AttributeError: Custom field name '' must begin with '#'

[Peter Blaise]

"... DNSB wrote: Instead of Check Book which is a rather simplistic check, try using epubcheck. As for the re-import? You do not save a backup copy before editing? ..."

.

Two things:

1 - I'm not asking for support as much as I'm trying to present to Kovid Goyal that the various EPUB checking algorithms in the Calibre Ebook program disagree with each other: make all Calibre Ebook check routines equivalent.

2 - 'Help'-wise, yes, once I've completely edited an EPUB, I'm chagrinned that those edits are lost and I have to re-do them, hence the query if it's possible to re-import THE SUPOSEDLY OK BUT REALLY DAMAGED EPUB without it's apparently munged metadata, other wise, I either re-edit after re-importing the original, or in this case, Amazon converted it to MOBI ( and KFX ) no problem - in other words, instead of Calibre saying "I can't open this for editing", how about Calibre saying "This appears damaged, please wait a moment while I fix it".

So this thread is more of a report to Kovid Goyal than it is a request for specific EPUB repair for me, more of a general "once an EPUB becomes corrupt, in spite of the Calibre Edit book Check book saying "no problems", but then the EPUB cannot be re-opened for editing, for polishing, for converting, only for reading ( funny that ), or for sending to Amazon, who converts it just fine, no problem.

Regarding epubcheck, I presume from W3.org, do you expect that program to FIX malformatted metadata?

Thanks.

.

[DNSB]

Originally Posted by Peter Blaise

Regarding epubcheck, I presume from W3.org, do you expect that program to FIX malformatted metadata?

To fix, no. To let you know that there is an issue? Yes. epubcheck, does not have a fix option, it simply lets you know where the error occurred and what it thinks the error is. Even for errors that are caught elsewhere and offer an automatic fix, I will hit fix manually. All too often, the results of an automatic fix have been unexpected, interesting and a pain to fix.

Expecting the same results from a standards compliancy checker and a renderer is a bit of a stretch. The calibre ebook-viewer is Chrome based and is very forgiving of errors as most web browsers tend to be. The calibre ebook-editor BooK Check tool is not as forgiving and epubcheck is basically that martinet who screamed at you over a misplaced comma.

I also use the W3C CSS validation service to check the CSS files for errors such as using a CSS3 (ePub3 uses that flavour) in an ePub2 (CSS 2.1 flavour). It helps to prevent unpleasant surprises for the end user.

[BetterRed]

Originally Posted by Peter Blaise

. . .

I'm not asking for support as much as I'm trying to present to Kovid Goyal that the various EPUB checking algorithms in the Calibre Ebook program disagree with each other: make all Calibre Ebook check routines equivalent.

. . .

Then, I suggest you report your issue as a Bug/Enhancement, see ==>> calibre - Bugs

You will get a direct response from Kovid there.

BR

[kovidgoyal]

Jesus, what a load of noise over nothing. There was a bug which caused calibre to choke on a particular type of malformed metadata in EPUB3 files. That was fixed, as I indicated in my post days ago. All you have to do is wait for the next calibre preview/release and you will be able to import your EPUB with no issues.

[Peter Blaise]

"BetterRed wrote: . . . I suggest you report your issue as a Bug/Enhancement, see ==>> *calibre - Bugs* (https://calibre-ebook.com/bugs) You will get a direct response from Kovid there. BR . . ."

Oh, great to know a https://calibre-ebook.com/bugs site exists.

"Kovid Goyal wrote: . . . There was a bug which caused calibre to choke on a particular type of malformed metadata in EPUB3 files . . . fixed . . . wait for the next calibre preview/release and you will be able to import your EPUB with no issues . . ."

There is no response to a search for "AttributeError: Custom field name '' must begin with '#'" at https://calibre-ebook.com/bugs - what "Bug" is this situation listed under?

Anyway, glad to know this was noticed and will be addressed via a subsequent release.

As ever, thank you, Kovid Goyal - and all the folks at https://calibre-ebook.com/about#contributors - for the superlative Calibre Ebook universe of programs for eBook reading, converting, library management, editing, emailing, researching, and more!

.

[PeterT]

The canonical location for fixes is the calibre GitHub repository. You maybe overlooked an earlier response from Kovid

Originally Posted by kovidgoyal

Somebody has tampered with that epub file leaving behind invalid metadata in its OPF. https://github.com/kovidgoyal/calibr...2b080895ad476f

That link he includes is to the change made to fix the issue which will be included in the next calibre release.

Sent from my Pixel 7a using Tapatalk

[Peter Blaise]

"PeterT wrote: The canonical location for fixes is the calibre GitHub repository. You maybe overlooked an earlier response from Kovid:

". . . Kovid Goyal wrote: Somebody has tampered with that epub file leaving behind invalid metadata in its OPF. https://github.com/kovidgoyal/calibr...2b080895ad476f . . ."

That link he includes is to the change made to fix the issue which will be included in the next calibre release. Sent from my Pixel 7a using Tapatalk"

How would an end user know what all that meant?

However:

"Kovid Goyal wrote: . . . There was a bug which caused calibre to choke on a particular type of malformed metadata in EPUB3 files . . . fixed . . . wait for the next calibre preview/release and you will be able to import your EPUB with no issues . . ."

. . . works for me.

Thanks.

.

[DNSB]

Originally Posted by Peter Blaise

How would an end user know what all that meant?

Familiarity with how Kovid Goyal posts fixes? Looking at links in posted messages?

[Quoth]

Originally Posted by DNSB

Even for errors that are caught elsewhere and offer an automatic fix, I will hit fix manually. All too often, the results of an automatic fix have been unexpected, interesting and a pain to fix.

Indeed life is too short for that. I do click on "fix filename" of a font file. That's rare.