Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

I18n README.md improvement #2962

Merged
merged 2 commits into from
Oct 23, 2017
Merged

I18n README.md improvement #2962

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
7181806
Make the i18n README more clear
JocelynDelalande Oct 21, 2017
bd48b42
Use local paths for i18n-related commands
JocelynDelalande Oct 21, 2017
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
121 changes: 64 additions & 57 deletions notebook/i18n/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,88 +1,88 @@
# Implementation Notes for Internationalization of Jupyter Notebook

This is a prototype implementation of i18n features for Jupyter notebook, and should not
yet be considered ready for production use. I have tried to focus on the public user
interfaces in the notebook for the first cut, while leaving much of the console messages
behind, as their usefulness in a translated environment is questionable at best.
The implementation of i18n features for jupyter notebook is still a work-in-progress:

### Using a prototype translated version
- User interface strings are (mostly) handled
- Console messages are not handled (their usefulness in a translated environment is questionable)
- Tooling has to be refined

In order to use this preliminary version, you need to do things after installing the
notebook as normal:
However…

1. Set the LANG environment variable in your shell to "xx_XX" or just "xx".
where "xx" is the language code you're wanting to run in. If you're
running on Windows, I've found the easiest way to do this is to use Windows PowerShell,
and run the command:
## How the language is selected ?

`${Env:LANG} = "xx_XX"`
1. `jupyter notebook` command reads the `LANG` environment variable at startup,
(`xx_XX` or just `xx` form, where `xx` is the language code you're wanting to
run in).

2. Set the preferred language for web pages in your browser to YourLanguage (xx). At the moment,
it has to be first in the list.
Hint: if running Windows, you can set it in PowerShell with `${Env:LANG} = "xx_XX"`.

3. Run the `jupyter notebook` command to start the notebook.
2. The preferred language for web pages in your browser settings (`xx`) is
also used. At the moment, it has to be first in the list.

### Message extraction:
## Contributing and managing translations

I have split out the translatable material for the notebook into 3 POT, as follows:
### Requirements

notebook/i18n/notebook.pot - Console and startup messages, basically anything that is
produced by Python code.

notebook/i18n/nbui.pot - User interface strings, as extracted from the Jinja2 templates
in notebook/templates/*.html

noteook/i18n/nbjs.pot - JavaScript strings and dialogs, which contain much of the visible
user interface for Jupyter notebook.

To extract the messages from the source code whenever new material is added, use the
`pybabel` command to extract messages from the source code as follows:
( assuming you are in the base directory for Jupyter notebook )
- *pybabel* (could be installed `pip install pybabel`)
- *po2json* (could be installed with `npm install -g po2json`)

`pybabel extract -F notebook/i18n/babel_notebook.cfg -o notebook/i18n/notebook.pot --no-wrap --project Jupyter .`
`pybabel extract -F notebook/i18n/babel_nbui.cfg -o notebook/i18n/nbui.pot --no-wrap --project Jupyter .`
`pybabel extract -F notebook/i18n/babel_nbjs.cfg -o notebook/i18n/nbjs.pot --no-wrap --project Jupyter .`
**All i18n-related commands are done from the related directory :**

(Note: there is a '.' at the end of these commands, and it has to be there...)
cd notebook/i18n/

After this is complete you have 3 POT files that you can give to a translator for your favorite language.
Babel's documentation has instructions on how to integrate this into your setup.py so that eventually
we can just do:
### Message extraction

`setup.py extract_messages`
The translatable material for notebook is split into 3 `.pot` files, as follows:

I hope to get this working at some point in the near future.
- *notebook/i18n/notebook.pot* - Console and startup messages, basically anything that is
produced by Python code.
- *notebook/i18n/nbui.pot* - User interface strings, as extracted from the Jinja2 templates
in *notebook/templates/\*.html*
- *noteook/i18n/nbjs.pot* - JavaScript strings and dialogs, which contain much of the visible
user interface for Jupyter notebook.

### Post translation procedures
To extract the messages from the source code whenever new material is added, use the
`pybabel` command:

After the source material has been translated, you should have 3 PO files with the same base names
as the POT files above. Put them in `notebook/i18n/${LANG}/LC_MESSAGES`, where ${LANG} is the language
code for your desired language ( i.e. German = "de", Japanese = "ja", etc. ). The first 2 files then
need to be converted from PO to MO format for use at runtime. There are many different ways to do
this, but pybabel has an option to do this as follows:
```shell
pybabel extract -F babel_notebook.cfg -o notebook.pot --no-wrap --project Jupyter .
pybabel extract -F babel_nbui.cfg -o nbui.pot --no-wrap --project Jupyter .
pybabel extract -F babel_nbjs.cfg -o nbjs.pot --no-wrap --project Jupyter .
```

`pybabel compile -D notebook -f -l ${LANG} -i notebook/i18n/${LANG}/LC_MESSAGES/notebook.po -o notebook/i18n/${LANG}/notebook.mo`
After this is complete you have 3 `.pot` files that you can give to a translator for your favorite language.

`pybabel compile -D nbui -f -l ${LANG} -i notebook/i18n/${LANG}/LC_MESSAGES/nbui.po -o notebook/i18n/${LANG}/nbui.mo`
### Messages compilation

The nbjs.po needs to be converted to JSON for use within the JavaScript code. I'm using po2json for this, as follows:
After the source material has been translated, you should have 3 `.po` files with the same base names
as the `.pot` files above. Put them in `notebook/i18n/${LANG}/LC_MESSAGES`, where `${LANG}` is the language
code for your desired language ( i.e. German = "de", Japanese = "ja", etc. ).

`po2json -p -F -f jed1.x -d nbjs notebook/i18n/${LANG}/LC_MESSAGES/nbjs.po notebook/i18n/${LANG}/LC_MESSAGES/nbjs.json`
*notebook.po* and *nbui.po* need to be converted from `.po` to `.mo` format for
use at runtime.

```shell
pybabel compile -D notebook -f -l ${LANG} -i ${LANG}/LC_MESSAGES/notebook.po -o ${LANG}/notebook.mo
pybabel compile -D nbui -f -l ${LANG} -i ${LANG}/LC_MESSAGES/nbui.po -o ${LANG}/nbui.mo
```

The conversions from PO to MO probably can and should be done during setup.py.
*nbjs.po* needs to be converted to JSON for use within the JavaScript code, with *po2json*, as follows:

When new languages get added, their language codes should be added to notebook/i18n/nbjs.json
under the "supported_languages" element.
po2json -p -F -f jed1.x -d nbjs ${LANG}/LC_MESSAGES/nbjs.po ${LANG}/LC_MESSAGES/nbjs.json

When new languages get added, their language codes should be added to *notebook/i18n/nbjs.json*
under the `supported_languages` element.

### Tips for Jupyter developers

The biggest "mistake" I found while doing i18n enablement was the habit of constructing UI messages
from English "piece parts". For example, code like:

```javascript
var msg = "Enter a new " + type + "name:"
```

`var msg = "Enter a new " + type + "name:"`

where "type" is either "file", "directory", or "notebook"....
where `type` is either "file", "directory", or "notebook"....

is problematic when doing translations, because the surrounding text may need to vary
depending on the inserted word. In this case, you need to switch it and use complete phrases,
Expand All @@ -104,9 +104,9 @@ create the plural form doesn't translate well. Some languages have as many as 5
plural forms for differing numbers, so using an API such as ngettext() is necessary in order
to handle these cases properly.

### Known issues
### Known issues and future evolutions

1. Right now there are two different places where the desired language is set. At startup time, the Jupyter console's messages pay attention to the setting of the ${LANG} environment variable
1. Right now there are two different places where the desired language is set. At startup time, the Jupyter console's messages pay attention to the setting of the `${LANG}` environment variable
as set in the shell at startup time. Unfortunately, this is also the time where the Jinja2
environment is set up, which means that the template stuff will always come from this setting.
We really want to be paying attention to the browser's settings for the stuff that happens in the
Expand All @@ -117,6 +117,13 @@ of languages in the UI ( never a good thing ).
2. We will need to decide if console messages should be translatable, and enable them if desired.
3. The keyboard shorcut editor was implemented after the i18n work was completed, so that portion
does not have translation support at this time.
4. Babel's documentation has instructions on how to integrate messages extraction
into your *setup.py* so that eventually we can just do:

Any questions or comments please let me know @JCEmmons on github ([email protected])
./setup.py extract_messages

I hope to get this working at some point in the near future.
5. The conversions from `.po` to `.mo` probably can and should be done using `setup.py install`.


Any questions or comments please let me know @JCEmmons on github ([email protected])