Just an academic interest: what this checks are so needed for? What is actually checked? The only problem I face is that sometimes I forget to insert a trailing '\n' in the end of multiline message where original msgid has it. Any other issues you know? -- Regards, Minton. -- To unsubscribe, e-mail: opensuse-translation+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-translation+owner@opensuse.org

Am Freitag, 6. April 2012, 11:03:30 schrieb Alexander Melentyev:

Just an academic interest: what this checks are so needed for? What is actually checked? The only problem I face is that sometimes I forget to insert a trailing '\n' in the end of multiline message where original msgid has it. Any other issues you know? msgfmt is able to perform quite a few checks, not all of them (if any) are checked serverside on commit.

msgfmt --check-format --check-header --check-domain --check-accelerators Regards, Michael

On Fri, Apr 6, 2012 at 3:01 PM, Carlos E. R. <robin.listas@telefonica.net> wrote:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

On 2012-04-06 09:03, Alexander Melentyev wrote:

...

Just an academic interest: what this checks are so needed for? What is actually checked? The only problem I face is that sometimes I forget to insert a trailing '\n' in the end of multiline message where original msgid has it. Any other issues you know?

There is one that might cause crashes: a %s more or less. If the source has an accelerator, the translated message must have it, too. Most of the checks are for looks of the text. Wrong xml tags can make the text be unreadable. kbabel also checks for "English text", which would warn you of forgotten strings. Another would be matching quotes, I think.

There are others that could be added, but the software don't support it: sometimes the messages has newlines, to make it fit into a box, but we have to guess what is that box size. This could be solved if the file added a token like

msgsize "------------------------"

and then the checker would verify that too. But this "msgsize" has not been invented yet, don't worry :-)

- -- Cheers / Saludos,

You can try Pology http://techbase.kde.org/Localization/Tools/Pology It has a lot of checks and has integration with Lokalize, so you can run Pology and make open the wrong files in Lokalize showing only the faulty strings. Bye -- To unsubscribe, e-mail: opensuse-translation+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-translation+owner@opensuse.org

On Tue, Apr 10, 2012 at 12:17 PM, Carlos E. R. <robin.listas@telefonica.net> wrote:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

On 2012-04-10 12:05, Leandro Regueiro wrote:

...

You can try Pology http://techbase.kde.org/Localization/Tools/Pology

It has a lot of checks and has integration with Lokalize, so you can run Pology and make open the wrong files in Lokalize showing only the faulty strings.

Pology User Manual pology.nedohodnik.net/doc/user/en_US/ Pology is a Python library and collection of scripts for in-depth processing of PO files. The library is designed for easy and robust writing of custom scripts in field ...

So, CLI utils? No point and click? No...

It opens Lokalize for pointing and clicking. -- To unsubscribe, e-mail: opensuse-translation+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-translation+owner@opensuse.org

On Tue, Apr 10, 2012 at 12:57 PM, Carlos E. R. <robin.listas@telefonica.net> wrote:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

On 2012-04-10 12:33, Leandro Regueiro wrote:

...

It opens Lokalize for pointing and clicking.

If you have a link to the howto documenting this, perhaps.

Since you want to check rules, you must be interested in this particular sieve: http://pology.nedohodnik.net//doc/user/en_US/ch-sieve.html#sv-check-rules The check-rules sieve allows to open the files in Lokalize if you pass the lokalize option to the posieve command. I also recommend you to read the introductory text to that particular chapter at the top of that webpage (sections 3.1 to 3.4, both included). I will put a link to our translation group wiki with information in galician about how to use Pology to run the checks, but right now the site seems to be not working. I will try again later. -- To unsubscribe, e-mail: opensuse-translation+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-translation+owner@opensuse.org

On Tue, Apr 10, 2012 at 2:12 PM, Carlos E. R. <robin.listas@telefonica.net> wrote:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

On 2012-04-10 13:31, Leandro Regueiro wrote:

...

On Tue, Apr 10, 2012 at 12:57 PM, Carlos E. R. <> wrote:

...

...

If you have a link to the howto documenting this, perhaps.

Since you want to check rules, you must be interested in this particular sieve: http://pology.nedohodnik.net//doc/user/en_US/ch-sieve.html#sv-check-rules The check-rules sieve allows to open the files in Lokalize if you pass the lokalize option to the posieve command. I also recommend you to read the introductory text to that particular chapter at the top of that webpage (sections 3.1 to 3.4, both included).

That link is a man page - useless. I need the exact command line to give and exactly how to integrate with lokalize. A howto document, not a list of options.

http://wiki.trasno.net/Uso_de_Pology_en_GNOME

Also, it appears that what it does is generate a file of errors that you can open in lokalize - how do I put that automatically back on the full file?

No. It opens the files with errors in Lokalize, but showing only the strings with errors. If you edit the strings with errors and save the file, you are editing the whole file and saving the whole file, but you only saw the strings with errors.

This is far more complex than clicking on the file in kbabel. There I correct the error and the message goes automatically from red to black: error solved. I only have to click save and done.

Here you run posieve with the appropiate options. Then you edit the affected strings and you save.

...

I will put a link to our translation group wiki with information in galician about how to use Pology to run the checks, but right now the site seems to be not working. I will try again later.

If you do it in English it would be of general use.

I already put the link to the galician docs upside, but it is simple. You download Pology, and since it is in a svn server you just have to run the appropiate command (available in the above page). Once you have Pology somewhere in your computer, you just have to open a command line and run the scripts that are in the bin subdirectory. Pology has several scripts, but here we are just going to focus in posieve. posieve is used to apply sieves to the .po files. Among the several available sieves the ones that may be more interesting are check-rules (for checking a bunch of user defined rules) (Pology comes with a lot of these rules and you can contribute more) and check-spell-ec (for checking the spell using enchant). posieve accepts several options, and each sieve has its particular options too. For specifying each particular sieve's options you have to preceed them with -s, which is the posieve option to specify options for the sieves. For example if you want to specify the lang:gl option for the check-rules sieve you have to write -slang:gl in the command line. For example for checking for errors in user defined rules and open the wrong strings in Lokalize you can run: posieve -b --msgfmt-check --no-skip check-rules -slang:gl -s"rulerx:^PT*" -saccel:_ -snorulerx:dual -slokalize core/nautilus/po/gl.po posieve is the command, -b is for skiping obsolete strings, --msgfmt-check is for checking file validity with msgfmt -c, --no-skip is for not skipping problematic strings, check-rules is the name of the sieve used -slang:gl is for applying the rules for galician (just change the language code) -s"rulerx:^PT*" is for specifying just the rules that start with "PT" (we are applying a regular expression to match the rule's names) (we have several classes of rules and we group them by the name) -saccel:_ tells the sieve to strip the _ accelerators from each string before checking the rules (you can use other accelerators such as &) -snorulerx:dual tells the sieve to not use the rules with names that match this regular expression -slokalize is for opening the strings with errors in lokalize core/nautilus/po/gl.po is a file, so just put here the path to another file (you can pass here several files, which is useful to check hundreds of files in a row) You have a complete list of options for this particular sieve in http://pology.nedohodnik.net//doc/user/en_US/ch-sieve.html#sv-check-rules The list of options for posieve is available in http://pology.nedohodnik.net//doc/user/en_US/ch-sieve.html#sec-svoptions Bye -- To unsubscribe, e-mail: opensuse-translation+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-translation+owner@opensuse.org

Am Mittwoch, 11. April 2012, 13:54:03 schrieb Carlos E. R.:

I need a guide for dummies, I'm not going to study hard when I can just click in kbabel and get it done. Hey Carlos,

I'm looking at the Kbabel source (at least that's what I think it is ;D). As it looks to me, Kbabel DID NOT use msgfmt to compare the syntax of original/translation, but rather implemented an own algorithm to do so (basicly counting the number of &s in each). Do you know what any other checks Kbabel performes? I don't have it installed anymore, but I think it also checked whether the amount of "\n"'s is the same, right? (although msgfmt actually just cares about whether they both end with an "\n", since some languages could use an additional extra line or two for better readability). Maybe I can implement something like that in Lokalize, or at least point the right people at the right things. Regards, Michael

On Thu, Apr 12, 2012 at 5:31 PM, Carlos E. R. <robin.listas@telefonica.net> wrote:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

On 2012-04-12 10:00, Michael Skiba wrote:

...

Am Mittwoch, 11. April 2012, 13:54:03 schrieb Carlos E. R.:

...

I need a guide for dummies, I'm not going to study hard when I can just click in kbabel and get it done. Hey Carlos,

I'm looking at the Kbabel source (at least that's what I think it is ;D). As it looks to me, Kbabel DID NOT use msgfmt to compare the syntax of original/translation, but rather implemented an own algorithm to do so (basicly counting the number of &s in each).

I thought it did its own checking, yes. But it also calls msgfmt. You are checking one of the tests only :-)

It also calls msgfmt when trying to save the file - which is also a problem if you don't know how to solve the error and want to email somebody else to have a look at it. I know it calls msgfmt because when there is an error it tells you so and the dialog has the capture of the command output.

...

Do you know what any other checks Kbabel performes?

I made a photo, I'll transcribe the test names.

 Check syntax    --> msgfmt --statistics  Translations containing English  Check accelerators  Whitespace translations  Check plural forms  Check translated message length  Check translated messages with a set of regular expressions  Check equations  Check tags  Look for translated context info  Check arguments  Check punctuation

When doing "All tests", one errors with "Error loading data (file not found)". I don't remember which test it refers to, I'll call them one by one to find out [...] Ok, it is the "Check translated messages with a set of regular expressions" test.

...

I don't have it installed anymore, but I think it also checked whether the amount of "\n"'s is the same, right? (although msgfmt actually just cares about whether they both end with an "\n", since some languages could use an additional extra line or two for better readability).

Yes, you can add as many as you need, but it errors if the last one is different.

...

Maybe I can implement something like that in Lokalize, or at least point the right people at the right things.

That would be fantastic.

Instead of implement this in Lokalize maybe will be better to integrate it with other tools like Pology or Translate Toolkit pofilter tool: http://translate.sourceforge.net/wiki/toolkit/pofilter_tests Bye -- To unsubscribe, e-mail: opensuse-translation+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-translation+owner@opensuse.org