Anlage des Repos

archive/Get-PDFContentAndWriteToDB/Get-PDFContentAndWriteToDB.ps1

@@ -0,0 +1,219 @@
+#PowerShell 4.0 Script
+
+#Digital Data
+#Ludwig-Rinn-Strasse 16
+#35452 Heuchelheim
+#Tel.: 0641 / 202360
+#E-Mail: info@digitaldata.works
+
+#Version Number 1.0.0.0
+#Version Date 20.11.2020
+
+#Requires –Version 4.0
+
+#-----------------------------------------------------------------------------------------------------#
+######################################## check for arguments ##########################################
+#-----------------------------------------------------------------------------------------------------#
+
+#-----------------------------------------------------------------------------------------------------#
+################################## add additional buildin assemblys ###################################
+#-----------------------------------------------------------------------------------------------------#
+
+#-----------------------------------------------------------------------------------------------------#
+############################################ set variables ############################################
+#-----------------------------------------------------------------------------------------------------#
+
+[string]$PDFExtractorDLL				= "E:\itextsharp\itextsharp.dll"
+
+[string]$SourcePath						= "E:\itextsharp\in"
+[string]$ArchivPath						= "E:\itextsharp\out"
+[string]$ErrorPath						= "E:\itextsharp\error"
+
+[string]$SourceFileExtension			= "*.pdf"
+
+[string]$DBSQLConnectServer				= "172.24.12.41\tests"
+[string]$DBSQLConnectDatabase			= "DD_Test"
+[string]$DBSQLConnectUser				= "sa"
+[string]$DBSQLConnectPassword			= "dd"
+[string]$DBSQLConnectIntegratedSecurity	= $false
+[string]$DBSQLQueryHead 				= "INSERT INTO [itextsharp] (FILENAME, FILECONTENT)" 
+
+[int]$Counter = 0
+
+#-----------------------------------------------------------------------------------------------------#
+########################################### preparing part ############################################
+#-----------------------------------------------------------------------------------------------------#
+
+#Clear Console Content 
+Clear-Host
+
+#get all relevant files
+[Array]$Items = Get-ChildItem -Path $SourcePath -Filter $SourceFileExtension
+
+#-----------------------------------------------------------------------------------------------------#
+############################################# main part ###############################################
+#-----------------------------------------------------------------------------------------------------#
+
+IF ($Items.Count -ge 1) {
+
+	Write-Host "Found" $Items.Count $SourceFileExtension.Replace("*.","")"Files!"
+
+	TRY {
+	
+		$DBSQLConnection = New-Object System.Data.SqlClient.SqlConnection
+	    $DBSQLConnection.ConnectionString = "Server = $DBSQLConnectServer; uid=$DBSQLConnectUser; pwd=$DBSQLConnectPassword; Database = $DBSQLConnectDatabase; Integrated Security = $DBSQLConnectIntegratedSecurity"
+	   	
+		$DBSQLCommand = New-Object System.Data.SqlClient.SqlCommand
+		$DBSQLCommand.Connection = $DBSQLConnection
+
+	} #end try
+		
+	CATCH {
+	
+		Write-Host "Cannot prepare db connect!"
+		Write-Host $DBSQLConnectServer
+		Write-Host $DBSQLConnectDatabase
+		Write-Host $DBSQLConnectUser
+		Write-Host $DBSQLConnectPassword
+		Write-Host $DBSQLConnectIntegratedSecurity
+		Write-Host $DBSQLQueryHead		
+		Write-Error $Error[0]
+		Remove-Variable * -ErrorAction SilentlyContinue
+		$Error.Clear()
+		EXIT
+	
+	} #end catch
+
+	FOREACH ($Item in $Items) {
+
+		[int]$Counter++ | Out-Null
+
+		Write-Host ""
+		Write-Host "================================="
+		Write-Host "---------------------------------"
+		Write-Host "Item" $Counter "of" $Items.count
+		Write-Host "Processing File:" $Item
+
+		TRY {
+
+			$PDFExtractJob = Start-Job   -ScriptBlock {
+			
+				$PDFExtractorDLL = $args[0]
+				$ItemFullName 	 = $args[1]
+
+				Add-Type -path $PDFExtractorDLL
+				$PDFReader = [iTextSharp.text.pdf.parser.PdfTextExtractor]
+				$PDFExtract = $NULL
+				$PDFExtract = $PDFReader::GetTextFromPage($ItemFullName,1)
+						
+				$PDFExtract = $PDFExtract.TrimStart()
+				$PDFExtract = $PDFExtract.TrimEnd()
+				$PDFExtract = $PDFExtract.Replace("'","")
+				$PDFExtract = $PDFExtract.Replace('"','')
+				$PDFExtract = $PDFExtract.Replace("‘","")
+				$PDFExtract = $PDFExtract.Replace("’","")
+				$PDFExtract = $PDFExtract.Replace("·","")
+			    
+				Write-Output $PDFExtract
+				
+			} -ArgumentList $PDFExtractorDLL, $($Item.FullName) 
+
+			
+		$PDFExtract = Receive-Job -Job $PDFExtractJob -Wait
+			
+		} #end try
+		
+		CATCH {
+		
+			Write-Host "Cannot read file fulltext!"
+			Write-Error $Error[0]
+			Remove-Variable * -ErrorAction SilentlyContinue
+			$Error.Clear()
+
+		} #end catch
+		
+		IF ($PDFExtract.Length -gt 20) {
+		
+			Write-Host "File has a valid Fulltext!"
+
+			TRY {
+
+				$DBSQLQuery = $NULL
+				$DBSQLQuery = $DBSQLQueryHead
+				$DBSQLQuery = $DBSQLQuery + "VALUES ('$($item.BaseName)', '$PDFExtract');" 
+				
+				Write-Host "Executing SQL Query..."
+				
+				$DBSQLCommand = New-Object System.Data.SqlClient.SqlCommand
+				$DBSQLCommand.Connection = $DBSQLConnection
+		
+				$DBSQLCommand.CommandText = $DBSQLQuery
+
+		        $DBSQLConnection.Open()
+		        $DBSQLCommand.ExecuteNonQuery() | Out-Null
+		        $DBSQLConnection.Close()
+				
+				Move-Item -Path $($Item.fullname) -Destination $ArchivPath -Force
+				Write-Host "... done!"
+			
+			} #end try
+			
+			CATCH {
+				
+				$LastErrorQuery = $DBSQLQuery
+				Move-Item -path $($Item.fullname) -Destination $ErrorPath -Force
+			
+				$DBSQLConnection.Close()
+				Write-Host "Cannot insert to db!"
+				Write-Host "Moving file to error path."
+				Write-Error $Error[0]
+				$Error.Clear()
+			
+			} #end catch
+			
+			Write-Host "---------------------------------"
+			Write-Host "================================="			
+
+		} #end if
+		
+		ELSE {
+		
+			Write-Host "File has no or an invalid Fulltext!"
+		
+		} #end else
+
+	} #end foreach
+
+} #end of 
+
+ELSE {
+
+	Write-Host "No files to process!"
+
+} #end else
+
+Remove-Variable * -ErrorAction SilentlyContinue
+$Error.Clear()
+
+
+#Sources:
+#https://vasol.eu/working-with-pdfs-with-powershell-in-run-net-script-activites/
+
+#Create SQL TB
+#SET ANSI_NULLS ON
+#GO
+#
+#SET QUOTED_IDENTIFIER ON
+#GO
+#
+#CREATE TABLE [dbo].[itextsharp](
+#	[GUID] [bigint] IDENTITY(1,1) NOT NULL,
+#	[FILENAME] [varchar](50) NOT NULL,
+#	[FILECONTENT] [varchar](max) NOT NULL,
+#	[STATE] [int] NULL,
+#	[ADDED_WHEN] [datetime] NULL
+#) ON [PRIMARY] TEXTIMAGE_ON [PRIMARY]
+#GO
+#
+#ALTER TABLE [dbo].[itextsharp] ADD  CONSTRAINT [DF_itextsharp_ADDED_WHEN]  DEFAULT (getdate()) FOR [ADDED_WHEN]
+#GO