Error importing mobi files

[Nomadin]

I am receiving the following error when I attempt to add mobi files to Calibre. This is just one example, it happens on every one.

WARNING: Failed to read metadata: Failed to read metadata from the following:

The_Buried_Pyramid.mobi
Traceback (most recent call last):
File "site-packages\calibre\ebooks\metadata\worker.py", line 54, in read_metadata_
File "shutil.py", line 52, in copyfile
IOError: [Errno 13] Permission denied: 'C:\\ProgramData\\Microsoft\\Windows\\Start Menu\\Programs\\calibre - E-book Management'

Any ideas where I should look first? The file is added to Calibre, but I have to add all of the metadata manually.



(Windows 7, Calibre 0.6.39)

[kovidgoyal]

Do you have a plugin for removing drm from mobi files?

[kilikini]

I was also receiving this error while adding prc/mobi files and using the mobidedrm plugin, but after updating the plugin to the latest version of mobidedrm, 0.10, the error no longer occurs and calibre is again able to read the metadata successfully.

[Nomadin]

Yes, I de-protected my Kindle books for future protection. I guess that script is interfering?

[Nomadin]

Originally Posted by kilikini

I was also receiving this error while adding prc/mobi files and using the mobidedrm plugin, but after updating the plugin to the latest version of mobidedrm, 0.10, the error no longer occurs and calibre is again able to read the metadata successfully.

Thanks for the lead! For the moment, I disabled the plugin and the import worked perfectly.

Thank you both for the incredibly quick reply!

[pdurrant]

Originally Posted by Nomadin

Thanks for the lead! For the moment, I disabled the plugin and the import worked perfectly.

I have seen a few reports of this problem, but I don't know what could be causing it.

I attach a dummy import plug-in. It does nothing except make a temporary copy of the file and returns that to Calibre. (In a very similar manner to MobiDeDRM, but without doing any manipulation of the file.)

Perhaps people who've had problems with the MobiDeDRM plug-in could try this plug-in (removing any other mobi/prc/azw plugin first) and see if they still have a problem with import. If they do, perhaps Kovid could take a look and suggest what the problem is with this (completely innocuous) code.

[HarryT]

It looks as if it's attempting to create a file in a directory in which it doesn't have write access.

[pdurrant]

Originally Posted by HarryT

It looks as if it's attempting to create a file in a directory in which it doesn't have write access.

If you look at the dummy plug-in, you'll see that it just copies the input file to an output file using the Calibre-provided temporary_file method.

http://calibre-ebook.com/user_manual...filetypeplugin

Now, it might be that the code isn't implementing a FileTypePlugin correctly. Or perhaps there's a problem with Calibre. that's why I created this dummy plug-in. Now I need

(1) someone who's had a problem with Mobipocket import plug-ins to try this dummy plug-in and see if there's still a problem
(2) If there is still a problem, for someone who knows more about Calibre plug-ins that me to take a look at the plug-in and see if there's something wrong with it
(3) If there isn't, for someone to try to work out what's wrong with the Calibre/Plug-in interaction.

[pdurrant]

Originally Posted by pdurrant

Calibre-provided temporary_file method.

Hmm.. I've been looking at this bit of code and the documentation, and I can see a potential problem or two.

Here's the guts of the plugin:

def run(self, path_to_ebook):
    of = self.temporary_file('.mobi')
    data_file = file(path_to_ebook, 'rb').read()
    file(of.name, 'wb').write(data_file)
    return of.name

Now the documentation describes temporary_file as returning a file-like object. It doesn't say whether this file object is open, and if it is, whether it's open in binary of text mode.

So the plug-in uses the file path from the open to make a new file object, open it for binary writing, and writes the input file to the temporary file.

* Should the plug-in be using the of.name in this way, and opening the file as binary write itself?
* Should the plug-in be closing the temporary_file in seom explicit way?

[kovidgoyal]

Temporary files are always opened in w+b (write, random access, binary)

of = self.temporary_file('.mobi')
of.write(data)
of.close()
return of.name

[pdurrant]

Originally Posted by kovidgoyal

Temporary files are always opened in w+b (write, random access, binary)

Ah - thank you. I'll get someone to give it a go.

[pdurrant]

Originally Posted by kovidgoyal

Temporary files are always opened in w+b (write, random access, binary)

That seems to have done the trick. Thanks again.

A couple more questions:

Suppose a Filetype plugin was unable to successfully import a file for some reason.

1. What would be the best way to report this error to the user?
2. How would is be best to tell Calibre that the import had failed?

[GRiker]

Here's a template that lets you display a simple dialog:

import os
from calibre.customize import FileTypePlugin

class HelloWorld(FileTypePlugin):

    name                = 'HelloWorld'
    description         = 'A shell for creating plugins'
    supported_platforms = ['windows', 'osx', 'linux']
    author              = 'your_name_here'
    version             = (1, 0, 0)
    file_types          = set(['your_file_type_here'])
    on_import           = True 

    def run(self, path_to_ebook):
        from calibre.gui2 import is_ok_to_use_qt
        from PyQt4.Qt import QMessageBox

        if is_ok_to_use_qt():
            QMessageBox.question( None, 
                        'Hello World Plugin', 
                        "path_to_ebook: %s" % path_to_ebook,
                         QMessageBox.Ok)

        return path_to_ebook

I believe if you return None calibre understands that the import failed.

G

[pdurrant]

Originally Posted by GRiker

Here's a template that lets you display a simple dialog:

I believe if you return None calibre understands that the import failed.

Thank you. That looks simple enough.

[pdurrant]

Originally Posted by pdurrant

Thank you. That looks simple enough.

I have a message box appearing now - thank you.

But returning None doesn't stop Calibre trying to import the file. Perhaps when a script can't do the intended conversion, it would be best just to return the original file path.