Options
  • Locked Locked
  • Replies 7 replies
  • Subscribers 5 subscribers
  • Views 670 views
  • Users 0 members are here
Related
This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

OCR mail attachment

Former Member
Former Member
In your opinion, what's the best/cleanest way to OCR mail-attachments (TIF files from MFC device) that are imported through an external mail source. So, just an e-mail with a tif-file that need to be indexed.

While the External File Import source has the option to OCR the files that are being imported, the mail connector does not have such thing.

We can use a workflow action to OCR, but this seems to have an unpredictable result where OCR isn't performed as it should, it only creates a PDF, but not a searchable one.

I want to prevent that we have to create a solution where mail source imports have to be downloaded/extreacted by an event handler to a file system folder to be imported again in a regular external file import job, it's a bit cumbersome.
  • Maybe your best best would be to have some utility to detach the TIFF to a folder and have it as the source to your import (that's how we solved this issue...)
  • Former Member
    Former Member

    Maybe your best best would be to have some utility to detach the TIFF to a folder and have it as the source to your import (that's how we solved this issue...)

    That's a scenario I thought of, but to be honest, I don't understand why the OCR option is not a default option in the mail import source, as it is in the file import source.

    Any scripts or hints for utils are very welcome.
  • We can use a workflow action to OCR, but this seems to have an unpredictable result where OCR isn't performed as it should, it only creates a PDF, but not a searchable one.


    Did you run the OCR operation in a workflow state action script or just use the built-in PDF conversion option in the workflow state? The PDF conversion doesn't do OCR as far as I know, you need to trigger OCR in a script. Here's an example:

    Option Explicit

    ' Prepare the files of the object for modification by script.
    Dim files
    Set files = Vault.ObjectFileOperations.GetFilesForModificationInEventHandler( ObjVer )

    ' Prepare OCR options.
    Dim opts
    Set opts = CreateObject( "MFilesAPI.OCROptions" )
    opts.PrimaryLanguage = MFOCRLanguageEnglishUS
    opts.SecondaryLanguage = MFOCRLanguageFinnish

    ' Perform OCR on each of the convertible files.
    Dim file
    For Each file In files

    	' Is the file in a convertible file format?
    	If file.Extension = "tif" Or _
    		file.Extension = "tiff" Or _
    		file.Extension = "jpg" Or _
    		file.Extension = "jpeg" Or _
    		file.Extension = "pdf" Then

    		' Convert this file to searchable PDF.
    		Vault.ObjectFileOperations.PerformOCROperation ObjVer, file.FileVer, _
    				opts, MFOCRZoneRecognitionModeNoZoneRecognition, Nothing, True

    	End If

    Next
  • Former Member
    Former Member
    Hi Joonas, the script looks promising, I'll give it a go. Thanks!
  • Former Member
    Former Member
    Hi Joonas,

    1:

    Basically the OCR is working, however sometimes I get eventlog errors regarding a timeout in the OCR Action Script when OCR'ing somewhat larger tif files (4MB+).

    Vault.ObjectFileOperations.PerformOCROperation ObjVer, file.FileVer, opts, MFOCRZoneRecognitionModeNoZoneRecognition, Nothing, True


    Is there some tuning possible, because these aren't excessive file sizes.

    2:

    the script sometimes crashes on PDF files that have a digital origin, like a tekst document that was printed to PDF format.

    How would one check if a pdf file is image only, image+text (already OCR'ed), or only digital text ?
  • 1. The OCR tuning options are listed here: Registry settings for scanning and OCR in M-Files

    If it seems the workflow action scripts are taking too long to complete and tuning doesn't help, the next option would be to create a VAF background operation that searches the vault for objects to OCR (based on the workflow state, for instance): developer.m-files.com/.../

    2. There is no function in M-Files API to check if a PDF is already searchable. What you could do programmatically is to check if the PDF file contains strings like "FontName", that typically indicates that a PDF is searchable.
  • There is no function in M-Files API to check if a PDF is already searchable. What you could do programmatically is to check if the PDF file contains strings like "FontName", that typically indicates that a PDF is searchable.


    Bit of of a thread drift but I'm sure this has been already been discussed multiple times here - it wold be a very welcome addition (ideally a built in property "is searchable").




              

© M-Files Corporation 2024 | Privacy Policy