Html To Excel .NET converter - C# / VB.NET Sample Codes

Imports DTG.Html

Module Module1

Dim linktext As String

Sub Main()
        ' Parse the html file
        Dim html As HtmlDoc = HtmlDoc.ParseHTML("../../../htmlparser.html")

' The function seeks for links
        ScanHtmlTree("a", html.Nodes)
        Console.ReadLine()
    End Sub

Sub ScanHtmlTree(ByVal TagName As String, ByVal nodes As HtmlNodeCollection)
        For Each node As HtmlNode In nodes
            If node.IsElement Then

Dim tag As HtmlTag = CType(node, HtmlTag)
                If tag.Name = TagName Then
                    If tag.Attributes.IndexOf("href") >= 0 Then

Console.WriteLine("Link: " + tag.Attributes("href").Value)
                        linktext = ""
                        GetText(tag.Nodes)
                        Console.WriteLine("Text: " + linktext)
                        Console.WriteLine()

End If
                End If

ScanHtmlTree(TagName, tag.Nodes)
            End If
        Next
    End Sub

Sub GetText(ByVal nodes As HtmlNodeCollection)
        For Each node As HtmlNode In nodes
            If node.IsElement Then
                Dim tag As HtmlTag = CType(node, HtmlTag)
                GetText(tag.Nodes)
            Else
                linktext += node.ToString()
            End If
        Next
    End Sub

End Module