[PDFBOX-5627] Fonts are not subsetted when saving incrementally

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.0.28, 3.0.0 PDFBox
• Fix Version/s: 2.0.30, 3.0.0 PDFBox
• Component/s: PDModel
• Assignee: Tilman Hausherr
• Reporter: DvdM
• Created: 28/Jun/23 07:04
• Updated: 18/Aug/23 05:46
• Resolved: 02/Jul/23 15:48

Description

I've noticed saving documents incrementally does not trigger fonts to be subsetted.

The function PDDocument.save() contains the following code:

// subset designated fonts
for (PDFont font : fontsToSubset)
{
    font.subset();
}
fontsToSubset.clear();

However, this code to subset fonts does not seem te be included in the PDDocument.saveIncremental(...) and PDDocument.saveIncrementalForExternalSigning(...) functions. When loading a font and indicating it should be subset, the font will be used in the appearance stream as if subsetted, but it won't be in the end, resulting in a corrupt file.

Solution

I propose to also call subset when signing incrementally.

I use this for files to which I add multiple signatures. In between signatures, text fields are filled in using PDFBox. Since the fields are not user-editable, fonts can be subsetted since the value cannot be changed afterwards.

Reproduce

I have attached 2 files to reproduce this issue. In both files you can change saveIncremental() by save() or manually call font.subset() and the output will be valid.

Note: both files are minimal reproducible files just designed to trigger this issue, that's why the implementations don't make sense (especially the first file).