Wrong font rendering in EPG eventlist

[Benny]

Greetings,

small bug in eventlist.cpp:
https://gitorious.org/neutrino-mp/marti ... ee35b58e78

Code: Alles auswählen

snprintf(beginnt, sizeof(beginnt), "%s %d %s", g_Locale->getText(LOCALE_WORD_IN), seit, unit_short_minute);
			int w = g_Font[SNeutrinoSettings::FONT_TYPE_EVENTLIST_ITEMSMALL]->getRenderWidth(beginnt) + 10;
			g_Font[SNeutrinoSettings::FONT_TYPE_EVENTLIST_ITEMSMALL]->RenderString(x+width-fwidth2-5- 20- w, ypos+ fheight1+3, fwidth2, beginnt, color);

String (at line 765 in eventlist.cpp) should be rendered as UTF8:

Code: Alles auswählen

g_Font[SNeutrinoSettings::FONT_TYPE_EVENTLIST_ITEMSMALL]->RenderString(x+width-fwidth2-5- 20- w, ypos+ fheight1+3, fwidth2, beginnt, color, 0, true);

martii, can you explain to me - why some string rendered as UTF8, others - is not? Ie, why not render all of them in UTF8 by default?

Screenshots:

[Benny]

Some styling for EPG view/eventlist (not really necessary, just for better "look & feel"):

[martii]

Some styling for EPG view/eventlist (not really necessary, just for better "look & feel"):

Applied, thanks!

[martii]

Hi Benny,

martii, can you explain to me - why some string rendered as UTF8, others - is not? Ie, why not render all of them in UTF8 by default?

usually these are just oversights. Nowadays pretty much everything should conform to UTF8 (and changing the default to UTF8 could be a good idea).

Cheers,

martii

[Benny]

usually these are just oversights ...and changing the default to UTF8 could be a good idea.

So there no any hidden traps with it, in Neutrino code?

Ok, thanks for explanation.

[martii]

So there no any hidden traps with it, in Neutrino code?

Well, I wouldn't vouch for that.

Anyway ... the font renderer now defaults to UTF-8: https://gitorious.org/neutrino-mp/marti ... af51de8909

Cheers,

martii

[Benny]

Anyway ... the font renderer now defaults to UTF-8

Oh, this was really quick decision, thank you. If there any pitfalls still exist, soon we will know about it =)

Update: quick test after repo update - at first glance, everything looks OK, but:

1) unrelated to, seems old small "defect", in streaminfo2.cpp line 479 should look like:

Code: Alles auswählen

 spaceoffset = array[i]+10; //add 10 pix to x_offset

2) related to, in eventlist.cpp line 768 should look like:

Code: Alles auswählen

g_Font[SNeutrinoSettings::FONT_TYPE_EVENTLIST_ITEMSMALL]->RenderString(x+width-fwidth2-5- 20- w, ypos+ fheight1+3, w, beginnt, color);

[Benny]

martii, if you going to fix streaminfo2.cpp anyway, please, look at audiotype output code - it's broken, or full-feature audio-attributes list not supported on SPARK hardware?
It's always outputs "MPEG stereo ()", regardless of selected audio track.

Code: Alles auswählen

//AUDIOTYPE
	int type, layer, freq, mode, lbitrate;
	audioDecoder->getAudioInfo(type, layer, freq, lbitrate, mode);

	snprintf (buf, sizeof(buf), "%s:", g_Locale->getText (LOCALE_STREAMINFO_AUDIOTYPE));
	g_Font[font_info]->RenderString (xpos, ypos, box_width, buf, COL_INFOBAR_TEXT);

	if(type == 0) {
		const char *mpegmodes[4] = { "stereo", "joint_st", "dual_ch", "single_ch" };
		snprintf (buf, sizeof(buf), "MPEG %s (%d)", mpegmodes[mode], freq);
	} else {
		const char *ddmodes[8] = { "CH1/CH2", "C", "L/R", "L/C/R", "L/R/S", "L/C/R/S", "L/R/SL/SR", "L/C/R/SL/SR" };
		snprintf (buf, sizeof(buf), "DD %s (%d)", ddmodes[mode], freq);
	}

I tried to debug it myself, but stuck at source/libstb-hal/libspark/audio.cpp - due to lack of knowledge in c++ and SPARK hardware.

Code: Alles auswählen

void cAudio::getAudioInfo(int &type, int &layer, int &freq, int &bitrate, int &mode)
{
	lt_debug("%s\n", __FUNCTION__);
	type = 0;
	layer = 0;
	freq = 0;
	bitrate = 0;
	mode = 0;
.......
}

As far i understand, getAudioInfo() not working as expected, so someone just "zeroed" its output, and this is why in Stream Info we always get "MPEG stereo ()".

[martii]

Hi Benny,

getAudioInfo() is just a dummy method (and the parts that are commented out are (IMHO) from the TripleDragon variant which I assume *does* work). Player2 doesn't provide most the data that method should return, so seife was absolutely right in just zeroing the values.

I'm currently short on time. I may have a closer look at the other issues later this week.

Cheers,

martii

[Benny]

I'm currently short on time. I may have a closer look at the other issues later this week.

No problem.
I just wanted verify that
a) you know about this issue,
b) maybe, "AudioType" should be removed from StreamInfo, until (if ever) fix is ready?

Post update: attached patches for minor changes, mentioned above.

[martii]

Hi Benny,

I've applied two of your patches but skipped the streaminfo2.cpp one (incrementing the offset inside the loop looks wrong to me). Plus, streaminfo2 now shows the stock audio channel description instead of the previous MPEG... default. (I see no way to get any sensible information from the hardware, and probing the stream with ffmpeg introduces a delay that's plainly unacceptable).

Cheers,

martii

[Benny]

I've applied two of your patches

Thank you.

Plus, streaminfo2 now shows the stock audio channel description instead of the previous MPEG... default.

Thanks a lot, this is better then nothing, and much better then previous misinform output.

Of course, getting Dbox2-like audio stream info would be more useful, but i personally will not ask for impossible =)

[Benny]

martii, great work on extended audio info - thanks.

After i tried it myself, i understand, what do you mean under "a delay that's plainly unacceptable" - i personally don't mind, but someone less patient may throw RC into the wall or TV =)
Can "audio PIDs extended info" be painted in separate place, ie not in main paint loop - in this case, delay not affect paint time for bitrate graph/average bitrate info, and overall visually noticeable paint delay should be less annoying?

Sorry, if you didn't finished with StreamInfo, and work still in progress, but anyway, some comments on current implementation:
1) In MP+WebTV mode long URL looks really ugly. Should it be cropped/truncated?
2) In MP+file mode "URL" value is blank. Can we use filename path there, something like "URL: file://some_file.avi" or "URL: file://mnt/nfs/some_file.avi" ?
I tried, but can't invent something smarter then

Code: Alles auswählen

	if (mp) {
		// url
		r.key = g_Locale->getText (LOCALE_STREAMINFO_URL);
		r.key += ": ";
		r.val =channel->getUrl();
		if (strlen(r.val.c_str()) == 0) {   // if value is empty - hide URL label. Should we add path/filename here instead?
		r.key = "";
		}
		if (strlen(r.val.c_str()) > 40) {  // if string too long, crop it. Should it be cropped at render time via "width" option?
		r.val = r.val.substr(0,37);
		r.val += "...";
		}
		r.col = COL_INFOBAR_TEXT;
		v.push_back(r);
		scaling = 8000;
	} else {

If someone prefer old-style StreamInfo, patch also provided, see attached files.

[martii]

Hi Benny,

stream probing is now in parallel to determining the bitrate. The other issues should be resolved, too.

Cheers,

martii

[Benny]

stream probing is now in parallel to determining the bitrate. The other issues should be resolved, too.

Great, thank you.