phares3757/pdf-text-stripper
3
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Compare commits

base: phares3757:02-26-a
Branches Tags
phares3757:master phares3757:02-26-c phares3757:02-26-a
..
compare: phares3757:02-26-c
Branches Tags
phares3757:02-26-c phares3757:02-26-a phares3757:master

1 Commits
02-26-a ... 02-26-c

Author SHA1 Message Date
Mike Phares
cb333ec871 Use of Splitter 
dotnet 8.0.100
 2025-02-26 12:36:41 -07:00
2 changed files with 34 additions and 11 deletions
Expand all files Collapse all files

2
PDF-Text-Stripper/PDF-Text-Stripper.csproj
View File

@ -8,7 +8,7 @@
<PackageId>Infineon.Mesa.PDF.Text.Stripper</PackageId> <PackageId>Infineon.Mesa.PDF.Text.Stripper</PackageId>
<RuntimeIdentifier>win-x86</RuntimeIdentifier> <RuntimeIdentifier>win-x86</RuntimeIdentifier>
<TargetFrameworks>net48</TargetFrameworks> <TargetFrameworks>net48</TargetFrameworks>
<Version>4.8.0.1</Version> <Version>4.8.0.2</Version>
</PropertyGroup> </PropertyGroup>
<ItemGroup Condition=" '$(TargetFramework)' == 'net48' "> <ItemGroup Condition=" '$(TargetFramework)' == 'net48' ">
<Reference Include="Microsoft.CSharp" /> <Reference Include="Microsoft.CSharp" />

43
PDF-Text-Stripper/Program.cs
View File

@ -8,16 +8,39 @@ namespace PDF_Text_Stripper;
public class Program public class Program
{ {
private static void PDFTextStripper(string file) private static void PortableDocumentFormatWriteText(string sourceFileName)
{ {
string altFileName = Path.ChangeExtension(file, ".txt"); object item;
if(File.Exists(altFileName)) string pageText;
File.Delete(altFileName); string pagePDFFile;
org.apache.pdfbox.pdmodel.PDDocument pdfDocument = org.apache.pdfbox.pdmodel.PDDocument.load(file); string pageTextFile;
org.apache.pdfbox.util.PDFTextStripper stripper = new(); java.io.File file = new(sourceFileName);
string text = stripper.getText(pdfDocument); org.apache.pdfbox.util.Splitter splitter = new();
pdfDocument.close(); string sourcePath = Path.GetDirectoryName(sourceFileName) ?? throw new Exception();
File.WriteAllText(altFileName, text); string sourceFileNameWithoutExtension = Path.GetFileNameWithoutExtension(sourceFileName);
org.apache.pdfbox.pdmodel.PDDocument pdDocument = org.apache.pdfbox.pdmodel.PDDocument.load(file);
java.util.List list = splitter.split(pdDocument);
java.util.ListIterator iterator = list.listIterator();
org.apache.pdfbox.util.PDFTextStripper dataStripper = new();
for (short i = 1; i < short.MaxValue; i++)
{
if (!iterator.hasNext())
break;
item = iterator.next();
pagePDFFile = string.Concat(sourcePath, @"\", sourceFileNameWithoutExtension, "_", i, ".pdf");
if (File.Exists(pagePDFFile))
File.Delete(pagePDFFile);
pageTextFile = Path.ChangeExtension(pagePDFFile, ".txt");
if (File.Exists(pageTextFile))
File.Delete(pageTextFile);
if (item is not org.apache.pdfbox.pdmodel.PDDocument pd)
continue;
pageText = dataStripper.getText(pd);
pd.save(pagePDFFile);
pd.close();
File.WriteAllText(pageTextFile, pageText);
}
pdDocument.close();
} }
public static void Secondary(List<string> args) public static void Secondary(List<string> args)
@ -28,7 +51,7 @@ public class Program
try try
{ {
if (args.Any() && File.Exists(args[0])) if (args.Any() && File.Exists(args[0]))
PDFTextStripper(args[0]); PortableDocumentFormatWriteText(args[0]);
else else
throw new Exception(args[0]); throw new Exception(args[0]);
} }