info as comments in OPF lost

[AlanHK]

I was looking at a batch of epubs and wondering how they were made. Opening in Sigil I could not find any declarations of the software used to make them in file headers or OPF file.

Then I opened one epub as a zip and looked in the OPF. This had a line in the metadata section:

<!-- Created by Jutoh 3.13.5 at 21/04/2022 17:22:31 - config "Epub" -->

Opening the epub in Sigil, this is not shown, and if the epub is saved, it's lost.

Sigil leaves its own mark like this:
<meta content="0.9.13" name="Sigil version" />
<meta property="dcterms:modified">2022-04-02T17:43:56Z</meta>

And Calibre like:
<meta name="calibre:timestamp" content="2017-05-05T17:14:23.743000+00:00" />
<dc:contributor opf:role="bkp">calibre (2.83.0) [https://calibre-ebook.com]</dc:contributor>

"bkp" being "Book producer".
Annoying that there is no standard for the creating software, when there is a myriad of esoteric creator/contributor codes:
https://www.loc.gov/marc/relators/relaterm.html
Maybe that's why Jutoh never bothered to try.

So, I know that it's perfectly correct for Sigil to ignore such HTML comments, but in the interest of preserving info that someone thought was important enough to put there, maybe on opening these can be converted to valid metadata. Sigil already does a lot of fixes on non-standard files on first opening.

This seems to work:
<meta content="Created by Jutoh 3.13.5 at 21/04/2022 17:22:31 - config &quot;Epub&quot;" name="comment" />

[KevinH]

The OPF is and must be machine parseable and regenerable at all times. That is how Sigil's tools work to change the opf each and every time a file is renamed, split, merged, things added, things moved, things deleted etc. Inside Sigil, the opf is actually parsed and stored in data structures to make this possible and reasonably fast. All xml comments are removed from the opf after the first Sigil command that touches the OPF as they are free form and have no place in the data structure. Sigil has always worked liked this. End Users never see the OPF directly nor are they meant to.

You can set an environment variable to turn off Sigil adding its own metadata. And the date modified must always be added according to the epub3 spec each time the epub is modified.

And converting comments to metadata is no solution and is technically incorrect. The OPF as with the nav and ncx are truly meant to be machine parseable. And Comments meant for humans do not belong in files only meant to be machine parseable. Jutoh users or whomever adds that comment can add that metadata if so desired and Sigil will keep it but Sigil will not add it.

And fwiw, that Jutoh "comment" is no longer correct as soon as Sigil open the OPF, cleans and parses it to set up the file. Sigil is the generator of that file from that moment onward.

[KevinH]

Jon, please at least read what I wrote before saying something incorrect about Sigil.

As I explained above, Comments in the opf are valid and allowed on input to Sigil but as soon as Sigil is used to do something that forces it to modify the OPF or when the epub is saved, then (since the OPF is meant to be machine read and not read by the end user) comments are removed. And it allows Sigil to perform common OPF housekeeping tasks faster.

This is how Sigil has always worked.

Sigil is not treating opf comments like errors, it is ignoring them as the spec allows and for valid performance reasons.

And if you want respond to *unasked* questions about how better to use Calibre, then please do so in the Calibre forum and not here in the Sigil forum.

[AlanHK]

Originally Posted by KevinH

And fwiw, that Jutoh "comment" is no longer correct as soon as Sigil open the OPF, cleans and parses it to set up the file. Sigil is the generator of that file from that moment onward.

When Sigil opens imperfect epubs, it tries to repair them.
Consider converting comments in OPFs to something valid that retains the information on opening.
"Cleaning" the OPF doesn't have to mean deleting every imperfect element.

Originally Posted by KevinH

The OPF as with the nav and ncx are truly meant to be machine parseable.

But also human readable and often edited.
I manually remove junk or fix stuff in them all the time.

The "<meta content" I used passes epubcheck, but I don't care what tag it has, the important thing to me is that the information is not lost before I even know it was there.

Originally Posted by KevinH

You can set an environment variable to turn off Sigil adding its own metadata. And the date modified must always be added according to the epub3 spec each time the epub is modified.

I mentioned that not to complain but to point out that every editor uses a different method to do this.
Jutoh unfortunately chose a method that isn't kosher.

I have occasionally noticed other programs putting info in HTML comments in the OPF. Only rarely do I peek at the OPF outside Sigil, so I can't say just how common this is, but it is not just Jutoh.

[JSWolf]

Originally Posted by AlanHK

I have occasionally noticed other programs putting info in HTML comments in the OPF. Only rarely do I peek at the OPF outside Sigil, so I can't say just how common this is, but it is not just Jutoh.

When using KindleUnpack to convert KF8 to ePub, a lot of things are put in a comment.

According to epubcheck, comments in the OPF are valid. So why blow them away?

[KevinH]

Yes and as one of the authors of KindleUnpack those comments can happily go away. I added them originally as a debug tool while KindleUnpack was still evolving to support KF8. I will remove them in a future release of KindleUnpack.

And those comments are never seen by any end user or used by any reader and will continue to be removed upon regeneration of the opf by Sigil for performance reasons which is allowed by the spec.

Do try and read what I wrote earlier as to why this is done. I already explained the reasons. Sigil controls and rewrites the opf period.

So no I am not implementing a comment to metadata converter for Sigil. Do it as a standalone python tool if you want.

I am getting really tired of getting badgered about requests for things in Sigil I have already turned down and explained why. Please stop.

[DiapDealer]

Originally Posted by KevinH

I am getting really tired of getting badgered about requests for things in Sigil I have already turned down and explained why. Please stop.

And just in case Kevin's request wasn't perfectly clear:

Moderator Notice
Please stop.