public static List<string> GetNewsToday()
{
List<string> tmp = new List<string>();
TextExtraction t = new TextExtraction();
ZetaCrawlerEntities db = new ZetaCrawlerEntities();
string[] strSplit = { "/" };
var pBIDV = from q in db.BIDVObjects
where ((q.IsToEmail == false) && (q.LastModified.Value.Day == DateTime.Now.Day))
select q;
foreach (var item in pBIDV)
{
tmp.Add("<br />" + "[" + item.LastModified.ToString() + "] " + "<b>" + t.GetBetween2Words("Trích yếu: ", "Độ khẩn", item.Summary) + "</b><br />Xem chi tiết : " + item.OriginalUrl + "<br />");
item.IsToEmail = true;
}
var pASPNET = from q in db.ASPNETObjects
where ((q.IsToEmail == false) &&(q.LastModified.Value.Day == DateTime.Now.Day) )
select q;
foreach (var item in pASPNET)
{
tmp.Add("<br />" + "[" + item.LastModified.ToString() + "] " + "<b>" + item.Summary + "</b><br />Xem chi tiết : " + item.OriginalUrl + "<br />");
item.IsToEmail = true;
}
db.SaveChanges();
return tmp;
}
public ActionResult FileUploaderAjax(PDFFile membervalues)
{
//string FileName = Path.GetFileNameWithoutExtension(membervalues.PDFFileName.FileName);
////To Get File Extension
//string FileExtension = Path.GetExtension(membervalues.PDFFileName.FileName);
////Add Current Date To Attached File Name
//FileName = DateTime.Now.ToString("yyyyMMdd") + "-" + FileName.Trim() + FileExtension;
string errorLogPathStr = ConfigurationManager.AppSettings["ErrorLogPath"];
string errorLogPath = Server.MapPath(errorLogPathStr);
bool deleteUploadedFile;
try
{
if (Request.Files.Count > 0)
{
var file = Request.Files[0];
if (file != null && file.ContentLength > 0)
{
var fileName = Path.GetFileName(file.FileName);
string UploadPathConfig = ConfigurationManager.AppSettings["UploadPath"].ToString();
deleteUploadedFile = Convert.ToBoolean(ConfigurationManager.AppSettings["DeleteUploadedFile"].ToString());
string UploadPath = Server.MapPath(UploadPathConfig);
var path = Path.Combine(UploadPath, fileName);//+"_"+Guid.NewGuid().ToString()
if (!System.IO.File.Exists(path))
{
file.SaveAs(path);
}
TextExtraction textExtraction = new TextExtraction();
Dictionary <string, Int64> keyValuePairs = textExtraction.CalculatePages(path, errorLogPath);
ViewBag.PDFInfo = keyValuePairs;
ViewBag.FilePath = path;
if (deleteUploadedFile && System.IO.File.Exists(path))
{
System.IO.File.Delete(path);
}
return(Json(new { Total = keyValuePairs["TotalPages"], Colored = keyValuePairs["ColorPages"], BlackAndWhite = keyValuePairs["BWPages"] }, JsonRequestBehavior.AllowGet));
}
}
}
catch (Exception ex)
{
CreateLog(ex, errorLogPath);
}
//Get Upload path from Web.Config file AppSettings.
//Its Create complete path to store in server.
//membervalues.ImagePath = UploadPath + FileName;
//To copy and save file into server.
//membervalues.PDFFileName.SaveAs(membervalues.ImagePath);
return(View());
}
public void MergeFrom(AnnotationPayload other)
{
if (other == null)
{
return;
}
if (other.AnnotationSpecId.Length != 0)
{
AnnotationSpecId = other.AnnotationSpecId;
}
if (other.DisplayName.Length != 0)
{
DisplayName = other.DisplayName;
}
switch (other.DetailCase)
{
case DetailOneofCase.Translation:
if (Translation == null)
{
Translation = new global::Google.Cloud.AutoML.V1.TranslationAnnotation();
}
Translation.MergeFrom(other.Translation);
break;
case DetailOneofCase.Classification:
if (Classification == null)
{
Classification = new global::Google.Cloud.AutoML.V1.ClassificationAnnotation();
}
Classification.MergeFrom(other.Classification);
break;
case DetailOneofCase.ImageObjectDetection:
if (ImageObjectDetection == null)
{
ImageObjectDetection = new global::Google.Cloud.AutoML.V1.ImageObjectDetectionAnnotation();
}
ImageObjectDetection.MergeFrom(other.ImageObjectDetection);
break;
case DetailOneofCase.TextExtraction:
if (TextExtraction == null)
{
TextExtraction = new global::Google.Cloud.AutoML.V1.TextExtractionAnnotation();
}
TextExtraction.MergeFrom(other.TextExtraction);
break;
case DetailOneofCase.TextSentiment:
if (TextSentiment == null)
{
TextSentiment = new global::Google.Cloud.AutoML.V1.TextSentimentAnnotation();
}
TextSentiment.MergeFrom(other.TextSentiment);
break;
}
_unknownFields = pb::UnknownFieldSet.MergeFrom(_unknownFields, other._unknownFields);
}
/// <summary>
/// </summary>
/// <param name="crawler">
/// The crawler.
/// </param>
/// <param name="propertyBag">
/// The property bag.
/// </param>
public void Process(Crawler crawler, PropertyBag propertyBag)
{
CultureInfo contentCulture = (CultureInfo)propertyBag["LanguageCulture"].Value;
string cultureDisplayValue = "N/A";
if (!contentCulture.IsNull())
{
cultureDisplayValue = contentCulture.DisplayName;
}
TextExtraction t = new TextExtraction();
lock (this)
{
BIDVObject item = new BIDVObject();
item.OriginalUrl = propertyBag.Step.Uri.ToString();
if (!IsDuplicate(item.OriginalUrl))
{
item.Title = propertyBag.Title;
item.StatusDescription = propertyBag.StatusDescription;
item.ResponseUri = propertyBag.ResponseUri.ToString();
item.Text = propertyBag.Text;
item.Depth = propertyBag.Step.Depth;
item.LastModified = propertyBag.LastModified;
item.OriginalReferrerUrl = propertyBag.OriginalReferrerUrl.ToString();
item.Server = propertyBag.Server;
string description = t.GetBetween2Words("Chi tiết văn bản", "Xem toàn màn hình", item.Text.Replace("\r"," ").Replace("\n"," "));
item.Summary = t.RemoveWhiteSpace(description);
string strNgayPhatHanh = t.GetBetween2Words("Ngày phát hành", "Số đi", item.Summary);
strNgayPhatHanh = strNgayPhatHanh.Replace(' ', '/').Remove(0, ("Ngày phát hành").Length);
string[] strSplit = { "/" };
int day = int.Parse(strNgayPhatHanh.Split(strSplit, StringSplitOptions.None)[1]);
int month = int.Parse(strNgayPhatHanh.Split(strSplit, StringSplitOptions.None)[2]);
int year = int.Parse(strNgayPhatHanh.Split(strSplit, StringSplitOptions.None)[3]);
//Clean the text field is null
item.Text = null;
item.IsToEmail = false;
db.AddToBIDVObjects(item);
item.ContentEncoding = propertyBag.ContentEncoding;
item.ContentType = propertyBag.ContentType;
//item.Length = propertyBag.Text.IsNull() ? 0 : propertyBag.Text.Length;
//item.CultureDisplayValue = cultureDisplayValue;
}
}
try
{
db.SaveChanges();
}
catch (Exception ex)
{
throw new Exception(ex.Message);
}
}
/// <summary>
/// </summary>
/// <param name="crawler">
/// The crawler.
/// </param>
/// <param name="propertyBag">
/// The property bag.
/// </param>
public void Process(Crawler crawler, PropertyBag propertyBag)
{
CultureInfo contentCulture = (CultureInfo)propertyBag["LanguageCulture"].Value;
string cultureDisplayValue = "N/A";
if (!contentCulture.IsNull())
{
cultureDisplayValue = contentCulture.DisplayName;
}
TextExtraction t = new TextExtraction();
lock (this)
{
BIDVObject item = new BIDVObject();
//item.Id = Guid.NewGuid();
//item.Url = propertyBag.Step.Uri.ToString();
//if (item.Url.StartsWith("http://bidvportal.vn/eDocman"))
//{
// item.Title = propertyBag.Title;
// string strTarget = t.GetMinimumString(propertyBag.Text, "Chi tiết văn bản", "Nội dung văn bản");
// item.Text = strTarget;
// string strNgayPhatHanh = t.GetMinimumString(strTarget, "Ngày phát hành", "Số đi");
// item.NgayPhatHanh = strNgayPhatHanh.Replace(' ','/');
// string strSubject = t.GetMinimumString(strTarget, "Trích yếu", "Độ khẩn");
// //item.Subject = strSubject;
// //item.ContentEncoding = propertyBag.ContentEncoding;
// //item.ContentType = propertyBag.ContentType;
// //item.Length = propertyBag.Text.IsNull() ? 0 : propertyBag.Text.Length;
// item.Depth = propertyBag.Step.Depth;
// //item.CultureDisplayValue = cultureDisplayValue;
// string[] strSplit = { "/" };
// int day = int.Parse(item.NgayPhatHanh.Split(strSplit, StringSplitOptions.None)[0]);
// int month = int.Parse(item.NgayPhatHanh.Split(strSplit, StringSplitOptions.None)[1]);
// int year = int.Parse(item.NgayPhatHanh.Split(strSplit, StringSplitOptions.None)[2]);
// if ((DateTime.Now.Year == year) && (DateTime.Now.Month == month) && (DateTime.Now.Day == day))
// {
// //db.AddToItems(item);
// }
//}
}
try
{
db.SaveChanges();
}
catch (Exception ex)
{
Console.WriteLine("=====================================================");
Console.WriteLine(ex.Message);
}
}
public override int GetHashCode()
{
int hash = 1;
if (detailCase_ == DetailOneofCase.Translation)
{
hash ^= Translation.GetHashCode();
}
if (detailCase_ == DetailOneofCase.Classification)
{
hash ^= Classification.GetHashCode();
}
if (detailCase_ == DetailOneofCase.ImageObjectDetection)
{
hash ^= ImageObjectDetection.GetHashCode();
}
if (detailCase_ == DetailOneofCase.VideoClassification)
{
hash ^= VideoClassification.GetHashCode();
}
if (detailCase_ == DetailOneofCase.VideoObjectTracking)
{
hash ^= VideoObjectTracking.GetHashCode();
}
if (detailCase_ == DetailOneofCase.TextExtraction)
{
hash ^= TextExtraction.GetHashCode();
}
if (detailCase_ == DetailOneofCase.TextSentiment)
{
hash ^= TextSentiment.GetHashCode();
}
if (detailCase_ == DetailOneofCase.Tables)
{
hash ^= Tables.GetHashCode();
}
if (AnnotationSpecId.Length != 0)
{
hash ^= AnnotationSpecId.GetHashCode();
}
if (DisplayName.Length != 0)
{
hash ^= DisplayName.GetHashCode();
}
hash ^= (int)detailCase_;
if (_unknownFields != null)
{
hash ^= _unknownFields.GetHashCode();
}
return(hash);
}
/// <summary>
/// </summary>
/// <param name="crawler">
/// The crawler.
/// </param>
/// <param name="propertyBag">
/// The property bag.
/// </param>
public void Process(Crawler crawler, PropertyBag propertyBag)
{
CultureInfo contentCulture = (CultureInfo)propertyBag["LanguageCulture"].Value;
string cultureDisplayValue = "N/A";
if (!contentCulture.IsNull())
{
cultureDisplayValue = contentCulture.DisplayName;
}
TextExtraction t = new TextExtraction();
lock (this)
{
ASPNETObject item = new ASPNETObject();
item.OriginalUrl = propertyBag.Step.Uri.ToString();
if (!IsDuplicate(item.OriginalUrl))
{
item.Title = propertyBag.Title;
item.StatusDescription = propertyBag.StatusDescription;
item.ResponseUri = propertyBag.ResponseUri.ToString();
item.Text = null;
item.Depth = propertyBag.Step.Depth;
item.LastModified = propertyBag.LastModified;
item.OriginalReferrerUrl = propertyBag.OriginalReferrerUrl.ToString();
item.Server = propertyBag.Server;
//Clean the text field is null
db.AddToASPNETObjects(item);
item.ContentEncoding = propertyBag.ContentEncoding;
item.ContentType = propertyBag.ContentType;
item.IsToEmail = false;
item.Summary = propertyBag.Title;
//item.Length = propertyBag.Text.IsNull() ? 0 : propertyBag.Text.Length;
//item.CultureDisplayValue = cultureDisplayValue;
}
}
try
{
db.SaveChanges();
}
catch (Exception ex)
{
throw new Exception(ex.Message);
}
}
public ActionResult FileUploader(PDFFile membervalues)
{
//string FileName = Path.GetFileNameWithoutExtension(membervalues.PDFFileName.FileName);
////To Get File Extension
//string FileExtension = Path.GetExtension(membervalues.PDFFileName.FileName);
////Add Current Date To Attached File Name
//FileName = DateTime.Now.ToString("yyyyMMdd") + "-" + FileName.Trim() + FileExtension;
string errorLogPathStr = ConfigurationManager.AppSettings["ErrorLogPath"];
string errorLogPath = Server.MapPath(errorLogPathStr);
if (Request.Files.Count > 0)
{
var file = Request.Files[0];
if (file != null && file.ContentLength > 0)
{
var fileName = Path.GetFileName(file.FileName);
string UploadPathConfig = ConfigurationManager.AppSettings["UploadPath"].ToString();
string UploadPath = Server.MapPath(UploadPathConfig);
var path = Path.Combine(UploadPath, fileName);//+"_"+Guid.NewGuid().ToString()
if (!System.IO.File.Exists(path))
{
file.SaveAs(path);
}
TextExtraction textExtraction = new TextExtraction();
Dictionary <string, Int64> keyValuePairs = textExtraction.CalculatePages(path, errorLogPath);
ViewBag.PDFInfo = keyValuePairs;
ViewBag.FilePath = path;
}
}
//Get Upload path from Web.Config file AppSettings.
//Its Create complete path to store in server.
//membervalues.ImagePath = UploadPath + FileName;
//To copy and save file into server.
//membervalues.PDFFileName.SaveAs(membervalues.ImagePath);
return(View());
}
public static List<string> GetNewsToday()
{
List<string> tmp = new List<string>();
TextExtraction t = new TextExtraction();
ZetaCrawlerEntities db = new ZetaCrawlerEntities();
string[] strSplit = { "/" };
var p = from q in db.BIDVObjects
where ((q.LastModified.Value.Day >= DateTime.Today.Day) && (q.LastModified.Value.Month == DateTime.Today.Month) && (q.LastModified.Value.Year == DateTime.Today.Year))
select q;
foreach (var item in p)
{
tmp.Add("<br />" + "[" + item.LastModified.ToString() + "] " + "<b>" + t.GetBetween2Words("Trích yếu: ", "Độ khẩn", item.Summary) + "</b><br />Xem chi tiết : " + item.OriginalUrl + "<br />");
}
return tmp;
}
private static void DisplayMergerExamples()
{
bool exitToProducts = false;
while (!exitToProducts)
{
Console.WriteLine("Select a DynamicPDF Merger for .NET example to run:");
Console.WriteLine(" A : Field Level Flattening");
Console.WriteLine(" B : Form Field Reader");
Console.WriteLine(" C : Form Flattening");
Console.WriteLine(" D : Merge PDFs");
Console.WriteLine(" E : Place PDFs");
Console.WriteLine(" F : Stamp PDF");
Console.WriteLine(" G : Text Extraction");
Console.WriteLine(" H : Select Pages");
Console.WriteLine(" I : Merge Invoices");
Console.WriteLine(" J : AcroForm Fill");
Console.WriteLine(" K : Form Fill");
Console.WriteLine();
Console.WriteLine("Press 'Backspace' for the main products menu");
Console.WriteLine("Press 'Esc' to exit application");
Console.WriteLine();
ConsoleKeyInfo runKey = Console.ReadKey();
Console.WriteLine();
Console.WriteLine();
string exampleName = string.Empty;
string fileName = string.Empty;
switch (runKey.Key)
{
case ConsoleKey.A:
exampleName = "Field Level Flattening";
fileName = "FieldLevelFlattening.pdf";
Console.WriteLine("Example " + exampleName + " is Running...");
FieldLevelFlattening.Run(fileName);
break;
case ConsoleKey.B:
exampleName = "Form Field Reader";
fileName = "FormFieldReader.pdf";
Console.WriteLine("Example " + exampleName + " is Running...");
FormFieldReader.Run(fileName);
break;
case ConsoleKey.C:
exampleName = "Form Flattening";
fileName = "FormFlattening.pdf";
Console.WriteLine("Example " + exampleName + " is Running...");
FormFlattening.Run(fileName);
break;
case ConsoleKey.D:
exampleName = "Merge PDFs";
fileName = "MergePDFs.pdf";
Console.WriteLine("Example " + exampleName + " is Running...");
MergePDFs.Run(fileName);
break;
case ConsoleKey.E:
exampleName = "Place PDFs";
fileName = "PlacePDFs.pdf";
Console.WriteLine("Example " + exampleName + " is Running...");
PlacePDFs.Run(fileName);
break;
case ConsoleKey.F:
exampleName = "Stamp PDF";
fileName = "StampPDF.pdf";
Console.WriteLine("Example " + exampleName + " is Running...");
StampPDF.Run(fileName);
break;
case ConsoleKey.G:
exampleName = "Text Extraction";
Console.WriteLine("Example " + exampleName + " is Running...");
Console.WriteLine();
Console.WriteLine(TextExtraction.Run());
break;
case ConsoleKey.H:
exampleName = "Select Pages";
fileName = "SelectPages.pdf";
Console.WriteLine("Select Pages to Merge from Document A");
Console.WriteLine(" A1: Page 1");
Console.WriteLine(" A2: Page 2");
Console.WriteLine(" A3: Page 3");
Console.WriteLine(" A4: Page 4");
Console.WriteLine();
Console.WriteLine("Select Pages to Merge from Document B");
Console.WriteLine(" B1: Page 1");
Console.WriteLine(" B2: Page 2");
Console.WriteLine(" B3: Page 3");
Console.WriteLine();
Console.WriteLine("Select Pages to Merge from Document C");
Console.WriteLine(" C1: Page 1");
Console.WriteLine(" C2: Page 2");
Console.WriteLine();
Console.WriteLine("Select Pages to Merge from Document D");
Console.WriteLine(" D1: Page 1");
Console.WriteLine(" D2: Page 2");
Console.WriteLine(" D3: Page 3");
Console.WriteLine(" D4: Page 4");
Console.WriteLine(" D5: Page 5");
Console.WriteLine(" D6: Page 6");
Console.WriteLine();
Console.WriteLine("Please enter the page number(s) to merge. Use a comma ',' to seperate multiple entries (Ex: A1,B2,D5):");
Console.WriteLine();
string selectedPages = Console.ReadLine().ToUpper();
Console.WriteLine("Example " + exampleName + " is Running...");
SelectPages.Run(selectedPages.Split(','), fileName);
break;
case ConsoleKey.I:
exampleName = "Merge Invoices PDF";
fileName = "MergerInvoice.pdf";
Console.WriteLine("Please enter the invoice number(s) to include. Use a comma ',' to seperate multiple entries (Ex: 10248,10249,10250):");
Console.WriteLine();
Console.WriteLine("10248");
Console.WriteLine("10249");
Console.WriteLine("10250");
Console.WriteLine("10251");
Console.WriteLine("10252");
Console.WriteLine("10360");
Console.WriteLine("10979");
Console.WriteLine("11077");
Console.WriteLine();
string invoiceNumbers = Console.ReadLine();
Console.WriteLine("Example " + exampleName + " is Running...");
byte[] pdf = MergerInvoice.Run(invoiceNumbers.Split(','));
File.WriteAllBytes(fileName, pdf);
break;
case ConsoleKey.J:
exampleName = "AcroForm Fill";
fileName = "AcroFormFill.pdf";
Console.WriteLine("Example " + exampleName + " is Running...");
AcroFormFill.Run(fileName);
break;
case ConsoleKey.K:
exampleName = "Form Fill";
fileName = "FormFill.pdf";
Console.WriteLine("Example " + exampleName + " is Running...");
FormFill.Run(fileName);
break;
case ConsoleKey.Escape:
System.Environment.Exit(0);
break;
case ConsoleKey.Backspace:
exitToProducts = true;
break;
default:
Console.WriteLine();
Console.WriteLine("Key not recognized.");
break;
}
if (fileName != string.Empty)
{
DisplayOutputPathWithOptionToOpen(fileName);
}
}
}
