PDF Portfolio attachments display order #1030

erand96 · 2021-04-23T13:57:29Z

erand96
Apr 23, 2021

Good morning!

I am working on a script to manipulate PDF Portfolios. One issue that I'm having relates to the order that the attachments are displayed (Name, date, order.....etc). Is this "sorted by" criteria found anywhere in the PDF metadata?

Ideally, I would extract the "sorted by" criteria and then use that to reorganize all of the attachments into a singular PDF file in the same order.

Let me know if I need to provide any additional information!

Answered by JorjMcKie

Apr 24, 2021

I have made a few changes to PyMuPDF's embedded files handling.
Note that the portfolio information for each embedded file is contained in object encoded in xref "collection".
If you look at this example file, the following things will be possible in v1.18.13:

>>> doc=fitz.open("portfolio-sample.pdf")
>>> n=doc.embfile_count()
>>> from pprint import pprint
>>> for i in range(n):
	pprint(doc.embfile_info(i))
	print("-"*10)

	
{'checksum': b'c59313c2a309cb9a235fe2809464c39266c3bc',
 'collection': 50,
 'creationDate': "D:20120928172443-04'00'",
 'desc': '',
 'filename': 'facility_budget.xlsx',
 'length': 9633,
 'modDate': "D:20120928172458-04'00'",
 'name': '<0>facility_budget.xlsx',
 'size': …

View full answer

JorjMcKie · 2021-04-23T14:13:38Z

JorjMcKie
Apr 23, 2021
Maintainer

What is a "portfolio" in PDF?

an EmbeddedFiles object?
an annotation?
???

(Standard) metadata is a ultimately well-defined term: a Python dictionary with a fixed set of keys.

4 replies

erand96 Apr 23, 2021
Author

Hi Jorj,

Thank you for the quick reply, hope you're having a good Friday.

A Portfolio is essentially one PDF file that contains multiple attachments which can have various formats. Adobe's overview can be found here.

Below is a screenshot of a portfolio that I currently have open in Adobe.

Additional information on my full project can be found in discussion post #1010

JorjMcKie Apr 23, 2021
Maintainer

I see. A PDF portfolio is a superset of EmbeddedFiles with lots of configurable metadata information.
It used to be supported until some version of MuPDF, but no longer is.
All you can have with PyMuPDF are embedded files, implemented by a set of Document´ methods starting with names like embfile_*. Features include the full set of usual maintenance tasks, but the only metadata per file are what you see in method embfile_add(...)` - no date, time, author, etc. or whatever.

erand96 Apr 23, 2021
Author

That makes sense, thank you for the explanation. Any ideas of other utilities that could help me extract that info?

JorjMcKie Apr 23, 2021
Maintainer

Not at the moment, sorry.
You may be tempted to use PyMuPDF's low-level object access functions doc.xref_get_keys() and doc.xref_get_key() to dig your own way through the strcuture of a given PDF's portfolio items.
Of course you must know how this structure works on a technical level: which objects may exist, how they point to other objects, etc. The starting point should always be the sub-dictionary / array contained in the PDF's catalog, which is accessible via the xref returned by doc.pdf_catalog(). Here is the layout of a PDF catalog showing the 3 " EmbeddedFiles` of this PDF:

1 0 obj
<<
  /Type /Catalog
  /Pages 2 0 R
  /Names <<
    /EmbeddedFiles <<
      /Names [ (file1) <<
            /CI <<
            >>
            /EF <<
              /F 3 0 R
            >>
            /F (file1)
            /UF (file1)
            /Desc (file1)
            /Type /Filespec
          >> (file2) <<
            /CI <<
            >>
            /EF <<
              /F 4 0 R
            >>
            /F (file2)
            /UF (file2)
            /Desc (file2)
            /Type /Filespec
          >> (file3) <<
            /CI <<
            >>
            /EF <<
              /F 5 0 R
            >>
            /F (file3)
            /UF (file3)
            /Desc (file3)
            /Type /Filespec
          >> ]
    >>
  >>
  /PageMode /UseAttachments
>>
endobj

So the array containing the embedded files has the size 2 * number of embedded files. The first of each item pair is the unique item's name, and the second a PDF dictionary describing the file itself.
You can directly access that array with the said functions like so:

xref = doc.pdf_catalog()
t, array = doc.xref_get_key(xref, ""Names/EmbeddedFiles/Names")
# which puts the following string in variable 'array':
[(file1)<</CI ... >>]

JorjMcKie · 2021-04-24T16:02:54Z

JorjMcKie
Apr 24, 2021
Maintainer

I have made a few changes to PyMuPDF's embedded files handling.
Note that the portfolio information for each embedded file is contained in object encoded in xref "collection".
If you look at this example file, the following things will be possible in v1.18.13:

>>> doc=fitz.open("portfolio-sample.pdf")
>>> n=doc.embfile_count()
>>> from pprint import pprint
>>> for i in range(n):
	pprint(doc.embfile_info(i))
	print("-"*10)

	
{'checksum': b'c59313c2a309cb9a235fe2809464c39266c3bc',
 'collection': 50,
 'creationDate': "D:20120928172443-04'00'",
 'desc': '',
 'filename': 'facility_budget.xlsx',
 'length': 9633,
 'modDate': "D:20120928172458-04'00'",
 'name': '<0>facility_budget.xlsx',
 'size': 12019,
 'ufilename': 'facility_budget.xlsx'}
----------
{'checksum': b'6e0c1707112538c2aa65cb9a143f5864e280ba21',
 'collection': 48,
 'creationDate': "D:20120928172410-04'00'",
 'desc': '',
 'filename': 'globalcorp_projectupdate.pptx',
 'length': 772137,
 'modDate': "D:20120928172433-04'00'",
 'name': '<0>globalcorp_projectupdate.pptx',
 'size': 1179462,
 'ufilename': 'globalcorp_projectupdate.pptx'}
----------
{'checksum': b'c384c5a1c3aee2809e74',
 'collection': 198,
 'creationDate': "D:20120928172335-04'00'",
 'desc': '',
 'filename': 'memo.docx',
 'length': 223867,
 'modDate': "D:20120928172359-04'00'",
 'name': '<0>memo.docx',
 'size': 227780,
 'ufilename': 'memo.docx'}
----------
{'checksum': b'33416be2809a5249c2b738c2b0e2809e5b2c767414e28098',
 'collection': 44,
 'creationDate': "D:20120726135901-04'00'",
 'desc': '',
 'filename': 'new_secure_network.pdf',
 'length': 451454,
 'modDate': "D:20120726140215-04'00'",
 'name': '<0>new_secure_network.pdf',
 'size': 535348,
 'ufilename': 'new_secure_network.pdf'}
----------
{'checksum': b'68c2a846c3ad0dc3a1c3a661c38fc2a8c5a1efac81c2b2c39d6ecb9a',
 'collection': 42,
 'creationDate': "D:20120928172712-04'00'",
 'desc': '',
 'filename': 'rendering.jpg',
 'length': 601541,
 'modDate': "D:20120928172824-04'00'",
 'name': '<0>rendering.jpg',
 'size': 614801,
 'ufilename': 'rendering.jpg'}
----------
>>> # look at 'collection' of last file:
>>> print(doc.xref_object(42))
<<
  /adobe:DisplayName (Office Rendering)
  /adobe:Order 5
  /adobe:Summary 43 0 R
>>
>>> # let us see what is in 'adobe:Summary':
>>> print(doc.xref_object(43))
<<
  /D (Architects vision of one of the social areas of the new location.)
  /RichText (<TextFlow whiteSpaceCollapse="preserve" xmlns="http://ns.adobe.com/textLayout/2008"><p><span>Architects vision of one of the social areas of the new location.</span></p></TextFlow>)
>>
>>>

1 reply

erand96 Apr 26, 2021
Author

Hi Jorj,

Thank you for the in-depth response. That looks very promising! The particular line that caught my eye was

/adobe:Order 5

I believe this is would help me out greatly in this project. If I can get the "order" field, then I would just need to identify the sorting method used in the document. Different files are sorted by different criteria, so I need to go through each file, identify the sorting method, and then save the attachments in that same order. I'm hoping to dig through the metadata and eventually find a field that displays the "sort by" criteria.

Below is just a quick example of what I'm talking about with the "sort by" field. Thanks!

JorjMcKie · 2021-04-26T14:09:54Z

JorjMcKie
Apr 26, 2021
Maintainer

There is a pre-version 1.18.13 in the PyMuPDF-Wheels repo. Look in either the Linux or the OSX branch for your config. This version has the mehods I was using for that little demo. Sent from Mail<https://go.microsoft.com/fwlink/?LinkId=550986> for Windows 10 From: ***@***.***> Sent: Montag, 26. April 2021 09:59 To: ***@***.***> Cc: Jorj X. ***@***.***>; ***@***.***> Subject: Re: [pymupdf/PyMuPDF] PDF Portfolio attachments display order (#1030) Hi Jorj, Thank you for the in-depth response. That looks very promising! The particular line that caught my eye was /adobe:Order 5 I believe this is would help me out greatly in this project. If I can get the "order" field, then I would just need to identify the sorting method used in the document. Different files are sorted by different criteria, so I need to go through each file, identify the sorting method, and then save the attachments in that same order. I'm hoping to dig through the metadata and eventually find a field that displays the "sort by" criteria. Below is just a quick example of what I'm talking about with the "sort by" field. Thanks! [image]<https://user-images.githubusercontent.com/82478006/116094490-31ad0200-a66d-11eb-9d85-8846037714b6.png> — You are receiving this because you commented. Reply to this email directly, view it on GitHub<#1030 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AB7IDIRODK2IXD3CUTETS73TKVWTFANCNFSM43OWMBMA>.

2 replies

erand96 May 10, 2021
Author

Just to follow up on this, I was able to build code around the example provided above and extract the order of the attachments just how I wanted. I associated the extracted order with a letter that was then inserted before the file name. This allowed the merging function to merge all of the attachments alphabetically which is the only way it can merge.

If there is interest, I can post more of the code later this week once it is finished up.

Thanks for the help!

JorjMcKie May 10, 2021
Maintainer

Congratulations!
Sure I am interested. Please feel free to submit a Pull Request in this repo. So it gets published as your contribution.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

PDF Portfolio attachments display order #1030

{{title}}

Replies: 3 comments 7 replies

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

Select a reply

PDF Portfolio attachments display order #1030

erand96 Apr 23, 2021

Replies: 3 comments · 7 replies

JorjMcKie Apr 23, 2021 Maintainer

erand96 Apr 23, 2021 Author

JorjMcKie Apr 23, 2021 Maintainer

erand96 Apr 23, 2021 Author

JorjMcKie Apr 23, 2021 Maintainer

JorjMcKie Apr 24, 2021 Maintainer

erand96 Apr 26, 2021 Author

JorjMcKie Apr 26, 2021 Maintainer

erand96 May 10, 2021 Author

JorjMcKie May 10, 2021 Maintainer

erand96
Apr 23, 2021

Replies: 3 comments 7 replies

JorjMcKie
Apr 23, 2021
Maintainer

erand96 Apr 23, 2021
Author

JorjMcKie Apr 23, 2021
Maintainer

erand96 Apr 23, 2021
Author

JorjMcKie Apr 23, 2021
Maintainer

JorjMcKie
Apr 24, 2021
Maintainer

erand96 Apr 26, 2021
Author

JorjMcKie
Apr 26, 2021
Maintainer

erand96 May 10, 2021
Author

JorjMcKie May 10, 2021
Maintainer