public ComprovantePDF(string path)
{
var parser = new PDFTextParser(new ParserContext(path));
Texto = parser.Parse();
try {
_textAsArray = Texto.Split(_newline).Select(r => r.Trim()).Where(r => !string.IsNullOrEmpty(r))
.ToArray();
if (_textAsArray.First() == "Comprovante")
{
LerTransferencia();
}
else
{
LerPagamento();
}
}
catch {
}
finally {
if (!DadosOK)
{
var data = System.IO.Path.GetFileName(path).Substring(0, 10);
var DataArquivo = DateTime.ParseExact(data, "yyyy-MM-dd",
CultureInfo.InvariantCulture);
Agendamento = Agendamento.Year > 2000 ? Agendamento : DataArquivo;
Pagamento = Pagamento.Year > 2000 ? Pagamento : DataArquivo;
Valor = Valor == 0 ? 1 : Valor;
}
}
}
public void TestReadBigPDFFile()
{
string path = TestDataSample.GetPdfPath("Word97-2007BinaryFileFormat(doc)Specification.pdf");
var parser = new PDFTextParser(new ParserContext(path));
string result = parser.Parse();
Assert.IsTrue(true);
}
public void TestParsePlainTextFromPDF()
{
string path = TestDataSample.GetPdfPath("Sample1.PDF");
var parser = new PDFTextParser(new ParserContext(path));
string result = parser.Parse();
Assert.IsTrue(result.StartsWith("LA MARCHE"));
}
private string GetTextFromPdf(string file_path)
{
var context = new ParserContext(file_path);
var parser = new PDFTextParser(context);
var content = parser.Parse();
return(content.Replace("\n", "\r\n"));
}
public void TestParsePlainTextFromSample5()
{
string path = TestDataSample.GetPdfPath("Sample5.PDF");
var parser = new PDFTextParser(new ParserContext(path));
string result = parser.Parse();
string[] results = result.Split('\n');
Assert.AreEqual("License income by market (%)", results[0]);
Assert.AreEqual("Philadelphia, Atlanta, Dallas, San Diego, and New", results[1]);
}
public void TestParsePlainTextFromSample1()
{
string path = TestDataSample.GetPdfPath("Sample1.PDF");
var parser = new PDFTextParser(new ParserContext(path));
string result = parser.Parse();
Assert.IsTrue(result.StartsWith("LA MARCHE"));
ContainText(result, "Toute discussion stratégique sur nos actions nécessite un rappel de ce que nous avons fait en");
ContainText(result, "l’an 2000 et depuis. Au niveau mondial, en l’an 2000, nous avons mené une campagne de");
ContainText(result, "Une structure pour nous amener à 2005");
ContainText(result, "Lors de la 4e rencontre qui aura lieu en Inde, nous avons deux objectifs majeurs");
}
static void FunWithCraftObjects()
{
string dirPath = @"E:\Documents\Tabletop RPGs\Numenera\APPs\";
string filename = dirPath + "TEST_Installations.txt";
string testfilename = dirPath + "TEST_Installations.txt";
string keywordsFile = dirPath + "KEYWORDS_Craft_Objects.txt";
// first - create Info
var info = new PDFTextFileInfo
{
ItemsFileName = testfilename,
TableKeyword = ""
};
PDFTextParser.LoadKeywordsFromFile(info, keywordsFile);
var lines = File.ReadAllLines(testfilename);
// second - get string objects
var parsedObjects = PDFTextParser.SplitItemsToObjects(info, lines);
File.Delete(dirPath + "TEST_OUTPUT_InBetween.txt");
foreach (var obj in parsedObjects)
{
File.AppendAllLines(dirPath + "TEST_OUTPUT_InBetween.txt", obj);
}
// create Dictionary
var dic = PDFTextParser.CreateDictionariesFromObjects(parsedObjects, info.KeywordsList);
// and make XML from it
var xmlInfo = new PDFTextXmlInfo()
{
XmlFileName = dirPath + "TEST_OUTPUT_Installations.xml",
ObjectsName = "CraftObjects",
ObjectName = "CraftObject",
Source = "Destiny"
};
PDFTextXmlCreator.CreateXML(xmlInfo, dic, info.TableKeyword);
}
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);
}
