public void ParentNodeOfNotRootElementIsNotNull()
{
var doc = new HtmlDocumentNode();
var a = new HtmlElementNode("a");
doc.AppendChild(a);
Assert.Equal(doc, a.ParentNode);
}
private static void PopulatePropertiesElement(HtmlElementNode node, ListView listView)
{
InitializeListView(AttributeColumns, listView);
foreach (var att in node.Attributes)
{
var item = listView.Items.Add(att.Key);
item.SubItems.Add(att.Value != null ? att.Value.Value : "(null)");
}
}
public void PreviousSiblingIsNull()
{
var doc = new HtmlDocumentNode();
var a = new HtmlElementNode("a");
doc.AppendChild(a);
var previousSibling = a.PreviousSibling;
Assert.Null(previousSibling);
}
public void NextSiblingIsNull()
{
var doc = new HtmlDocumentNode();
var a = new HtmlElementNode("a");
doc.AppendChild(a);
var nextSibling = a.NextSibling;
Assert.Null(nextSibling);
}
public void PreviousSiblingIsBr()
{
var doc = new HtmlDocumentNode();
var a = new HtmlElementNode("a");
doc.AppendChild(new HtmlElementNode("br"));
doc.AppendChild(a);
var previousSibling = a.PreviousSibling;
Assert.Equal("br", previousSibling.Name);
}
public void NextSiblingIsBr()
{
var doc = new HtmlDocumentNode();
var a = new HtmlElementNode("a");
doc.AppendChild(a);
doc.AppendChild(new HtmlElementNode("br"));
var nextSibling = a.NextSibling;
Assert.Equal("br", nextSibling.Name);
}
/// <summary>
/// Returns true if this selector matches the specified <see cref="HtmlElementNode"/>.
/// </summary>
public bool IsMatch(HtmlElementNode node)
{
// Compare tag
if (!string.IsNullOrWhiteSpace(Tag) && !string.Equals(Tag, node.TagName, HtmlRules.TagStringComparison))
{
return(false);
}
// Compare attributes
foreach (AttributeSelector selector in Attributes)
{
if (!selector.IsMatch(node))
{
return(false);
}
}
return(true);
}
/// <summary> /// Recursively finds all the matching field elements from the given node. /// </summary> /// <param name="node">Root node to search.</param> /// <returns>All the matching field elements from the given node.</returns> internal IEnumerable <HtmlElementNode> FindValue(HtmlElementNode node) => (Selectors != null) ? Selectors.Find(node) : Enumerable.Empty <HtmlElementNode>();
internal override string GetValueFromNode(HtmlElementNode node) => node.Attributes[AttributeName]?.Value ?? string.Empty;
public static HtmlDocument BuildXmlDocument()
{
HtmlDocument document = new();
// XML header
document.RootNodes.Add(new XmlHeaderNode(new HtmlAttributeCollection
{
new HtmlAttribute("xml"),
new HtmlAttribute("version", "1.0"),
new HtmlAttribute("encoding", "UTF-8")
}));
document.RootNodes.Add(new HtmlTextNode("\r\n"));
// Catalog element
HtmlElementNode catalogNode = document.RootNodes.Add(new HtmlElementNode("catalog"));
catalogNode.Children.Add(new HtmlTextNode("\r\n "));
// Item element
HtmlElementNode xmlNode = new("plant");
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("common", null, new HtmlNodeCollection {
new HtmlTextNode("Bloodroot")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("botanical", null, new HtmlNodeCollection {
new HtmlTextNode("Sanguinaria canadensis")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("zone", null, new HtmlNodeCollection {
new HtmlTextNode("4")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("light", null, new HtmlNodeCollection {
new HtmlTextNode("Mostly Shady")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("price", null, new HtmlNodeCollection {
new HtmlTextNode("$2.44")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
catalogNode.Children.Add(xmlNode);
catalogNode.Children.Add(new HtmlTextNode("\r\n "));
// Item element
xmlNode = new("plant");
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("common", null, new HtmlNodeCollection {
new HtmlTextNode("Columbine")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("botanical", null, new HtmlNodeCollection {
new HtmlTextNode("Aquilegia canadensis")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("zone", null, new HtmlNodeCollection {
new HtmlTextNode("3")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("light", null, new HtmlNodeCollection {
new HtmlTextNode("Mostly Shady")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("price", null, new HtmlNodeCollection {
new HtmlTextNode("$9.37")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
catalogNode.Children.Add(xmlNode);
catalogNode.Children.Add(new HtmlTextNode("\r\n "));
// Item element
xmlNode = new("plant");
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("common", null, new HtmlNodeCollection {
new HtmlTextNode("Marsh Marigold")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("botanical", null, new HtmlNodeCollection {
new HtmlTextNode("Caltha palustris")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("zone", null, new HtmlNodeCollection {
new HtmlTextNode("4")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("light", null, new HtmlNodeCollection {
new HtmlTextNode("Mostly Sunny")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("price", null, new HtmlNodeCollection {
new HtmlTextNode("$6.81")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
catalogNode.Children.Add(xmlNode);
catalogNode.Children.Add(new HtmlTextNode("\r\n "));
// Item element
xmlNode = new("plant");
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("common", null, new HtmlNodeCollection {
new HtmlTextNode("Dutchman's-Breeches")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("botanical", null, new HtmlNodeCollection {
new HtmlTextNode("Dicentra cucullaria")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("zone", null, new HtmlNodeCollection {
new HtmlTextNode("3")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("light", null, new HtmlNodeCollection {
new HtmlTextNode("Mostly Shady")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("price", null, new HtmlNodeCollection {
new HtmlTextNode("$6.44")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
catalogNode.Children.Add(xmlNode);
catalogNode.Children.Add(new HtmlTextNode("\r\n "));
// Item element
xmlNode = new("plant");
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("common", null, new HtmlNodeCollection {
new HtmlTextNode("Ginger, Wild")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("botanical", null, new HtmlNodeCollection {
new HtmlTextNode("Asarum canadense")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("zone", null, new HtmlNodeCollection {
new HtmlTextNode("3")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("light", null, new HtmlNodeCollection {
new HtmlTextNode("Mostly Shady")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
xmlNode.Children.Add(new HtmlElementNode("price", null, new HtmlNodeCollection {
new HtmlTextNode("$9.03")
}));
xmlNode.Children.Add(new HtmlTextNode("\r\n "));
catalogNode.Children.Add(xmlNode);
catalogNode.Children.Add(new HtmlTextNode("\r\n"));
document.RootNodes.Add(new HtmlTextNode("\r\n"));
return(document);
}
/// <summary>
/// Parses the given HTML string into a number of root nodes.
/// </summary>
/// <param name="html">The HTML text to parse.</param>
public IEnumerable <HtmlNode> ParseChildren(string html)
{
TextParser parser = new TextParser(html);
HtmlElementNode rootNode = new HtmlElementNode("[Root]");
HtmlElementNode parentNode = rootNode;
string tag;
bool selfClosing;
// Loop until end of input
while (!parser.EndOfText)
{
if (parser.Peek() == HtmlRules.TagStart)
{
// Test for CDATA segments, which we store but do not parse. This includes comments.
CDataDefinition definition = HtmlRules.CDataDefinitions.FirstOrDefault(dd => parser.MatchesCurrentPosition(dd.StartText, dd.IgnoreCase));
if (definition != null)
{
parentNode.Children.Add(ParseCDataNode(parser, definition));
continue;
}
// Closing tag
if (parser.Peek(1) == HtmlRules.ForwardSlash)
{
parser.MoveAhead(2);
tag = parser.ParseWhile(c => HtmlRules.IsTagCharacter(c));
if (tag.Length > 0)
{
if (parentNode.TagName.Equals(tag, HtmlRules.TagStringComparison))
{
// Should never have matched parent if the top-level node
Debug.Assert(!parentNode.IsTopLevelNode);
parentNode = parentNode.ParentNode;
}
else
{
// Handle mismatched closing tag
int tagPriority = HtmlRules.GetTagPriority(tag);
while (!parentNode.IsTopLevelNode && tagPriority > HtmlRules.GetTagPriority(parentNode.TagName))
{
parentNode = parentNode.ParentNode;
}
if (parentNode.TagName.Equals(tag, HtmlRules.TagStringComparison))
{
Debug.Assert(!parentNode.IsTopLevelNode);
parentNode = parentNode.ParentNode;
}
}
}
parser.MoveTo(HtmlRules.TagEnd);
parser.MoveAhead();
continue;
}
// Open tag
if (ParseTag(parser, out tag))
{
HtmlTagFlag flags = HtmlRules.GetTagFlags(tag);
if (flags.HasFlag(HtmlTagFlag.HtmlHeader))
{
parentNode.Children.Add(ParseHtmlHeader(parser));
}
else if (flags.HasFlag(HtmlTagFlag.XmlHeader))
{
parentNode.Children.Add(ParseXmlHeader(parser));
}
else
{
// Parse attributes
HtmlAttributeCollection attributes = ParseAttributes(parser);
// Parse rest of tag
if (parser.Peek() == HtmlRules.ForwardSlash)
{
parser.MoveAhead();
parser.MovePastWhitespace();
selfClosing = true;
}
else
{
selfClosing = false;
}
parser.MoveTo(HtmlRules.TagEnd);
parser.MoveAhead();
// Add node
HtmlElementNode node = new HtmlElementNode(tag, attributes);
while (!HtmlRules.TagMayContain(parentNode.TagName, tag) && !parentNode.IsTopLevelNode)
{
Debug.Assert(parentNode.ParentNode != null);
parentNode = parentNode.ParentNode;
}
parentNode.Children.Add(node);
if (flags.HasFlag(HtmlTagFlag.CData))
{
// CDATA tags are treated as elements but we store and do not parse the inner content
if (!selfClosing)
{
if (ParseToClosingTag(parser, tag, out string content) && content.Length > 0)
{
node.Children.Add(new HtmlCDataNode(string.Empty, string.Empty, content));
}
}
}
else
{
if (selfClosing && flags.HasFlag(HtmlTagFlag.NoSelfClosing))
{
selfClosing = false;
}
if (!selfClosing && !flags.HasFlag(HtmlTagFlag.NoChildren))
{
parentNode = node; // Node becomes new parent
}
}
}
continue;
}
}
// Text node
int start = parser.Index;
// Text must be at least 1 character (handle '<' that is not part of a tag)
parser.MoveAhead();
parser.MoveTo(HtmlRules.TagStart);
Debug.Assert(parser.Index > start);
parentNode.Children.Add(new HtmlTextNode(parser.Extract(start, parser.Index)));
}
//
return(rootNode.Children);
}
private static string ShortDescriptionElement(HtmlElementNode node) => $"<{node.TagName}>";
private static string LongDescriptionElement(HtmlElementNode node) => string.Empty;
public HtmlNodeCollection(HtmlElementNode parentNode)
{
ParentNode = parentNode;
}
/// <summary>
/// Parses the given HTML string into a collection of root nodes and their
/// children.
/// </summary>
/// <param name="html">The HTML text to parse.</param>
public IEnumerable <HtmlNode> ParseChildren(string?html, bool ignoreHtmlRules = false)
{
HtmlElementNode rootNode = new("[TempContainer]");
HtmlElementNode parentNode = rootNode;
Parser.Reset(html);
bool selfClosing;
string?tag;
// Loop until end of input
while (!Parser.EndOfText)
{
if (Parser.Peek() == HtmlRules.TagStart)
{
// CDATA segments (blocks we store but don't parse--includes comments)
CDataDefinition?definition = HtmlRules.CDataDefinitions.FirstOrDefault(dd => Parser.MatchesCurrentPosition(dd.StartText, dd.StartComparison));
if (definition != null)
{
parentNode.Children.Add(ParseCDataNode(definition));
continue;
}
// Closing tag
if (Parser.Peek(1) == HtmlRules.ForwardSlash)
{
Parser.Index += 2;
tag = Parser.ParseWhile(HtmlRules.IsTagCharacter);
if (tag.Length > 0)
{
if (parentNode.TagName.Equals(tag, HtmlRules.TagStringComparison))
{
// Should never have matched parent if the top-level node
if (!parentNode.IsTopLevelNode)
{
parentNode = parentNode.ParentNode;
}
}
else
{
// Handle mismatched closing tag
int tagPriority = HtmlRules.GetTagNestLevel(tag);
while (!parentNode.IsTopLevelNode && tagPriority > HtmlRules.GetTagNestLevel(parentNode.TagName))
{
parentNode = parentNode.ParentNode;
}
if (parentNode.TagName.Equals(tag, HtmlRules.TagStringComparison))
{
if (!parentNode.IsTopLevelNode)
{
parentNode = parentNode.ParentNode;
}
}
}
}
Parser.SkipTo(HtmlRules.TagEnd);
Parser.Next();
continue;
}
// Open tag
if (ParseTag(out tag))
{
HtmlTagFlag flags = ignoreHtmlRules ? HtmlTagFlag.None : HtmlRules.GetTagFlags(tag);
if (flags.HasFlag(HtmlTagFlag.HtmlHeader))
{
parentNode.Children.Add(ParseHtmlHeader());
}
else if (flags.HasFlag(HtmlTagFlag.XmlHeader))
{
parentNode.Children.Add(ParseXmlHeader());
}
else
{
// Parse attributes
HtmlAttributeCollection attributes = ParseAttributes();
// Parse rest of tag
if (Parser.Peek() == HtmlRules.ForwardSlash)
{
Parser.Next();
Parser.SkipWhiteSpace();
selfClosing = true;
}
else
{
selfClosing = false;
}
Parser.SkipTo(HtmlRules.TagEnd);
Parser.Next();
// Add node
HtmlElementNode node = new(tag, attributes);
while (!HtmlRules.TagMayContain(parentNode.TagName, tag) && !parentNode.IsTopLevelNode)
{
parentNode = parentNode.ParentNode;
}
parentNode.Children.Add(node);
if (flags.HasFlag(HtmlTagFlag.CData))
{
// CDATA tags are treated as elements but we store and do not parse the inner content
if (!selfClosing)
{
if (ParseToClosingTag(tag, out string?content) && content.Length > 0)
{
node.Children.Add(new HtmlCDataNode(string.Empty, string.Empty, content));
}
}
}
else
{
if (selfClosing && flags.HasFlag(HtmlTagFlag.NoSelfClosing))
{
selfClosing = false;
}
if (!selfClosing && !flags.HasFlag(HtmlTagFlag.NoChildren))
{
parentNode = node; // Node becomes new parent
}
}
}
continue;
}
}
// Text node: must be at least 1 character (handles '<' that was not a tag)
string text = Parser.ParseCharacter();
text += Parser.ParseTo(HtmlRules.TagStart);
parentNode.Children.Add(new HtmlTextNode(text));
}
// Return top-level nodes from nodes just parsed
return(rootNode.Children);
}
/// <summary> /// Extracts this field value from the given node. /// </summary> internal abstract string GetValueFromNode(HtmlElementNode node);
internal override string GetValueFromNode(HtmlElementNode node) => node.Text;
public static HtmlDocument BuildHtmlDocument()
{
HtmlDocument document = new();
// HTML header
document.RootNodes.Add(new HtmlHeaderNode(new HtmlAttributeCollection {
new HtmlAttribute("html")
}));
document.RootNodes.Add(new HtmlTextNode("\r\n"));
// HTML element
HtmlElementNode htmlNode = document.RootNodes.Add(new HtmlElementNode("html"));
htmlNode.Children.Add(new HtmlTextNode("\r\n "));
// Head element
HtmlElementNode headNode = htmlNode.Children.Add(new HtmlElementNode("head"));
headNode.Children.Add(new HtmlTextNode("\r\n "));
// Title element
HtmlElementNode node = headNode.Children.Add(new HtmlElementNode("title"));
node.Children.Add(new HtmlTextNode("Title"));
// Meta element
headNode.Children.Add(new HtmlTextNode("\r\n "));
headNode.Children.Add(new HtmlElementNode("meta", new HtmlAttributeCollection
{
new HtmlAttribute("name", "description"),
new HtmlAttribute("content", "This is my test meta description node!")
}));
headNode.Children.Add(new HtmlTextNode("\r\n "));
// Body element
htmlNode.Children.Add(new HtmlTextNode("\r\n "));
HtmlElementNode bodyNode = htmlNode.Children.Add(new HtmlElementNode("body"));
// Comment
bodyNode.Children.Add(new HtmlTextNode("\r\n "));
bodyNode.Children.Add(new HtmlCDataNode("<!--", "-->", " Here's a comment! "));
// First paragraph
bodyNode.Children.Add(new HtmlTextNode("\r\n "));
node = bodyNode.Children.Add(new HtmlElementNode("p", new HtmlAttributeCollection
{
new HtmlAttribute("id", "par1")
}));
node.Children.Add(new HtmlTextNode("\r\n "));
node.Children.Add(new HtmlTextNode("This is my first paragraph"));
node.Children.Add(new HtmlTextNode("\r\n "));
// Second paragraph
bodyNode.Children.Add(new HtmlTextNode("\r\n "));
node = bodyNode.Children.Add(new HtmlElementNode("p", new HtmlAttributeCollection
{
new HtmlAttribute("id", "par2")
}));
node.Children.Add(new HtmlTextNode("\r\n "));
node.Children.Add(new HtmlTextNode("This is my second paragraph"));
node.Children.Add(new HtmlTextNode("\r\n "));
bodyNode.Children.Add(new HtmlTextNode("\r\n "));
htmlNode.Children.Add(new HtmlTextNode("\r\n"));
return(document);
}
