public void GetIndices()
{
HtmlPath path = HtmlPath.Parse("/BODY[0]/DIV[5]/DIV[0]/DIV[1]/TABLE[7]/TBODY[0]/TR[6]/TD[1]");
var e = myDocument.GetElementByPath(path);
Assert.AreEqual(1, HtmlTable.GetColumnIndex(e));
Assert.AreEqual(6, HtmlTable.GetRowIndex(e));
Assert.AreEqual(e, HtmlTable.GetEmbeddingTD(e));
Assert.AreEqual(e.Parent, HtmlTable.GetEmbeddingTR(e));
}
/// <summary>
/// Gets the HtmlTable the given path is pointing to.
/// If the path is pointing into a table, the embedding table is returned.
/// If the path is not pointing to a table element null is returned.
/// </summary>
public static HtmlTable GetByPath(IHtmlDocument doc, HtmlPath path)
{
var start = doc.GetElementByPath(path);
if (start == null)
{
return(null);
}
return(GetByElement(start));
}
/// <summary>
/// Gets the HtmlTable the given path is pointing to.
/// If the path is pointing into a table, the embedding table is returned.
/// If the path is not pointing to a table element null is returned.
/// </summary>
public static HtmlTable GetTableByPath(this IHtmlDocument doc, HtmlPath path)
{
var start = doc.GetElementByPath(path);
if (start == null)
{
return(null);
}
return(start.FindEmbeddingTable());
}
/// <summary>
/// Gets the text of the element specified by the given <see cref="HtmlPath"/>.
/// </summary>
public static string GetTextByPath(this IHtmlDocument doc, HtmlPath path)
{
var e = doc.GetElementByPath(path);
if (e == null)
{
return(null);
}
return(e.InnerText);
}
public DataTable ExtractTable()
{
var pathSeriesDescriptor = myDescriptor as PathSeriesDescriptor;
if (pathSeriesDescriptor != null)
{
var table = ExtractTable(myDocument, HtmlPath.Parse(pathSeriesDescriptor.Path));
return(TableFormatter.ToFormattedTable(pathSeriesDescriptor, table));
}
var pathTableDescriptor = myDescriptor as PathTableDescriptor;
if (pathTableDescriptor != null)
{
var table = ExtractTable(myDocument, HtmlPath.Parse(pathTableDescriptor.Path));
return(TableFormatter.ToFormattedTable(pathTableDescriptor, table));
}
var pathCellDescriptor = myDescriptor as PathCellDescriptor;
if (pathCellDescriptor != null)
{
var table = ExtractTable(myDocument, HtmlPath.Parse(pathCellDescriptor.Path));
var value = TableFormatter.GetValue(pathCellDescriptor, table);
// XXX: this is really ugly - i have to create a table just to satisfy the interface :(
return(CreateTableForScalar(pathCellDescriptor.ValueFormat.Type, value));
}
var pathSingleValueDescriptor = myDescriptor as PathSingleValueDescriptor;
if (pathSingleValueDescriptor != null)
{
var e = myDocument.GetElementByPath(HtmlPath.Parse(pathSingleValueDescriptor.Path));
var str = e == null ? null : e.InnerText;
var value = pathSingleValueDescriptor.ValueFormat.Convert(str);
// XXX: this is really ugly - i have to create a table just to satisfy the interface :(
return(CreateTableForScalar(pathSingleValueDescriptor.ValueFormat.Type, value));
}
throw new NotSupportedException("Format not supported for Html documents: " + myDescriptor.GetType());
}
public void GetRoot()
{
HtmlPath path = HtmlPath.Parse("/BODY[0]/DIV[5]/DIV[0]/DIV[1]/TABLE[7]/TBODY[0]/TR[6]/TD[1]");
var e = myDocument.GetElementByPath(path);
Assert.AreEqual("TD", e.TagName);
Assert.AreEqual("HTML", e.GetRoot().TagName);
}
