240 lines
11 KiB
C#
240 lines
11 KiB
C#
using Microsoft.Extensions.Logging;
|
|
using System.Diagnostics;
|
|
|
|
namespace File_Folder_Helper.Helpers;
|
|
|
|
internal static class HelperPdfStripperWrapper
|
|
{
|
|
|
|
private record Input(string? DestinationDirectory,
|
|
string? Key,
|
|
char? Parser);
|
|
|
|
private const char _Linc = 'L';
|
|
private const char _Ghost = 'G';
|
|
private const char _Kofax = 'K';
|
|
private const char _Stripper = 'S';
|
|
|
|
private static string GetTextFromPDF(string pdfTextStripperFileName, string sourceFileNamePdf, string destinationFileName)
|
|
{
|
|
string result;
|
|
ProcessStartInfo processStartInfo = new(pdfTextStripperFileName, $"s \"{sourceFileNamePdf}\"")
|
|
{
|
|
UseShellExecute = false,
|
|
RedirectStandardError = true,
|
|
RedirectStandardOutput = true,
|
|
};
|
|
Process? process = Process.Start(processStartInfo);
|
|
_ = process?.WaitForExit(30000);
|
|
if (!File.Exists(destinationFileName))
|
|
result = string.Empty;
|
|
else
|
|
result = File.ReadAllText(destinationFileName);
|
|
return result;
|
|
}
|
|
|
|
private static string GetLincTextFromPDF(string lincPDFCFileName, string sourceFileNamePdf, string destinationFileName)
|
|
{
|
|
string result;
|
|
if (File.Exists(destinationFileName))
|
|
File.Delete(destinationFileName);
|
|
string arguments = string.Concat("-i \"", sourceFileNamePdf, "\" -o \"", destinationFileName, "\"");
|
|
Process? process = Process.Start(lincPDFCFileName, arguments);
|
|
_ = process?.WaitForExit(30000);
|
|
if (!File.Exists(destinationFileName))
|
|
result = string.Empty;
|
|
else
|
|
result = File.ReadAllText(destinationFileName);
|
|
return result;
|
|
}
|
|
|
|
private static string GetGhostTextFromPDF(string ghostPCLFileName, string sourceFileNamePdf, string destinationFileName)
|
|
{
|
|
string result;
|
|
if (File.Exists(destinationFileName))
|
|
File.Delete(destinationFileName);
|
|
string arguments = $"-dSAFER -dBATCH -dNOPAUSE -dFIXEDMEDIA -dFitPage -dAutoRotatePages=/All -dDEVICEWIDTHPOINTS=792 -dDEVICEHEIGHTPOINTS=612 -sOutputFile=\"{destinationFileName}\" -sDEVICE=pdfwrite \"{sourceFileNamePdf}\"";
|
|
Process? process = Process.Start(ghostPCLFileName, arguments);
|
|
_ = process?.WaitForExit(30000);
|
|
if (!File.Exists(destinationFileName))
|
|
result = string.Empty;
|
|
else
|
|
result = File.ReadAllText(destinationFileName);
|
|
return result;
|
|
}
|
|
|
|
private static string GetKofaxTextFromPDF(string kofaxFileName, string sourceFileNamePdf, string destinationFileName)
|
|
{
|
|
string result;
|
|
if (File.Exists(destinationFileName))
|
|
File.Delete(destinationFileName);
|
|
string arguments = $"-i\"{sourceFileNamePdf}\" -o\"{destinationFileName}\" -ttxt";
|
|
Process? process = Process.Start(kofaxFileName, arguments);
|
|
_ = process?.WaitForExit(30000);
|
|
if (!File.Exists(destinationFileName))
|
|
result = string.Empty;
|
|
else
|
|
result = File.ReadAllText(destinationFileName);
|
|
return result;
|
|
}
|
|
|
|
private static Input GetInput(List<string> args)
|
|
{
|
|
string? d = null;
|
|
string? k = null;
|
|
string? p = null;
|
|
for (int i = 1; i < args.Count; i++)
|
|
{
|
|
if (args[i].Length == 2 && i + 1 < args.Count)
|
|
{
|
|
if (args[i][1] == 'd')
|
|
d = Path.GetFullPath(args[i + 1]);
|
|
else if (args[i][1] == 'k')
|
|
k = args[i + 1].Trim();
|
|
else if (args[i][1] == 'p')
|
|
p = args[i + 1];
|
|
i++;
|
|
}
|
|
}
|
|
string directoryName = Path.GetFileName(args[0]);
|
|
if (!string.IsNullOrEmpty(d) && !d.EndsWith(directoryName))
|
|
d = Path.Combine(d, directoryName);
|
|
if (!string.IsNullOrEmpty(d) && !Directory.Exists(d))
|
|
_ = Directory.CreateDirectory(d);
|
|
char? parser = p is null ? null : p == "S" ? _Stripper : p == "G" ? _Ghost : p == "K" ? _Kofax : p == "L" ? _Linc : null;
|
|
return new(d, k, parser);
|
|
}
|
|
|
|
private static void ParseSave(ILogger log, string pdfTextStripperFileName, string ghostPCLFileName, string kofaxFileName, string lincPDFCFileName, string destinationDirectory, char parser, string[] files)
|
|
{
|
|
string text;
|
|
string destinationFileName;
|
|
if (!Directory.Exists(destinationDirectory))
|
|
_ = Directory.CreateDirectory(destinationDirectory);
|
|
foreach (string file in files)
|
|
{
|
|
destinationFileName = Path.Combine(destinationDirectory, $"{file}.txt");
|
|
if (parser == _Kofax)
|
|
text = GetKofaxTextFromPDF(kofaxFileName, file, destinationFileName);
|
|
else if (parser == _Linc)
|
|
text = GetLincTextFromPDF(lincPDFCFileName, file, destinationFileName);
|
|
else if (parser == _Ghost)
|
|
text = GetGhostTextFromPDF(ghostPCLFileName, file, destinationFileName);
|
|
else if (parser == _Stripper)
|
|
text = GetTextFromPDF(pdfTextStripperFileName, file, destinationFileName);
|
|
else
|
|
throw new NotImplementedException();
|
|
log.LogInformation("<{file}> == {length}", Path.GetFileName(file), text.Length);
|
|
}
|
|
}
|
|
|
|
private static void ParseStrip(ILogger log, string destinationDirectory, string key, string[] files)
|
|
{
|
|
string[] lines;
|
|
string fileName;
|
|
string[] segments;
|
|
List<string> collection = [];
|
|
foreach (string file in files)
|
|
{
|
|
lines = File.ReadAllLines(file);
|
|
fileName = Path.GetFileName(file);
|
|
foreach (string line in lines)
|
|
{
|
|
segments = line.Split(':');
|
|
if (segments.Length < 2)
|
|
continue;
|
|
if (segments[0].Trim() != key)
|
|
continue;
|
|
collection.Add($"{fileName}\t{line}");
|
|
}
|
|
log.LogInformation("<{fileName}>", fileName);
|
|
}
|
|
if (collection.Count > 0)
|
|
File.WriteAllLines(Path.Combine(destinationDirectory, $"{key}.txt"), collection);
|
|
}
|
|
|
|
internal static void ParseSave(ILogger log, List<string> args)
|
|
{
|
|
// <ItemGroup>
|
|
// <PackageReference Include="Infineon.Mesa.PDF.Text.Stripper" Version="4.8.0.1"><NoWarn>NU1701</NoWarn></PackageReference>
|
|
// </ItemGroup>
|
|
// <ItemGroup>
|
|
// <None Include="\\mestsa003.infineon.com\EC_EAFRepository\Staging\DeploymentStorage\GhostPCL\gpcl6win64\gpcl6dll64.dll">
|
|
// <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
|
|
// </None>
|
|
// <None Include="\\mestsa003.infineon.com\EC_EAFRepository\Staging\DeploymentStorage\GhostPCL\gpcl6win64\gpcl6win64.exe">
|
|
// <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
|
|
// </None>
|
|
// <None Include="\\mestsa003.infineon.com\EC_EAFRepository\Staging\DeploymentStorage\LincPDFC\v2.6.6.21\LincPDFC.exe">
|
|
// <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
|
|
// </None>
|
|
// </ItemGroup>
|
|
string sourceDirectory = Path.GetFullPath(args[0]);
|
|
Input input = GetInput(args);
|
|
if (string.IsNullOrEmpty(input.DestinationDirectory))
|
|
log.LogInformation("-d <{DestinationDirectory}> wasn't supplied!", nameof(input.DestinationDirectory));
|
|
else
|
|
{
|
|
if (input.Parser is null)
|
|
log.LogInformation("-p <{Parser}> wasn't supplied!", nameof(input.Parser));
|
|
else
|
|
{
|
|
string pdfTextStripperFileName = Path.Combine(AppContext.BaseDirectory, "PDF-Text-Stripper.exe");
|
|
if (input.Parser == _Stripper && !File.Exists(pdfTextStripperFileName))
|
|
log.LogInformation("exe <{pdfTextStripperFileName}> doesn't exist!", pdfTextStripperFileName);
|
|
else
|
|
{
|
|
string ghostPCLFileName = Path.Combine(AppContext.BaseDirectory, "gPcl6win64.exe");
|
|
if (input.Parser == _Ghost && !File.Exists(ghostPCLFileName))
|
|
log.LogInformation("exe <{ghostPCLFileName}> doesn't exist!", ghostPCLFileName);
|
|
else
|
|
{
|
|
string kofaxFileName = "C:/Program Files (x86)/Kofax/Power PDF 50/batchConverter.com";
|
|
if (input.Parser == _Kofax && !File.Exists(kofaxFileName))
|
|
log.LogInformation("exe <{kofaxFileName}> doesn't exist!", kofaxFileName);
|
|
else
|
|
{
|
|
string lincPDFCFileName = Path.Combine(AppContext.BaseDirectory, "LincPDFC.exe");
|
|
if (input.Parser == _Kofax && !File.Exists(lincPDFCFileName))
|
|
log.LogInformation("exe <{lincPDFCFileName}> doesn't exist!", lincPDFCFileName);
|
|
else
|
|
{
|
|
string[] files = Directory.GetFiles(sourceDirectory, "*.pdf", SearchOption.TopDirectoryOnly);
|
|
if (files.Length == 0)
|
|
log.LogInformation("Length == {length}", files.Length);
|
|
else
|
|
ParseSave(log, pdfTextStripperFileName, ghostPCLFileName, kofaxFileName, lincPDFCFileName, input.DestinationDirectory, input.Parser.Value, files);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
internal static void ParseStrip(ILogger log, List<string> args)
|
|
{
|
|
if (DateTime.Now > new DateTime(2023, 9, 15))
|
|
log.LogInformation("This helper was a short term helper!");
|
|
else
|
|
{
|
|
Input input = GetInput(args);
|
|
if (string.IsNullOrEmpty(input.DestinationDirectory))
|
|
log.LogInformation("-d <{DestinationDirectory}> wasn't supplied!", nameof(input.DestinationDirectory));
|
|
else
|
|
{
|
|
if (string.IsNullOrEmpty(input.Key))
|
|
log.LogInformation("-k <{key}> wasn't supplied!", nameof(input.Key));
|
|
else
|
|
{
|
|
string[] files = Directory.GetFiles(input.DestinationDirectory, "*.txt", SearchOption.TopDirectoryOnly);
|
|
if (files.Length == 0)
|
|
log.LogInformation("Length == {length}", files.Length);
|
|
else
|
|
ParseStrip(log, input.DestinationDirectory, input.Key, files);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
} |