/// <summary> /// Recursively searches the given nodes for ones matching the specified compiled selectors. /// </summary> /// <param name="nodes">The nodes to be searched.</param> /// <param name="selectors">Compiled selectors that describe the nodes to find.</param> /// <returns>The matching nodes.</returns> public static IEnumerable <HtmlElementNode> Find(this IEnumerable <HtmlNode> nodes, SelectorCollection selectors) => selectors.Find(nodes);
/// <summary>
/// Searches the given nodes for ones matching the specified selector.
/// </summary>
/// <param name="nodes">The nodes to be searched.</param>
/// <param name="selector">Selector that describes the nodes to find.</param>
/// <returns>The matching nodes.</returns>
public static IEnumerable <HtmlElementNode> Find(IEnumerable <HtmlNode> nodes, string selector)
{
SelectorCollection selectors = Selector.ParseSelector(selector);
return(selectors.Find(nodes));
}
/// <summary>
/// Parses the given selector text and returns the corresponding data structures.
/// </summary>
/// <param name="selectorText">The selector text to be parsed.</param>
/// <remarks>
/// Returns multiple <see cref="Selector"/>s when the selector contains commas.
/// </remarks>
/// <returns>The parsed selector data structures.</returns>
public static SelectorCollection ParseSelector(string selectorText)
{
SelectorCollection selectors = new SelectorCollection();
if (!string.IsNullOrWhiteSpace(selectorText))
{
TextParser parser = new TextParser(selectorText);
parser.MovePastWhitespace();
while (!parser.EndOfText)
{
// Test next character
char ch = parser.Peek();
if (IsNameCharacter(ch) || ch == '*')
{
// Parse tag name
Selector selector = selectors.GetLast(true);
if (ch == '*')
{
selector.Tag = null; // Match all tags
}
else
{
selector.Tag = parser.ParseWhile(c => IsNameCharacter(c));
}
}
else if (SpecialCharacters.TryGetValue(ch, out string name))
{
// Parse special attributes
parser.MoveAhead();
string value = parser.ParseWhile(c => IsValueCharacter(c));
if (value.Length > 0)
{
SelectorAttribute attribute = new SelectorAttribute
{
Name = name,
Value = value,
Mode = SelectorAttributeMode.Contains
};
Selector selector = selectors.GetLast(true);
selector.Attributes.Add(attribute);
}
}
else if (ch == '[')
{
// Parse attribute selector
parser.MoveAhead();
parser.MovePastWhitespace();
name = parser.ParseWhile(c => IsNameCharacter(c));
if (name.Length > 0)
{
SelectorAttribute attribute = new SelectorAttribute
{
Name = name
};
// Parse attribute assignment operator
parser.MovePastWhitespace();
if (parser.Peek() == '=')
{
attribute.Mode = SelectorAttributeMode.Match;
parser.MoveAhead();
}
else if (parser.Peek() == ':' && parser.Peek(1) == '=')
{
attribute.Mode = SelectorAttributeMode.RegEx;
parser.MoveAhead(2);
}
else
{
attribute.Mode = SelectorAttributeMode.ExistsOnly;
}
// Parse attribute value
if (attribute.Mode != SelectorAttributeMode.ExistsOnly)
{
parser.MovePastWhitespace();
if (HtmlRules.IsQuoteChar(parser.Peek()))
{
attribute.Value = parser.ParseQuotedText();
}
else
{
attribute.Value = parser.ParseWhile(c => IsValueCharacter(c));
}
}
Selector selector = selectors.GetLast(true);
selector.Attributes.Add(attribute);
}
// Close out attribute selector
parser.MovePastWhitespace();
Debug.Assert(parser.Peek() == ']');
if (parser.Peek() == ']')
{
parser.MoveAhead();
}
}
else if (ch == ',')
{
// Multiple selectors
parser.MoveAhead();
parser.MovePastWhitespace();
selectors.Add(new Selector());
}
else if (ch == '>')
{
// Whitespace indicates child selector
parser.MoveAhead();
parser.MovePastWhitespace();
Debug.Assert(selectors.Any());
Selector selector = selectors.AddChildSelector();
selector.ImmediateChildOnly = true;
}
else if (char.IsWhiteSpace(ch))
{
// Handle whitespace
parser.MovePastWhitespace();
// ',' and '>' change meaning of whitespace
if (parser.Peek() != ',' && parser.Peek() != '>')
{
selectors.AddChildSelector();
}
}
else
{
// Unknown syntax
Debug.Assert(false);
parser.MoveAhead();
}
}
}
selectors.RemoveEmpty();
return(selectors);
}
