61 lines
2.9 KiB
C#
61 lines
2.9 KiB
C#
using DocumentOperator.Application.Common.DTOs;
|
|
|
|
namespace DocumentOperator.Application.Common.Interfaces;
|
|
|
|
public interface IPdfProcessor
|
|
{
|
|
/// <summary>
|
|
/// Validates a PDF and extracts metadata.
|
|
/// </summary>
|
|
/// <param name="pdfStream">
|
|
/// PDF document stream. Must be readable and positioned at the beginning (Position = 0).
|
|
/// Non-seekable streams are supported. Caller is responsible for disposal.
|
|
/// </param>
|
|
/// <returns>PDF metadata (page count, size, version, attachments)</returns>
|
|
/// <exception cref="Domain.Common.Exceptions.BadRequestException">
|
|
/// Thrown when stream is empty, invalid, or not positioned at the beginning
|
|
/// </exception>
|
|
Task<PdfMetadata> ValidateAsync(Stream pdfStream);
|
|
|
|
/// <summary>
|
|
/// Validates a PDF/A document and checks conformance level.
|
|
/// </summary>
|
|
/// <param name="pdfStream">
|
|
/// PDF document stream. Must be readable and positioned at the beginning (Position = 0).
|
|
/// Non-seekable streams are supported. Caller is responsible for disposal.
|
|
/// </param>
|
|
/// <returns>PDF/A metadata including conformance level and validation errors/warnings</returns>
|
|
/// <exception cref="Domain.Common.Exceptions.BadRequestException">
|
|
/// Thrown when stream is empty, invalid, or not positioned at the beginning
|
|
/// </exception>
|
|
Task<PdfAMetadata> ValidatePdfAAsync(Stream pdfStream);
|
|
|
|
/// <summary>
|
|
/// Checks for embedded files (attachments) in a PDF document and returns detailed metadata.
|
|
/// </summary>
|
|
/// <param name="pdfStream">
|
|
/// PDF document stream. Must be readable and positioned at the beginning (Position = 0).
|
|
/// Non-seekable streams are supported. Caller is responsible for disposal.
|
|
/// </param>
|
|
/// <returns>Attachment information (count, file names, MIME types, sizes)</returns>
|
|
/// <exception cref="Domain.Common.Exceptions.BadRequestException">
|
|
/// Thrown when stream is empty, invalid, or not positioned at the beginning
|
|
/// </exception>
|
|
Task<AttachmentInfo> CheckAttachmentsAsync(Stream pdfStream);
|
|
|
|
/// <summary>
|
|
/// Extracts all embedded files from a PDF document and returns them as a ZIP archive.
|
|
/// </summary>
|
|
/// <param name="pdfStream">
|
|
/// PDF document stream. Must be readable and positioned at the beginning (Position = 0).
|
|
/// Non-seekable streams are supported. Caller is responsible for disposal.
|
|
/// </param>
|
|
/// <returns>ZIP archive containing all extracted attachments as byte array</returns>
|
|
/// <exception cref="Domain.Common.Exceptions.BadRequestException">
|
|
/// Thrown when stream is empty, invalid, or not positioned at the beginning
|
|
/// </exception>
|
|
/// <exception cref="Domain.Common.Exceptions.NotFoundException">
|
|
/// Thrown when PDF contains no attachments
|
|
/// </exception>
|
|
Task<byte[]> ExtractAttachmentsAsync(Stream pdfStream);
|
|
} |