One-Click Install Packages for KOReader & Plato

[turnroot]

Originally Posted by NiLuJe

@turnroot:

Thank you, I somehow had missed that part!

[elinkser]

I replaced my OCP-KOReader-v2022.08 with OCP-KOReader-v2023.06 and had the pleasant surprise of USB keyboard working on my Clara HD in the KOReader editor (although some important keys like Return, Caps, and direction keys were not being recognized correctly on two different keyboards.)

So that was pretty exciting, but then KOReader "bombed out" on a PDF file, so I reinstalled the previous version which I was perfectly content with.


I'm not sure I used the approved install method, as I just manually copied new files over the previous versions using a dual-pane window manager.
So just leaving this out there in case someone else had the same thing happen...

My crash log looked like:


---------------------------------------------
launching...
_ _____ ____ _
| |/ / _ \| _ \ ___ __ _ __| | ___ _ __
| ' / | | | |_) / _ \/ _` |/ _` |/ _ \ '__|
| . \ |_| | _ < __/ (_| | (_| | __/ |
|_|\_\___/|_| \_\___|\__,_|\__,_|\___|_|

It's a scroll... It's a codex... It's KOReader!
[*] Current time: 07/04/23-15:27:30[*] Version: v2023.06

ffi.load: libs/libutf8proc.so.2
ffi.load: blitbuffer
07/04/23-15:27:30 INFO initializing for device Kobo_nova
07/04/23-15:27:30 INFO framebuffer resolution: {
h = 1448,
w = 1072
} --[[table: 0x76ad3f20]]
[ko-input] Forked off fake event generator (pid: 5834)
07/04/23-15:27:30 WARN Device:getKeyRepeat: EVIOCGREP ioctl failed: Function not implemented
ffi.load: libs/libmupdf.so
ffi.load: libs/libwrap-mupdf.so
ffi.load: libs/libfreetype.so.6
ffi.load: libs/libharfbuzz.so.0
ffi.load: libs/libzstd.so.1
07/04/23-15:27:31 INFO Loading plugins from directory: plugins
ffi.load: sqlite3
07/04/23-15:27:35 INFO Terminal: spawn done
xxxxxxxxxxxx enable alternate buffer
xxxxxxxxxxxx enable alternate buffer
xxxxxxxxxxxx disable alternate buffer
xxxxxxxxxxxx enable alternate buffer
xxxxxxxxxxxx disable alternate buffer
07/04/23-15:32:26 INFO opening file /mnt/onboard/.adds/koreader/kobobooks/books/Kipling_ Rudyard - Just So Stories.epub
07/04/23-15:32:27 INFO Inhibiting user input
07/04/23-15:32:28 INFO Restoring user input handling
07/04/23-15:32:50 INFO opening file /mnt/onboard/.adds/koreader/kobobooks/books/Animal Farm by George Orwell.pdf
ffi.load: libs/liblept.so.5
ffi.load: libs/libk2pdfopt.so.2
Segmentation fault
!!!!
Uh oh, something went awry... (Crash n°1: 07/04/23 @ 15:32:51)
Running FW 4.35.20400 on Linux 4.1.15-00136-g12655eaaef89 (#68 SMP PREEMPT Fri May 6 14:04:51 CST 2022)
Attempting to restart KOReader . . .
!!!!
---------------------------------------------

[NiLuJe]

I wouldn't mind taking a look at that specific file (bundled with its sdr folder, in case it's settings-specific).

As for the affected keyboards, anything particular about them? (verbose debug logs would be helpful there.)

IIRC, I pulled a few keycodes from a random old (PS/2!) Logitech board... but an AZERTY one. I wouldn't think it could affect KP scan codes, but... who knows ;o).

[elinkser]

Originally Posted by NiLuJe

I wouldn't mind taking a look at that specific file (bundled with its sdr folder, in case it's settings-specific).

As for the affected keyboards, anything particular about them? (verbose debug logs would be helpful there.)

IIRC, I pulled a few keycodes from a random old (PS/2!) Logitech board... but an AZERTY one. I wouldn't think it could affect KP scan codes, but... who knows ;o).

Well I retried the installation of OCP-KOReader-v2023.06 but this time saving my files, deleting the koreader directory, and doing a fresh install from the zipfile.

And opening PDF files worked, until I copied back one too many of my backed up files - probably a settings file, since copying back my saved libs and scripts didn't mess anything up.
***
Did clean reinstall and 3 power cycles, and it opened PDF files each time.
So I'm pretty sure it was a "me" problem!

As for the the keyboards...

These are the dmesg outputs when hotplugging the two USB keyboards to the Clara HD while in KOReader:

931-usb plugged 257
932-usb 1-1: new full-speed USB device number 2 using ci_hdrc
933-PMU:_config_ricoh619_charger_params set SDP 500mA charging.
934:input: HP HP Wireless Mini Keyboard as /devices/platform/soc/2100000.aips-bus/2184000.usb/ci_hdrc.0/usb1/1-1/1-1:1.0/0003:03F0:A107.0001/input/input2
935:hid-generic 0003:03F0:A107.0001: input: USB HID v1.11 Keyboard [HP HP Wireless Mini Keyboard] on usb-ci_hdrc.0-1/input0
936:input: HP HP Wireless Mini Keyboard as /devices/platform/soc/2100000.aips-bus/2184000.usb/ci_hdrc.0/usb1/1-1/1-1:1.1/0003:03F0:A107.0002/input/input3
937:hid-generic 0003:03F0:A107.0002: input: USB HID v1.11 Mouse [HP HP Wireless Mini Keyboard] on usb-ci_hdrc.0-1/input1
---
1599-usb 1-1: new full-speed USB device number 3 using ci_hdrc
1600-input: iPazzPort iPazzPort as /devices/platform/soc/2100000.aips-bus/2184000.usb/ci_hdrc.0/usb1/1-1/1-1:1.0/0003:0168:1898.0003/input/input4
1601-hid-generic 0003:0168:1898.0003: input: USB HID v1.11 Mouse [iPazzPort iPazzPort] on usb-ci_hdrc.0-1/input0
1602-input: iPazzPort iPazzPort as /devices/platform/soc/2100000.aips-bus/2184000.usb/ci_hdrc.0/usb1/1-1/1-1:1.1/0003:0168:1898.0004/input/input5
1603:hid-generic 0003:0168:1898.0004: input: USB HID v1.11 Keyboard [iPazzPort iPazzPort] on usb-ci_hdrc.0-1/input1
---

These were the keycodes obtained for the HP Keyboard (obtained via "xev -event keyboard" when in an Alpine Linux chroot on the same device):

https://www.mobileread.com/forums/sh...5&postcount=47


Note that in the case of the chroot, simply installing the eudev hardware database, resolved the misinterpreted key issue:

https://www.mobileread.com/forums/sh...0&postcount=48

Xorg.0.log output from the Alpine chroot:

[ 61.766] (**) HP HP Wireless Mini Keyboard: Applying InputClass "evdev keyboard catchall"
[ 61.766] (**) HP HP Wireless Mini Keyboard: Applying InputClass "libinput keyboard catchall"
[ 61.766] (II) Using input driver 'libinput' for 'HP HP Wireless Mini Keyboard'
[ 61.766] (**) HP HP Wireless Mini Keyboard: always reports core events
[ 61.766] (**) Option "Device" "/dev/input/event2"
[ 61.766] (II) libinput: HP HP Wireless Mini Keyboard: is a virtual subdevice
[ 61.767] (**) Option "config_info" "udev:/sys/devices/platform/soc/2100000.aips-bus/2184000.usb/ci_hdrc.0/usb1/1-1/1-1:1.0/0003:03F0:A107.0001/input/input2/event2"


I couldn't find which database entry was being used.
The following command lists a bunch of HP devices including keyboards, but not the 03F0:A107.

# grep -n -C 1 03F0 /etc/udev/hwdb.d/*


So both keyboards are working perfectly in the chroot, but I can't identify any specific database entry associated with them.

[NiLuJe]

X11 keycodes are utterly meaningless in this context, they're a different thing altogether.

I need evdev keycodes, which would be dumped in verbose debug logs .

(or evtest/evemu-record).

[elinkser]

This guru says they're just offset by 8:
https://unix.stackexchange.com/quest...es-to-keycodes

And I believe him, after comparing evtest and xev results on a few keys.

But are we trying to rebuild the database gadget by gadget?

Would it not be easier on you to just reveal the magic procedure for each person to make their own entry?

i.e. teach them to fish, as opposed to becoming permanent supplier?

[NiLuJe]

Yes and no, it's trivial enough that it's easier to just merge the missing keys upstream directly, even for such a niche feature.

c.f., plugins/externalkeyboard.koplugin/event_map_keyboard.lua

[elinkser]

***

When upgrading to OCP-KOReader-v2023.06 and copying libs from my old KOReader setup, KOReader would crash when opening pdf files.

The crash was because of libs installed for the w3m browser:
https://www.mobileread.com/forums/sh...91&postcount=3

w3m libs not needed:
libgcc_s.so.1
libgctba.so.1
* this lib would crash KOReader when opening pdf files *

Ironically, if I don't copy those two libs to the new KOReader installation, w3m seems to work fine anyway.

***

If I set up udev rules for the HP keyboard as follows:

# mkdir /etc/udev/hwdb.d

# vi /etc/udev/hwdb.d/10-my-modifiers.hwdb

evdev:input:b0003v03F0pA107* # HP Wireless Mini Keyboard
 KEYBOARD_KEY_70028=enter  # bind to enter
 KEYBOARD_KEY_70052=up # bind to up

# udevadm control --reload-rules

# udevadm trigger

Then when I press return on the USB keyboard, it clicks whatever key the focus is on the onscreen keyboard!
i.e.the udev rule accomplished nothing!
* EDIT: even if I try to bind other keys, e.g. comma, the udev rule has no effect.

However, the arrow keys on the USB keyboard work to scroll focus on the onscreen keyboard. So if I scroll focus to the return key on the onscreen keyboard, then return on the USB keyboard works by this kludge.
On my other USB keyboard, the enter key works perfectly, but the arrow keys also scroll the onscreen keyboard.
*EDIT: LOL! It turns out that the "# udevadm trigger" was mainly what "triggered" the misdirected focus on the onscreen keyboard. You hopefully will not have this problem if you don't shoot yourself in the foot like me.
In any case, exiting the terminal and going back in resolves the issue.

So actually, these USB keyboards are pretty usable in KOReader.
And there is a nice visual cue in the file manager that you are in host mode and keyboard connected.

*I forgot to mention you have to put the Clara HD in host mode, with powered connector as explained here:
https://www.mobileread.com/forums/sh...d.php?t=340418

So this is a nice capability.
Great work NiLuje!

***

[NiLuJe]

Shipping a *shared* GCC RT is... not recommended without careful consideration and/or isolation, so, yeah, here be dragons.

[elinkser]

Originally Posted by NiLuJe

Shipping a *shared* GCC RT is... not recommended without careful consideration and/or isolation, so, yeah, here be dragons.

Thankfully they are not real dragons!

Hey, I have a question about the event_map_keyboard.lua file.
I would like to add "shift" capability to non-alphabetic characters, e.g. the slash key to be recognized as a question mark when the shift key is pressed.
Or even for all the numeric/symbol keys in the top row.

Is that capability provided there or am I misunderstanding something?

[NiLuJe]

I don't think so, but I haven't looked at it too closely.

EDIT: Yeah, no, definitely not. The shift modifier is actually handled by the InputText widget, and that's just basically a dumb tolower call if !shifted .

[elinkser]

Thanks, I appreciate the info.

[Buhaj47]

I have not updated KFMon since last year I think, is the update recommended in that case? I only update KOreader OTA.

[NiLuJe]

@Buhaj47: Nope.

[curiousdreamer]

Hello, not sure if this is the right place to post this, but I installed the one-click package from this thread and have NickelMenu and KFMon working fine (Plato and KOReader also function fine).

Somehow, the .png images appear in My Books in Nickel/default Kobo interface. I don't want to delete them or move them just yet as I heard they might have some info on them that helps keep the programs running.

I checked the /Kobo/KOBOereader.conf and

[FeatureSettings]
ExcludeSyncFolders=(\\.(?!kobo|adobe).+|([^.][^/]*/)+\\..+)

shows up in there so I am not sure what else to do.

What should I do? Completely uninstall and reinstall this one-click package? Will I have to reinstall the tweaks I put in NickelMenu (thus backing up the config file beforehand)?