make CIDSet generation optional again for TrueTypeFontUnicode (subset = true)

[zyro23]

Is your feature request related to a problem? Please describe.
#815 (release 1.3.31) added generating a CIDSet for TrueTypeFontUnicode if subset is true:

• https://github.com/LibrePDF/OpenPDF/pull/815/files#diff-ee32c0d231c530fb36681df08f997140dc74ccbf4da1f3927039f08cf058a978R399

however, the referenced PDF/A-1 validation rule (6.3.5-3) is not required for PDF/A version 2 and 3 compliance.

a pdf without CIDSet streams that contains only a subset of the used fonts validates fine with PDF/A versions 2 and 3.

but if a CIDSet is present, all CIDs of the font have to be present which means effectively the complete font being embedded:

• https://github.com/veraPDF/veraPDF-validation-profiles/wiki/PDFA-Parts-2-and-3-rules#rule-621142-2

If the FontDescriptor dictionary of an embedded CID font contains a CIDSet stream, then it shall identify all CIDs which are present in the font program, regardless of whether a CID in the font is referenced or used by the PDF or not.

Describe the solution you'd like
please allow to opt out of CIDSet generation for TrueTypeFontUnicode (Identity-H, embedded= true, subset = true) again.

Describe alternatives you've considered

• disabling PdfWriter.pdfxConformance has no effect because the subset check takes precedence (see referenced pr)
• setting subset = false "works" but is no viable option - that results in pdf file size differences of megabytes instead of a few kb because the complete fonts are embedded

Additional context
versions up to 1.3.30 are working fine.

thank you!

[andreasrosdal]

@bsanchezb May you investigate this please?

[bsanchezb]

Hi,

Our scenario is that an end-user provides a document and wants it to be signed with a visual electronic signature. The initial document format is not known, but the signature creation tool should be able to keep PDF/A conformance when possible.

The PR #815 keeps the conformance by default, while also adding CIDSet for other document types, when font is used. But it does not break conformance for those documents neither.

I see that @zyro23 sees this information as redundant in their case.

The only alternative option I can see, because OpenPdf is not capable of identifying PDF/A type of the document, is to add a separate boolean variable, something like "includeCIDSet", which may be used to explicitly modify the behavior of the CIDSet creation, instead of using "subset". Or we can even use "subset && includeCIDSet" in order to include the CIDSet.

I would keep the value of "includeCIDSet" as TRUE by default, as this will protect users from breaking PDF/A-1 conformant documents.

KR,
Aleksandr

[zyro23]

completed?

i think the reasoning for an additional boolean flag makes sense (being used instead of subset).
ok for me if it defaults to true - at least that will not break the current behavior again.

i can come up with a pr and would make the flag accessible via TrueTypeFontUnicode only (not exposing it via getter/setter up to BaseFont like subset).

sounds ok? if so, please re-open.

[andreasrosdal]

Please submit a pull request, thank you.