.CDATA:
case XmlNodeType.Text:
{
ParseTerminalNode(reader, parent, finder, flavor);
break;
}
default:
{
reader.Read();
break;
}
}
}
