Unable to open an epub file

[iieeann]

Hi,

I have not been able to open a specific epub file (10MB) successfully with Sigil as well as Calibre. Sigil can't show any viewable page while Caibre only show 1% of the pages. When i change the epub extension to zip, the extracted files seem incomplete as well (4.3MB), 99% of the text files are missing.

I use an Android epub reader Moon+ Reader Pro to open the file, it works perfectly fine.

Sigil error


Calibre Viewer error


Calibre Editor screen (no error reported)


If the epub file is corrupted, Moon+ Reader won't be able to open it. What is wrong here?

The epub file is attached as below.

Moderator Notice
Copyrighted content removed by moderator

[DiapDealer]

For Sigil's part: it is the insane xhtml filenames that are the problem. Plus, it appears there may be tons of improperly encoded characters for the UTF-8/16 encoding.

I'm going to give the benefit of the doubt in assuming you haven't posted content that infringes on the rights of any potential copyright holders, here. Just know that your upload will be deleted as soon as it's determined otherwise.

[DiapDealer]

The fact that Moon Reader will open it has no bearing on whether or not it's a valid, spec-compliant, epub

[KevinH]

The file names are ridiculous and something about them is making Sigil think they are not the same as listed in the opf manifest. I have not figured out why (yet). If you load the epub into Sigil, it will tell you that it is seeing these files as unmanifested. So the path in the opf manifest does not seem to match the path as created when extracting file from the zip archive.

If you unzip the epub, you can actually use AddExisting to import the entire set of files and save it as an epub, and then it will work.

So it may be an encoding issue in the content.opf itself as it appears it was edited by some tool that is insane enough to add a BOM (byte order mark) to a utf-8 file! There no big endian vs little endian when it is a stream of just bytes (ie. in a utf-8 file) and no ByteOrder mark is ever needed or ever used anymore.

I did notice the Sigil 1.7 metadata at some point in the content.opf so at some point this was edited with Sigil but Sigil will not write out BOM so something has edited it after that.

[elibrarian]

I don't have a solution, but some clues for more knowledgeable people around here perhaps: The book is in Traditional Chinese (Google/Chrome tells me), and I suspect that the ridiculous file names probably was too at some point of their existence. Given the problems with spaces, commas etc. in filenames inside epubs, we've discussed ad nauseam here on Mobileread for years, I think the problem might lie here.

The Moon reader seems to be originating from Hong Kong, and thus chinese probably are not at problem for it.

Just my 2 cents - as usual.

Regards,

Kim

[kovidgoyal]

Do an EPUB to AZW3 to EPUB conversion in calibre that will take care of most issues. Or even just an EPUB to EPUB though that will not fix the invalid filenames.

[Sarmat89]

The easiest way is to use Macintosh or Linux machine/VM to edit the file and mass rename all 900+ files to some Latin letters.

[KevinH]

Or just load the epub in Sigil then use AddExisting to add the .xhtml files from an unzipped version of the epub. Then you can use Sigil's regex rename and batch renaming to create better filenames while properly handling any internal links.

[Sarmat89]

That file cannot be extracted in Windows, because it does not allow : and * in the Win32-compatible filenames.

[Binchen]

Thats a chinese book:

Dabong's shift worker
The little man who sells newspapers
WeChat public number-Asheng reads books

The filenames contains characters which are not allowed as Sarmat already mentioned.

Chapter 197
Dinner Party and Loquat

A dinner party? Well, the governor's lord has been in the city for so long, it is impossible for the Yunzhou officialdom not to know ...... Xu Qi'an's mental state is much better, anyway, he can't sleep, it is boring to stay at the post, so he said.

"Good, your Excellency wait for a moment."

He put on his boots, tied the gong placed on the bed to his chest, and hung the black and gold long knife on his back waist, and opened the door of his room.

Inspector Zhang stood outside the door, wearing a scarlet official robe, with a straight posture and a distinguished air.

I asume thats acommercial book. If its copyrighted i dont know. Maybe the strange filenames are coming from a failed unicode conversion or may be some protection or failed try to get rif of a protection.

[DiapDealer]

If it's a commercial book that's not being given away for free, it needs to be taken down.

The filenames make zero sense in any conversion, or liberation sense. There's no reason for them to be converted to extremely long meaningless strings comprised of nothing but two characters (both of which are illegal on Windows). There's nothing human parseable in "::***::*:*::**::**::**:::***:::****::::****:.xhtm l" in any language so something like f00001.xhtml, f00002.xhtml, etc... would work equally well (in any language) and not break.

My advice is to throw it back and buy a better copy.

[Binchen]

I am pretty sure this is a copyrighted commercial book, the author is Xiao Lang Jun
and there currently plans to adopt this to a cinematic film for the chinese market.

Klick

[DiapDealer]

Down it comes, then.

[KevinH]

And I was able to track down the issue for Sigil on macOS and Linux. When reading in a manifest item, it looks at the href for a ":" which should only exist in the url's scheme and if it exists, the href is considered to be external and not actually present in the epub. Using external resources in the manifest is allowed in some case under epub3.

So all and all a very strange beast set up to fail on almost very platform.

[iieeann]

yes get a better copy is the solution. There are 2 formats available, i get the mobi format and convert it to epub, then the file is now editable.