public InternalTag AddSyntaxError(string description)
{
if (this.SyntaxErrors != null && this.SyntaxErrors.Length > 0)
return this; // don't add error if there already is one
InternalTag tag = (InternalTag)MemberwiseClone();
tag.SyntaxErrors = new InternalSyntaxError[] { new InternalSyntaxError(0, Length, description) };
return tag;
}
bool ReadCurrentPosition()
{
attributes = null;
attributeIndex = -1;
inAttributeValue = false;
while (true)
{
var obj = objectIterator.CurrentObject;
if (obj == null)
{
readState = ReadState.EndOfFile;
elementNodeType = XmlNodeType.None;
return(false);
}
else if (objectIterator.IsAtElementEnd)
{
elementNodeType = XmlNodeType.EndElement;
return(true);
}
else if (obj is InternalElement)
{
// element start
elementNodeType = XmlNodeType.Element;
InternalTag startTag = ((InternalTag)obj.NestedObjects[0]);
if (startTag.NestedObjects != null)
{
attributes = startTag.NestedObjects.OfType <InternalAttribute>().ToList();
}
return(true);
}
else if (obj is InternalText)
{
InternalText text = (InternalText)obj;
if (text.ContainsOnlyWhitespace)
{
elementNodeType = XmlNodeType.Whitespace;
}
else
{
elementNodeType = XmlNodeType.Text;
}
return(true);
}
else if (obj is InternalTag)
{
// start/end tags can be skipped as the parent InternalElement already handles them,
// TODO all other tags (xml decl, comments, ...)
}
else
{
throw new NotSupportedException();
}
objectIterator.MoveInto();
}
}
public InternalElement(InternalTag tag)
{
this.Name = tag.Name;
}
void ProcessObject(InternalObject obj, int indentationLevel, ConfigurationList oldConfigurations, ref ConfigurationList newConfigurations)
{
newConfigurations.Clear();
InternalTag tag = obj as InternalTag;
for (int i = 0; i < oldConfigurations.count; i++)
{
Configuration c = oldConfigurations.configurations[i];
if (c.Cost == InfiniteCost)
{
continue;
}
if (tag != null && tag.IsStartTag)
{
// Push start tag
newConfigurations.Add(
c.OpenTags.Push(tag.Name, indentationLevel),
c.Document.Push(obj),
c.Cost
);
}
else if (tag != null && tag.IsEndTag)
{
// We can ignore this end tag
newConfigurations.Add(
c.OpenTags,
c.Document.Push(StartTagPlaceholder).Push(obj),
c.Cost + IgnoreEndTagCost
);
// We can match this end tag with one of the currently open tags
var openTags = c.OpenTags;
var documentWithInsertedEndTags = c.Document;
uint newCost = c.Cost;
while (!openTags.IsEmpty)
{
uint matchCost = 0;
if (openTags.IndentationLevel >= 0 && indentationLevel >= 0)
{
matchCost += (uint)Math.Abs(openTags.IndentationLevel - indentationLevel);
}
if (openTags.Name != tag.Name)
{
matchCost += MismatchedNameCost;
}
newConfigurations.Add(
openTags.Pop(),
documentWithInsertedEndTags.Push(obj),
newCost + matchCost
);
newCost += MissingEndTagCost;
openTags = openTags.Pop();
documentWithInsertedEndTags = documentWithInsertedEndTags.Push(EndTagPlaceholder);
}
}
else
{
newConfigurations.Add(
c.OpenTags,
c.Document.Push(obj),
c.Cost
);
}
}
}
/// <summary>
/// Context: "<"
/// </summary>
void ReadTag()
{
AssertHasMoreData();
int tagStart = this.CurrentLocation;
InternalTag tag = new InternalTag();
var frame = BeginInternalObject(tag);
// Read the opening bracket
// It identifies the type of tag and parsing behavior for the rest of it
tag.OpeningBracket = ReadOpeningBracket();
if (tag.IsUnknownBang && !TryPeekWhiteSpace())
{
OnSyntaxError(tagStart, this.CurrentLocation, "Unknown tag");
}
if (tag.IsStartOrEmptyTag || tag.IsEndTag || tag.IsProcessingInstruction)
{
// Read the name
TryMoveToNonWhiteSpace();
tag.RelativeNameStart = this.CurrentRelativeLocation;
string name;
if (TryReadName(out name))
{
if (!IsValidName(name))
{
OnSyntaxError(this.CurrentLocation - name.Length, this.CurrentLocation, "The name '{0}' is invalid", name);
}
}
else
{
OnSyntaxError("Element name expected");
}
tag.Name = name;
}
else
{
tag.Name = string.Empty;
}
bool isXmlDeclr = tag.Name == "xml" && tag.IsProcessingInstruction;
int oldObjectCount = objects.Count;
if (tag.IsStartOrEmptyTag || tag.IsEndTag || isXmlDeclr)
{
// Read attributes for the tag
while (HasMoreData())
{
// Chech for all forbiden 'name' characters first - see ReadName
TryMoveToNonWhiteSpace();
if (TryPeek('<'))
{
break;
}
string endBr;
int endBrStart = this.CurrentLocation; // Just peek
if (TryReadClosingBracket(out endBr)) // End tag
{
GoBack(endBrStart);
break;
}
// We have "=\'\"" or name - read attribute
int attrStartOffset = this.CurrentLocation;
ReadAttribute();
if (tag.IsEndTag)
{
OnSyntaxError(attrStartOffset, this.CurrentLocation, "Attribute not allowed in end tag.");
}
}
}
else if (tag.IsDocumentType)
{
ReadContentOfDTD();
}
else
{
int start = this.CurrentLocation;
if (tag.IsComment)
{
ReadText(TextType.Comment);
}
else if (tag.IsCData)
{
ReadText(TextType.CData);
}
else if (tag.IsProcessingInstruction)
{
ReadText(TextType.ProcessingInstruction);
}
else if (tag.IsUnknownBang)
{
ReadText(TextType.UnknownBang);
}
else
{
throw new InternalException(string.Format(CultureInfo.InvariantCulture, "Unknown opening bracket '{0}'", tag.OpeningBracket));
}
// Backtrack at complete start
if (IsEndOfFile() || (tag.IsUnknownBang && TryPeek('<')))
{
GoBack(start);
objects.RemoveRange(oldObjectCount, objects.Count - oldObjectCount);
}
}
// Read closing bracket
string bracket;
TryReadClosingBracket(out bracket);
tag.ClosingBracket = bracket;
// Error check
int brStart = this.CurrentLocation - (tag.ClosingBracket ?? string.Empty).Length;
int brEnd = this.CurrentLocation;
if (tag.Name == null)
{
// One error was reported already
}
else if (tag.IsStartOrEmptyTag)
{
if (tag.ClosingBracket != ">" && tag.ClosingBracket != "/>")
{
OnSyntaxError(brStart, brEnd, "'>' or '/>' expected");
}
}
else if (tag.IsEndTag)
{
if (tag.ClosingBracket != ">")
{
OnSyntaxError(brStart, brEnd, "'>' expected");
}
}
else if (tag.IsComment)
{
if (tag.ClosingBracket != "-->")
{
OnSyntaxError(brStart, brEnd, "'-->' expected");
}
}
else if (tag.IsCData)
{
if (tag.ClosingBracket != "]]>")
{
OnSyntaxError(brStart, brEnd, "']]>' expected");
}
}
else if (tag.IsProcessingInstruction)
{
if (tag.ClosingBracket != "?>")
{
OnSyntaxError(brStart, brEnd, "'?>' expected");
}
}
else if (tag.IsUnknownBang)
{
if (tag.ClosingBracket != ">")
{
OnSyntaxError(brStart, brEnd, "'>' expected");
}
}
else if (tag.IsDocumentType)
{
if (tag.ClosingBracket != ">")
{
OnSyntaxError(brStart, brEnd, "'>' expected");
}
}
else
{
throw new InternalException(string.Format(CultureInfo.InvariantCulture, "Unknown opening bracket '{0}'", tag.OpeningBracket));
}
// Attribute name may not apper multiple times
if (objects.Count > oldObjectCount)
{
// Move nested objects into tag.NestedObjects:
tag.NestedObjects = new InternalObject[objects.Count - oldObjectCount];
objects.CopyTo(oldObjectCount, tag.NestedObjects, 0, tag.NestedObjects.Length);
objects.RemoveRange(oldObjectCount, objects.Count - oldObjectCount);
// Look for duplicate attributes:
HashSet <string> attributeNames = new HashSet <string>();
foreach (var obj in tag.NestedObjects)
{
InternalAttribute attr = obj as InternalAttribute;
if (attr != null && !attributeNames.Add(attr.Name))
{
int attrStart = tagStart + attr.StartRelativeToParent;
OnSyntaxError(attrStart, attrStart + attr.Name.Length, "Attribute with name '{0}' already exists", attr.Name);
}
}
}
EndInternalObject(frame);
}
void StoreObject(InternalObject obj)
{
objects.Add(obj);
// Now combine properly-nested elements:
if (elementNameStack == null)
{
return; // parsing tag soup
}
InternalTag tag = obj as InternalTag;
if (tag == null)
{
return;
}
if (tag.IsEmptyTag)
{
// the tag is its own element
objects[objects.Count - 1] = new InternalElement(tag)
{
Length = tag.Length,
LengthTouched = tag.LengthTouched,
IsPropertyNested = true,
StartRelativeToParent = tag.StartRelativeToParent,
NestedObjects = new [] { tag.SetStartRelativeToParent(0) }
};
}
else if (tag.IsStartTag)
{
elementNameStack.Push(tag.Name);
}
else if (tag.IsEndTag && elementNameStack.Count > 0)
{
// Now look for the start element:
int startIndex = objects.Count - 2;
bool ok = false;
string expectedName = elementNameStack.Pop();
if (tag.Name == expectedName)
{
while (startIndex > 0)
{
var startTag = objects[startIndex] as InternalTag;
if (startTag != null)
{
if (startTag.IsStartTag)
{
ok = (startTag.Name == expectedName);
break;
}
else if (startTag.IsEndTag)
{
break;
}
}
startIndex--;
}
}
if (ok)
{
// We found a correct nesting, let's create an element:
InternalObject[] nestedObjects = new InternalObject[objects.Count - startIndex];
int oldStartRelativeToParent = objects[startIndex].StartRelativeToParent;
int pos = 0;
int maxLengthTouched = 0;
for (int i = 0; i < nestedObjects.Length; i++)
{
nestedObjects[i] = objects[startIndex + i].SetStartRelativeToParent(pos);
maxLengthTouched = Math.Max(maxLengthTouched, pos + nestedObjects[i].LengthTouched);
pos += nestedObjects[i].Length;
}
objects.RemoveRange(startIndex, nestedObjects.Length);
objects.Add(
new InternalElement((InternalTag)nestedObjects[0])
{
HasEndTag = true,
IsPropertyNested = true,
Length = pos,
LengthTouched = maxLengthTouched,
StartRelativeToParent = oldStartRelativeToParent,
NestedObjects = nestedObjects
});
}
else
{
// Mismatched name - the nesting isn't properly;
// clear the whole stack so that none of the currently open elements are closed as properly-nested.
elementNameStack.Clear();
}
}
}
public InternalElement(InternalTag tag)
{
this.Name = tag.Name;
}
/// <summary>
/// Context: "<"
/// </summary>
void ReadTag()
{
AssertHasMoreData();
int tagStart = this.CurrentLocation;
InternalTag tag = new InternalTag();
var frame = BeginInternalObject(tag);
// Read the opening bracket
// It identifies the type of tag and parsing behavior for the rest of it
tag.OpeningBracket = ReadOpeningBracket();
if (tag.IsUnknownBang && !TryPeekWhiteSpace())
OnSyntaxError(tagStart, this.CurrentLocation, "Unknown tag");
if (tag.IsStartOrEmptyTag || tag.IsEndTag || tag.IsProcessingInstruction) {
// Read the name
TryMoveToNonWhiteSpace();
tag.RelativeNameStart = this.CurrentRelativeLocation;
string name;
if (TryReadName(out name)) {
if (!IsValidName(name)) {
OnSyntaxError(this.CurrentLocation - name.Length, this.CurrentLocation, "The name '{0}' is invalid", name);
}
} else {
OnSyntaxError("Element name expected");
}
tag.Name = name;
} else {
tag.Name = string.Empty;
}
bool isXmlDeclr = tag.Name == "xml" && tag.IsProcessingInstruction;
int oldObjectCount = objects.Count;
if (tag.IsStartOrEmptyTag || tag.IsEndTag || isXmlDeclr) {
// Read attributes for the tag
while (HasMoreData()) {
// Chech for all forbiden 'name' characters first - see ReadName
TryMoveToNonWhiteSpace();
if (TryPeek('<')) break;
string endBr;
int endBrStart = this.CurrentLocation; // Just peek
if (TryReadClosingBracket(out endBr)) { // End tag
GoBack(endBrStart);
break;
}
// We have "=\'\"" or name - read attribute
int attrStartOffset = this.CurrentLocation;
ReadAttribute();
if (tag.IsEndTag)
OnSyntaxError(attrStartOffset, this.CurrentLocation, "Attribute not allowed in end tag.");
}
} else if (tag.IsDocumentType) {
ReadContentOfDTD();
} else {
int start = this.CurrentLocation;
if (tag.IsComment) {
ReadText(TextType.Comment);
} else if (tag.IsCData) {
ReadText(TextType.CData);
} else if (tag.IsProcessingInstruction) {
ReadText(TextType.ProcessingInstruction);
} else if (tag.IsUnknownBang) {
ReadText(TextType.UnknownBang);
} else {
throw new InternalException(string.Format(CultureInfo.InvariantCulture, "Unknown opening bracket '{0}'", tag.OpeningBracket));
}
// Backtrack at complete start
if (IsEndOfFile() || (tag.IsUnknownBang && TryPeek('<'))) {
GoBack(start);
objects.RemoveRange(oldObjectCount, objects.Count - oldObjectCount);
}
}
// Read closing bracket
string bracket;
TryReadClosingBracket(out bracket);
tag.ClosingBracket = bracket;
// Error check
int brStart = this.CurrentLocation - (tag.ClosingBracket ?? string.Empty).Length;
int brEnd = this.CurrentLocation;
if (tag.Name == null) {
// One error was reported already
} else if (tag.IsStartOrEmptyTag) {
if (tag.ClosingBracket != ">" && tag.ClosingBracket != "/>") OnSyntaxError(brStart, brEnd, "'>' or '/>' expected");
} else if (tag.IsEndTag) {
if (tag.ClosingBracket != ">") OnSyntaxError(brStart, brEnd, "'>' expected");
} else if (tag.IsComment) {
if (tag.ClosingBracket != "-->") OnSyntaxError(brStart, brEnd, "'-->' expected");
} else if (tag.IsCData) {
if (tag.ClosingBracket != "]]>") OnSyntaxError(brStart, brEnd, "']]>' expected");
} else if (tag.IsProcessingInstruction) {
if (tag.ClosingBracket != "?>") OnSyntaxError(brStart, brEnd, "'?>' expected");
} else if (tag.IsUnknownBang) {
if (tag.ClosingBracket != ">") OnSyntaxError(brStart, brEnd, "'>' expected");
} else if (tag.IsDocumentType) {
if (tag.ClosingBracket != ">") OnSyntaxError(brStart, brEnd, "'>' expected");
} else {
throw new InternalException(string.Format(CultureInfo.InvariantCulture, "Unknown opening bracket '{0}'", tag.OpeningBracket));
}
// Attribute name may not apper multiple times
if (objects.Count > oldObjectCount) {
// Move nested objects into tag.NestedObjects:
tag.NestedObjects = new InternalObject[objects.Count - oldObjectCount];
objects.CopyTo(oldObjectCount, tag.NestedObjects, 0, tag.NestedObjects.Length);
objects.RemoveRange(oldObjectCount, objects.Count - oldObjectCount);
// Look for duplicate attributes:
HashSet<string> attributeNames = new HashSet<string>();
foreach (var obj in tag.NestedObjects) {
InternalAttribute attr = obj as InternalAttribute;
if (attr != null && !attributeNames.Add(attr.Name)) {
int attrStart = tagStart + attr.StartRelativeToParent;
OnSyntaxError(attrStart, attrStart + attr.Name.Length, "Attribute with name '{0}' already exists", attr.Name);
}
}
}
EndInternalObject(frame);
}
internal AXmlTag(AXmlObject parent, int startOffset, InternalTag internalObject)
: base(parent, startOffset, internalObject)
{
this.internalObject = internalObject;
}
internal AXmlTag(AXmlObject parent, int startOffset, InternalTag internalObject)
: base(parent, startOffset, internalObject)
{
this.internalObject = internalObject;
}
bool ReadCurrentPosition()
{
attributes = null;
attributeIndex = -1;
inAttributeValue = false;
while (true)
{
var obj = objectIterator.CurrentObject;
if (obj == null)
{
readState = ReadState.EndOfFile;
elementNodeType = XmlNodeType.None;
return(false);
}
else if (objectIterator.IsAtElementEnd)
{
if (IsEmptyElement)
{
// Don't report EndElement for empty elements
nsManager.PopScope();
}
else
{
elementNodeType = XmlNodeType.EndElement;
return(true);
}
}
else if (obj is InternalElement)
{
// element start
elementNodeType = XmlNodeType.Element;
InternalTag startTag = ((InternalTag)obj.NestedObjects[0]);
nsManager.PushScope();
if (startTag.NestedObjects != null)
{
attributes = startTag.NestedObjects.OfType <InternalAttribute>().ToList();
for (int i = 0; i < attributes.Count; i++)
{
var attr = attributes[i];
if (attr.Name.StartsWith("xmlns:", StringComparison.Ordinal))
{
nsManager.AddNamespace(AXmlObject.GetLocalName(attr.Name), attr.Value);
}
else if (attr.Name == "xmlns")
{
nsManager.AddNamespace(string.Empty, attr.Value);
}
}
}
return(true);
}
else if (obj is InternalText)
{
InternalText text = (InternalText)obj;
if (text.ContainsOnlyWhitespace)
{
elementNodeType = XmlNodeType.Whitespace;
}
else
{
elementNodeType = XmlNodeType.Text;
}
return(true);
}
else if (obj is InternalTag)
{
InternalTag tag = (InternalTag)obj;
if (tag.IsStartOrEmptyTag || tag.IsEndTag)
{
// start/end tags can be skipped as the parent InternalElement already handles them
}
else if (tag.IsComment && !settings.IgnoreComments)
{
elementNodeType = XmlNodeType.Comment;
return(true);
}
else if (tag.IsProcessingInstruction && !settings.IgnoreProcessingInstructions)
{
if (tag.Name == "xml")
{
elementNodeType = XmlNodeType.XmlDeclaration;
attributes = tag.NestedObjects.OfType <InternalAttribute>().ToList();
}
else
{
elementNodeType = XmlNodeType.ProcessingInstruction;
}
return(true);
}
else if (tag.IsCData)
{
elementNodeType = XmlNodeType.CDATA;
return(true);
}
else
{
// TODO all other tags
}
}
else
{
throw new NotSupportedException();
}
objectIterator.MoveInto();
}
}