CreationDate metadata shows wrong timezone offset for UTC

[stevesimmons]

Dates such as the PDF metadata's CreationDate have the general format "D:".

The datetime format strings in PDFDate.serialize() (included below) are correct for non-UTC: the timezone is shown as "+01'00'" or "-01'00'" for times ahead/behind of UTC.

However for UTC, the code in PDFDate.serialize() is given on line 318 of fpdf.syntax as f"D:{self.date:%Y%m%d%H%M%SZ%H'%M'}". This incorrectly takes the timestamp's hour and seconds as the timezone.

I think in this UTC case, the serialized date should either finish at the "Z", or have a timezone of "+00'00'". Printing "Z00'00'" for UTC looks wrong (I've only ever seen "Z" terminated strings", and printing "Z15:22" (my current time) looks very wrong indeed.

The incorrect line of fpdf.syntax is here:

class PDFDate:
    ...
    def serialize(self, _security_handler=None, _obj_id=None):
        if self.with_tz:
            assert self.date.tzinfo
            if self.date.tzinfo == timezone.utc:
                out_str = f"D:{self.date:%Y%m%d%H%M%SZ%H'%M'}"   # This should finish at 'Z'                            
                # out_str = f"D:{self.date:%Y%m%d%H%M%SZ}"   # Instead finish at 'Z' like this            
                # out_str = f"D:{self.date:%Y%m%d%H%M%S+00'00'}"   # Or make UTC explicit with "+00'00'"
else:
                out_str = f"D:{self.date:%Y%m%d%H%M%S%z}"   
                out_str = out_str[:-2] + "'" + out_str[-2:] + "'"
        else:
            out_str = f"D:{self.date:%Y%m%d%H%M%S}"

[Lucas-C]

Thank you for reporting this bug @stevesimmons 👍

Your analysis seems accurate to me.

Would you like to sumbit a Pull Request in order to fix this?

We already have a unit test for this case in test/metadata/test_set_date.py
I guess it should be fixed.
FYI, we have some documentation on how to regenerate reference PDF files used in unit tests there: https://py-pdf.github.io/fpdf2/Development.html#assert_pdf_equal-writing-new-tests

[Lucas-C]

@allcontributors please add @stevesimmons for bug

[allcontributors]

@Lucas-C

I've put up a pull request to add @stevesimmons! 🎉