public void readfrompdffile()
{
string clearedfilepath = requiredPath.Replace("file:\\", "").Replace("\\", "/");
string file = clearedfilepath + "/terms-and-conditions.pdf";
string outFile = clearedfilepath + "/terms-and-conditions.txt";
try
{
if (!File.Exists(file))
{
file = Path.GetFullPath(file);
if (!File.Exists(file))
{
Console.WriteLine("Please give in the path to the PDF file.");
}
}
PDFParser pdfParser = new PDFParser();
pdfParser.ExtractText(file, outFile);
}
catch (Exception exc)
{
Console.WriteLine(exc);
}
}
public ActionResult callExtraxtText(int serviceId)
{
Service service = db.Services.Find(serviceId);
if (service != null)
{
if (service.PDF != null)
{
string inFileName = service.PDF.PDFPath;
if (inFileName != null)
{
inFileName = ConfigurationManager.AppSettings["pdfArchive"] + @"\PDFs\" + inFileName;
PDFParser reader = new PDFParser();
string text = reader.ExtractText(serviceId, inFileName);
PDF pdf = service.PDF;
if (text != "false")
{
pdf.FacebookDescription = text;
pdf.FacebookTitle = service.FirstName + " " + service.LastName + "'s Memorial Folder";
db.SaveChanges();
return(Json("success"));
}
}
}
}
return(Json("Error"));
}
public static void ParsePdf()
{
PDFParser pdfParser = new PDFParser();
string address = @"http://ec.europa.eu/health/documents/community-register/html/h_direct_anx.htm#412_et";
string outfile = "outfile.txt";
bool result = pdfParser.ExtractText(address, outfile);
}
static void Main(string[] args)
{
PDFParser pdfParser = new PDFParser();
pdfParser.ExtractText(@"C:\Users\unknown\Desktop\oyak.pdf",
@"C:\Users\unknown\Desktop\output.txt");
Console.ReadKey();
}
public static string DownloadFile(string url)
{
string fileLocation = ConfigurationManager.AppSettings["downloadLocation"];
string fileName = DateTime.Now.Hour.ToString() + DateTime.Now.Minute.ToString() + DateTime.Now.Second.ToString() + DateTime.Now.Millisecond.ToString();
string fullFileName = fileLocation + @"\" + fileName + ".pdf";
string result = "";
try
{
using (WebClient client = new WebClient())
{
client.DownloadFile(url, fullFileName);
}
PDFParser pdfParser = new PDFParser();
pdfParser.ExtractText(fullFileName, System.IO.Path.GetFileNameWithoutExtension(fullFileName) + ".txt");
try
{
//using (StreamReader sr = new StreamReader(System.IO.Path.GetFileNameWithoutExtension(fullFileName) + ".txt"))
//{
// // Read the stream to a string, and write the string to the console.
// result = sr.ReadToEnd();
//}
//string output= System.IO.Path.GetFileNameWithoutExtension(fullFileName) + ".txt";
// var bytes = File.ReadAllBytes(fullFileName);
// File.WriteAllText(output, ConvertToText(bytes), Encoding.UTF8);
var bytes = File.ReadAllBytes(fullFileName);
result = ConvertToText(bytes);
}
catch (Exception ex)
{
ErrorUtil.logError(ex, "");
}
}
catch (Exception ex)
{
ErrorUtil.logError(ex, "");
}
return(result);
}
public void convertFile(string fileToConvert)
{
O2Thread.mtaThread(
() => {
//fileToConvert.error();
if (fileToConvert.fileExists())
{
if (fileToConvert.extension(".pdf"))
{
textBox.set_Text("...processing pdf file: " + fileToConvert);
var pdfParser = new PDFParser();
var tempFile = PublicDI.config.getTempFileInTempDirectory(".txt");
pdfParser.ExtractText(fileToConvert, tempFile);
textBox.set_Text(tempFile.contents().fixCRLF());
Files.deleteFile(tempFile);
}
else
{
textBox.set_Text(fileToConvert.contents());
}
}
});
}
static void Main(string[] args)
{
ITextParse.ExtractText("employe-1.pdf", "iparse.txt");
//using (PdfReader reader = new PdfReader("letter.pdf")) //Index was outside the bounds of the array.
//using (PdfReader reader = new PdfReader("employe-1.pdf")) //ok
//using (PdfReader reader = new PdfReader("feuille_de_paie.pdf")) //Rebuild failed: trailer not found.; Original message: PDF startxref not found.
using (iTextSharp.text.pdf.PdfReader reader = new iTextSharp.text.pdf.PdfReader("modele-bulletin-de-salaire.pdf")) //ok
{
StringBuilder text = new StringBuilder();
for (int page = 1; page <= reader.NumberOfPages; page++)
{
iTextSharp.text.pdf.parser.ITextExtractionStrategy strategy = new iTextSharp.text.pdf.parser.SimpleTextExtractionStrategy();
//ITextExtractionStrategy strategy = new LocationTextExtractionStrategy();
string currentText = iTextSharp.text.pdf.parser.PdfTextExtractor.GetTextFromPage(reader, page, strategy);
currentText = Encoding.UTF8.GetString(ASCIIEncoding.Convert(Encoding.Default, Encoding.UTF8, Encoding.Default.GetBytes(currentText)));
text.Append(currentText);
}
System.IO.StreamWriter file = new System.IO.StreamWriter("itextsharp.txt");
file.WriteLine(text);
file.Close();
//return text.ToString();
}
//PdfSharp
using (var _document = PdfReader.Open("letter.pdf", PdfDocumentOpenMode.ReadOnly)) //ok
//using (var _document = PdfReader.Open("employe-1.pdf", PdfDocumentOpenMode.ReadOnly)) //ok
//using (var _document = PdfReader.Open("feuille_de_paie.pdf", PdfDocumentOpenMode.ReadOnly)) //Non-negative number required.
//using (var _document = PdfReader.Open("modele-bulletin-de-salaire.pdf", PdfDocumentOpenMode.ReadOnly)) //ok
{
System.IO.StreamWriter file = new System.IO.StreamWriter("pdfsharp.txt");
foreach (PdfPage page in _document.Pages)
{
var text = ExtractText(page);
foreach (string s in text)
{
file.Write(s);
}
}
file.Close();
}
//iTextSharp too
PDFParser parser = new PDFParser();
//parser.ExtractText("letter.pdf", "pdfparser.txt"); //error
parser.ExtractText("employe-1.pdf", "pdfparser.txt"); //ok
//parser.ExtractText("feuille_de_paie.pdf", "pdfparser.txt"); //error
//parser.ExtractText("modele-bulletin-de-salaire.pdf", "pdfparser.txt"); //error
//PdfSharp
//string text2 = PdfTextExtractor.GetText("letter.pdf"); //ok
//string text2 = PdfTextExtractor.GetText("employe-1.pdf"); //ok
//string text2 = PdfTextExtractor.GetText("feuille_de_paie.pdf"); //Non-negative number required.
string text2 = PdfTextExtractor.GetText("modele-bulletin-de-salaire.pdf"); //ok
System.IO.StreamWriter file2 = new System.IO.StreamWriter("PdfTextExtractor.txt");
file2.Write(text2);
file2.Close();
}
/// <summary>
/// Get specify file PDF content
/// </summary>
/// <param name="filePath"></param>
/// <returns></returns>
public static string GetPdfContent(string filePath)
{
PDFParser pdfParser = new PDFParser();
return(pdfParser.ExtractText(filePath));;
}
// Insert logic for processing found files here.
public static void ProcessFile(string path, string tipo, string dpto, bool actualizar)
{
bool indexatexto = false, indexaaudio = false, indexaimagen = false, indexahipertexto = false, indexavideo = false;
Random rnd = new Random();
switch (tipo)
{
case "texto":
indexatexto = true;
break;
case "hipertexto":
indexahipertexto = true;
break;
case "video":
indexavideo = true;
break;
case "imagen":
indexaimagen = true;
break;
case "audio":
indexaaudio = true;
break;
case "":
indexatexto = true;
indexahipertexto = true;
indexavideo = true;
indexaimagen = true;
indexaaudio = true;
break;
}
if (Herramientas.EsHiperTexto(path) && indexatexto && path.Contains(dpto))
{
Regex trimmer = new Regex(@"\s\s+");
ScrapingBrowser Browser = new ScrapingBrowser();
Browser.AllowAutoRedirect = true; // Browser has settings you can access in setup
Browser.AllowMetaRedirect = true;
HtmlNode html = GetNodes(new Uri(path));
var titulo = html.CssSelect("title").FirstOrDefault().InnerText;
var body = html.CssSelect("body").FirstOrDefault().InnerText;
body = Regex.Replace(body, "<.*?>", string.Empty);
body = Regex.Replace(body, @"(?:(?:\r?\n)+ +){2,}", @"\n");
var f = new FileInfo(path);
var fileLengthInKB = f.Length / 1024.0;
Hipertexto h = new Hipertexto();
h.nombreArchivo = titulo;
h.textoContenido = body;
h.tamanoArchivo = fileLengthInKB;
Uri u = new Uri(path);
string ext = System.IO.Path.GetExtension(path);
string auxiliar = "http://localhost/servidores/";
h.urlRuta = u.AbsoluteUri;
var puerto = u.Port;
int pos = h.urlRuta.IndexOf("servidores/") + 11;
string aux2 = h.urlRuta.Substring(pos);
int pos2 = aux2.IndexOf("/");
h.urlRuta = auxiliar + aux2;
string servidor = aux2.Substring(0, pos2);
int pos3 = aux2.IndexOf(servidor + "/") + servidor.Length + 1;
string depart = aux2.Substring(pos3);
int pos4 = depart.IndexOf("/");
depart = depart.Substring(0, pos4);
depart = depart.Replace("%20", " ");
h.departamento = depart;
h.urlRuta = auxiliar + aux2;
h.tamanoArchivo = fileLengthInKB;
h.formato = ext;
h.idServidor = servidor;
h.nombreArchivo = System.IO.Path.GetFileName(path);
h.fechaCreacionArchivo = (DateTime)File.GetCreationTime(path);
h.fechaModificacionArchivo = (DateTime)File.GetLastWriteTime(path);
h.fechaUltimaLectura = (DateTime)File.GetLastAccessTime(path);
h.fechaUltimaActualizacion = DateTime.Now;
h.hits = rnd.Next(1, 55);
if (File.Exists(path))
{
h.estadoActividad = 1;
}
Hipertexto existente = OperacionesElasticSearch.ExisteHipertexto(h);
if (existente == null)
{
OperacionesElasticSearch.InsertarHiperTexto(h);
}
else
{
OperacionesElasticSearch.actualizarHipertexto(existente, h);
}
}
else if (path.EndsWith(".txt") && indexatexto && path.Contains(dpto))
{
var f = new FileInfo(path);
var fileLengthInKB = f.Length / 1024.0;
Uri u = new Uri(path);
Texto t = new Texto();
string ext = System.IO.Path.GetExtension(path);
string auxiliar = "http://localhost/servidores/";
t.urlRuta = u.AbsoluteUri;
var puerto = u.Port;
int pos = t.urlRuta.IndexOf("servidores/") + 11;
string aux2 = t.urlRuta.Substring(pos);
t.urlRuta = auxiliar + aux2;
int pos2 = aux2.IndexOf("/");
string servidor = aux2.Substring(0, pos2);
int pos3 = aux2.IndexOf(servidor + "/") + servidor.Length + 1;
string depart = aux2.Substring(pos3);
int pos4 = depart.IndexOf("/");
depart = depart.Substring(0, pos4);
depart = depart.Replace("%20", " ");
string textoContenido = System.IO.File.ReadAllText(path);
string user = System.IO.File.GetAccessControl(path).GetOwner(typeof(System.Security.Principal.NTAccount)).ToString();
t.estadoActividad = 1;
t.departamento = depart;
t.urlRuta = auxiliar + aux2;
t.tamanoArchivo = fileLengthInKB;
t.idServidor = servidor;
t.textoContenido = textoContenido;
t.titulo = t.nombreArchivo;
t.formato = ext;
t.nombreArchivo = System.IO.Path.GetFileName(path);
t.fechaCreacionArchivo = (DateTime)File.GetCreationTime(path);
t.fechaModificacionArchivo = (DateTime)File.GetLastWriteTime(path);
t.fechaUltimaLectura = (DateTime)File.GetLastAccessTime(path);
t.fechaUltimaActualizacion = DateTime.Now;
t.hits = rnd.Next(1, 55);
t.autorArchivo = user;
Texto existente = OperacionesElasticSearch.ExisteTexto(t);
if (existente == null)
{
OperacionesElasticSearch.InsertarTexto(t);
}
else
{
OperacionesElasticSearch.actualizarTexto(existente, t);
}
}
else if (Herramientas.EsWord(path) && indexatexto && path.Contains(dpto))
{
var f = new FileInfo(path);
var fileLengthInKB = f.Length / 1024.0;
var applicationWord = new Microsoft.Office.Interop.Word.Application();
applicationWord.Visible = false;
Word.Document w = applicationWord.Documents.Open(@path, ReadOnly: true);
Word.Range ContentTypeProperties = w.Content;
Microsoft.Office.Interop.Word.Application word = new Microsoft.Office.Interop.Word.Application();
object miss = System.Reflection.Missing.Value;
object readOnly = true;
Microsoft.Office.Interop.Word.Document docs = word.Documents.Open(path, ref miss, ref readOnly, ref miss, ref miss, ref miss, ref miss, ref miss, ref miss, ref miss, ref miss, ref miss, ref miss, ref miss, ref miss, ref miss);
//Get Author Name
object wordProperties = docs.BuiltInDocumentProperties;
Type typeDocBuiltInProps = wordProperties.GetType();
Object Authorprop = typeDocBuiltInProps.InvokeMember("Item", System.Reflection.BindingFlags.Default | System.Reflection.BindingFlags.GetProperty, null, wordProperties, new object[] { "Author" });//query for author properties
Type typeAuthorprop = Authorprop.GetType();
//string strAuthor = typeAuthorprop.InvokeMember("Value", System.Reflection.BindingFlags.Default | System.Reflection.BindingFlags.GetProperty, null, Authorprop, new object[] { }).ToString();//get author name
string textoContenido = "";
for (int i = 0; i < docs.Paragraphs.Count; i++)
{
textoContenido += " \r\n " + docs.Paragraphs[i + 1].Range.Text.ToString();
}
Uri u = new Uri(path);
Texto t = new Texto();
string ext = System.IO.Path.GetExtension(path);
string auxiliar = "http://localhost/servidores/";
t.urlRuta = u.AbsoluteUri;
var puerto = u.Port;
int pos = t.urlRuta.IndexOf("servidores/") + 11;
string aux2 = t.urlRuta.Substring(pos);
t.urlRuta = auxiliar + aux2;
int pos2 = aux2.IndexOf("/");
string servidor = aux2.Substring(0, pos2);
int pos3 = aux2.IndexOf(servidor + "/") + servidor.Length + 1;
string depart = aux2.Substring(pos3);
int pos4 = depart.IndexOf("/");
depart = depart.Substring(0, pos4);
depart = depart.Replace("%20", " ");
t.departamento = depart;
t.urlRuta = auxiliar + aux2;
t.tamanoArchivo = fileLengthInKB;
t.idServidor = servidor;
t.textoContenido = textoContenido;
t.titulo = t.nombreArchivo;
t.formato = ext;
t.nombreArchivo = System.IO.Path.GetFileName(path);
t.fechaCreacionArchivo = (DateTime)File.GetCreationTime(path);
t.fechaModificacionArchivo = (DateTime)File.GetLastWriteTime(path);
t.fechaUltimaLectura = (DateTime)File.GetLastAccessTime(path);
t.fechaUltimaActualizacion = DateTime.Now;
t.hits = rnd.Next(1, 55);
if (File.Exists(path))
{
t.estadoActividad = 1;
}
w.Close();
Texto existente = OperacionesElasticSearch.ExisteTexto(t);
if (existente == null)
{
OperacionesElasticSearch.InsertarTexto(t);
}
else
{
OperacionesElasticSearch.actualizarTexto(existente, t);
}
}
else if (Herramientas.EsExcel(path) && indexatexto && path.Contains(dpto))
{
/*Microsoft.Office.Interop.Excel.Application app = new Microsoft.Office.Interop.Excel.Application();
* Microsoft.Office.Interop.Excel.Workbook wb = app.Workbooks.Open(@path, ReadOnly: true);
*
* var f = new FileInfo(path);
* var fileLengthInKB = f.Length / 1024.0;
*
* //Create COM Objects. Create a COM object for everything that is referenced
* Excel.Application xlApp = new Excel.Application();
* Excel.Workbook xlWorkbook = xlApp.Workbooks.Open(path);
* Excel._Worksheet xlWorksheet = xlWorkbook.Sheets[0];
* Excel.Range xlRange = xlWorksheet.UsedRange;
* int rowCount = xlRange.Rows.Count;
* int colCount = xlRange.Columns.Count;
* //iterate over the rows and columns and print to the console as it appears in the file
* //excel is not zero based!!
* //Get Author Name
* String autor = wb.Author;
*
* String textoContenido = "";
* for (int i = 1; i <= rowCount; i++)
* {
* for (int j = 1; j <= colCount; j++)
* {
* //new line
* if (j == 1)
* Console.Write("\r\n");
*
* //write the value to the console
* if ((Excel.Range)xlRange.Cells[i, j] != null && xlRange.Cells[i, j].Value2 != null)
* textoContenido += xlRange.Cells[i, j].Value2.ToString() + " ";
*
* //add useful things here!
* }
* }
*
* xlWorkbook.Close();
* Uri u = new Uri(path);
*
* Texto t = new Texto();
* t.urlRuta = u.AbsoluteUri;
* string ext = System.IO.Path.GetExtension(path);
*
* string auxiliar = "http://localhost/servidorIntranet/";
* h.urlRuta = u.AbsoluteUri;
* var puerto = u.Port;
* int pos = h.urlRuta.IndexOf("servidores/") + 11 ;
* string aux2 = h.urlRuta.Substring(pos);
* t.urlRuta = auxiliar + aux2;
* int pos2 = aux2.IndexOf("/");
* string servidor = aux2.Substring(0, pos2);
* int pos3 = aux2.IndexOf(servidor + "/") + servidor.Length + 1;
*
* string depart = aux2.Substring(pos3);
* int pos4 = depart.IndexOf("/");
* depart = depart.Substring(0, pos4);
* depart = depart.Replace("%20", " ");
*
* string depart = aux2.Substring(pos3);
* int pos4 = depart.IndexOf("/");
* depart = depart.Substring(0, pos4);
* depart = depart.Replace("%20", " ");
* t.departamento = depart;
*
* t.idServidor = servidor;
* t.urlRuta = u.AbsoluteUri;
* t.tamanoArchivo = fileLengthInKB;
* t.textoContenido = textoContenido;
* t.titulo = t.nombreArchivo;
* t.formato = ext;
* t.nombreArchivo = System.IO.Path.GetFileName(path);
* t.fechaCreacionArchivo = (DateTime)File.GetCreationTime(path);
* t.fechaModificacionArchivo = (DateTime)File.GetLastWriteTime(path);
* t.fechaUltimaLectura = (DateTime)File.GetLastAccessTime(path);
* t.fechaUltimaActualizacion = DateTime.Now;
* t.hits = 0;
*
* if (File.Exists(path))
* t.estadoActividad = 1;
*
* Texto existente = OperacionesElasticSearch.ExisteTexto(t);
* if (existente == null)
* OperacionesElasticSearch.InsertarTexto(t);
* else
* OperacionesElasticSearch.actualizarTexto(existente, t);
*
*/
}
else if (Herramientas.EsPDF(path) && indexatexto && path.Contains(dpto))
{
var text = new TextExtractor().Extract(path).Text;
text = Regex.Replace(text, @"\s+", " ");
text = text.Replace("\r", "");
text = text.Replace("\n", "");
PDFParser pdfParser = new PDFParser();
var f = new FileInfo(path);
var fileLengthInKB = f.Length / 1024.0;
// extract the text
String resultado = "";
pdfParser.ExtractText(path, "C:\\Users\\cesar\\Desktop\\DocumentosIndeaxar\\salida.txt");
resultado = pdfParser.ToString();
String autor = "";
String textoContenido = "";
String titulo = "";
using (PdfReader reader = new PdfReader(path)){
//titulo = reader.Info["Title"];
//String ayt = reader.Info["Author"];
titulo = "";
StringBuilder text2 = new StringBuilder();
for (int i = 1; i <= reader.NumberOfPages; i++)
{
text2.Append(PdfTextExtractor.GetTextFromPage(reader, i));
}
textoContenido = text.ToString();
}
Texto t = new Texto();
Uri u = new Uri(path);
t.urlRuta = u.AbsoluteUri;
string auxiliar = "http://localhost/servidores/";
t.urlRuta = u.AbsoluteUri;
var puerto = u.Port;
int pos = t.urlRuta.IndexOf("servidores/") + 11;
string aux2 = t.urlRuta.Substring(pos);
t.urlRuta = auxiliar + aux2;
int pos2 = aux2.IndexOf("/");
string servidor = aux2.Substring(0, pos2);
int pos3 = aux2.IndexOf(servidor + "/") + servidor.Length + 1;
string depart = aux2.Substring(pos3);
int pos4 = depart.IndexOf("/");
depart = depart.Substring(0, pos4);
depart = depart.Replace("%20", " ");
t.idServidor = servidor;
t.departamento = depart;
t.textoContenido = textoContenido;
t.nombreArchivo = path.Substring(0, path.IndexOf(".pdf"));
t.titulo = titulo;
t.tamanoArchivo = fileLengthInKB;
string ext = System.IO.Path.GetExtension(path);
t.formato = ext;
t.nombreArchivo = path.Substring(0, path.IndexOf(ext));
t.nombreArchivo = System.IO.Path.GetFileName(path);
t.fechaCreacionArchivo = (DateTime)File.GetCreationTime(path);
t.fechaModificacionArchivo = (DateTime)File.GetLastWriteTime(path);
t.fechaUltimaLectura = (DateTime)File.GetLastAccessTime(path);
t.fechaUltimaActualizacion = DateTime.Now;
t.hits = rnd.Next(1, 55);
if (File.Exists(path))
{
t.estadoActividad = 1;
}
Texto existente = OperacionesElasticSearch.ExisteTexto(t);
if (existente == null)
{
OperacionesElasticSearch.InsertarTexto(t);
}
else
{
OperacionesElasticSearch.actualizarTexto(existente, t);
}
}
else if (Herramientas.EsPowerPoint(path) && indexatexto && path.Contains(dpto))
{
Microsoft.Office.Interop.PowerPoint.Application PowerPoint_App = new Microsoft.Office.Interop.PowerPoint.Application();
Microsoft.Office.Interop.PowerPoint.Presentations multi_presentations = PowerPoint_App.Presentations;
Microsoft.Office.Interop.PowerPoint.Presentation presentation = multi_presentations.Open(path);
var f = new FileInfo(path);
var fileLengthInKB = f.Length / 1024.0;
string textoContenido = "";
for (int i = 0; i < presentation.Slides.Count; i++)
{
foreach (var item in presentation.Slides[i + 1].Shapes)
{
var shape = (Powerpoint.Shape)item;
if (shape.HasTextFrame == MsoTriState.msoTrue)
{
if (shape.TextFrame.HasText == MsoTriState.msoTrue)
{
var textRange = shape.TextFrame.TextRange;
var text = textRange.Text;
textoContenido += text + " ";
}
}
}
}
//Get Author Name
object wordProperties = presentation.BuiltInDocumentProperties;
Type typeDocBuiltInProps = wordProperties.GetType();
Object Authorprop = typeDocBuiltInProps.InvokeMember("Item", System.Reflection.BindingFlags.Default | System.Reflection.BindingFlags.GetProperty, null, wordProperties, new object[] { "Author" }); //query for author properties
Type typeAuthorprop = Authorprop.GetType();
string autor = typeAuthorprop.InvokeMember("Value", System.Reflection.BindingFlags.Default | System.Reflection.BindingFlags.GetProperty, null, Authorprop, new object[] { }).ToString(); //get author name
Texto t = new Texto();
t.textoContenido = textoContenido;
Uri u = new Uri(path);
t.urlRuta = u.AbsoluteUri;
string ext = System.IO.Path.GetExtension(path);
t.formato = ext;
t.nombreArchivo = path.Substring(0, path.IndexOf(ext));
string auxiliar = "http://localhost/servidores/";
t.urlRuta = u.AbsoluteUri;
var puerto = u.Port;
int pos = t.urlRuta.IndexOf("servidores/") + 11;
string aux2 = t.urlRuta.Substring(pos);
t.urlRuta = auxiliar + aux2;
int pos2 = aux2.IndexOf("/");
string servidor = aux2.Substring(0, pos2);
int pos3 = aux2.IndexOf(servidor + "/") + servidor.Length + 1;
string depart = aux2.Substring(pos3);
int pos4 = depart.IndexOf("/");
depart = depart.Substring(0, pos4);
depart = depart.Replace("%20", " ");
t.departamento = depart;
t.hits = rnd.Next(1, 55);
PowerPoint_App.Quit();
presentation.Close();
t.idServidor = servidor;
textoContenido = textoContenido.Trim();
t.nombreArchivo = System.IO.Path.GetFileName(path);
t.fechaCreacionArchivo = (DateTime)File.GetCreationTime(path);
t.fechaModificacionArchivo = (DateTime)File.GetLastWriteTime(path);
t.fechaUltimaLectura = (DateTime)File.GetLastAccessTime(path);
t.fechaUltimaActualizacion = DateTime.Now;
if (File.Exists(path))
{
t.estadoActividad = 1;
}
Texto existente = OperacionesElasticSearch.ExisteTexto(t);
if (existente == null)
{
OperacionesElasticSearch.InsertarTexto(t);
}
else
{
OperacionesElasticSearch.actualizarTexto(existente, t);
}
}
else if (Herramientas.EsImagen(path) && indexaimagen && path.Contains(dpto))
{
var f = new FileInfo(path);
var fileLengthInKB = f.Length / 1024.0;
string ext = System.IO.Path.GetExtension(path);
List <string> eti = new List <string>();
eti.Add("imagen");
eti.Add("foto");
String titulo = path.Substring(0, path.IndexOf(ext));
FileInfo file = new FileInfo(path);
int tamanio = (int)file.Length;
Bitmap img = new Bitmap(path);
int altura = img.Height;
int anchura = img.Width;
Imagen im = new Imagen();
im.pixelesAltura = altura;
im.pixelesAnchura = anchura;
Uri u = new Uri(path);
im.urlRuta = u.AbsoluteUri;
string auxiliar = "http://localhost/servidores/";
im.urlRuta = u.AbsoluteUri;
var puerto = u.Port;
int pos = im.urlRuta.IndexOf("servidores/") + 11;
string aux2 = im.urlRuta.Substring(pos);
im.urlRuta = auxiliar + aux2;
int pos2 = aux2.IndexOf("/");
string servidor = aux2.Substring(0, pos2);
int pos3 = aux2.IndexOf(servidor + "/") + servidor.Length + 1;
string depart = aux2.Substring(pos3);
int pos4 = depart.IndexOf("/");
depart = depart.Substring(0, pos4);
depart = depart.Replace("%20", " ");
im.departamento = depart;
im.idServidor = servidor;
ext = System.IO.Path.GetExtension(path);
im.formato = ext;
im.nombreArchivo = System.IO.Path.GetFileName(path);
im.fechaCreacionArchivo = (DateTime)File.GetCreationTime(path);
im.fechaModificacionArchivo = (DateTime)File.GetLastWriteTime(path);
im.fechaUltimaLectura = (DateTime)File.GetLastAccessTime(path);
im.fechaUltimaActualizacion = DateTime.Now;
im.etiquetas = eti;
im.hits = rnd.Next(1, 55);
if (File.Exists(path))
{
im.estadoActividad = 1;
}
Imagen existente = OperacionesElasticSearch.ExisteImagen(im);
if (existente == null)
{
OperacionesElasticSearch.InsertarImagen(im);
}
else
{
OperacionesElasticSearch.actualizarImagen(existente, im);
}
}
else if (Herramientas.EsAudio(path) && indexaaudio)
{
var fi = new FileInfo(path);
var fileLengthInKB = fi.Length / 1024.0;
string ext = System.IO.Path.GetExtension(path);
string titulo = path.Substring(0, path.IndexOf(ext));
TagLib.File f = TagLib.File.Create(path, TagLib.ReadStyle.Average);
var duracion = (int)f.Properties.Duration.TotalSeconds;
List <string> eti = new List <string>();
eti.Add("audio");
eti.Add("sonido");
Audio au = new Audio();
Uri u = new Uri(path);
au.urlRuta = u.AbsoluteUri;
string auxiliar = "http://localhost/servidores/";
au.urlRuta = u.AbsoluteUri;
var puerto = u.Port;
int pos = au.urlRuta.IndexOf("servidores/") + 11;
string aux2 = au.urlRuta.Substring(pos);
au.urlRuta = auxiliar + aux2;
int pos2 = aux2.IndexOf("/");
string servidor = aux2.Substring(0, pos2);
int pos3 = aux2.IndexOf(servidor + "/") + servidor.Length + 1;
string depart = aux2.Substring(pos3);
int pos4 = depart.IndexOf("/");
depart = depart.Substring(0, pos4);
depart = depart.Replace("%20", " ");
au.departamento = depart;
au.duracion = duracion;
au.etiquetas = eti;
au.formato = ext;
au.nombreArchivo = System.IO.Path.GetFileName(path);
au.fechaCreacionArchivo = (DateTime)File.GetCreationTime(path);
au.fechaModificacionArchivo = (DateTime)File.GetLastWriteTime(path);
au.fechaUltimaLectura = (DateTime)File.GetLastAccessTime(path);
au.fechaUltimaActualizacion = DateTime.Now;
au.hits = rnd.Next(1, 55);
au.idServidor = servidor;
if (File.Exists(path))
{
au.estadoActividad = 1;
}
Audio existente = OperacionesElasticSearch.ExisteAudio(au);
if (existente == null)
{
OperacionesElasticSearch.InsertarAudio(au);
}
else
{
OperacionesElasticSearch.actualizarAudio(existente, au);
}
}
else if (Herramientas.EsVideo(path) && indexavideo)
{
string ext = System.IO.Path.GetExtension(path);
string titulo = path.Substring(0, path.IndexOf(ext));
List <string> eti = new List <string>();
eti.Add("video");
var fi = new FileInfo(path);
int duracion = 0;
string calidad = "";
var fileLengthInKB = fi.Length / 1024.0;
if (ext == ".mp4")
{
TagLib.File f = TagLib.File.Create(path, TagLib.ReadStyle.Average);
duracion = (int)f.Properties.Duration.TotalSeconds;
if (f.Properties.VideoHeight != 0 && f.Properties.VideoWidth != 0)
{
int height = (int)f.Properties.VideoHeight;
int width = (int)f.Properties.VideoWidth;
calidad = height + "x" + width;
}
}
Uri u = new Uri(path);
Video v = new Video();
v.urlRuta = u.AbsoluteUri;
string auxiliar = "http://localhost/servidores/";
v.urlRuta = u.AbsoluteUri;
var puerto = u.Port;
int pos = v.urlRuta.IndexOf("servidores/") + 11;
string aux2 = v.urlRuta.Substring(pos);
v.urlRuta = auxiliar + aux2;
int pos2 = aux2.IndexOf("/");
string servidor = aux2.Substring(0, pos2);
int pos3 = aux2.IndexOf(servidor + "/") + servidor.Length + 1;
string depart = aux2.Substring(pos3);
int pos4 = depart.IndexOf("/");
depart = depart.Substring(0, pos4);
depart = depart.Replace("%20", " ");
depart = depart.Replace("%20", " ");
v.departamento = depart;
v.duracion = duracion;
v.etiquetas = eti;
v.calidad = calidad;
v.idServidor = servidor;
v.nombreArchivo = System.IO.Path.GetFileName(path);
v.fechaCreacionArchivo = (DateTime)File.GetCreationTime(path);
v.fechaModificacionArchivo = (DateTime)File.GetLastWriteTime(path);
v.fechaUltimaLectura = (DateTime)File.GetLastAccessTime(path);
v.fechaUltimaActualizacion = DateTime.Now;
v.formato = ext;
v.hits = rnd.Next(1, 55);
if (File.Exists(path))
{
v.estadoActividad = 1;
}
Video existente = OperacionesElasticSearch.ExisteVideo(v);
if (existente == null)
{
OperacionesElasticSearch.InsertarVideo(v);
}
else
{
OperacionesElasticSearch.actualizarVideo(existente, v);
}
}
}
public static void CheckAttachmentsForDocOrPDFText(ActiveRecord record)
{
//walk the field list for this record looking for attachments
foreach (var fieldName in record.GetFieldNames())
{
if (fieldName.Contains("Attachment") && fieldName.DoesntContain("RawText"))
{
//if (record.Fields.Attachment.IsDirty) {
if (ActiveFieldBase.IsDirtyObj(record[fieldName].ValueObject, record[fieldName].OriginalValueObject))
{
if (record[fieldName].ToString().Contains(".doc") || record[fieldName].ToString().EndsWith(".pdf") || record[fieldName].ToString().EndsWith(".rtf"))
{
if (!record.FieldExists(fieldName + "RawText"))
{
(new Sql("ALTER TABLE ", record.GetTableName().SqlizeName(), " ADD [" + fieldName + "RawText] nvarchar (MAX);")).Execute();
}
string output = "";
if (record[fieldName].ToString().ToLower().EndsWith(".doc"))
{
OfficeFileReader.OfficeFileReader objOFR = new OfficeFileReader.OfficeFileReader();
if (objOFR.GetText(Web.MapPath(Web.Attachments) + record[fieldName].ToString(), ref output) > 0)
{
//ok
}
}
else if (record[fieldName].ToString().ToLower().EndsWith(".docx"))
{
BewebCore.ThirdParty.ReadWordDocText.DocxToText objOFR = new DocxToText(Web.MapPath(Web.Attachments) + record[fieldName].ToString());
if ((output = objOFR.ExtractText()).Length > 0)
{
//ok
}
}
else if (record[fieldName].ToString().Contains(".pdf"))
{
PdfToText.PDFParser pdf = new PDFParser();
if (pdf.ExtractText(Web.MapPath(Web.Attachments) + record[fieldName].ToString(), ref output))
{
//ok
}
}
else if (record[fieldName].ToString().Contains(".rtf"))
{
#if RTFProcessingAvailable
//Create the RTF tree object
RtfTree tree = new RtfTree();
//Load and parse RTF document
tree.LoadRtfFile(Web.MapPath(Web.Attachments) + record[fieldName].ToString());
output = tree.Text;
#else
throw new Exception("rtf library not included");
#endif
}
if (output.Trim() != "")
{
(new Sql("update ", record.GetTableName().SqlizeName(), "set " + fieldName + "RawText=", output.SqlizeText(), " where ",
record.GetPrimaryKeyName().SqlizeName(), "=", record.ID_Field.Sqlize(), "")).Execute();
}
}
else
{
//no doc any more
if (record.FieldExists(fieldName + "RawText"))
{
(new Sql("update ", record.GetTableName().SqlizeName(), "set " + fieldName + "RawText=null where ",
record.GetPrimaryKeyName().SqlizeName(), "=", record.ID_Field.Sqlize(), "")).Execute();
}
}
}
}
}
}