public ToxySpreadsheet LoadData(DataContext context)
{
ParserContext pc = new ParserContext(context.Path);
ISpreadsheetParser parser=ParserFactory.CreateSpreadsheet(pc);
var ss=parser.Parse();
return ss;
}
public async Task<bool> Process(ICrawler crawler, PropertyBag propertyBag)
{
if (propertyBag.StatusCode != HttpStatusCode.OK
|| propertyBag.Response == null)
{
return true;
}
string extension = MapContentTypeToExtension(propertyBag.ContentType);
if (extension.IsNullOrEmpty())
{
return true;
}
propertyBag.Title = propertyBag.Step.Uri.PathAndQuery;
using (TempFile temp = new TempFile())
{
temp.FileName += "." + extension;
using (FileStream fs = new FileStream(temp.FileName, FileMode.Create, FileAccess.Write, FileShare.Read, 0x1000))
{
await fs.WriteAsync(propertyBag.Response, 0, propertyBag.Response.Length);
}
ParserContext context = new ParserContext(temp.FileName);
ITextParser parser = ParserFactory.CreateText(context);
propertyBag.Text = parser.Parse();
}
return true;
}
static object CreateObject(ParserContext context, Type itype)
{
string ext = GetFileExtention(context.Path);
var types = parserMapping[ext];
object obj = null;
bool isFound = false;
foreach (Type type in types)
{
obj = Activator.CreateInstance(type, context);
if (itype.IsAssignableFrom(obj.GetType()))
{
isFound = true;
break;
}
}
if (!isFound)
throw new InvalidDataException(ext + " is not supported");
return obj;
}
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;
}
}
public static ISpreadsheetParser CreateSpreadsheet(ParserContext context)
{
object obj = CreateObject(context, typeof(ISpreadsheetParser));
ISpreadsheetParser parser = (ISpreadsheetParser)obj;
return parser;
}
public static ITextParser CreateText(ParserContext context)
{
object obj = CreateObject(context, typeof(ITextParser));
ITextParser parser = (ITextParser)obj;
return parser;
}
public static VCardParser CreateVCard(ParserContext context)
{
object obj = CreateObject(context, typeof(VCardParser));
VCardParser parser = (VCardParser)obj;
return parser;
}
public static IDomParser CreateDom(ParserContext context)
{
object obj = CreateObject(context, typeof(IDomParser));
IDomParser parser = (IDomParser)obj;
return parser;
}
private void OpenFile(string filepath, string encoding)
{
if (string.IsNullOrWhiteSpace(filepath))
{
tbPath.Clear();
return;
}
tbPath.Text = filepath;
FileInfo fi = new FileInfo(filepath);
ParserContext context = new ParserContext(filepath);
context.Encoding = Encoding.GetEncoding(encoding);
string extension = fi.Extension.ToLower();
tbExtension.Text = extension;
panel1.Visible = false;
switch (extension)
{
case ".txt":
case ".html":
case ".htm":
case ".pdf":
AppendRichTextBox();
var tparser = ParserFactory.CreateText(context);
richTextBox1.Text = tparser.Parse();
tbParserType.Text = tparser.GetType().Name;
break;
case ".rtf":
case ".docx":
AppendRichTextBox();
IDocumentParser docparser = ParserFactory.CreateDocument(context);
ToxyDocument doc = docparser.Parse();
tbParserType.Text = docparser.GetType().Name;
richTextBox1.Text = doc.ToString();
break;
case ".csv":
case ".xlsx":
case ".xls":
AppendSpreadsheetGrid();
ISpreadsheetParser ssparser = ParserFactory.CreateSpreadsheet(context);
ss = ssparser.Parse();
tbParserType.Text = ssparser.GetType().Name;
//DataSet ds = ss.ToDataSet();
//dataGridView1.DataSource = ds.Tables[0].DefaultView;
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;
dataGridView1.DataSource =vcards.ToDataTable().DefaultView;
break;
default:
AppendRichTextBox();
richTextBox1.Text = "Unknown document";
tbParserType.Text = "";
break;
}
}
private void openToolStripMenuItem_Click(object sender, EventArgs e)
{
OpenFileDialog dialog = new OpenFileDialog();
dialog.Multiselect = false;
dialog.Filter = "Business Card Files (*.vcf)|*.vcf";
if (dialog.ShowDialog() == System.Windows.Forms.DialogResult.OK)
{
string path= dialog.FileName;
ParserContext context=new ParserContext(path);
VCardParser vparser=ParserFactory.CreateVCard(context);
bcs = vparser.Parse();
FillListView(bcs, checkBox1.Checked, checkBox2.Checked);
}
}
public static ISlideshowParser CreateSlideshow(ParserContext context)
{
object obj = CreateObject(context, typeof(ISlideshowParser), "CreateSlideshow");
ISlideshowParser parser = (ISlideshowParser)obj;
return parser;
}
public static IMetadataParser CreateMetadata(ParserContext context)
{
object obj = CreateObject(context, typeof(IMetadataParser), "CreateMetadata");
IMetadataParser parser = (IMetadataParser)obj;
return parser;
}
public static IDocumentParser CreateDocument(ParserContext context)
{
object obj = CreateObject(context, typeof(IDocumentParser), "CreateDocument");
IDocumentParser parser = (IDocumentParser)obj;
return parser;
}
public static IEmailParser CreateEmail(ParserContext context)
{
object obj = CreateObject(context, typeof(IEmailParser), "CreateEmail");
IEmailParser parser = (IEmailParser)obj;
return parser;
}
public ExcelTableTemplate Parse(string filepath)
{
var templateMetadata = new TemplateMetaData();
if (filepath.StartsWith("/"))
filepath = HostingEnvironment.MapPath(filepath);
FilePath = filepath;
ParserContext context = new ParserContext(filepath);
ISpreadsheetParser parser = ParserFactory.CreateSpreadsheet(context);
ToxySpreadsheet ss = parser.Parse();
foreach (var table in ss.Tables)
{
_currTable = table;
for (_currRowIndex = 0; _currRowIndex < table.Rows.Count; _currRowIndex++)
{
var row = table.Rows[_currRowIndex];
for (_currColIndex = 0; _currColIndex < row.Cells.Count; _currColIndex++)
{
var cell = row.Cells[_currColIndex];
var tabletoken = ParseTable(cell);
if (tabletoken != null)
{
templateMetadata.Tables.Add(tabletoken);
}
else
{
var cellToken = ParseCell(cell);
if (cellToken != null)
{
templateMetadata.Cells.Add(cellToken);
}
}
}
}
}
return null;
}
