IEnumerable <AXmlObject> Split(AXmlElement elem)
{
int myIndention = GetIndentLevel(elem);
// Has start tag and no end tag ? (other then empty-element tag)
if (elem.HasStartOrEmptyTag && elem.StartTag.IsStartTag && !elem.HasEndTag && myIndention != -1)
{
int lastAccepted = 0; // Accept start tag
while (lastAccepted + 1 < elem.Children.Count)
{
AXmlObject nextItem = elem.Children[lastAccepted + 1];
if (nextItem is AXmlText)
{
lastAccepted++; continue; // Accept
}
else
{
// Include all more indented items
if (GetIndentLevel(nextItem) > myIndention)
{
lastAccepted++; continue; // Accept
}
else
{
break; // Reject
}
}
}
// Accepted everything?
if (lastAccepted + 1 == elem.Children.Count)
{
yield return(elem);
yield break;
}
AXmlParser.Log("Splitting {0} - take {1} of {2} nested", elem, lastAccepted, elem.Children.Count - 1);
AXmlElement topHalf = new AXmlElement();
topHalf.HasStartOrEmptyTag = elem.HasStartOrEmptyTag;
topHalf.HasEndTag = elem.HasEndTag;
topHalf.AddChildren(elem.Children.Take(1 + lastAccepted)); // Start tag + nested
topHalf.StartOffset = topHalf.FirstChild.StartOffset;
topHalf.EndOffset = topHalf.LastChild.EndOffset;
TagReader.OnSyntaxError(topHalf, topHalf.LastChild.EndOffset, topHalf.LastChild.EndOffset,
"Expected '</{0}>'", topHalf.StartTag.Name);
AXmlParser.Log("Constructed {0}", topHalf);
trackedSegments.AddParsedObject(topHalf, null);
yield return(topHalf);
for (int i = lastAccepted + 1; i < elem.Children.Count; i++)
{
yield return(elem.Children[i]);
}
}
else
{
yield return(elem);
}
}
public AXmlDocument ReadDocument()
{
AXmlDocument doc = new AXmlDocument()
{
Parser = parser
};
// AXmlParser.Log("Flat stream: {0}", PrintObjects(tags));
List <AXmlObject> valid = MatchTags(tags);
// AXmlParser.Log("Fixed stream: {0}", PrintObjects(valid));
IEnumerator <AXmlObject> validStream = valid.GetEnumerator();
validStream.MoveNext(); // Move to first
while (true)
{
// End of stream?
try {
if (validStream.Current == null)
{
break;
}
} catch (InvalidCastException) {
break;
}
doc.AddChild(ReadTextOrElement(validStream));
}
if (doc.Children.Count > 0)
{
doc.StartOffset = doc.FirstChild.StartOffset;
doc.EndOffset = doc.LastChild.EndOffset;
}
// Check well formed
foreach (AXmlTag xmlDeclaration in doc.Children.OfType <AXmlTag>().Where(t => t.IsProcessingInstruction && t.Name.ToLower() == "xml"))
{
if (xmlDeclaration.StartOffset != 0)
{
TagReader.OnSyntaxError(doc, xmlDeclaration.StartOffset, xmlDeclaration.StartOffset + 5,
"XML declaration must be at the start of document");
}
}
int elemCount = doc.Children.OfType <AXmlElement>().Count();
if (elemCount == 0)
{
TagReader.OnSyntaxError(doc, doc.EndOffset, doc.EndOffset,
"Root element is missing");
}
if (elemCount > 1)
{
AXmlElement next = doc.Children.OfType <AXmlElement>().Skip(1).First();
TagReader.OnSyntaxError(doc, next.StartOffset, next.StartOffset,
"Only one root element is allowed");
}
foreach (AXmlTag tag in doc.Children.OfType <AXmlTag>())
{
if (tag.IsCData)
{
TagReader.OnSyntaxError(doc, tag.StartOffset, tag.EndOffset,
"CDATA not allowed in document root");
}
}
foreach (AXmlText text in doc.Children.OfType <AXmlText>())
{
if (!text.ContainsOnlyWhitespace)
{
TagReader.OnSyntaxError(doc, text.StartOffset, text.EndOffset,
"Only whitespace is allowed in document root");
}
}
AXmlParser.Log("Constructed {0}", doc);
trackedSegments.AddParsedObject(doc, null);
return(doc);
}
AXmlElement ReadElement(IEnumerator <AXmlObject> objStream)
{
AXmlElement element = new AXmlElement();
element.IsProperlyNested = true;
// Read start tag
AXmlTag startTag = ReadSingleObject(objStream) as AXmlTag;
AXmlParser.DebugAssert(startTag != null, "Start tag expected");
AXmlParser.DebugAssert(startTag.IsStartOrEmptyTag || startTag == StartTagPlaceholder, "Start tag expected");
if (startTag == StartTagPlaceholder)
{
element.HasStartOrEmptyTag = false;
element.IsProperlyNested = false;
TagReader.OnSyntaxError(element, objStream.Current.StartOffset, objStream.Current.EndOffset,
"Matching openning tag was not found");
}
else
{
element.HasStartOrEmptyTag = true;
element.AddChild(startTag);
}
// Read content and end tag
if (startTag == StartTagPlaceholder || // Check first in case the start tag is null
element.StartTag.IsStartTag)
{
while (true)
{
AXmlTag currTag = objStream.Current as AXmlTag; // Peek
if (currTag == EndTagPlaceholder)
{
TagReader.OnSyntaxError(element, element.LastChild.EndOffset, element.LastChild.EndOffset,
"Expected '</{0}>'", element.StartTag.Name);
ReadSingleObject(objStream);
element.HasEndTag = false;
element.IsProperlyNested = false;
break;
}
else if (currTag != null && currTag.IsEndTag)
{
if (element.HasStartOrEmptyTag && currTag.Name != element.StartTag.Name)
{
TagReader.OnSyntaxError(element, currTag.StartOffset + 2, currTag.StartOffset + 2 + currTag.Name.Length,
"Expected '{0}'. End tag must have same name as start tag.", element.StartTag.Name);
}
element.AddChild(ReadSingleObject(objStream));
element.HasEndTag = true;
break;
}
AXmlObject nested = ReadTextOrElement(objStream);
if (nested is AXmlElement)
{
if (!((AXmlElement)nested).IsProperlyNested)
{
element.IsProperlyNested = false;
}
element.AddChildren(Split((AXmlElement)nested).ToList());
}
else
{
element.AddChild(nested);
}
}
}
else
{
element.HasEndTag = false;
}
element.StartOffset = element.FirstChild.StartOffset;
element.EndOffset = element.LastChild.EndOffset;
AXmlParser.Assert(element.HasStartOrEmptyTag || element.HasEndTag, "Must have at least start or end tag");
AXmlParser.Log("Constructed {0}", element);
trackedSegments.AddParsedObject(element, null); // Need all elements in cache for offset tracking
return(element);
}