97 replies to this topic

[ims]

Thanks. Now plugin can save true filenames and e2 use it

[Huevos]

By the way I've changed our algorithm slightly just in case there are any characters that have been overlooked. Result is the same but test is whitelist rather than blacklist. So easier to read. Also been pushed to OpenWebIF.

	if not pngname: # picon by channel name
		name = ServiceReference(serviceName).getServiceName()
		name = unicodedata.normalize('NFKD', unicode(name, 'utf_8')).encode('ASCII', 'ignore')
		name = re.sub('[^a-z0-9]', '', name.replace('&', 'and').replace('+', 'plus').replace('*', 'star').lower())
		if len(name) > 0:
			pngname = findPicon(name)

Edited by Huevos, 13 November 2014 - 21:46.

[Huevos]

@ Huevos - my plugin SetPicon can save all used picon from actual bouquet as service_name.png.

F.eg:

Eurosport_HD.png

Eurosport__HD.png

Fishing_and_Hunting.png

eeee

but e2 does not accept it with your changes. Something I overlooked ? Names all in lower ?

Reason it was done lower case is because sometimes the same channel is written differently on different satellites. For example on 13E, DAS ERSTE, and 19.2E, Das Erste. Also BBC One HD on 28E, or BBC ONE HD on terrestrial. Space was just got rid of completely because sometimes it is tab, sometimes a space, sometimes multiple spaces etc. Easiest thing was just get rid of the spaces, and also easy for everyone to understand. And 'star', 'plus', & 'and' were converted to words, which allows the file names to be completely alphanumeric.

[rhurbarb]

Hi ,Sorry are you saying this file is what i need to show pictures on my lcd screen,since updating to latest build i lost my lcd screen pictures which i used to get for streaming scenic webcam sites. 

[Huevos]

Hi ,Sorry are you saying this file is what i need to show pictures on my lcd screen,since updating to latest build i lost my lcd screen pictures which i used to get for streaming scenic webcam sites. 

No, this is nothing to do with displaying websites. It is for displaying picons by service name on the infobar and channel list. And it doesn't remove any function. It just adds an extra fallback.

[littlesat]

Why doing if len(name) > 0 while if name is doing the same?

[Huevos]

Why doing if len(name) > 0 while if name is doing the same?

Littlesat, I'm not a Python coder. So to be sure I was looking for a zero-length string I wrote it like that. In some other languages string "0" returns false.

 

if "0":
    print 'True'
else:
    print 'False'

 

In Python that returns True.

In PHP it returns False even though it is not a zero-length string.

 

Whereas len(str) > 0 returns True in both.

 

Another example:

"0" == 0

Returns False in Python and True in PHP.

[littlesat]

So in puthon you can do if name.... Python != php.

Edited by littlesat, 23 November 2014 - 09:17.

[Erik Slagter]

Indeed python is weird

[Huevos]

All I'm saying is if you write in more than one language it is easy to get confused when things are different.

 

So if I explicitly test for a zero-length string I know that is what I am getting.

 

Lets say I made a mistake further back in the code and name was now holding a boolean True. In this case, using, "if name:" would return true, but in my code would raise an exception.

 

No point arguing about it. Neither is wrong.

[Erik Slagter]

Don't get me wrong, I agree with you!

[littlesat]

Even with len(name) it wile rise an exception when name = True... So no difference here.....
In fact python makes it easy for you here...

[Huevos]

Even with len(name) it wile rise an exception when name = True... So no difference here.....
In fact python makes it easy for you here...

Littlesat that is exactly what I am saying. len(name) raises an exception if the data type is wrong. if name: does not raise that exception, so the chance to catch a bug is missed.

[rhurbarb]

 

Hi ,Sorry are you saying this file is what i need to show pictures on my lcd screen,since updating to latest build i lost my lcd screen pictures which i used to get for streaming scenic webcam sites. 

No, this is nothing to do with displaying websites. It is for displaying picons by service name on the infobar and channel list. And it doesn't remove any function. It just adds an extra fallback.

 

Thanks Huevos

 

[Taapat]

Very nice option but it works only for users who use latin characters.
First of all, as I wrote in another topic, it will not work if in the name of the channel is, for example, the Cyrillic symbol. Maybe someone has an idea how to use such names. Maybe could be used translit from here: https://github.com/j2a/pytils to replace the symbols, but I think, that replace it one by one, is slowly. If it would be acceptable for one picon, then build the channel list with picons, not seem a good idea. Maybe someone else has a better and faster solution?

 

Second, if in the user bouquet is a wrong entry, then if it worked before with no problems, now is GSOD.
Today I saw one crashlog where in the user IPTV bouquet is strange entries, coming from the wrong convertible, and which now causes GSOD. For example:

#SERVICE 4097:0:0:0:0:0:0:0:0:0:http%3a//109.87.127.138%3a8888/udp/238.0.0.4%3a1234:ÈÍÒÅÐ

Causes:

File "/usr/lib/enigma2/python/Components/Renderer/Picon.py", line 75, in getPiconName
    name = unicodedata.normalize('NFKD', unicode(name, 'utf_8')).encode('ASCII', 'ignore')
  File "/usr/lib/python2.7/encodings/utf_8.py", line 16, in decode
    return codecs.utf_8_decode(input, errors, True)
UnicodeDecodeError: 'utf8' codec can't decode byte 0xc8 in position 0: invalid continuation byte

Of course, we can say that the user correct their bouquets, but very possible that someone may be sat channel with a similar problem.

Therefore, it seems to me that the manipulation of the encoding needs to be included in the try, but maybe also here is a better solution?

[Huevos]

Taapat, your bouquets files should only contain UTF-8 characters. If you have used the wrong charset that is not the fault of this code, but the person creating the settings.

[Erik Slagter]

I agree with Huevos, but can't we make it failsafe and just ignore the non-UTF-noise?

[athoik]

I agree with Huevos, but can't we make it failsafe and just ignore the non-UTF-noise?

Can you try this:
 

    name = unicodedata.normalize('NFKD', name.decode('utf-8', 'ignore')).encode('ASCII', 'ignore')

Edited by athoik, 30 November 2014 - 09:10.

[Huevos]

Works for me... but I don't have broken settings.

[Pr2]

Hi,

 

I am now creating my picons based on the name instead of the service reference (when possible) and this give me the following idea:

 

- Can you improve the picon assignement by name by ignoring the HD reference in the TV channel name when the picon is not there.

 

For exemple TF1 HD can use the picon tf1.png if tf1hd.png doesn't exist.

 

What do you think of this improvement?

This will avoid to duplicate picons and create symlink for SD and HD channels since most of them use the same logos.

 

Pr2