Get the Untranslated strings, I have the app installed. bench --site mysite.localhost get-untranslated es-CO ./es-CO-untrans.csvAlso get-untranslated did not work and show an error until I applied this PR changes to bench
Find the files ./es-CO-untrans.csv in /sites, make a copy of it named es-CO.csv.
Edit es-CO.csv changing each line I want to translate from english to spanish. as:
File before:
my number
my other number
File after:
mi numero
mi otro numero
Update the translations bench update-translations es-CO es-CO-untrans.csv es-CO.csv
Add my language to /apps/language.txt althought this seems uneccesary seems I’m not bootstrapping a new language.
Check if the translation were created in my custom app, apps/custom_app/custom_app/translations/es-CO.csv.
Check the content of the added translation.
b'DocType: Package',b'my number',b'mi numero'
b'DocType: Package',b'my other number',b'mi otro numero'
Now I restart the bench, clear chache, set site to not cache at all, change my language, change it back, migrate the bench, update, etc. Nothing works, I still get the strings in english.
And It just doesn’t work. The only way I can get translations to work is by Using the GUI on desk#List/Translation/List. Is also so not user friendly that bench get-untranslated seems to be broken for a year an someone submitted a PR and issue, and both are ignored.
Environment
erpnext 12.10.1
frappe 12.8.1
# This is my application in which I'm testing
package_management 0.0.1
So the question remains, how does one create translations for a custom app?
Hello! Been trying to get attention into this issue. Seems like a very crucial part of the framework. Decided to write my app in English to then translate and ended up with tons of problems in this aspect. I also commented in [Release Notes] bench v5.2.0 - #5 by Javier_Lopez, but still no answer! Might be helpful If you comment as well
It’s really worrying when core developers ignore PRs for obvious problems.
I have only recently begun a customization and a new module. Like you, I thought it would make sense to do it all in English . It never occurred to me that I18N and L11N would simply not work.
Is it possible you have missed something, for your current version perhaps?
About missing steps, I followed everything I could find and tried anything I could think of and I didn’t work. And haven’t been able to get answers, asked around on the Not Official Telegram Channel and no one commented on it. Also red all I could find in the forum but nothing seems to help.
But I lean on in being a bench problem, since you can’t even get the untranslated strings, that just tells me there’s probably no test for it, and not many people using it I guess.
Well they seem to have something bootstrapped for ERPNext languages see: https://translate.erpnext.com/ I don’t know if it uses bench. That’s not what you do in a custom app. I hope I’m wrong an I’m just doing something dumb. So I can fix it and help documenting the issue.
Looking to follow up on the subject, the above mentioned issue has been closed and a PR has been merged. Fixing the csv generation to translate, but sadly translation is still not possible.
Now that the steps of generating the untranslated strings work bench --site mysite.localhost get-untranslated es-CO ./es-CO-untrans.csv when you make the neccesay translations and go to update there’s an error:
So update-translations es-CO es-CO-untrans.csv es-CO.csv doesn’t work, here’s the new bug report:
I tested in my instance and It worked, finally got some translations in my app, I’m excited, I hope the PR I submitted is good enough, and translations might work!!
I believe the main issue in all this is that the code is over 5 years old, and using some python modules like csv that have had api changes when moving to python 3 and some things have not been migrated properly.
Something I noticed now that translations are working, is that this should not be the output of bench update-translations file, It should show without the b' which are python way of representing encoded strings.
b'DocType: Package',b'my number',b'mi numero'
b'DocType: Package',b'my other number',b'mi otro numero'
@MartinHBramwell Just letting you know in case you’re still interested!
I’m happy to say the PR was merged and translations are working well for me! The problem was that update-translations would generate incorrect translations when it worked, as can be seen in step 7 of the original post. The correct output is similar but without b' and all strings properly encoded.
This should still be broken. I bet (sadly), that you can not add new strings to your app, and run the same (official) cycle of bench get-untanslated... → translating → bench update-translations again.
I would be highly interested, if this works for you. For me it doesn’t.
The first time it works, and after that it will just override, whatever was already translated.
@Javier_Lopez
I do have a dirty, and hacky codefix, which I did not yet bother polishing for a mergerequest, since I am still not sure, if I do something wrong. In case you want to try, and can reproduce my Problem, feel free to experiment: Commits · alias/frappe · GitHub (Bunch of new patches since Oct. 27 where I merged your fixes)
Hi I am just wondering how do you mange those untranslated strings outside of your custom app? When I do get-untranslated I get >4k untranslated strings and I have not capacity to translate all of them but only for my custom app. How do I get ride of these untranslated strings?
