PDF writer: Document properties

Hi @freekdw

I've asked our development team and unfortunately, FME does not natively support writing PDF document properties/metadata. While the PDF reader is able to read document properties, the PDF writer uses a different library that does not support document properties as it was implemented long before the reader.

If you wish to use FME to update PDFs but still carry through the metadata from PDF input files, you will need to post-process the PDF output to add metadata.

I have not tested this, but quick searching suggests it may be possible using PDF utility applications (eg.this page here) or Python libraries (eg. as mentioned here). You would want to combine these with either the SystemCaller or PythonCaller depending on which method used.

Hi @debbiatsafe,

Thanks for asking. Too bad it is not natively supported but good to hear there are options. I'm not that familiar with this kind of coding but I'll have a look into it.

Hey @warrengis,

Wonderful! Thank you so much for sharing this elegant code! This will help me getting the files published.

I am encountering some 'NoneType' objects in my files. It's working fine on some random pdf's I grabbed from the web. It causes a fail of the pythoncode.

2020-04-30 15:53:41| 1.6| 0.5|ERROR |Python Exception <AttributeError>: 'NoneType' object has no attribute 'update'

2020-04-30 15:53:41| 1.6| 0.0|ERROR |Error encountered while calling function `add_metadata'

2020-04-30 15:53:41| 1.6| 0.0|FATAL |PythonCaller (PythonFactory): PythonFactory failed to process feature

2020-04-30 15:53:41| 1.6| 0.0|ERROR |A fatal error has occurred. Check the logfile above for details

Any ideas?

4_JorisIvensplein.pdf

It looks like the structure of the pdf that is causing the error. I don't know how to fix that exactly with pdfrw. After I opened it in my PDF program and saved it back out, it worked.

I did get it to work with another python library pikepdf. Here is the documentation: https://pikepdf.readthedocs.io/en/latest/tutorial.html 

from pikepdf import Pdf
pdf = Pdf.open(r'37028-4-jorisivensplein.pdf')
with pdf.open_metadata() as meta:
    meta['dc:title'] = "Let's change the title"
pdf.save('output.pdf')

 4-jorisivensplein.pdf

from pikepdf import Pdf
pdf = Pdf.open(r'37028-4-jorisivensplein.pdf')
with pdf.open_metadata() as meta:
    meta['dc:title'] = "Let's change the title"
pdf.save('output.pdf')

Hi @warrengis,

I made it work as well. Thanks for your guidance. After installing the pikepdf package the python caller contains the following code:

import fme
import fmeobjects
from pikepdf import Pdf

def add_metadata(feature):
    pdf = Pdf.open(FME_MacroValues['input_pdf'])
    with pdf.open_metadata() as meta:
        meta['dc:title'] = feature.getAttribute('title')
        meta['dc:subject'] = feature.getAttribute('subject')
        meta['dc:creator'] = feature.getAttribute('creator')
        meta['dc:description'] = feature.getAttribute('description')
        meta['xmp:CreateDate'] = feature.getAttribute('creation_date')
        meta['xmp:ModifyDate'] = feature.getAttribute('creation_date')
        meta['xmp:CreatorTool'] = feature.getAttribute('producer')
    pdf.save(FME_MacroValues['output_pdf'])

creation_date is formatted like '%Y-%m-%dT%H:%M:%S%Ez'. The workspace errors on the xmp: metadata but still writes it to the output pdf. I also find the naming of dc:subject and dc:description are not very intuitive. Ah well... We made it.

Only 1 object from the pdf has to enter the pythoncaller so a sampler or maxFeatures can be set to speed up the process.

import fme
import fmeobjects
from pikepdf import Pdf

def add_metadata(feature):
    pdf = Pdf.open(FME_MacroValues['input_pdf'])
    with pdf.open_metadata() as meta:
        meta['dc:title'] = feature.getAttribute('title')
        meta['dc:subject'] = feature.getAttribute('subject')
        meta['dc:creator'] = feature.getAttribute('creator')
        meta['dc:description'] = feature.getAttribute('description')
        meta['xmp:CreateDate'] = feature.getAttribute('creation_date')
        meta['xmp:ModifyDate'] = feature.getAttribute('creation_date')
        meta['xmp:CreatorTool'] = feature.getAttribute('producer')
    pdf.save(FME_MacroValues['output_pdf'])

Glad you got it working!