Files
monorepo/LCU.API.Internal.FileOperations/LCU.Api.Internal.FileOperations/Services/PdfServices/PdfParserService.cs
T
2026-04-27 10:50:07 -05:00

47 lines
1.9 KiB
C#

using System;
using System.IO;
using System.Text;
using PdfSharp.Pdf;
using PdfSharp.Pdf.IO;
using Serilog;
using Newtonsoft.Json;
using LCU.Library.Internal.Exceptions;
using LCU.Library.Internal.FileOperationsModels.Models;
namespace LCU.Api.Internal.FileOperations.Services.PdfServices;
public class PdfParserService : IPdfParserService
{
public string ParsePdf(ParsePdfRequest parsePdfRequest)
{
try
{
var fullPath = Path.Combine(parsePdfRequest.RootPath, parsePdfRequest.PdfName);
var inputDocument = PdfReader.Open(fullPath, PdfDocumentOpenMode.Import);
var startIndex = parsePdfRequest.PageStart - 1;
var stopIndex = parsePdfRequest.PageStop - 1;
var outputDocument = new PdfDocument();
for (int page = startIndex; page <= stopIndex; page++)
{
outputDocument.Version = inputDocument.Version;
outputDocument.Info.Title = inputDocument.Info.Title;
outputDocument.Info.Creator = inputDocument.Info.Creator;
outputDocument.AddPage(inputDocument.Pages[page]);
}
string name = Path.GetFileNameWithoutExtension(parsePdfRequest.PdfName);
var newPdfName = string.IsNullOrWhiteSpace(parsePdfRequest.ParsedPdfname) ? $"{name}.{parsePdfRequest.PageStart}.pdf" : parsePdfRequest.ParsedPdfname;
var newPath = Path.Combine(parsePdfRequest.RootPath, newPdfName);
Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
outputDocument.Save(newPath);
return newPdfName;
}
catch (Exception exception)
{
Log.Error(exception, $"Failure to Parse Pdf. ParsePdfRequest:{JsonConvert.SerializeObject(parsePdfRequest)}");
throw new ServiceException($"Failure to Parse Pdf. ExceptionMessage:{exception.Message}");
}
}
}