met08resihgcv/Adaptation/FileHandlers/pcl/Convert.cs

using Adaptation.Shared;
using System;
using System.Collections.Generic;
using System.Collections.ObjectModel;
using System.Diagnostics;
using System.IO;

namespace Adaptation.FileHandlers.pcl;

internal class Convert
{

    /// <summary>
    /// Convert the raw data file to parsable file format - in this case from PCL to PDF
    /// </summary>
    /// <param name="sourceFile">source file to be converted to PDF</param>
    /// <returns></returns>
    private static string ConvertSourceFileToPdf(Logistics logistics, string ghostPCLFileName)
    {
        string result = Path.ChangeExtension(logistics.ReportFullPath, ".pdf");
        if (!File.Exists(result))
        {
            //string arguments = string.Concat("-i \"", sourceFile, "\" -o \"", result, "\"");
            string arguments = string.Concat("-dSAFER -dBATCH -dNOPAUSE -dFIXEDMEDIA -dFitPage -dAutoRotatePages=/All -dDEVICEWIDTHPOINTS=792 -dDEVICEHEIGHTPOINTS=612 -sOutputFile=\"", result, "\" -sDEVICE=pdfwrite \"", logistics.ReportFullPath, "\"");
            //Process process = Process.Start(configData.LincPDFCFileName, arguments);
            Process process = Process.Start(ghostPCLFileName, arguments);
            _ = process.WaitForExit(30000);
            if (!File.Exists(result))
                throw new Exception("PDF file wasn't created");
        }
        return result;
    }

    private static ReadOnlyCollection<string> GetTextFromPDF(string pdfTextStripperFileName, string sourceFileNamePdf, string altHeaderFileName)
    {
        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(altHeaderFileName))
            result = Array.Empty<string>();
        else
            result = File.ReadAllLines(altHeaderFileName);
        return new(result);
    }

    internal static ReadOnlyCollection<string> PDF(Logistics logistics, string ghostPCLFileName, string pdfTextStripperFileName, List<FileInfo> fileInfoCollection)
    {
        ReadOnlyCollection<string> result;
        string sourceFileNamePdf = ConvertSourceFileToPdf(logistics, ghostPCLFileName);
        fileInfoCollection.Add(new FileInfo(sourceFileNamePdf));
        string altHeaderFileName = Path.ChangeExtension(logistics.ReportFullPath, ".txt");
        if (File.Exists(altHeaderFileName))
        {
            result = new(File.ReadAllLines(altHeaderFileName));
            fileInfoCollection.Add(new FileInfo(altHeaderFileName));
        }
        else
        {
            try
            {
                //Pdfbox, IKVM.AWT.WinForms
                org.apache.pdfbox.pdmodel.PDDocument pdfDocument = org.apache.pdfbox.pdmodel.PDDocument.load(sourceFileNamePdf);
                org.apache.pdfbox.util.PDFTextStripper stripper = new();
                string text = stripper.getText(pdfDocument);
                pdfDocument.close();
                File.AppendAllText(altHeaderFileName, text);
                fileInfoCollection.Add(new FileInfo(altHeaderFileName));
                result = new(text.Split(new string[] { Environment.NewLine }, StringSplitOptions.None));
            }
            catch (MissingMethodException)
            {
                if (!File.Exists(pdfTextStripperFileName))
                    throw;
                result = GetTextFromPDF(pdfTextStripperFileName, sourceFileNamePdf, altHeaderFileName);
                if (result.Count == 0)
                    throw;
                fileInfoCollection.Add(new FileInfo(altHeaderFileName));
            }
        }
        return result;
    }

}