public void TestReadRTF_Html()
{
string path = TestDataSample.GetRTFPath("htmlrtf2.rtf");
var parser = new RTFTextParser(new ParserContext(path));
string result = parser.Parse();
Assert.IsNotNullOrEmpty(result);
}
public void TestReadRTF_FormattedText()
{
string path = TestDataSample.GetRTFPath("Formated text.rtf");
var parser = new RTFTextParser(new ParserContext(path));
string result = parser.Parse();
string[] lines = result.Replace("\r\n", "\n").Split('\n');
Assert.AreEqual(lines.Length, 11);
Assert.AreEqual("11111111111", lines[0]);
Assert.AreEqual("22222222222", lines[1]);
Assert.AreEqual("张三李四王五", lines[2]);
Assert.AreEqual("RTF Sample , Author : yuans , contact : [email protected] , site : http://www.cnblogs.com/xdesigner .", lines[7]);
}
public List <Attatchment> FetchAttachementsData()
{
string index = _source.index;
Elastic elastic = new Elastic(index);
List <Attatchment> lstAttachments = new List <Attatchment>();
int id = elastic.GetMaxId() + 1;
Console.WriteLine("===> Max id:" + id);
int maxRetries = _source.retries ?? 5;
WebClient wc = new WebClient();
bool end = false;
int retry = 0;
do
{
var tempFileName = Path.GetTempFileName();
try
{
string url = _source.url ?? "{0}";
url = string.Format(url, id++);
Console.WriteLine("=> Get " + url);
wc.DownloadFile(url, tempFileName);
var mimeType = wc.ResponseHeaders["content-type"];
Console.WriteLine("=> Mimetype " + mimeType);
var fileName = wc.ResponseHeaders["Content-Disposition"].Substring(wc.ResponseHeaders["Content-Disposition"].IndexOf("filename=") + 9).Replace("\"", "");
Console.WriteLine("=> Filename " + fileName);
if (string.IsNullOrEmpty(mimeType))
{
end = true;
}
var body = string.Empty;
if (fileName.ToLower().EndsWith(".pdf"))
{
var pdf = new PDFTextParser(new Toxy.ParserContext(tempFileName));
body = pdf.Parse();
}
else if (fileName.ToLower().EndsWith(".docx"))
{
System.Text.Encoding.RegisterProvider(System.Text.CodePagesEncodingProvider.Instance);
var docx = new Word2007TextParser(new Toxy.ParserContext(tempFileName));
body = docx.Parse();
}
else if (fileName.ToLower().EndsWith(".rtf"))
{
System.Text.Encoding.RegisterProvider(System.Text.CodePagesEncodingProvider.Instance);
var rtf = new RTFTextParser(new Toxy.ParserContext(tempFileName));
body = rtf.Parse();
}
else if (fileName.ToLower().EndsWith(".doc"))
{
System.Text.Encoding.RegisterProvider(System.Text.CodePagesEncodingProvider.Instance);
var doc = new Word2003TextParser(new Toxy.ParserContext(tempFileName));
body = doc.Parse();
}
if (!string.IsNullOrEmpty(body))
{
Attatchment attatch = new Attatchment(id.ToString(), fileName, url, mimeType, body, DateTime.Now);
elastic.SaveItem(attatch);
}
retry = 0;
//lstAttachments.Add(attatch);
}
catch (Exception ex)
{
Console.WriteLine(ex);
retry++;
}
try { File.Delete(tempFileName); }
catch { } // best effort
tempFileName = null;
} while (!end && retry < maxRetries);
return(lstAttachments);
}
