Save to disk : adding pubdate to the filename

[Bertrand]

I'm trying to add the pubdate to every book when I save to disk :
For now, my template is as follow :
{author_sort[0]}/{author_sort} - {series}{series_index:0>2s| | - } {title} ({pubdate})

which gives : D/Doe, John/Title (2021)

But, obviously, some books have no pubdate, and I get this :

B/Bench, John/Title of the book (101)

(What is this "101" number ?)

So my goal is :
if pubdate = 101 do nothing
else add "(pubdate)"

I have tried various variations based on the series model but I can't get anywhere.

Thanks for your help.

[theducks]

101 is the basis (base) of a Date data type everyplace in Calibre. I believe this is part of the open database package Calibre is built upon

[Bertrand]

Thanks theducks for the explanation.


Maybe someone will have an idea for the last part of my question.

[theducks]

Originally Posted by Bertrand

Thanks theducks for the explanation.


Maybe someone will have an idea for the last part of my question.

You could always us 2 templates for now
Search:

not pubdate:true

gets you those books with no pubdate
The not is because of the underlying 101

[chaley]

Originally Posted by Bertrand

I have tried various variations based on the series model but I can't get anywhere.

Thanks for your help.

It is supposed to be empty, not 101. The fact that it isn't is a bug. I will submit a fix.

[ownedbycats]

What kind of time formatting uses 101 as an epoch anyways? Google is failing me...

[chaley]

Originally Posted by ownedbycats

What kind of time formatting uses 101 as an epoch anyways? Google is failing me...

Some value must be chosen for undefined. That value must work in sqlite, in python (2 & 3), and in other libraries, which means that NULL, None, and empty don't work.

Many years ago (probably > 15) the choice was made to use 1 Jan 0101 as the undefined date. It is up to the user to check. Somewhere lost in time that check disappeared from save-to-disk template handing, which BTW although it looks a bit like SFM it is *not* using the calibre template language. Again, historical. There are a few cases like this that predate the template language or require processing where the template processor doesn't exist.

I often run across code I wrote before templates existed. And before a consolodated metadata structure existed. And before consistent custom columns existed. And ...

[ownedbycats]

Normally, I'd expect the Unix Epoch, but I suspect it was chosen against this considering the purpose of Calibre and the number of books published before 1970...

[chaley]

Originally Posted by ownedbycats

Normally, I'd expect the Unix Epoch, but I suspect it was chosen against this considering the purpose of Calibre and the number of books published before 1970...

That epoch is measured in seconds since X. That scheme doesn't work with many database date types where you must store a 'real' date or empty. You could make it work by storing a large natural number (64 bit) that represents seconds (plus or minus) from some base date, but doing so would make the database hard to use. Reading a date would always require calculation.

FWIW: I was at BTL when we started worrying about whether seconds-from-1970 would be storable with the integer types we had (32 bits signed). We hadn't done anything about it before I left (1979), but we knew we had made a mistake.