WIP: PDF Attachment Extractor

GUIs.Test.ZUGFeRDTest/Form1.Designer.vb

@@ -32,6 +32,7 @@ Partial Class Form1
         Me.Button5 = New System.Windows.Forms.Button()
         Me.GroupBox1 = New System.Windows.Forms.GroupBox()
         Me.GroupBox2 = New System.Windows.Forms.GroupBox()
+        Me.Button7 = New System.Windows.Forms.Button()
         Me.GroupBox3 = New System.Windows.Forms.GroupBox()
         Me.txtPropName = New System.Windows.Forms.TextBox()
         Me.GroupBox4 = New System.Windows.Forms.GroupBox()
@@ -60,7 +61,7 @@ Partial Class Form1
         Me.ListBox1.FormattingEnabled = True
         Me.ListBox1.Location = New System.Drawing.Point(378, 12)
         Me.ListBox1.Name = "ListBox1"
-        Me.ListBox1.Size = New System.Drawing.Size(526, 407)
+        Me.ListBox1.Size = New System.Drawing.Size(526, 472)
         Me.ListBox1.TabIndex = 1
         '
         'Button2
@@ -122,19 +123,29 @@ Partial Class Form1
         'GroupBox2
         '
         Me.GroupBox2.Controls.Add(Me.Button2)
+        Me.GroupBox2.Controls.Add(Me.Button7)
         Me.GroupBox2.Controls.Add(Me.Button3)
         Me.GroupBox2.Location = New System.Drawing.Point(12, 110)
         Me.GroupBox2.Name = "GroupBox2"
-        Me.GroupBox2.Size = New System.Drawing.Size(360, 90)
+        Me.GroupBox2.Size = New System.Drawing.Size(360, 155)
         Me.GroupBox2.TabIndex = 8
         Me.GroupBox2.TabStop = False
         Me.GroupBox2.Text = "Run Functions on a single file (needs Breakpoint)"
         '
+        'Button7
+        '
+        Me.Button7.Location = New System.Drawing.Point(6, 77)
+        Me.Button7.Name = "Button7"
+        Me.Button7.Size = New System.Drawing.Size(221, 23)
+        Me.Button7.TabIndex = 3
+        Me.Button7.Text = "Extract PDF Attachments"
+        Me.Button7.UseVisualStyleBackColor = True
+        '
         'GroupBox3
         '
         Me.GroupBox3.Controls.Add(Me.Button4)
         Me.GroupBox3.Controls.Add(Me.txtMD5Checksum)
-        Me.GroupBox3.Location = New System.Drawing.Point(12, 206)
+        Me.GroupBox3.Location = New System.Drawing.Point(12, 271)
         Me.GroupBox3.Name = "GroupBox3"
         Me.GroupBox3.Size = New System.Drawing.Size(360, 85)
         Me.GroupBox3.TabIndex = 9
@@ -152,7 +163,7 @@ Partial Class Form1
         '
         Me.GroupBox4.Controls.Add(Me.Button6)
         Me.GroupBox4.Controls.Add(Me.txtPropName)
-        Me.GroupBox4.Location = New System.Drawing.Point(12, 297)
+        Me.GroupBox4.Location = New System.Drawing.Point(12, 362)
         Me.GroupBox4.Name = "GroupBox4"
         Me.GroupBox4.Size = New System.Drawing.Size(360, 122)
         Me.GroupBox4.TabIndex = 11
@@ -172,7 +183,7 @@ Partial Class Form1
         '
         Me.AutoScaleDimensions = New System.Drawing.SizeF(6.0!, 13.0!)
         Me.AutoScaleMode = System.Windows.Forms.AutoScaleMode.Font
-        Me.ClientSize = New System.Drawing.Size(916, 435)
+        Me.ClientSize = New System.Drawing.Size(916, 492)
         Me.Controls.Add(Me.GroupBox4)
         Me.Controls.Add(Me.GroupBox3)
         Me.Controls.Add(Me.GroupBox2)
@@ -204,4 +215,5 @@ Partial Class Form1
     Friend WithEvents txtPropName As TextBox
     Friend WithEvents GroupBox4 As GroupBox
     Friend WithEvents Button6 As Button
+    Friend WithEvents Button7 As Button
 End Class

GUIs.Test.ZUGFeRDTest/Form1.vb

@@ -85,6 +85,7 @@ Public Class Form1
         Dim args As New WorkerArgs()
         args = LoadFolderConfig(args)
         args = LoadPropertyMapFor(args, "DEFAULT")
+        args.GDPictureKey = "21182889975216572111813147150675976632"
 
         Dim job As New Jobs.ImportZUGFeRDFiles(_logConfig, _firebird)
 
@@ -138,7 +139,7 @@ Public Class Form1
     End Function
 
     Private Sub Button5_Click(sender As Object, e As EventArgs) Handles Button5.Click
-        Process.Start("\\dd-sto01\DD-STO01-A2\SharedObjects\Public\Projekte\Test\Import\ZUGFerD\Email_in")
+        Process.Start("\\dd-sto01\DD-DFSR01\SharedObjects\Public\Projekte\Test\Import\ZUGFerD\Email_in")
     End Sub
 
     Private Sub Button6_Click(sender As Object, e As EventArgs) Handles Button6.Click
@@ -160,4 +161,15 @@ Public Class Form1
             End Try
         End If
     End Sub
+
+    Private Sub Button7_Click(sender As Object, e As EventArgs) Handles Button7.Click
+        Dim oExtractor = New Jobs.PDFAttachments(_logConfig, "21182889975216572111813147150675976632")
+
+        Dim oResult = OpenFileDialog1.ShowDialog()
+
+        If oResult = DialogResult.OK Then
+            oExtractor.Extract(OpenFileDialog1.FileName, AllowedExtensions:=New List(Of String) From {"docx", "doc", "pdf", "xls", "xlsx", "ppt", "pptx", "txt"})
+        End If
+
+    End Sub
 End Class

Modules.Jobs/EDMI/ZUGFeRD/ImportZUGFeRDFiles.vb

@@ -42,6 +42,9 @@ Public Class ImportZUGFeRDFiles
         </ul></p>
     """
 
+    ' List of allowed extensions for PDF/A Attachments
+    Private AllowedExtensions = New List(Of String) From {"docx", "doc", "pdf", "xls", "xlsx", "ppt", "pptx", "txt"}
+
     Private _logger As Logger
     Private _logConfig As LogConfig
     Private _zugferd As ZUGFeRDInterface
@@ -318,6 +321,7 @@ Public Class ImportZUGFeRDFiles
     Public Sub Start(Arguments As Object) Implements IJob.Start
         Dim oArgs As WorkerArgs = Arguments
         Dim oPropertyExtractor = New PropertyValues(_logConfig)
+        Dim oAttachmentExtractor = New PDFAttachments(_logConfig, oArgs.GDPictureKey)
 
         _logger.Debug("Starting Job {0}", [GetType].Name)
 
@@ -404,10 +408,16 @@ Public Class ImportZUGFeRDFiles
                                             _logger.Warn("Unexpected Error occurred while extracting ZUGFeRD Information from file {0}", oFile.FullName)
                                             Throw ex
                                     End Select
-
-
                                 End Try
 
+                                Dim oAttachments = oAttachmentExtractor.Extract(oFile.FullName, AllowedExtensions)
+                                If oAttachments Is Nothing Then
+                                    _logger.Warn("Attachments for file [{0}] could not be extracted", oFile.FullName)
+                                Else
+                                    oFileAttachmentFiles.AddRange(oFileGroupFiles)
+                                    oFileAttachmentFiles.AddRange(oAttachments)
+                                End If
+
                                 oMD5CheckSum = CreateMD5(oFile.FullName)
                                 If oMD5CheckSum <> String.Empty Then
                                     Dim oCheckCommand = $"SELECT * FROM TBEDM_ZUGFERD_HISTORY_IN WHERE GUID = (SELECT MAX(GUID) FROM TBEDM_ZUGFERD_HISTORY_IN WHERE UPPER(MD5HASH) = UPPER('{oMD5CheckSum}'))"

Modules.Jobs/EDMI/ZUGFeRD/PDFAttachments.vb

@@ -1,9 +1,84 @@
-Public Class PDFAttachments
+Imports System.Collections.Generic
-    Public Sub New(GdPictureKey As String)
+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)
+    Public Function Extract(FileName As String, AllowedExtensions As List(Of String)) As List(Of FileInfo)
-        Using oGDPicturePDF As New GDPicturePDF
+        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

Modules.Jobs/EDMI/ZUGFeRD/WorkerArgs.vb

@@ -9,6 +9,7 @@ Public Class WorkerArgs
     Public AttachmentsSubDirectory As String
     Public PropertyMap As Dictionary(Of String, XmlItemProperty)
     Public InsertIntoSQLServer As Boolean
+    Public GDPictureKey As String
 
     Public Sub New()
         WatchDirectories = New List(Of String)
@@ -19,5 +20,6 @@ Public Class WorkerArgs
         AttachmentsSubDirectory = Nothing
         PropertyMap = New Dictionary(Of String, XmlItemProperty)
         InsertIntoSQLServer = False
+        GDPictureKey = String.Empty
     End Sub
 End Class

Modules.Jobs/Jobs.vbproj

@@ -109,6 +109,9 @@
     <Reference Include="FirebirdSql.Data.FirebirdClient, Version=6.4.0.0, Culture=neutral, PublicKeyToken=3750abcc3150b00c, processorArchitecture=MSIL">
       <HintPath>..\packages\FirebirdSql.Data.FirebirdClient.6.4.0\lib\net452\FirebirdSql.Data.FirebirdClient.dll</HintPath>
     </Reference>
+    <Reference Include="GdPicture.NET.14">
+      <HintPath>D:\ProgramFiles\GdPicture.NET 14\Redist\GdPicture.NET (.NET Framework 4.5)\GdPicture.NET.14.dll</HintPath>
+    </Reference>
     <Reference Include="Microsoft.CSharp" />
     <Reference Include="NLog, Version=4.0.0.0, Culture=neutral, PublicKeyToken=5120e14c03d0593c, processorArchitecture=MSIL">
       <HintPath>..\packages\NLog.4.6.8\lib\net45\NLog.dll</HintPath>