[Possible Bug] Calibre Publication Date after updating metadata

[Katsunami]

Hi,

While cleaning up the books in my library, I've found something strange.

When I put a publication date into a book using Sigil, it looks like this:

<dc:date opf:event="publication">2005-01-01</dc:date>

This is fine: when importing a book into Calibre, it sees the publication date and displays it accordingly. (There is also a "<dc:date opf:event="modification"></dc:date>, which Calibre does not seem to use.)

After updating metadata, either through Modify EPUB, Polish Books, or Save to Disk, the publication date looks like this:

<dc:date>2004-12-31T23:00:00+00:00</dc:date>

In one way or another, it seems Calibre is able to import this book again, and still display the publication date as being 2005.

If the book is then edited again with Sigil, Sigil doesn't see the publication date anymore. It doesn't show in the tag editor. Often, but not always (strangely enough), Sigil actually deletes the <dc:date> part from the OPF-file.

If the book is then imported into Calibre again, the publication date is gone.

As this problem occurs with Save to Disk, Polish Books and Modify EPUB, I expect this problem to be in the Calibre API. I think I'd be best if Calibre saved the publication date in the same way as Sigil, if at all possible, when updating metadata.

(In case you ask: I'm very particular about my EPUB files being as perfect as possible, with correct tags, dates, a high resolution cover, my preferred layout, and so on. I store a copy of all of my "perfected and finished" books outside Calibre. I use both a "workplace library" and Sigil in conjuction to make the ebooks as I want them.)

[theducks]

Is your (system) timezone -1?

Maybe Calibre is adjusting based upon that

Add: The original was a raw date (no TZ info)

[Katsunami]

My timezone is (UTC+01:00) Amsterdam, Berlin, Bern, Rome, Stockholm, Vienna...

(Because of Daylight Saving Time, we're now actually +2 ahead of UTC.)

If I put in that the date of publication is 1-1-2005, I think it's illogical to save it as 31-12-2004, 23:00 in the assumption that calibre is going to Add +1 or more hours based on my time zone If I'd move to the west at some point in my life, it would substract hours, changing the publication date to 2004.

It would be safest to set the publication date at 30-06-2005.

I don't care about month, day, or hour of publication. The "raw" version is what Sigil produces when you add date: publication to the metadata tag-editor.

Sigil can't read Calibre's <dc:date> format (the publication date does not appear in the tag editor), and sometimes, but not always (I have not yet been able to determine when exactly) it'll completely remove Calibre's <dc:date> entries from the OPF file.

[chaley]

Originally Posted by Katsunami

Sigil can't read Calibre's <dc:date> format (the publication date does not appear in the tag editor), and sometimes, but not always (I have not yet been able to determine when exactly) it'll completely remove Calibre's <dc:date> entries from the OPF file.

The date 2004-12-31T23:00:00+00:00 is exactly the same date as 2005-01-01T00:00:00+01:00, where the +01:00 is your (and my) home time zone. If no time zone is shown, then your home time zone applies.

All calibre dates are timezone aware. You could argue that some dates should be time zone naive (unknown time zone), but that doesn't change the fact that they are not. It is not calibre's problem that sigil cannot read timezone-aware UTC dates, which is what calibre is producing. It should either keep the timezone or convert the date to local time + naive (no time zone).

[Katsunami]

Maybe then both should be updated to fix this problem.

Sigil should then be updated to correctly handle time-zone aware dates, but I think Calibre may still need to be updated.

First, if I want to put in a publication date, and I put in 1-1-2005, 0:00, I just think it's illogical to save it somewhere in 2004, and then compensate with my time zone when displaying the date. It makes the date change when I change my timezone.

If a book is officially published at 1-1-2005, 0:00 (date/time at the book's publishing location), then it should be possible to just put that date/time into the metadata without regard to timezones. If Calibre can't do that now (if it can, I'm not aware of it), it would be best to make this an option.

Second, I assume Sigil does not include opf:event="publication" in <dc:date> just for fun. At least, FlightCrew validation does not see this as a problem. It's used to recognize the publication date from the creation and modification date of the ebook. I think Calibre should include an option to preserve opf:event="publication" (or creation, or modification) if it's there.

[theducks]

Originally Posted by Katsunami

Maybe then both should be updated to fix this problem.

Sigil should then be updated to correctly handle time-zone aware dates, but I think Calibre may still need to be updated.

First, if I want to put in a publication date, and I put in 1-1-2005, 0:00, I just think it's illogical to save it somewhere in 2004, and then compensate with my time zone when displaying the date. It makes the date change when I change my timezone.

If a book is officially published at 1-1-2005, 0:00 (date/time at the book's publishing location), then it should be possible to just put that date/time into the metadata without regard to timezones. If Calibre can't do that now (if it can, I'm not aware of it), it would be best to make this an option.

Second, I assume Sigil does not include opf:event="publication" in <dc:date> just for fun. At least, FlightCrew validation does not see this as a problem. It's used to recognize the publication date from the creation and modification date of the ebook. I think Calibre should include an option to preserve opf:event="publication" (or creation, or modification) if it's there.

Here is a timestamp from a book I just edited in Sigil

<dc:date>1980-01-01T13:30:00+00:00</dc:date>

But you are correct that Sigil created ones to not include a TZ

<dc:date opf:event="modification">2013-09-08</dc:date>

You might raise a bug report at Sigil

[Katsunami]

Is that first timestamp created by Sigil's metadata editor, or is it a preserved Calibre-created timestamp?

Now... Should opf:event="publication" be included by Sigil and Calibre, or not?

[theducks]

Originally Posted by Katsunami

Is that first timestamp created by Sigil's metadata editor, or is it a preserved Calibre-created timestamp?

Now... Should opf:event="publication" be included by Sigil and Calibre, or not?

Preserved. All calibre's timestamps are preserved (the older ones had an Flightcrew error. Just taping F8 and saving, fixed those)

[BetterRed]

Originally Posted by Katsunami

First, if I want to put in a publication date, and I put in 1-1-2005, 0:00, I just think it's illogical to save it somewhere in 2004, and then compensate with my time zone when displaying the date.

- see attachment - Sigil or similar not used - EPUB converted from DOCX

EDIT : the pubdate of 1853 09 was entered directly into Published cell for the book

The 2nd attachment shows a pamphlet that might have been published in London on 2nd of September 1752, the metadata in the ePUB shows 1st September 1752. But in London there was no 1st of September in 1752, it was one of the 11 days lost when the British Empire changed from the Julian to the Gregorian calender. Pelham was incumbent British PM, my imagination is that the pamphlet would have been a satire involving the PM taking time-out at Bedlam