Add support for PDF/A-1b #34

Backbone81 · 2017-06-07T08:20:20Z

This pull request complements prawnpdf/prawn#1029

pointlessone · 2017-07-12T09:30:48Z

.gitignore

@@ -1,2 +1,3 @@
 Gemfile.lock
 coverage/*
+/.idea


This should go into your global .gitignore

pointlessone · 2017-07-12T09:31:45Z

lib/pdf/core/document_state.rb

+          info: options[:info]
+        }
+        store_params[:print_scaling] = options[:print_scaling] if options[:print_scaling]
+        store_params[:enable_pdfa_1b] = options[:enable_pdfa_1b] if options[:enable_pdfa_1b]


This feels a bit repetitive. Maybe we can use options.select here?

pointlessone · 2017-07-12T09:41:02Z

lib/pdf/core/xmp_metadata.rb

+        render_xmp if @xmp_creator_tool || @xmp_create_date || @xmp_modify_date
+        render_pdf if @pdf_keywords || @pdf_producer
+        render_dc if @dc_title || @dc_creator || @dc_description
+        @xml_doc.root.to_xml


I admire your perseverance but it feels like this whole class could be replaced by a string interpolation. It would be simpler, faster, and would spare us a dependency.

Speaking of which… Prawn is pure Ruby. All its deps are pure Ruby. It's one of our selling points. means that Nokogiri is not going to get added to Prawn deps. If you're convinced you must use it for this feature you should release it in its own gem instead. We'll be happy to to help you with integration.

I don't have to use nokogiri. It was my first shot at the problem because I use it elsewhere. I accept that no external dependencies should be added and will look into changing it to string interpolations.

pointlessone · 2017-07-12T09:47:54Z

lib/pdf/core/object_store.rb

@@ -19,6 +19,18 @@ def initialize(opts = {})

        @info  ||= ref(opts[:info] || {}).identifier
        @root  ||= ref(Type: :Catalog).identifier
+
+        # PDF/A-1b requirement: XMP metadata
+        @xmp_metadata ||= ref(Type: :Metadata, Subtype: :XML).identifier


I would like this to be optional. The documents we generate now should not get bigger for no reason with new version of Prawn. Let's include this only in documents that actually have some metadata.

Done. The XMP metadata is now only added to PDF/A-1b documents.

pointlessone · 2017-07-12T09:55:52Z

lib/pdf/core/sRGB2014.icc.license

@@ -0,0 +1,10 @@
+ICC sRGB v2 Profile


The profile is not code. Please put it in the data directory and please don't forget to add it to the packaged gem as well.

Also as far as I can tell copyright information and profile name is included in the profile itself. The license doesn't seem to require this file distribution along with the profile so maybe drop it. Include the links in commit message for history. That should be enough.

pointlessone · 2017-07-12T10:00:31Z

lib/pdf/core/renderer.rb

@@ -216,10 +217,12 @@ def render_xref(output)
      # Write out the PDF Trailer, as per spec 3.4.4
      #
      def render_trailer(output)
+        trailer_id = PDF::Core::ByteString.new(SecureRandom.random_bytes(16))


Let's make is deterministic. We're trying to reduce random changes to the generated documents.

With 'deterministic' are you thinking of hashing some document content? Because the PDF spec (10.3 File Identifiers) suggest to use the current time beside some other data. But that would not solve the problem with the idempotent test you mentioned somewhere else.

Indeed, the time is mentioned. But it doesn't seem like a good indicator of content stability. After all, are two documents different if their content is identical? Also a bit earlier in that chapter one can find this:

The first byte string is a permanent identifier based on the contents of the file at the time it was originally created and does not change when the file is incrementally updated. The second byte string is a changing identifier based on the file’s contents at the time it was last updated. When a file is first written, both identifiers are set to the same value.

I would just hash the body. It would work great for our purpose. Also since Prawn doesn't really support incremental updates we can reuse the value for the second part of ID.

I'm looking at getting both PDF/A-1b support in, and updating doc encryption to modern levels. In the process, I found that there's a potential issue here that will likely force a design decision for Prawn.

Specifically, the first trailer[:ID] value is used in calculating the encryption key for the PDF document. This is missing from the current Prawn implementation (which always assumes an empty :iD value), but is clearly laid out in the spec. So this ID value needs to be present when the encryption dictionary is built (currently when encrypt_document is called) and invariant thereafter.

My current favored approach is:

Calling encrypt_document should initialize the trailer[:iD] value in the exact same way as the PDF/A-1b code does, with the caveat that they may not wind up with the same value if the state of the document is altered between the two calls. Also, need to confirm that the current method for PDF/A-1b excludes an encryption dictionary from the ID calculation.

The PDF/A-1b code only sets trailer[:ID] if it isn't already set. Otherwise it uses the already present value

I think this is simple, involves the minimum change for Prawn and PDF-Core, and is consistent with the spec intentions.

Other options might include moving all encryption_dictionary calculation to the render phase or requiring that encrypt_document be the last call to a document before rendering. On the face of it both of those seem less desirable so I haven't yet thought through the implications.

@pointlessone Thoughts?

ICC profile was downloaded from http://www.color.org/srgbprofiles.xalter#v2 ICC profile license at http://www.color.org/profiles2.xalter#license

petergoldstein · 2022-02-10T01:36:21Z

@pointlessone what's your current thinking on this PR? Your concerns appear to have been addressed and it (combined with the matching prawn PR) seems like a significant value add. Would it make sense to rebase and see where it stands?

@Backbone81 Not sure if you're engaged on this anymore (it's been a while) but if you have any thoughts they'd be appreciated.

Backbone81 · 2022-02-10T09:20:11Z

@petergoldstein It has been quite some time. I did need that feature for a project I was working on five years ago and finished my project with a prawnpdf fork containing the current state from those two merge requests. I have since then moved on to other projects and do not have access to the ISO specifications for PDF and PDF/A-1b any more.

I would be fine with somebody else taking my changes and integrating them into the current state of prawnpdf. I think it would be a valuable feature for applications in the financial and insurance industry which do have a regulatory need for standards like PDF/A-1b.

petergoldstein · 2022-02-11T15:52:23Z

@Backbone81 Let me see if I can rebase this and then we can get @pointlessone 's feedback. Thanks for the initial contribution.

jlduran · 2022-02-11T15:55:18Z

@Backbone81 Let me see if I can rebase this and then we can get @pointlessone 's feedback. Thanks for the initial contribution.

We have been using this patch for a while now, with a few minor tweaks:

Labtec@83606ea

EDIT: We used veraPDF for PDF/A-1b validation.

pointlessone

I will have to check if I have a copy of PDF/A standard and if these changes represent the standard sufficiently.

pointlessone · 2022-02-11T16:32:55Z

lib/pdf/core/object_store.rb

+          xmp_metadata.stream = Stream.new
+          xmp_metadata.stream << xmp_metadata_content.render


xmp_metadata is a reference. It's smart enough to create a stream when needed so this can be a bit simpler:

Suggested change

xmp_metadata.stream = Stream.new

xmp_metadata.stream << xmp_metadata_content.render

xmp_metadata << xmp_metadata_content.render

pointlessone · 2022-02-11T16:34:52Z

lib/pdf/core/object_store.rb

+        icc_profile_name = 'sRGB2014.icc'.freeze
+
+        icc_profile_stream = ref(N: 3)
+        icc_profile_stream.stream = Stream.new


This is not needed.

Added support for PDF/A-1b

033eaa5

Backbone81 mentioned this pull request Jun 7, 2017

Add support for PDF/A-1b prawnpdf/prawn#1029

Open

pointlessone requested changes Jul 12, 2017

View reviewed changes

Backbone81 added 9 commits July 12, 2017 14:37

Moved some ignores from local ignore file to global ignore file.

c94f468

Made trailer ID deterministic

0cae866

Trailer ID generation does not use seek any more.

815c711

Replaced nokogiri with string interpolation for XMP metadata generation

65532fb

Moved ICC profile to data directory.

b60414d

ICC profile was downloaded from http://www.color.org/srgbprofiles.xalter#v2 ICC profile license at http://www.color.org/profiles2.xalter#license

Code cleanup including rubocop warnings

ada2c8e

XMP metadata is only added to PDF/A-1b compliant documents

c6f3661

Added XMP packet wrapper

1ca0ac9

Added missing timezone information to XMP timestamps

21ecc4e

pointlessone reviewed Feb 11, 2022

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add support for PDF/A-1b #34

Add support for PDF/A-1b #34

Backbone81 commented Jun 7, 2017

pointlessone Jul 12, 2017

Backbone81 Jul 12, 2017

pointlessone Jul 12, 2017

Backbone81 Jul 14, 2017

pointlessone Jul 12, 2017

Backbone81 Jul 12, 2017

Backbone81 Jul 13, 2017

pointlessone Jul 12, 2017

Backbone81 Jul 14, 2017

pointlessone Jul 12, 2017

Backbone81 Jul 14, 2017

pointlessone Jul 12, 2017

Backbone81 Jul 12, 2017

pointlessone Jul 12, 2017

Backbone81 Jul 13, 2017

petergoldstein Feb 11, 2022

petergoldstein commented Feb 10, 2022

Backbone81 commented Feb 10, 2022

petergoldstein commented Feb 11, 2022

jlduran commented Feb 11, 2022 •

edited

Loading

pointlessone left a comment

pointlessone Feb 11, 2022

pointlessone Feb 11, 2022

		xmp_metadata.stream = Stream.new
		xmp_metadata.stream << xmp_metadata_content.render

	xmp_metadata.stream = Stream.new
	xmp_metadata.stream << xmp_metadata_content.render
	xmp_metadata << xmp_metadata_content.render

Add support for PDF/A-1b #34

Are you sure you want to change the base?

Add support for PDF/A-1b #34

Conversation

Backbone81 commented Jun 7, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

petergoldstein commented Feb 10, 2022

Backbone81 commented Feb 10, 2022

petergoldstein commented Feb 11, 2022

jlduran commented Feb 11, 2022 • edited Loading

pointlessone left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jlduran commented Feb 11, 2022 •

edited

Loading