EPG Translator (Discussion thread)

[birdman]

Updated version 2.0rc5 now in the download thread, which should fix EventViews (and bumps the encode text length up to 7000).

[ja5]

Traceback (most recent call last):
File "/usr/lib/enigma2/python/Components/ActionMap.py", line 57, in action
File "/usr/lib/enigma2/python/Plugins/Extensions/EPGTranslator/plugin.py", line 813, in EPGTr_ToggleMode
self.setEvent(self.event)
File "/usr/lib/enigma2/python/Plugins/Extensions/EPGTranslator/plugin.py", line 909, in My_setEvent
self["epg_eventname"].setText(t_title)
KeyError: 'epg_eventname'
22:25:38.1137 [ E ] python/python.cpp:210 call [ePyObject] (PyObject_CallObject(<bound method ActionMap.action of <Components.ActionMap.ActionMap instance at 0xa63071e8>>,('VirtualKeyboardActions', 'showVirtualKeyboard'))
failed)

I got this once in openatv but rebooted and didnt happen again strange.

Tried 4 different skins in openatv but no eventview popup translate, screen bar shows something happend but no translation.

no translation openeight and pli. unless you can think of a work around then have to use 2.0

Openvix works really well now, no problem with long epg etc
Even tried a few other skins and all good.

rc5 best so far, well done mate

[birdman]

Openvix works really well now, no problem with long epg etc
Even tried a few other skins and all good.

rc5 best so far, well done mate

Thanks for that.
I'll take a look at OpenATV to see what I can find. If self["epg_eventname"].setText(t_title) is a KeyError then there is a significant difference there.

Since the Vix bits that are meant to be working are now working I might take a break from updates for a few days.
(Unless I come up with a possible solution for OpenATV - which looks possible after a quick check of that crash dump extract....)

[ja5]

birdman

not sure how i did it but twice i got

Traceback (most recent call last):
< 1568.4833> 00:24:51.2751 File "/usr/lib/enigma2/python/Components/ActionMap.py", line 57, in action
< 1568.4835> 00:24:51.2753 File "/usr/lib/enigma2/python/Screens/EventView.py", line 96, in nextEvent
< 1568.4836> 00:24:51.2754 File "/usr/lib/enigma2/python/Screens/EpgSelectionGrid.py", line 208, in eventViewCallback
< 1568.4837> 00:24:51.2755 File "/usr/lib/enigma2/python/Plugins/Extensions/EPGTranslator/plugin.py", line 841, in My_setEvent
< 1568.4882> 00:24:51.2800 uref = make_uref(event.getEventId(), self.currentService.getServiceName())
< 1568.4883> 00:24:51.2801 AttributeError: 'NoneType' object has no attribute 'getEventId'
< 1568.4884> 00:24:51.2802 [ePyObject] (CallObject(<bound method ActionMap.action of <Components.ActionMap.ActionMap instance at 0xaebbc620>>,('EventViewActions', 'nextEvent')) failed

[Skin] Error in screen 'EPGSelection' widget 'widget' [Skin] {ViX-Night-HD/skin.xml}: Component with name 'number' was not found in skin of screen 'EPGSelection'! Please contact the skin's author!!
[Skin] Processing screen 'EventView', position=(150, 460), size=(980 x 250) for module 'EventViewEPGSelect'.
[Screen] Warning: Skin is missing element 'epg_description' in <class 'Screens.EventView.EventViewEPGSelect'>.
[Screen] Warning: Skin is missing element 'epg_eventname' in <class 'Screens.EventView.EventViewEPGSelect'>.
[Screen] Warning: Skin is missing element 'channel' in <class 'Screens.EventView.EventViewEPGSelect'>.

works well in eventview popup then seems to stop translating (random), then change channel and works again or clearing cache works
Same thing doesnt happen in the plugin

changed to simple ten eighty and no crash

[birdman]

< 1568.4882> 00:24:51.2800 uref = make_uref(event.getEventId(), self.currentService.getServiceName())
< 1568.4883> 00:24:51.2801 AttributeError: 'NoneType' object has no attribute 'getEventId'

Ah, I'd missed that in setEvent() it checks that it actually has an Event.
I'll update my code to make the same check.

[birdman]

2.0-rc6 is available on the download thread.

This fixes the EventView to use generic (rather than an accidentally Vix-specific) source for translational data, which it puts into all know data fields (if they exist for the distro).

[birdman]

2.0-rc7 is now in the download thread.

It fixes a crash if a description is empty, and also fixes EventView windows, which were cacheing the untranslated values, not the translated ones (a result of an incomplete variable name tidy-up).

[ja5]

2.0-rc7 is now in the download thread.

It fixes a crash if a description is empty, and also fixes EventView windows, which were cacheing the untranslated values, not the translated ones (a result of an incomplete variable name tidy-up).

Testing on few images and can report all bugs iv reported seem gone. No random non translation also.
I will test again and see if anything shows up but so far very good.

Um i changed the self.EPGTr_translating = false to true and auto translate to english and text change to default language.
Seems to work well and no problems

Brilliant work

[birdman]

Along with looking at translations into Arabic (no-one may be using hem, but it annoys me that they are wrong when I can probably fix them).

Well, I can help a little but the problem appears to be in the widget handling itself (libfribidi?).

• It doesn't actually display right-to-left text properly at all (as it butts up to the left margin, when it should start on the right).
• If there are any non-rtol characters present such as something which was not translated) then the word ordering goes haywire. Including, apparently, displaying the lines in the wrong order (last to first).

I can help with the second part by:

• putting the start time and duration (which throw it) into a single bar at the top and then omitting it form both the original and translated text (this actually frees up additional display space).
• not putting any [] props into the translation (as these really throw it off).

Juts need to tidy up the Skin_small data for this...

[birdman]

I've put 2.01-rc8 on the download thread.

This moves the programme time and duration to its own window above the descriptions, so that it only gets displayed once.
It allows () as well as [] groups in the property matching.
And it does not display such groups when displaying right-to-left text as their untranslated nature means they usually mess up the display.

[ja5]

I've put 2.01-rc8 on the download thread.

This moves the programme time and duration to its own window above the descriptions, so that it only gets displayed once.
It allows () as well as [] groups in the property matching.
And it does not display such groups when displaying right-to-left text as their untranslated nature means they usually mess up the display.

Sorry to report again but some channels showing some untranslated and translated at the same time

[birdman]

Sorry to report again but some channels showing some untranslated and translated at the same time

That's disappointing, and odd.
The only changes were as noted, and it's difficult to see how any of that could have led to failed translations.

Are there any entries in your debug logs that contain "[EPGTranslator-Plugin]"?

What do you mean by "at the same time" - there's only one window!?!

[ja5]

That's disappointing, and odd.
The only changes were as noted, and it's difficult to see how any of that could have led to failed translations.

Are there any entries in your debug logs that contain "[EPGTranslator-Plugin]"?

What do you mean by "at the same time" - there's only one window!?!

There is nothing in the log or i would have reported it

Even if its plugin or eventview it will show part translation bit like the random before, i checked rc7 same channels work fine.

[ja5]

I think i found the error

rc8
(?: # Start a group

changed same as rc7

(?:\[ # Start a group

So far all working again

[birdman]

I think i found the error

rc8
(?: # Start a group

changed same as rc7

(?:\[ # Start a group

So far all working again

That makes no sense to me.
Not clear whether you are referring to begin_props or end_props, but both were changed to allow [ or ( to start a "property" grouping.

So:

 (?:\[                      # Start a group

became

 (?:                        # Start a group
  (?:\[|\()                 # Opening [ or (

You are starting a group a second time, which should fail, as there is (should be) no matching closer.

Could you send the text (or a screen picture) of a description that suffered? My tests (admittedly in Perl, but these are Perl regexes and it's so much easier to debug with) all work fine.