WIP: PDF Attachment Extractor

Modules.Jobs/EDMI/ZUGFeRD/PDFAttachments.vb

@@ -1,9 +1,84 @@
-Public Class PDFAttachments
-    Public Sub New(GdPictureKey As String)
+Imports System.Collections.Generic
+Imports System.IO
+Imports DigitalData.Modules.Logging
+Imports GdPicture14
 
+Public Class PDFAttachments
+    Private Logger As Logger
+
+    Private Const ZUGFERD_XML_FILENAME = "ZUGFeRD-invoice.xml"
+
+    Public Sub New(LogConfig As LogConfig, GdPictureKey As String)
+        Logger = LogConfig.GetLogger
     End Sub
 
-    Public Shared Function Extract(FileName As String)
-        Using oGDPicturePDF As New GDPicturePDF
+    Public Function Extract(FileName As String, AllowedExtensions As List(Of String)) As List(Of FileInfo)
+        Dim oResults As New List(Of FileInfo)
+        Dim oExtensions = AllowedExtensions.ConvertAll(Of String)(New Converter(Of String, String)(Function(ext) ext.ToUpper))
+
+        Try
+            Using oGDPicturePDF As New GdPicturePDF()
+                If oGDPicturePDF.LoadFromFile(FileName, False) = GdPictureStatus.OK Then
+                    Dim oEmbeddedFileCount As Integer = oGDPicturePDF.GetEmbeddedFileCount()
+                    If oGDPicturePDF.GetStat() = GdPictureStatus.OK Then
+                        If oEmbeddedFileCount > 1 Then
+                            For index = 0 To oEmbeddedFileCount - 1
+                                Dim oFileName As String = oGDPicturePDF.GetEmbeddedFileName(index)
+
+                                If oGDPicturePDF.GetStat() = GdPictureStatus.OK Then
+                                    Dim oExtension = New FileInfo(oFileName).Extension.ToUpper.Substring(1)
+                                    If oFileName.ToUpper <> ZUGFERD_XML_FILENAME.ToUpper Then
+                                        If oExtensions.Contains(oExtension) Then
+                                            Dim FileSize As Integer = oGDPicturePDF.GetEmbeddedFileSize(index)
+
+                                            If oGDPicturePDF.GetStat() = GdPictureStatus.OK Then
+                                                Dim FileData As Byte() = New Byte(FileSize) {}
+                                                Dim status As GdPictureStatus = oGDPicturePDF.ExtractEmbeddedFile(index, FileData)
+
+                                                If status = GdPictureStatus.OK Then
+                                                    Dim oTempName As String = Path.Combine(Path.GetTempPath(), oFileName)
+                                                    Using oFileStream As New FileStream(oTempName, FileMode.OpenOrCreate)
+                                                        oFileStream.Write(FileData, 0, FileData.Length)
+                                                    End Using
+
+                                                    oResults.Add(New FileInfo(oTempName))
+                                                Else
+                                                    Logger.Error("The embedded file [{0}] has failed to extract. Status: {1}", oFileName, oGDPicturePDF.GetStat().ToString())
+                                                    Continue For
+                                                End If
+                                            Else
+                                                Logger.Error("An error occurred getting the file size for [{0}]. Status: {1}", oFileName, oGDPicturePDF.GetStat().ToString())
+                                                Continue For
+                                            End If
+                                        Else
+                                            Logger.Warn("File [{0}] was skipped because its extension [{1}] is not allowed.", oFileName, oExtension)
+                                            Continue For
+                                        End If
+                                    Else
+                                        Logger.Debug("File [{0}] was skipped because its name indicates the invoice data file.", oFileName)
+                                        Continue For
+                                    End If
+                                Else
+                                    Logger.Error("An error occurred getting the file name for [{0}]. Status: {1}", oFileName, oGDPicturePDF.GetStat().ToString())
+                                    Continue For
+                                End If
+                            Next
+                        End If
+                    Else
+                        Logger.Error("An error occurred getting the number of embedded files. Status: {0}", oGDPicturePDF.GetStat().ToString())
+                        Return Nothing
+                    End If
+                Else
+                    Logger.Error("The file [{0}] can't be loaded.", FileName)
+                    Return Nothing
+                End If
+            End Using
+
+            Return oResults
+        Catch ex As Exception
+            Logger.Warn("Unexpected Error while Extracting attachments from File [{0}]", FileName)
+            Logger.Error(ex)
+            Return Nothing
+        End Try
     End Function
 End Class