2135 replies to this topic

[Tech]

Merged.

[andy1]

Here is the latest norwegian bokmal with updated strings.

Hope you can add it.

Attached Files

•  Norwegian bokmal.zip   123.47KB   4 downloads

[Tech]

Pushed.

[Pr2]

Hi,

 

Another french translation, add missing string + some small improvements.

 

Can you please also discuss internally of the opportunity to use this alternative Makefile.am in the po folder?

I just added (all other OE-Alliance images do this):

 

line 18,19:

sed --in-place enigma2-py.pot --expression=s/CHARSET/UTF-8/

sed --in-place enigma2-xml.pot --expression=s/CHARSET/UTF-8/

 

diff --git a/po/Makefile.am b/po/Makefile.am
index 29fd4aa..151f55e 100644
--- a/po/Makefile.am
+++ b/po/Makefile.am
@@ -15,6 +15,8 @@ enigma2-xml.pot: $(srcdir)/xml2po.py $(top_srcdir)/data/*.xml 
        $(PYTHON) $^ > $@
 
 enigma2.pot: enigma2-py.pot enigma2-xml.pot
+       sed --in-place enigma2-py.pot --expression=s/CHARSET/UTF-8/
+       sed --in-place enigma2-xml.pot --expression=s/CHARSET/UTF-8/
        cat $^ | $(MSGUNIQ) --no-wrap --no-location -o $@ -
 
 %.po: enigma2.pot

 

Thanks,

 

Pr2

Attached Files

•  OpenPLi-2015-08-18-fr.po.zip   59.82KB   2 downloads
•  Makefile.am.zip   925bytes   1 downloads

[andy1]

Pushed.

Thank you 

[Tech]

.............................Can you please also discuss internally of the opportunity to use this alternative Makefile.am in the po folder?

Thanks,

 

Pr2

 

I pushed your changes, and started an internal discussion about your question.

[mirakels]

I don't think the sed commands are needed.

First of all the enigma2-xml.pot file is just a collection of messages found and is intended to be appended to the enigma2-py.pot. Hence it does not contain a CHARSET line.

 

And the sed on enigma2-py.pot is not needed as the pot file is used to generate .po files using msgmerge or msginit. And msginit already 'renames' CHARSET to UTF-8 nowadays.

 

I verified on the enigma2 po files and the all have "Content-Type: text/plain; charset=UTF-8\n"

 

So I think this patch on Makefie.am is not needed.

[blzr]

there's no charset defined anywhere in enigma2-xml.pot, so this line:

sed --in-place enigma2-xml.pot --expression=s/CHARSET/UTF-8/

is superfluous...

(other than that, I'm all for this change, btw. ready-to-commit patch on this waits for acceptance/push for quite a few months now, lost somewhere in the forum depths )

edit:
@mirakels
but when you generate manually pot file this charset error is a little annoying (and you know how often po files are regenerated in the repo lately... )

Edited by blzr, 20 August 2015 - 12:55.

[Pr2]

Hi,

 

I agree for the enigma2-xml.pot this is indeed useless.

 

@Mirakels,

 

Yes this patch is needed.

 

in the openpli local copy do a:

 

./configure --with-po

 

Then go in the po folder and type make.

 

On the current openpli there are 2 bugs that prevent the correct automatic .po creation:

 

Bug 1:

See here:

 

http://forums.openpl...eation-process/

 

The file ./data/skin_display_et8500.xml give errors with xml2po, of course if you don't check it you launch again the make and it seems to work because the enigma2-xml.pot already exist.

 

Bug 2:

You will see that during the merge the CHARSET gives and errors, so the solution is to manually edit the enigma2-py.pot file and change the CHARSET to UTF-8 then start again a make.

 

The OpenPLi work around for the nightly build is simply to skip the .pot files creation if you don't use --with-po option. So only the .mo creation occurs.

 

Of course that all generated .po files in the repository have UTF-8 inside since this is changed by hand to be able to generate them. 

 

Would be nice if everybody is able to create the .po file simply by using:

./configure --with-po 
cd po 
make

That's the goal of the proposed changes.

 

And with this changes since the .pot and po file creation will runs smoothly you can suppress the "--with-po" trick and also generate all the .po files during nightly build and evey translators can work more smoothly just by using the .po in the trunk, today most of them are outdated and many strings are missing in them.

 

Regards,

 

Pr2

Edited by Pr2, 20 August 2015 - 13:40.

[mirakels]

Hm, strange msgmerge dos not look at the chrset from the .po file it is updating...

Also strange you can give xgettext a input charset but it does not use that to set the 'output charset'

 

Added the sed to makefile.am!

 

Thanks again.

[Taapat]

And if we are talking about automatic translation, I would like to draw attention that in the enigma is some things that will not be parsed automatically.
In this commit is introduced the words "Question" and "Warning": https://github.com/O...857b182c47af43a
In this commit "Select location": https://github.com/O...d0aee9c75e509bf
For myself I have solved this with adding in the enigma root file translate.py which contains these words: https://bitbucket.or...te.py?at=master
Then they is in my pot file when automatically parsing.
I do not know what is better, do as I, or prevent problems with translation in these screens.

Edited by Taapat, 20 August 2015 - 16:03.

[Taapat]

And of course now when update all po/language files mirakels due to incomplete pot file delete these words from my Latvian language translation.

[blzr]

@mirakels

seems that something went not-so-well with po files' update...
I see some definitely obsolete strings, that re-appeared after the update, eg:

msgid "Movie List Configuration"
msgid "Load Length of Movies in Movielist"
msgid "Remember these settings for each folder"
etc. etc...

in general my home brewed regenerated po file is much smaller /?/

are you sure you updated po's against up-to-date repo?

I would suggest to revert...

[Taapat]

Yes updated files contain many phrases which can not be found in enigma.
It would be better revert this commit.

[mirakels]

that is eeird. eull check out later....

[Pr2]

Hi,

 

I perform a brand new git clone and test the changes, thanks to the xml2po.py and Makefile.am now the generations of the .po runs smoothly.

So these are the result created on my local git clone:

 

Doing a grep returns those .po file containing it:

grep "Load Length of Movies in Movielist" *.po
bg.po:#~ msgid "Load Length of Movies in Movielist"
cs.po:#~ msgid "Load Length of Movies in Movielist"
da.po:#~ msgid "Load Length of Movies in Movielist"
de.po:#~ msgid "Load Length of Movies in Movielist"
el.po:#~ msgid "Load Length of Movies in Movielist"
fi.po:#~ msgid "Load Length of Movies in Movielist"
fr.po:#~ msgid "Load Length of Movies in Movielist"
hu.po:#~ msgid "Load Length of Movies in Movielist"
nl.po:#~ msgid "Load Length of Movies in Movielist"
no.po:#~ msgid "Load Length of Movies in Movielist"
pt_BR.po:#~ msgid "Load Length of Movies in Movielist"
pt.po:#~ msgid "Load Length of Movies in Movielist"
sl.po:#~ msgid "Load Length of Movies in Movielist"
sv.po:#~ msgid "Load Length of Movies in Movielist"

So we have 14 entries with this string in.

 

While in the folder we have this:

ls *.po
ar.po  da.po  es.po  fr.po  hu.po  lt.po  no.po     ro.po  sr.po  uk.po
bg.po  de.po  et.po  fy.po  is.po  lv.po  pl.po     ru.po  sv.po
ca.po  el.po  fa.po  he.po  it.po  nb.po  pt_BR.po  sk.po  th.po
cs.po  en.po  fi.po  hr.po  ku.po  nl.po  pt.po     sl.po  tr.po

So we have 37 entries.

 

Be we saw that the string are commented out!

 

#~ msgid "Load Length of Movies in Movielist"

 

So they are no longer used to generate the resulting .mo file, there is probably an option to mention to delete those entries and not keep them when generating the .po files. Because today once a string is already translated even if it no longer exist it is kept commented out forever in the .po file.

 

Pr2

Edited by Pr2, 20 August 2015 - 19:15.

[Pr2]

For the Taapat problem can we fix it this way?

 

diff --git a/lib/python/Screens/MessageBox.py b/lib/python/Screens/MessageBox.py
index 278f6a8..d3d3970 100644
--- a/lib/python/Screens/MessageBox.py
+++ b/lib/python/Screens/MessageBox.py
@@ -43,7 +43,7 @@ class MessageBox(Screen):
  self["QuestionPixmap"].hide()
  if picon != self.TYPE_INFO:
  self["InfoPixmap"].hide()
- self.title = self.type < self.TYPE_MESSAGE and ["Question", "Information", "Warning", "Error"][self.type] or "Message"
+ self.title = self.type < self.TYPE_MESSAGE and [_("Question"), _("Information"), _("Warning"), _("Error")][self.type] or _("Message")
  if type == self.TYPE_YESNO:
  if list:
  self.list = list

[Taapat]

In this case, you also must remove the translation in self.setTitle (_ (self.title)): https://github.com/O...ssageBox.py#L82

[mirakels]

Arg, I had a old leftover version of MovieSelection.py in the 'search tree'.

I'll generate a new one.

 

But about the #~ messages. These are obsolete translations. Sometimes you want to keep these when src strings are slightly changed. The old message + translation will then be saved as a #~ entry. The translator could then use this old obsolete translation for the new one, avoindong lots of retyping.

 

It is easy to get rid of the obsolete messages if everybody agrees. It could mean a lot of extra work for the translators.

So let me know...

[Pr2]

Hi,

 

This is how I work as a translator, I have my local copy that I "git pull" and then start the process of .po creation, then I copy the fr.po (in my case) on my PC and I use PoEdit to translate the missing string and correct the one turned into "to be reviewed" or "fuzzy".

 

http://poedit.net/

 

So from my side I really don't care of the old translation, I think that PoEdit have its own internal database where it kept the old translation that I perform. It shows me a matching indication in % so I can reuse them when needed and correct them.

 

And since I try to follow quite closely the translation, to avoid to have many hundreds strings to translate, it is not a real problem if due to a change I need to translate again an old string.

 

So for me those obsolete translations can be removed from the generated .po files without problem.

 

Perhaps a suggestion (I don't know if it is foreseen or even feasible during the make command) but since the .po files in the OpenPLi repository are not part of the nightly build, the already translated .po files remains quite static in it.

So if other translator use the method as I do, generate there own .po files locally with make, we can imagine to create a clean .po file and a  .po.obsolete with the obsolete strings in.

 

So when a change is done, the translator can have a look into the .po.obsolete file locally to check if he can re-use a translation. Then he gives its translated .po file that will be committed into the trunk. So every .po file in the trunk will always be "clean".

 

Pr2