Esempio n. 1
0
        public void TestParseHtml()
        {
            string path = Path.GetFullPath(TestDataSample.GetHtmlPath("mshome.html"));

            ParserContext context = new ParserContext(path);
            IDomParser    parser  = (IDomParser)ParserFactory.CreateDom(context);
            ToxyDom       toxyDom = parser.Parse();

            List <ToxyNode> metaNodeList = toxyDom.Root.SelectNodes("//meta");

            Assert.AreEqual(7, metaNodeList.Count);

            ToxyNode aNode = toxyDom.Root.SingleSelect("//a");

            Assert.AreEqual(1, aNode.Attributes.Count);
            Assert.AreEqual("href", aNode.Attributes[0].Name);
            Assert.AreEqual("http://www.microsoft.com/en/us/default.aspx?redir=true", aNode.Attributes[0].Value);

            ToxyNode titleNode = toxyDom.Root.ChildrenNodes[0].ChildrenNodes[0].ChildrenNodes[0];

            Assert.AreEqual("title", titleNode.Name);
            Assert.AreEqual("Microsoft Corporation", titleNode.ChildrenNodes[0].InnerText);

            ToxyNode metaNode = toxyDom.Root.ChildrenNodes[0].ChildrenNodes[0].ChildrenNodes[7];

            Assert.AreEqual("meta", metaNode.Name);
            Assert.AreEqual(3, metaNode.Attributes.Count);
            Assert.AreEqual("name", metaNode.Attributes[0].Name);
            Assert.AreEqual("SearchDescription", metaNode.Attributes[0].Value);
            Assert.AreEqual("scheme", metaNode.Attributes[2].Name);
            Assert.AreEqual(string.Empty, metaNode.Attributes[2].Value);
        }
Esempio n. 2
0
        public ToxyDom Parse()
        {
            if (!File.Exists(Context.Path))
                throw new FileNotFoundException("File " + Context.Path + " is not found");

            XmlDocument doc = new XmlDocument();
            doc.Load(Context.Path);

            ToxyNode rootNode = ConvertToToxyNode(doc.DocumentElement);
            ToxyDom dom = new ToxyDom();
            dom.Root = rootNode;
            AppendChildren(rootNode, doc.DocumentElement);
            return dom;
        }
Esempio n. 3
0
        public ToxyDom Parse()
        {
            if (!File.Exists(Context.Path))
            {
                throw new FileNotFoundException("File " + Context.Path + " is not found");
            }

            XmlDocument doc = new XmlDocument();

            doc.Load(Context.Path);

            ToxyNode rootNode = ConvertToToxyNode(doc.DocumentElement);
            ToxyDom  dom      = new ToxyDom();

            dom.Root = rootNode;
            AppendChildren(rootNode, doc.DocumentElement);
            return(dom);
        }
Esempio n. 4
0
        private void ShowDocument(string filepath, string encoding, string extension)
        {
            ParserContext context = new ParserContext(filepath);

            context.Encoding = Encoding.GetEncoding(encoding);

            if (Mode == ViewMode.Text)
            {
                AppendRichTextBox();
                var tparser = ParserFactory.CreateText(context);
                rtbPanel.Text     = tparser.Parse();
                tbParserType.Text = tparser.GetType().Name;
            }
            else if (Mode == ViewMode.Structured)
            {
                switch (extension)
                {
                case ".csv":
                    AppendSpreadsheetGrid();
                    context.Properties.Add("HasHeader", "1");
                    ISpreadsheetParser csvparser = ParserFactory.CreateSpreadsheet(context);
                    ss = csvparser.Parse();
                    tbParserType.Text = csvparser.GetType().Name;
                    var table1 = ss.Tables[0];
                    ShowToGrid(table1);
                    cbSheets.Items.Clear();
                    foreach (var table in ss.Tables)
                    {
                        cbSheets.Items.Add(table.Name);
                    }
                    cbSheets.SelectedIndex = 0;
                    panel1.Visible         = true;
                    break;

                case ".xlsx":
                case ".xls":
                    AppendSpreadsheetGrid();
                    ISpreadsheetParser ssparser = ParserFactory.CreateSpreadsheet(context);
                    ss = ssparser.Parse();
                    tbParserType.Text = ssparser.GetType().Name;
                    var table0 = ss.Tables[0];
                    ShowToGrid(table0);
                    cbSheets.Items.Clear();
                    foreach (var table in ss.Tables)
                    {
                        cbSheets.Items.Add(table.Name);
                    }
                    cbSheets.SelectedIndex = 0;
                    panel1.Visible         = true;
                    break;

                case ".vcf":
                    AppendDataGridView();
                    var vparser = ParserFactory.CreateVCard(context);
                    ToxyBusinessCards vcards = vparser.Parse();
                    tbParserType.Text             = vparser.GetType().Name;
                    gridPanel.GridView.DataSource = vcards.ToDataTable().DefaultView;
                    break;

                case ".pptx":
                    //TODO: show slides
                    break;

                case ".xml":
                case ".htm":
                case ".html":
                    AppendTreePanel();
                    var      domparser = ParserFactory.CreateDom(context);
                    ToxyDom  htmlDom   = domparser.Parse();
                    TreeNode rootNode  = treePanel.Tree.Nodes.Add(htmlDom.Root.NodeString);
                    treePanel.Tree.BeginUpdate();
                    AppendTree(rootNode, htmlDom.Root);
                    treePanel.Tree.EndUpdate();
                    //rootNode.ExpandAll();
                    break;
                }
            }
            else
            {
                AppendPropertyListPanel();
                var          tparser   = ParserFactory.CreateMetadata(context);
                ToxyMetadata metadatas = tparser.Parse();
                plPanel.Clear();
                foreach (var data in metadatas)
                {
                    plPanel.AddItem(data.Name, data.Value.ToString());
                }
                tbParserType.Text = tparser.GetType().Name;
            }
        }