strange ocurance with epg

[stavo]

could some kind sole undertake a test for me, using the latest vix image, 10 80 skin not sure if its relevant, vu ultimo 4k stb, when i download the onyx skin from feeds and install, my epg on the bullsat package (39 east) turns into symbols you cant read it,it only happens on this package any one else had this.

Regards

Stavo

[abu baniaz]

Are other skins ok?

[stavo]

After some more testing it appears the three kiddac skins in the plugin downloads feeds are all the same for me when watching bullsat the epg displays symbols not words, I did however download & install 6 other random skins from the feeds and they all worked correctly seems very strange ? can some kind sole test the skin on there setup report the findings

regards

Stavo

[abu baniaz]

The fonts he uses in his skin may not be able to handle the items. You need to report it to him. Post a screenshot so he sees what you mean.

[ronand]

Its not strange at all. Not all fonts are unicode containing non-latin character sets. It may not bother most people but there will be some situations (such as yours) where it is required.

[stavo]

Thanks Abu for reasons why this is happening will get in touch with Kiddac, Ronand not realy clued up on fonts and such, my reasons for thinking it was a strange ocurance was to the fact it only happens on the bulsat package so now i no, thanks to you both for infornentive replys.

Regards

Stavo

[Joe_90]

Bulsat will be using the Cyrillic alphabet which has a different character set. Like this - ПРОГРАМИ, КОИТО НИЕ ВИ ПРЕДЛАГАМЕ СА ЧАСТ ОТ
Is that what you mean? Or are the characters substituted with something else? A screenshot would help.

[BrokenUnusableAccount]

I only have Freeview and 28.2°E so I don't see many EPG entries with anything other than ordinary ASCII in them.
However the odd thing does pop up with accented characters and I do so them sometimes display incorrectly, and sometimes change to become correct after being wrong.
My best guess so far is that what's happening is that something like perhaps the character ò shows correctly if it came from the Freesat EPG but shows as À if it came from the Sky EPG.
It could also be that the correct characters come from now/next rather than from the EPG.

For me changing to a different skin doesn't seem to make any difference.

[birdman]

However the odd thing does pop up with accented characters and I do so them sometimes display incorrectly, and sometimes change to become correct after being wrong.
My best guess so far is that what's happening is that something like perhaps the character ò shows correctly if it came from the Freesat EPG but shows as À if it came from the Sky EPG.That's more likely to be an encoding issue.

if someone starts with the data (text) in, say, a Windows-1252(?) setting and moves that to something expecting utf-8 without a re-encoding the result will, for some characters, be incorrect.
Or sometimes the result indicates that things have been encoded twice...

EPG data providers have bugs too.


For me changing to a different skin doesn't seem to make any difference.That would be because the fonts you are trying to use all have glyphs for all characters you are trying to use.

It's been noted above that the skins in question uses fonts with fewer (or fewer relevant) glyphs.

[LraiZer]

My best guess so far is that what's happening is that something like perhaps the character ò shows correctly if it came from the Freesat EPG but shows as À if it came from the Sky EPG.I suggest this guess is the wrong way around, Freesat shows wrong and OpenTV shows correct?

Freesat is just decoded with the freesat huffman code and this is all. OpenTV is decoded with the opentv huffman code, but then also sent to be recoded using the 8859-5 char table. As far as i remember at the time, this recode was tested to display the correct chars in the International bouquet and the welsh channels. I only tested this char set on the 28.2e transponder.

eg. firman sam and paw patrol on channel S4C:

Freesat: Sam TÃan
OpenTV: Sam Tân

Freesat: PatrÃol Pawennau
OpenTV: Patrôl Pawennau

Freesat: Ty MÃel
OpenTV: Ty Mêl

[birdman]

Freesat is just decoded with the freesat huffman code and this is all. OpenTV is decoded with the opentv huffman code, but then also sent to be recoded using the 8859-5 char table. As far as i remember at the time, this recode was tested to display the correct chars in the International bouquet and the welsh channels.lOdd, given that Welsh uses ISO 8859-14.

Why not just use Unicode for everything?

[Joe_90]

@stavo please post a screenshot of what you are seeing on Bulsat.

[BrokenUnusableAccount]

I suggest this guess is the wrong way around, Freesat shows wrong and OpenTV shows correct?
Yes could be.
Plus of course, as @birdman pointed out, one or other could be sending it out in the wrong character set.

[abu baniaz]

There was an issue/thread with autotimers not getting the right hits on PLI forum. One of the explanations was this:
https://withblue.ink/2019/03/11/why-you-need-to-normalize-unicode-strings.html

Issue was raised here:
https://github.com/oe-alliance/enigma2-plugins/issues/405

[birdman]

Having a utf-8 case-insensitive matcher might be useful, but will still be of no help at all if the original descriptions are wrongly encocded.