[Editor Plugin] EpubCheck

[Doitsu]

This thread is not intended for discussing the pros and cons of using EPUBCheck or best practices.

@ryder please ask questions about best practices in the ePub forum.

I'll ask the mods to hide future posts that are not about installing or using the Calibre EPUBCheck plugin.

[Morea]

Hello!

I get an error message when starting this plugin:

Traceback (most recent call last):
File "calibre_plugins.epub_check.main", line 387, in ask_user
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xfc in position 28: invalid start byte

The version of Calibre doesn't matter. I uninstalled and reinstalled the plugin with the same result.

Hope you can help.

Morea

[JSWolf]

Originally Posted by Morea

Hello!

I get an error message when starting this plugin:



The version of Calibre doesn't matter. I uninstalled and reinstalled the plugin with the same result.

Hope you can help.

Morea

Are you using the latest 64-bit version of Java? If it's any 32-bit version of Java, then you will have a problem.

[Morea]

Originally Posted by JSWolf

Are you using the latest 64-bit version of Java? If it's any 32-bit version of Java, then you will have a problem.

Yes, I am using OpenJDK Runtime Environment Temurin-21.0.5+11 (build 21.0.5+11-LTS) 64 bit.

BUT the user settings of my Calibre installation come from an older installation of Calibre - and perhaps that was a 32-bit installation...

[Doitsu]

Originally Posted by Morea

I get an error message when starting this plugin

The error message is usually displayed because EPUBCheck generated invalid UTF-8 output. This usually happens if the epub file contains files that aren't valid UTF-8 files.

To fix this problem:

• uninstall the plugin
• re-download the plugin if you didn't save it
• extract main.py from the zip file
• search for and change the following two lines:
  

              stdout = result[0].decode('utf-8')
              stderr = result[1].decode('utf-8')

  to:
  

              stdout = result[0].decode('utf-8', errors='ignore')
              stderr = result[1].decode('utf-8', errors='ignore')

• replace main.py in the zip file with the modified version
• re-install the plugin from the .zip file

If you're getting more Python errors, open the book with Calibre Editor, select Tools > Fix HTML - all files followed by Tools > Beautify all files, save the epub and re-run the EpubCheck plugin.

[JSWolf]

Originally Posted by Doitsu

The error message is usually displayed because EPUBCheck generated invalid UTF-8 output. This usually happens if the epub file contains files that aren't valid UTF-8 files.

To fix this problem:

• uninstall the plugin
• re-download the plugin if you didn't save it
• extract main.py from the zip file
• search for and change the following two lines:
  

              stdout = result[0].decode('utf-8')
              stderr = result[1].decode('utf-8')

  to:
  

              stdout = result[0].decode('utf-8', errors='ignore')
              stderr = result[1].decode('utf-8', errors='ignore')

• replace main.py in the zip file with the modified version
• re-install the plugin from the .zip file

If you're getting more Python errors, open the book with Calibre Editor, select Tools > Fix HTML - all files followed by Tools > Beautify all files, save the epub and re-run the EpubCheck plugin.

Does this mean there will be a new version?

[DNSB]

Personally, if I get an error message indicating that an ePub files contains encoding that is not UTF-8 or UTF-16 (I've seen 2 ePubs in my entire involvement with ereading that were UTF-16 encoded), that ePub is not standards compliant and should be fixed. As section 3.9 of the ePub 3.3 recommendation states:

MUST be encoded in UTF-8 or UTF-16 [unicode], with UTF-8 as the RECOMMENDED encoding.

While the ePub 2.0.1 Open Publication Structure document 1.4.1.2 states:

A conformant XHTML Content Document must meet these conditions:
it is a well-formed XML document (as defined by XML 1.0); and
it is encoded in UTF-8 or UTF-16; and

I will admit that I have seen several ePubs where the creator did not use UTF-8 or UTF-16 but again, they are not standards compliant and were often the result of an amateur converting book formats or converting an OCR scan of a pbook to an ePub document.

[Morea]

open the book with Calibre Editor, select Tools > Fix HTML - all files followed by Tools > Beautify all files, save the epub and re-run the EpubCheck plugin.

This didn't help, and it could not, since this problem even occurs with books that I know to be 100% error-free.

I did not try to fix the plugin's main.py, since I used the epubcheck plugin in the version 4.26 for some years and never got such a crash. I suppose there is something new in Windows 11 that was no problem in Windows 10 all these years...

[Morea]

Now I fixed the plugin's main.py as you described and: yes! It works like charm, with the old 4.26 epubcheck version and the new 5.12 one!

This is the fixed zip:
https://www.mobileread.com/forums/at...1&d=1741082768

Thank you!

[rumplestiltskin]

Getting a really strange error window when I run EpubCheck; way beyond my pay grade. Seems to be the same error with any ePub I open with the Edit command and then run the EpubCheck.

[DNSB]

Rather odd. What version of Java are you running? If you are running a 64 bit version of Java, you could try modifying the main.py file in the EPUBCHECK directory by modifying the 64 bit command line parameters line to read:

            # define epubcheck command line parameters
            if is32bit:
                args = [java_path, '-Dfile.encoding=UTF8', '-Xss1024k', '-jar', epc_path, '--version']
            else:
                args = [java_path, '-Dfile.encoding=UTF8', '-Xss2048k', '-jar', epc_path, '--version']

I will have to say that running a 64bit Java, I only found 1 time when I got a stack overflow error and that was a omnibus document that to put it mildly was poorly constructed. OTOH, it reported over 1000 errors not the no errors which your case seems to show. OTOH, neither does your image show the version of epubcheck being used.

[Doitsu]

There's no need to modify the source code. Simply change the is32bit parameter in the EpubCheck.json file to true.

  "is32bit": true,

Or simply uninstall Java and install the latest 64 bit Java version.

[rumplestiltskin]

Will try the fix(es) suggested in the morning. I'm writing this from another computer and EpubCheck is working correctly (once I installed the 64-bit Java - this machine actually didn't have -any- Java installed).

I did get a message about installing a new version of EpubCheck but, when I had Calibre check for one, it didn't do anything. I believe I have 4.26 installed.

I'll advise tomorrow afternoon how it all goes.

Thank you!

[rumplestiltskin]

My installed Java (on the machine displaying the stackOverflow) was v8 161 (which I do not believe was ID'd as 64-bit). I DL'd the new Java installer and it removed the old version and installed v8 451 which -is- 64-bit.

The EpubCheck ran without any Java errors and all seems to be fine. Again, thanks for the advice.