K5 FW 5.x ScreenSavers Hack

[NiLuJe]

@nemiere: Huh, that's weird.

Try to redirect the KFXMeta output to a file, by tweaking linkss/bin/cover-extract (with a proper code editor that won't mess up line-endings), by switching L#115 from

${PYTHON} ${LINKSS_BINDIR}/kfxmeta.py "${last_read_kfx}" "${LINKSS_TMPFS_CACHE}/"

to

${PYTHON} ${LINKSS_BINDIR}/kfxmeta.py "${last_read_kfx}" "${LINKSS_TMPFS_CACHE}/" &> /mnt/us/linkss/kfxmeta.log

And check what linkss/kfxmeta.log has to say about that after trying to reopen this book...

[nemiere]

Really weird. Contents of kfxmeta.log:

KFX Meta v1.1.N
Traceback (most recent call last):
  File "/mnt/us/linkss/bin/kfxmeta.py", line 421, in <module>
    main()
  File "/mnt/us/linkss/bin/kfxmeta.py", line 144, in main
    print('Decoding "{}"'.format(infile))
UnicodeEncodeError: 'ascii' codec can't encode character u'\u2019' in position 32: ordinal not in range(128)

But: I have opened kterm (inside KUAL) and have tried:

# cd /mnt/us
# ./linkss/bin/kfxmeta.py "documents/Fool’s Errand (The Tawny Man Trilogy, Book 1)_B005JE1KD8.sdr/assets/metadata.kfx" ./
KFX Meta v1.1.N
Decoding "documents/Fool’s Errand (The Tawny Man Trilogy, Book 1)_B005JE1KD8.sdr/assets/metadata.kfx"
Extracting cover image to cover_raw_B005JE1KD8.jpg

Perhaps it's relevant kterm's value of LANG:

# echo $LANG
es_ES.utf8

[knc1]

You mean it appears to be sensitive to the current locale?

[NiLuJe]

@nemiere: Yaaaay...

Can't be it, the watchdog inherits the same env, LANG is set properly there too.
Plus, lab126 has a ldpreload hack whose sole purpose is making sure the env is always set locale-wise.

Best guess is being headless confuses Python... let's try that:

env PYTHONIOENCODING="utf-8" ${PYTHON} ${LINKSS_BINDIR}/kfxmeta.py "${last_read_kfx}" "${LINKSS_TMPFS_CACHE}/" &> /mnt/us/linkss/kfxmeta.log

[nemiere]

Now it's working ok for my 3 books with non-ascii chars in filename! But I don't understand the whole thing behind this problem...

[NiLuJe]

@nemiere: Python 2's handling of unicode and encodings and how it's handled on different platforms/environment being a fantastic mess unless you take a crazy amount of extra precautions in your code? :]

[44x2]

As the title says the button of my kindle pw2 is greyed out after I copied the installation file
Please me

[eschwartz]

Per the instructions in the first post (yes, those instructions are there for a reason)...

Originally Posted by NiLuJe

Note that if you have KUAL set up and running, you're welcome to use our very own MR Package Installer, which may speed up the process if you're installing multiple things .
IMPORTANT: Remember that on FW >= 5.5.x, using MRPI is REQUIRED to install custom packages!

[nemiere]

Originally Posted by NiLuJe

@nemiere: Python 2's handling of unicode and encodings and how it's handled on different platforms/environment being a fantastic mess unless you take a crazy amount of extra precautions in your code? :]

Indeed. I have to study more on this area...

Anyway, @NiLuJe thanks alot for your work and help!

[NiLuJe]

@nemiere: Thanks for your detailed reports and tests :].

[knc1]

Originally Posted by NiLuJe

@nemiere: Python 2's handling of unicode and encodings and how it's handled on different platforms/environment being a fantastic mess unless you take a crazy amount of extra precautions in your code? :]

Isn't Python one of those languages where (for utf8 conversions) you have to set **both** the locale (done, by the pre-loaded .so) and the code-page for the non-utf8 (not done, by the pre-loaded .so)?

The code page (notice the single byte error report) for code conversions in es_ES.utf8 is not the same as for that in en_GB.utf8 .

[NiLuJe]

@knc1: It's a mess, but in this case, it's simply the fact that being headless for some reason (AFAICT, it shouldn't if the env is sane, which it is, but, hey...) broke the stdout encoding autodetection. With a tty, it correctly inherits the utf-8 from the env, without one, it defaults to none -> ascii, which breaks because the rest of the code is going to great pains to make sure everything is either an utf-8 str, or an unicode object.

(We're not alone, and Python 3 is apparently much saner (in that respect, at least))

[knc1]

Originally Posted by NiLuJe

@knc1: It's a mess, but in this case, it's simply the fact that being headless for some reason (AFAICT, it shouldn't if the env is sane, which it is, but, hey...) broke the stdout encoding autodetection. With a tty, it correctly inherits the utf-8 from the env, without one, it defaults to none -> ascii, which breaks because the rest of the code is going to great pains to make sure everything is either an utf-8 str, or an unicode object.

(We're not alone, and Python 3 is apparently much saner (in that respect, at least))

Which is what I tried to say.
none -> ascii is correct for non-latin-1 English and wrong for Spanish.

It was just my guess about what the source of the problem was that was wrong. (not passed vs not specified).

= = = =

Meanwhile, in my own backyard - (I don't do the big snake stuff) -
You would not believe what Lab126 has done to the "standardized" extension library location(s) in their Lua install.
About the only thing the same is 'in current directory', I guess they couldn't be bothered to screw that up also.

[nemiere]

Originally Posted by NiLuJe

(We're not alone)

Interesting summary of proper pythonic way of this mess.

[thirdfloornorth]

Can not get cover view to work

I just jailbroke my Paperwhite 2, 5.6.5. Installed Python, then the screen saver. Got the confirmation "Screen Saver Hack is Installed!"

Created a blank folder in the "linkss" folder named "cover".

Went into the settings in KUAL and told it "behavior: cover" and restarted the framework. Also rebooted the whole thing. Also, made sure my time was correct.

Tried opening a variety of books. Will not show the cover, still just the "screensaver hack installed!" image.

I know for "custom" images, your files have to follow certain guidelines for resolution and type. But what about for "last cover" covers that I put on the book myself using Calibre? Do they have to follow the same guidelines?

EDIT

Solution! It helps to read the ENTIRE instructions. For anyone else who has this problem... you probably forgot to unplug your Kindle while installing, OR you made the 'cover' folder when its best to use the built-in option to do so.

• Run the 'Uninstall' bin included in the Python installation zip
• Run the 'Uninstall' bin included in the Screen Saver installation zip
• Re-transfer the correct Python and Screen Saver installation bins to your Kindle
• UNPLUG YOUR KINDLE
• Re-install both Python and Screen Saver
• Go into the Screen Saver options in KUAL
• Go to Screen Savers Behavior
• Tell it you want Cover Mode
• Re-start the Framework