JBPatch

[kbaerwald]

@NiLuJe: version is 5.1.2 (1679530004) / JBPatch is 27/25 (v2.4.1)

[kbaerwald]

I repeated the value change process through JBPatch and looked into the txt file - no changes again. Also creation date and mod date (as seen from Windows operating system) have not changed. It is interesting: when I do a manual change of this txt file Windows reflects the actual time of change but after restarting the reader the mod date is set back to 01.04.2012 01:20.

In the end - now that I know how to change values manually I have a good workaround as I do not plan to change these values very often.

@ixtab: I am sorry but I cannot judge the content of the logfile from reading the screen. Is it available as a logfile somewhere on the reader so I can copy it and send it to you?


P.S. And ... I appreciate your quick response in this matter ... thank you!

[ixtab]

Originally Posted by kbaerwald

@ixtab: I am sorry but I cannot judge the content of the logfile from reading the screen. Is it available as a logfile somewhere on the reader so I can copy it and send it to you?

Yes, of course. The file is /tmp/jbpatch.log. But you'll probably need usbnetwork to retrieve the file...

[dung.saga]

Originally Posted by dung.saga

I've just decoded PDFReader-impl.jar with jad, and I've found something looks like PDF zoom parameters in source code of PDFPageImpl.class
...
I guess this related to zoom ratio in PDF reader. But I don't know how to test it with JBPatch yet. A job for next week .

Eclipse is too bloat for my taste. So, after some searching, I use JD-GUI instead.

Arggg!!! Amazon obfuscated many class names in recent firmware updates. I wasted some time trying to figure it out. This is much harder than I expected.

Thanks to ixtab, I grabbed the firmware 5.0.0 and extract the JARs from it. All the class and method names are back . But it seems that the decompiler can't differentiate among local variables, all of them are named "a" . Does anybody know how to fix this?

[dung.saga]

While scanning decompiled source code of PDFReader-impl.jar, I found a method named zoomToColumn in class PDFBookNavigator. So, I guess Kindle Touch can zoom in to a column of text in a PDF file. I tried to double tap and it works (but I can't scroll to next or previous page without zooming out).

Zooming function of PDFReader-impl really sucks, that's why I'm searching for a way to improve it. Pan-and-zoom mode isn't fine grain enough, and zoom-to-column mode doesn't allow switching page .

I still can't find the zooming step for pan-and-zoom mode. Right now it zoom in at 1x, 2x, 3x, 4x, ... I think I should try to imrove zoom-to-column mode instead.

[dung.saga]

I found something interesting in method mousePressed(MouseEvent a) in class ContentPanel (in ReaderSDK-impl.jar). A variable is increased or decreased by 40 (depends on value of a.getButton() is 8 or 7).
As I watched "showlog -f", I see Button=7 when I tried to pinch with 2 fingers, and Button=8 when I tried to stretch with 2 fingers.

So, if my guesses are correct, I can change the zoom factor by replacing 40 with some user-defined values . Next step is learning how to do it with jbpatch

[geekmaster]

You just confirmed that pinch = keycode 7 and stretch = keycode 8, as was reported some time ago:
https://www.mobileread.com/forums/sho....php?p=1932094

You gave good information on how to use this information in java. Perhaps you could try the other gestures reported in that previous thread, using your method.

We have SO MANY threads here, that a lot of useful information can get deeply burried. The gesture engine thread link above IS in the index wikis, but even those wikis are getting a bit large.

[ixtab]

Originally Posted by dung.saga

Eclipse is too bloat for my taste. So, after some searching, I use JD-GUI instead.

Arggg!!! Amazon obfuscated many class names in recent firmware updates. I wasted some time trying to figure it out. This is much harder than I expected.

Thanks to ixtab, I grabbed the firmware 5.0.0 and extract the JARs from it. All the class and method names are back . But it seems that the decompiler can't differentiate among local variables, all of them are named "a" . Does anybody know how to fix this?

No easy way to fix it. The decompiler decompiles correctly - the variables *are* all named "a" because of obfuscation. (Yes, even 5.0.0 is slightly obfuscated). You might be able to "rename" them by modifying the constants pool of the class file, and correctly "linking" the variables there. If you're lucky, there is some deobfuscator out there which can do that, but I don't remember seeing one. If you do find one, then I'm very interested in it, because yes, it's a PITA to read such code.

One more comment concerning 5.0.0/5.1.x/5.2.0: I'm also usually resorting to the 5.0.0 version to understand what's going on, and then try to matchmake with the newer version. Just remember that the code may change slightly between the versions (some parts also changed heavily), and that ultimately you have to figure out the patch for the code of the exact version that you want to support. Oh, and a hint: "jad -a" is your friend.

[askak]

@ixtab

I've been told to do a full debrick after my font hack problem but obviously I'm very afraid of it. I'm thinking if there's a way of manually uninstalling or restoring the files that your hack modifies.

The first time I uninstalled the Jbpatch, it let me install the fonthack, which I couldn't, my mistake that I didn't try to install Jbpatch right away.

I'm asking because I want to avoid the debrick if possible, it scares me to dead, my kindle works, only problem is I can't install few hacks, and also I'm afraid it could happen again after debricking, I mean, is not that I messed up deleting some file or something. You even said it was a strange error, right?

Thanks for any help, and by the way, the time I could enjoy your hack worked great, I specially like the margins, a lot of time, I was reading in landscape mode, very few words per line with the original margins.

[ixtab]

Originally Posted by askak

@ixtab

I've been told to do a full debrick after my font hack problem but obviously I'm very afraid of it. I'm thinking if there's a way of manually uninstalling or restoring the files that your hack modifies.

The first time I uninstalled the Jbpatch, it let me install the fonthack, which I couldn't, my mistake that I didn't try to install Jbpatch right away.

I'm asking because I want to avoid the debrick if possible, it scares me to dead, my kindle works, only problem is I can't install few hacks, and also I'm afraid it could happen again after debricking, I mean, is not that I messed up deleting some file or something. You even said it was a strange error, right?

Thanks for any help, and by the way, the time I could enjoy your hack worked great, I specially like the margins, a lot of time, I was reading in landscape mode, very few words per line with the original margins.

Hmm... I mean it's your device after all, but I also suggest to do a full debrick. Your device is definitely acting strange if you can install some hacks, but not others, if fonthack only works after uninstalling jbpatch, etc. There is nothing to be scared of in the debrick procedure (except possibly the length of reading and understanding the procedure, and the windows driver nightmare). It is completely impossible to permanently brick the Kindle Touch - you can always get it back to life with the debrick procedure. Just make sure to understand and follow the guide, and your device will be up and running within 10 minutes.

[ixtab]

@askak: one more thing, in case you didn't debrick yet: can you try to install JBPatch and report back what exactly it says when failing? The JBPatch installer will tell exactly which step failed. That may be valuable information.

[askak]

Thanks a lot for your time ixtab.

Installing the error just says "unable to copy jbpatch.jar". And trying to uninstall (have to try everything) "can't restore init.xargs, bailing out".

Hope it helps a little bit.

[NiLuJe]

@askak: Yeah, that's... not good. To be perfectly honest, we could spend (some? a lot of?) time to take a dive into the logs & the system to try to figure it out, but it'd quite frankly be a lot less painful if you had a bit of Linux background.

(FWIW, that smells like a ro rootfs, which shouldn't happen during the update process (might happen with a corrupt FS, but in that case, it should have triggered a fsck at boot, which should have fixed it... :?)).

So, yeah, the full debrick/reflash might really be the easiest/fastest solution .

[askak]

Originally Posted by NiLuJe

@askak: Yeah, that's... not good. To be perfectly honest, we could spend (some? a lot of?) time to take a dive into the logs & the system to try to figure it out, but it'd quite frankly be a lot less painful if you had a bit of Linux background.

(FWIW, that smells like a ro rootfs, which shouldn't happen during the update process (might happen with a corrupt FS, but in that case, it should have triggered a fsck at boot, which should have fixed it... :?)).

So, yeah, the full debrick/reflash might really be the easiest/fastest solution .

Well, I changed the filesystem to rw, because the wiki commands to backup weren't working, and if has to be changed to ro again I don't think I did . I tried to ssh a lot of times, so probably yes, I damaged the filesystem.

[icantbebothered]

Sorry to break the mellow atmosphere, but I have an idea for a patch.

It's a patch that keeps the black header thingy at the top of the page when reading, even after you turn the page. And I hope nobody waits for me to make it, because they might wait over a year.