public static string TestRead()
{
var path = System.IO.Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.Desktop), "TestReadGAC.pdf");
Document pdfDocument = new Aspose.Pdf.Document(path);
// instantiate TextFragment Absorber object
Aspose.Pdf.Text.TextFragmentAbsorber TextFragmentAbsorberAddress = new Aspose.Pdf.Text.TextFragmentAbsorber();
// search text within page bound
TextFragmentAbsorberAddress.TextSearchOptions.LimitToPageBounds = true;
// specify the page region for TextSearch Options
TextFragmentAbsorberAddress.TextSearchOptions.Rectangle = new Aspose.Pdf.Rectangle(0, 0, 600, 800);
// search text from first page of PDF file
pdfDocument.Pages[1].Accept(TextFragmentAbsorberAddress);
List <string> l = new List <string>();
foreach (Aspose.Pdf.Text.TextFragment tf in TextFragmentAbsorberAddress.TextFragments)
{
l.Add(tf.Text);
}
if (l.Count > 0)
{
return(l[0]);
}
else
{
return(string.Empty);
}
}
public static string TestReadGAC_BAK(string path)
{
List <string> l = new List <string>();
Document pdfDocument = new Aspose.Pdf.Document(path);
Aspose.Pdf.Text.TextFragmentAbsorber textFragmentAbsorberAddress = new Aspose.Pdf.Text.TextFragmentAbsorber();
double width = 50d;
double height = 30d;
double x = 20d;
double y = 585d;
for (int i = 0; i < 3; i++)
{
// textFragmentAbsorberAddress.TextSearchOptions.LimitToPageBounds = true;
textFragmentAbsorberAddress.TextSearchOptions.LimitToPageBounds = false;
var rectangle = new Aspose.Pdf.Rectangle
(
//llx: x,
//lly: y - i * height,
//urx: width,
//ury: height
llx: width,
lly: height,
urx: x,
ury: y - i * height
);
textFragmentAbsorberAddress.TextSearchOptions.Rectangle = rectangle;
pdfDocument.Pages[1].Accept(textFragmentAbsorberAddress);
foreach (Aspose.Pdf.Text.TextFragment tf in textFragmentAbsorberAddress.TextFragments)
{
l.Add(tf.Text);
}
}
l = l.Distinct().ToList();
if (l.Count > 0)
{
return(l[0]);
}
else
{
return(string.Empty);
}
}
/// <summary>
/// Search documents
/// </summary>
public Response Search(DocumentInfo[] docs, string sourceFolder, string query)
{
if (docs == null)
{
return(BadDocumentResponse);
}
if (docs.Length == 0 || docs.Length > MaximumUploadFiles)
{
return(MaximumFileLimitsResponse);
}
SetDefaultOptions(docs, "");
Opts.AppName = "Search";
Opts.MethodName = "Search";
Opts.FolderName = sourceFolder;
Opts.OutputType = ".pdf";
Opts.ResultFileName = "Search Results";
Opts.CreateZip = false;
var statusValue = "OK";
var statusCodeValue = 200;
var fileProcessingErrorCode = FileProcessingErrorCode.OK;
if (IsValidRegex(query))
{
try
{
return(Process((inFilePath, outPath, zipOutFolder) =>
{
foreach (var pdfDocument in docs)
{
var document = pdfDocument.PdfDocument;
var tfa = new Aspose.Pdf.Text.TextFragmentAbsorber(query,
new TextSearchOptions(IsValidRegex(query)));
tfa.Visit(document);
var found = tfa.TextFragments.Count > 0;
foreach (var textFragment in tfa.TextFragments)
{
textFragment.Page.Annotations.Add(
new HighlightAnnotation(textFragment.Page, textFragment.Rectangle)
{
Title = "Aspose.PDF Search Free App"
}
);
}
document.Save(outPath);
}
}));
//fileProcessingErrorCode = FileProcessingErrorCode.NoSearchResults;
}
catch (Exception ex)
{
statusCodeValue = 500;
statusValue = "500 " + ex.Message;
}
}
else
{
fileProcessingErrorCode = FileProcessingErrorCode.WrongRegExp;
}
return(new Response
{
Status = statusValue,
StatusCode = statusCodeValue,
FileProcessingErrorCode = fileProcessingErrorCode
});
}
/// <summary>
/// Redact documents
/// </summary>
public Response Redact(DocumentInfo[] docs, string sourceFolder, string searchQuery, string replaceText,
bool caseSensitive, bool text, bool comments, bool metadata)
{
if (docs == null)
{
return(BadDocumentResponse);
}
if (docs.Length == 0 || docs.Length > MaximumUploadFiles)
{
return(MaximumFileLimitsResponse);
}
SetDefaultOptions(docs, "");
Opts.AppName = "Redaction";
Opts.MethodName = "Redact";
Opts.ZipFileName = "Redacted documents";
Opts.FolderName = sourceFolder;
Opts.OutputType = ".pdf";
if (replaceText == null)
{
replaceText = "";
}
var statusValue = "OK";
var statusCodeValue = 200;
var fileProcessingErrorCode = FileProcessingErrorCode.OK;
var lck = new object();
var catchedException = false;
if (IsValidRegex(searchQuery))
{
var regex = new Regex(searchQuery, caseSensitive ? RegexOptions.None : RegexOptions.IgnoreCase);
var matchesFound = new int[docs.Length];
var tasks = Enumerable.Range(0, docs.Length).Select(i => Task.Factory.StartNew(() =>
{
try
{
var document = docs[i].PdfDocument;
if (text)
{
var tfa = new Aspose.Pdf.Text.TextFragmentAbsorber(searchQuery, new TextSearchOptions(true));
tfa.Visit(document);
foreach (var textFragment in tfa.TextFragments)
{
textFragment.Text = replaceText;
matchesFound[i] += 1;
}
}
if (comments)
{
foreach (var page in document.Pages)
{
foreach (var item in page.Annotations)
{
if (item is TextMarkupAnnotation)
{
item.Contents = Regex.Replace(item.Contents, searchQuery, replaceText);
matchesFound[i] += 1;
}
}
}
}
if (metadata)
{
Pdf.DocumentInfo docInfo = new Pdf.DocumentInfo(document);
docInfo.Title = System.Text.RegularExpressions.Regex.Replace(docInfo.Title, searchQuery, replaceText);
docInfo.Keywords = System.Text.RegularExpressions.Regex.Replace(docInfo.Keywords, searchQuery, replaceText);
matchesFound[i] += 2;
}
//if (text || comments)
//{
// var findings = new RedactionCallback(text, comments);
// var options = new FindReplaceOptions()
// {
// ReplacingCallback = findings,
// Direction = FindReplaceDirection.Forward,
// MatchCase = caseSensitive
// };
// docs[i].Range.Replace(regex, replaceText, options);
// matchesFound[i] += findings.MatchesFound;
//}
//if (metadata)
// matchesFound[i] += ProcessMetadata(docs[i].PdfDocument, regex, replaceText);
}
catch (Exception ex)
{
Console.WriteLine(ex.Message);
lock (lck)
catchedException = true;
}
})).ToArray();
Task.WaitAll(tasks);
if (!catchedException)
{
if (matchesFound.Sum() > 0)
{
return(Process((inFilePath, outPath, zipOutFolder) =>
{
foreach (var doc in docs)
{
SaveDocument(doc, outPath, zipOutFolder, new SaveFormatType()
{
SaveOptions = new PdfSaveOptions(), SaveType = SaveType.pdf
});
}
}));
}
fileProcessingErrorCode = FileProcessingErrorCode.NoSearchResults;
}
else
{
statusCodeValue = 500;
statusValue = "500 Exception during processing";
}
}
else
{
fileProcessingErrorCode = FileProcessingErrorCode.WrongRegExp;
}
return(new Response
{
Status = statusValue,
StatusCode = statusCodeValue,
FileProcessingErrorCode = fileProcessingErrorCode
});
}
