[Conversion Output] KePub Output Plugin

[Leonatus]

Yes, it's in all epubs that I have tried now. Strange is that on my other computer with the same specs (Windows 10, 32 bit, Calibre V. 2.77) it works fine. I didn't change anything at the plugin settings.

[BetterRed]

@Leonatus - JSWolf had some sort of problem with 2.77 & Kepub (which he edited out) - apparently it was resolved by doing a Windows restart

The thread's been deleted by one the greenies because it was a "false alarm", a better practice might be to close the thread and leave a message why so.

BR

[JSWolf]

Originally Posted by jgoguen

Is it all ePubs or just a specific book?

I'm wondering if there are errors in the ePub that the plugin cannot handle.

[JSWolf]

Have you rebooted the computer since installing Calibre 2.77? That's what solved my problem. Oh and after rebooting the computer, I also cleaned out the temp directory.

[Leonatus]

Originally Posted by JSWolf

Have you rebooted the computer since installing Calibre 2.77? That's what solved my problem. Oh and after rebooting the computer, I also cleaned out the temp directory.

Hm, I do not quite understand (once more). After installing a new version of Calibre, it's always necessary to reboot the computer, isn't it? That's what I have done this time also. And "cleaning out the temp directory": you mean simply delete the content?

As I said above, the problems occurs to all books (obviously) in the library, might they have resided there for years ore been added recently, edited or not.

[davidfor]

Originally Posted by Leonatus

Hm, I do not quite understand (once more). After installing a new version of Calibre, it's always necessary to reboot the computer, isn't it? That's what I have done this time also. And "cleaning out the temp directory": you mean simply delete the content?

I rarely need to restart after installing a new version of calibre. If you get prompted to, it usually means calibre was running when you did the install and something couldn't be replaced.

And yes, cleaning the temporary directory can be as simple as deleting everything in it. But, you won't be able to delete anything being used by a running application. Windows does have a way to do this. I think it is part of the disk cleanup tool.

As I said above, the problems occurs to all books (obviously) in the library, might they have resided there for years ore been added recently, edited or not.

The error looks like one people have reported when the path to the temporary directory has a non-standard character in it. In most cases, this is because the character is in the username. The solution is to use a different username, or set the environment variable "CALIBRE_TEMP_DIR" to point to a different directory.

[jgoguen]

Originally Posted by Leonatus

Yes, it's in all epubs that I have tried now. Strange is that on my other computer with the same specs (Windows 10, 32 bit, Calibre V. 2.77) it works fine. I didn't change anything at the plugin settings.

Are you able to convert an ePub to any other format, like AZW3 or MOBI?

[Leonatus]

@Davidfor: I doubt that the path to the temporary Directory is the culprit, for all the time, Things were alright, and
@jgoguen: conversion epub to epub, epub to azw3, epub to moby, all work fine.
Rebooting the computer did not solve the problem.

[Leonatus]

@Davidfor: but you're right, my user name Directory contains non-ASCII-characters. I only doubt if I can simply rename it, without causing system problems. Right-clicking on the Directory Name Shows no Option for renaming. I don't even know whére resides the CALIBRE_TEMP_DIR. In the Directory, I didn't find it. Could you give me a hint, please?

[jgoguen]

Restart calibre in debug mode, try to convert a book to KePub, then close calibre. That will display the debug log. Upload both the debug log and the book you tried to convert at https://www.dropbox.com/request/g0Q3WAW4jIcCYgkTFBt2 as separate files (don't make a ZIP or other archive, upload the two files separately).

[Leonatus]

Done. I uploaded the error protocol as well.

Sorry, the first debug protocol is not the right one. The second that I uploaded referrs to conversion.

[Leonatus]

I just connected the Kobo to the computer to download the News I usually receive via Calibre, but today the same error as at kepub conversion is shown. Would it be worth while de- and reinstalling Calibre?
The error Code referrs to the Kobo Touch Extended Driver:

calibre, version 2.77.0
FEHLER: Fehler: Fehler bei der Kommunikation mit dem Gerät

Fehler beim Einfügen 'kath.net [Sa, 14 Jan 2017]' von 'calibre' mit Fehler: 'All strings must be XML compatible: Unicode or ASCII, no NULL bytes or control characters' (file: calibre_plugins.kobotouch_extended.container, line: 178)

Traceback (most recent call last):
  File "site-packages\calibre\gui2\device.py", line 88, in run
  File "site-packages\calibre\gui2\device.py", line 587, in _upload_books
  File "calibre_plugins.kobotouch_extended.device.driver", line 335, in upload_books
  File "site-packages\calibre\devices\kobo\driver.py", line 1950, in upload_books
  File "calibre_plugins.kobotouch_extended.device.driver", line 245, in _modify_epub
InvalidEPub: Fehler beim Einfügen 'kath.net [Sa, 14 Jan 2017]' von 'calibre' mit Fehler: 'All strings must be XML compatible: Unicode or ASCII, no NULL bytes or control characters' (file: calibre_plugins.kobotouch_extended.container, line: 178)

After uninstalling Kobo Touch Extended Driver, transferring of the newspaper was possible, but trying to convert an epub to kepub showed the same error message.

[davidfor]

Originally Posted by Leonatus

I just connected the Kobo to the computer to download the News I usually receive via Calibre, but today the same error as at kepub conversion is shown. Would it be worth while de- and reinstalling Calibre?
The error Code referrs to the Kobo Touch Extended Driver:

calibre, version 2.77.0
FEHLER: Fehler: Fehler bei der Kommunikation mit dem Gerät

Fehler beim Einfügen 'kath.net [Sa, 14 Jan 2017]' von 'calibre' mit Fehler: 'All strings must be XML compatible: Unicode or ASCII, no NULL bytes or control characters' (file: calibre_plugins.kobotouch_extended.container, line: 178)

Traceback (most recent call last):
  File "site-packages\calibre\gui2\device.py", line 88, in run
  File "site-packages\calibre\gui2\device.py", line 587, in _upload_books
  File "calibre_plugins.kobotouch_extended.device.driver", line 335, in upload_books
  File "site-packages\calibre\devices\kobo\driver.py", line 1950, in upload_books
  File "calibre_plugins.kobotouch_extended.device.driver", line 245, in _modify_epub
InvalidEPub: Fehler beim Einfügen 'kath.net [Sa, 14 Jan 2017]' von 'calibre' mit Fehler: 'All strings must be XML compatible: Unicode or ASCII, no NULL bytes or control characters' (file: calibre_plugins.kobotouch_extended.container, line: 178)

After uninstalling Kobo Touch Extended Driver, transferring of the newspaper was possible, but trying to convert an epub to kepub showed the same error message.

That error is exactly the same error you saw with the conversion. As you were using the extended driver, it was doing the work of changing the epub to a kepub. The driver and kepub output plugin share code, and the error occurred in exactly the same line.

To test if this is the temporary directory, open a command prompt and do the following:

SET CALIBRE_TEMP_DIR=C:\TEMP
calibre-debug -g

You will need to create the directory or change it to one that already exists. The seconds command runs calibre in debug mode. You can then try the conversion and see what happens.

[jgoguen]

I'm not sure if it's the temporary directory since it's doing other write operations on files before it fails. Unfortunately, this looks to be Windows-specific (and possibly even 32-bit Windows-specific); on macOS 10.12.2 with calibre 2.77.0 the book I got converts perfectly fine.

I did see some older mailing list posts suggesting that lxml will break if you have anything other than Unicode or ASCII characters, and the file that's being referenced in the error you sent me has 0xfeff as the last (invisible) character on line 12. That character used to be a zero-width non-breaking space, but its use in that context is deprecated (use 0x2060 instead) and is currently standardised as the byte order marker which should only appear as the very first character in a file if it's present at all.

Since this is specific to Windows, there's not going to be much I can do. I can offer suggestions, and based on what you tell me I may be able to come up with something that works, but I can't promise anything. To start with, use calibre to edit the book (right-click, choose "Edit Book"), open the file "@public@vhost@g@gutenberg@html@files@14105@14 105-0-0.txt_split_000.html", and remove the last character on line 12; if you look at the bottom right you should see it say "ZERO WIDTH NO-BREAK SPACE : Line: 12 : 78". Backspace once, it should look like nothing changes but the bottom right should change. Save the book, try to convert, it'll probably fail again but it should mention a different file right before the UnicodeWarning line. If that's what happens, I may be able to work around it but it will be some additional pre-processing of each file. If that doesn't work, I guess it's back to the drawing board to figure out what to try next.

[kovidgoyal]

@jgoguen You can use

calibre.utils.cleantext.clean_xml_chars

to ensure that lxml does not choke on input.