Merge pull request #2962 from JocelynDelalande/i18n-readme-improvement

I18n README.md improvement

notebook/i18n/README.md

@@ -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,
@@ -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
@@ -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 (emmo@us.ibm.com)
+    ./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 (emmo@us.ibm.com)