Successfully reported this slideshow.
Your SlideShare is downloading. ×

Easy contributable internationalization process with Sphinx @ pyconmy2015

Ad
Ad
Ad
Ad
Ad
Ad
Ad
Ad
Ad
Ad

Check these out next

1 of 62 Ad

Easy contributable internationalization process with Sphinx @ pyconmy2015

Download to read offline

Sphinx can extract paragraphs from sphinx document and store them into gettext format translation catalog files.
Gettext format translation catalog is easy to translate from one language to other languages.
Also Sphinx support internationalization by using such catalog files.
You can use your favorite editors or services to translate your sphinx docs.
In this session, I'll explain 3 things; (1) entire process to translate sphinx docs. (2) automation mechanism for the process. (3) tips, tricks and traps for writing docs and translating.

Sphinx can extract paragraphs from sphinx document and store them into gettext format translation catalog files.
Gettext format translation catalog is easy to translate from one language to other languages.
Also Sphinx support internationalization by using such catalog files.
You can use your favorite editors or services to translate your sphinx docs.
In this session, I'll explain 3 things; (1) entire process to translate sphinx docs. (2) automation mechanism for the process. (3) tips, tricks and traps for writing docs and translating.

Advertisement
Advertisement

More Related Content

Slideshows for you (20)

Viewers also liked (20)

Advertisement

Similar to Easy contributable internationalization process with Sphinx @ pyconmy2015 (20)

More from Takayuki Shimizukawa (20)

Advertisement

Recently uploaded (20)

Easy contributable internationalization process with Sphinx @ pyconmy2015

  1. 1. Takayuki Shimizukawa Sphinx co-maintainer Sphinx-users.jp 1
  2. 2. こんにちは Kon-Nichi-Wa 2
  3. 3. Who am I @shimizukawa 1. Sphinx co-maintainer 2. Sphinx-users.jp 3. PyCon JP committee  BePROUD co, ltd. 3
  4. 4. http://pycon.jp/ 4
  5. 5. Agenda 1. Sphinx introduction & Setup for i18n 2. A easy way and Non easy ways to translate 3. Sphinx i18n feature 4. Automated translation process with several services 5. Tips, tricks, traps 5
  6. 6. 6
  7. 7. Question1 How many people do you use Open Source Software? 7
  8. 8. Question2 How many people may have contributed to the OSS? 8
  9. 9. Question3 Does the translation into familiar language contribute to other developers? 9
  10. 10. 10 Translator Product Author Translated Docs Readers English Docs Translators
  11. 11. Tools & Services Tools 1. sphinx - documentation generator 2. docutils - base of sphinx 3. sphinx-intl - support utility for sphinx i18n 4. transifex-client - file transfer tool for Transifex Services 1. Transifex - translation support service 2. Drone.io - continuous integration service 11
  12. 12. 12
  13. 13. What is Sphinx? Sphinx is a documentation generator Sphinx generates doc as several output formats from the reST text markup 13 1. Sphinx introduction & Setup for i18n Sphinx reSTreSTreStructuredText (reST) reST Parser HTML Builder ePub Builder LaTeX Builder texlive HTML theme Favorite Editor
  14. 14. The history of Sphinx (short ver.) 14 1. Sphinx introduction & Setup for i18n The father of Sphinx Too hard to maintenance ~2007 Easy to write Easy to maintenance 2007~
  15. 15. Sphinx Before and After Before  There was no standard ways to write documents  Sometime, we need converting markups into other formats After  Generate multiple output format from single source  Integrated html themes make read docs easier  API references can be integrated into narrative docs  Automated doc building and hosting by ReadTheDocs 15 1. Sphinx introduction & Setup for i18n
  16. 16. Many docs are written by Sphinx For examples  Python libraries/tools: Python, Sphinx, Flask, Jinja2, Django, Pyramid, SQLAlchemy, Numpy, SciPy, scikit-learn, pandas, fabric, ansible, awscli, …  Non python libraries/tools: Chef, CakePHP(2.x), MathJax, Selenium, Varnish 16 1. Sphinx introduction & Setup for i18n
  17. 17. Many docs are written by Sphinx For examples  Python libraries/tools: Python, Sphinx, Flask, Jinja2, Django, Pyramid, SQLAlchemy, Numpy, SciPy, scikit-learn, pandas, fabric, ansible, awscli, …  Non python libraries/tools: Chef, CakePHP(2.x), MathJax, Selenium, Varnish 17 1. Sphinx introduction & Setup for i18n
  18. 18. SPHINX docutils HTML Builder HTML theme (Jinja2) gettext Builder *.pot *.po I18N *.mo OmegaT Pootle Transifex Translation Tools, Services Favorite Editor Sphinx i18n feature (built-in) 18 1. Sphinx introduction & Setup for i18n reST Parser (directive / role) doctree (Intermediate) reSTreST reStructuredText (reST)
  19. 19. $ pip install sphinx Support tools for translation How to install Sphinx with i18n tools To build a sphinx document $ pip install sphinx-intl $ pip install transifex-client=0.8 19 1. Sphinx introduction & Setup for i18n
  20. 20. $ git clone https://github.com/shimizukawa/deepthought.git $ cd deepthought/doc $ make html ... Build finished. The HTML pages are in _build/html. "make html" command generates html files into _build/html. make html once 20 1. Sphinx introduction & Setup for i18n
  21. 21. Files & setup conf.py $ tree /path/to/deepthought +- deep_thought | +- __init__.py | +- api.py | +- calc.py | +- utils.py +- doc | +- _build/ | | +- html/ | +- _static/ | +- _template/ | +- conf.py | +- index.py | +- make.bat | +- Makefile +- setup.py 21 1. Sphinx introduction & Setup for i18n Document source Document build output Target library for doc1. ... 2. 3. language = 'ma' 4. locale_dirs = ['locale'] 5. doc/conf.py
  22. 22. 22
  23. 23. What is internationalization? 23 2. Easy contributable internationalization process with Sphinx I18N
  24. 24. What is easy contributable? 24 2. Easy contributable internationalization process with Sphinx
  25. 25. Not a easy way (example 1/3) The manual are provided only in the HTML files Rewriting reST files 25 2. Easy contributable internationalization process with Sphinx
  26. 26. Not a easy way (example 2/3) The manual are generated from reST files and docstrings in the source files  Rewriting reST files  Rewriting docstrins in the source files. 26 2. Easy contributable internationalization process with Sphinx
  27. 27. Not a easy way (example 3/3) OK, we are engineers. We can use git!  Learn git  Learn GitHub  git clone to get the code  Translation (Yes! This is what I want to do!)  git commit  git pull  Sometimes resolve conflict  git push 27 2. Easy contributable internationalization process with Sphinx
  28. 28. Not easy vs easy Not a easy way Easy way 1. Learn git 2. Learn GitHub 3. git clone to get the code 4. Translation 5. git commit, pull, push 6. Resolve conflict 7. Build html your self 1. No git 2. No Github 3. No file 4. Translation 5. Update Automatically 6. No conflict 7. You can get a HTML output w/o hand-build. 28 2. Easy contributable internationalization process with Sphinx
  29. 29. 29
  30. 30. Two i18n features of Sphinx  Output pot files: from reST  Input po files: to generate translated HTML 30 3. Sphinx i18n feature reST pot reST html po
  31. 31. Translation flow  Generate pot  Translate pot into po  Generate Translated HTML 31 3. Sphinx i18n feature reST pot reST html po pot po Translator Please translate Translate! Thanks for your Contribution!
  32. 32. #: ../../../deep_thought/utils.py:docstring of deep_thought.utils.dumps:1 msgid "Serialize ``obj`` to a JSON formatted :class:`str`." msgstr "" msgid "For example:" msgstr "" 32 Output pot files 3. Sphinx i18n feature $ make gettext ... Build finished. The message catalogs are in _build/gettext. $ ls _build/gettext api.pot examples.pot generated.pot index.pot generated.pot reST pot
  33. 33. Preparing po files to translate doc +- _build/ | +- gettext/ | +- api.pot | +- examples.pot | +- generated.pot | +- index.pot +- locale/ doc +- _build/ +- locale/ +- zh_cn/ | +- LC_MESSAGES/ | +- api.po | +- examples.po | +- generated.po | +- index.po +- ja/ Translatethem Translator pot po 33 3. Sphinx i18n feature
  34. 34. 34 Preparing po files to translate $ sphinx-intl update -p _build/gettext -l zh_cn Create: locale/zh_cn/LC_MESSAGES/api.po Create: locale/zh_cn/LC_MESSAGES/examples.po Create: locale/zh_cn/LC_MESSAGES/generated.po Create: locale/zh_cn/LC_MESSAGES/index.po At first, sphinx-intl copy pot files and rename them pot po sphinx-intl $ make gettext $ sphinx-intl update -p _build/gettext -l zh_cn Not Changed: locale/zh_cn/LC_MESSAGES/api.po Updated: locale/zh_cn/LC_MESSAGES/examples.po +3, -1 After change the document, sphinx-intl update differences 3. Sphinx i18n feature
  35. 35. Translate po files #: ../../../deep_thought/utils.py:docstring of deep_thought.utils.dumps:1 msgid "Serialize ``obj`` to a JSON formatted :class:`str`." msgstr "" generated.po pot po sphinx-intl Translator #: ../../../deep_thought/utils.py:docstring of deep_thought.utils.dumps:1 msgid "Serialize ``obj`` to a JSON formatted :class:`str`." msgstr "序列 ``obj`` 以JSON格式 :class:`str` 。" generated.po Translate by using Vim, Emacs, OmegaT, ... 35 3. Sphinx i18n feature
  36. 36. Input po files 36 3. Sphinx i18n feature reST html po Translated $ make html ... Build finished. The HTML pages are in _build/html.
  37. 37. Big picture 37 3. Sphinx i18n feature reST pot html po make gettext sphinx-intl Translator make html Doc Author Translation Catalog Translated catalog
  38. 38. 38
  39. 39. Entire process to translate sphinx docs 39 4. Automated translation process with several services reST pot html po make gettext sphinx-intl make html Doc Author Translation Catalog Translated catalog Translator Translator Translator Autor / Translator upload Translator clone Translator
  40. 40. To Be 40 4. Automated translation process with several services reST pot html po make gettext sphinx-intl make html Doc Author Translation Catalog Translated catalog upload Translators To Be Automated To Be Parallel clone
  41. 41. Translation tool types  Vim / Emacs (Editors)  Edit local files  Translation support plugins are available.  OmegaT (Translation Tools)  Edit local files.  Translation support features.  Transifex (Services)  Edit online  Translation support features. 41 4. Automated translation process with several services po Translators To Be Parallel
  42. 42. Be Parallel by using Transifex Transifex provides... As API: Upload pot Download po As Web console: Glossary Translation memory Recommendation Auto-translation 42 4. Automated translation process with several services po Translators Parallel pot Upload pot Auto Update sphinx-intl transifex-client po transifex-client Download
  43. 43. Translation on Transifex web console 43 4. Automated translation process with several services Original Text Translated Text (you should keep reST syntax) Suggestions from Translation Memory (TM) Original (pot) Translation (po) Copy orig to translation Machine translation Save Review (if needed) Translators Parallel 1 2 4 3 5 6
  44. 44. To Be Automated 44 4. Automated translation process with several services po Translators Parallel pot Upload pot Auto Update sphinx-intl transifex-client po transifex-client Download reST html make gettext make html Doc Author upload To Be Automated clone
  45. 45. To Be Automated 45 4. Automated translation process with several services pot Upload sphinx-intl transifex-client po transifex-client Download reST html make gettext make html upload clone 1 2 3 45 6 To Be Automated
  46. 46. The procedure for automation 1. clone repository 2. make gettext 3. Upload pot 4. Download po 5. make html 6. Upload html 46 4. Automated translation process with several services pot Upload sphinx-intl transifex-client po transifex-client Download reST html make gettext make html upload clone 1 2 3 45 6 To Be Automated
  47. 47. 1. pip install sphinx sphinx-intl transifex-client==0.8 2. git clone https://github.com/shimizukawa/deepthought.git 3. cd deepthought/doc 4. sphinx-intl create-transifexrc # create ~/.transifexrc 5. sphinx-intl create-txconfig # create .tx/config 6. make gettext 7. sphinx-intl -p _build/gettext update-txconfig-resources # update .tx/config 8. tx push -s # push pot files to transifex 9. tx pull -l zh_cn # pull po files from transifex 10. make html SPHINXOPTS="-D language=ma" Shell commands for automation run.sh $ export SPHINXINTL_TRANSIFEX_USERNAME=mice $ export SPHINXINTL_TRANSIFEX_PASSWORD=42 $ export SPHINXINTL_TRANSIFEX_PROJECT_NAME=deepthought-0_7 $ export SPHINXINTL_POT_DIR=_build/gettext $ run.sh 47 4. Automated translation process with several services
  48. 48. The drone.io 48 WebHook Deploy Clone repository Run shell script The drone.io is a continuous integration service. 4. Automated translation process with several services
  49. 49. GitHub + drone.io + S3 49 GitHub Amazon S3 Transifex 1. Clone repository 2. make gettext. 3. Upload pot 4. Download po 5. make html 6. Upload html 4. Automated translation process with several services 2 1 3
  50. 50. Be Automated 50 pot Upload sphinx-intl transifex-client po transifex-client Download reST html make gettext make html upload clone 1 2 3 45 6 To Be Automated Upload pot Download po upload WebHook clone 1 5 6 make html make gettext2 3 4 1 WebHook Automated 4. Automated translation process with several services
  51. 51. Automated by drone.io 51 Upload potDownload po upload WebHook clone 1 5 6 make html make gettext2 3 4 1 WebHook Automated 4. Automated translation process with several services
  52. 52. View from doc author  Doc Author doesn't require annoying procedure. Update Translation Source Doc Author Notify See Commit 52 4. Automated translation process with several services
  53. 53. View from doc translators  No git  No file  No conflict  Update Automatically  They can get a translated HTML output w/o hand-build. Translators Parallel See Translate Translated Pages 53 4. Automated translation process with several services
  54. 54. The entire automated process 54 Update Translation Source Doc Author Translators Parallel Notify See See Publish Translate Commit Download Translations Notify Automated 4. Automated translation process with several services
  55. 55. Summary Tools 1. sphinx - documentation generator 2. docutils - base of sphinx 3. sphinx-intl - support utility for sphinx i18n 4. transifex-client - file transfer tool for Transifex Services 1. Transifex - translation support service 2. Drone.io - continuous integration service 55 4. Automated translation process with several services
  56. 56. 56
  57. 57. TIP: Drone.io limits 15mins for a build  Drone.io limits 15mins for a build.  Install from wheel package may help you 57 5. Tips, tricks, traps 1. curl -L -s https://example.com/wheelhouse.tgz | tar vzxf - 2. export PIP_FIND_LINKS=./wheelhouse 3. pip install sphinx sphinx-intl transifex-client==0.8 run.sh ex. https://bitbucket.org/sphinxjp/docutils-translation/
  58. 58. TRAP: Version of transifex-client  transifex-client 0.11b3 is not stable (for me)  Especially for Windows users  If you have encountered trouble with using newer version, please try: 58 5. Tips, tricks, traps $ pip install "transifex-client=0.8"
  59. 59.  Drone.io only list-up repositories which you have admin permission.  Prepare a empty repository to create a drone.io project. 59 TRICK: Preparing Drone.io project 5. Tips, tricks, traps
  60. 60. Examples  Sphinx-1.3 doc for "ja" translation https://drone.io/bitbucket.org/shimizukawa/sphinx-doc13/admin  Sphinx-1.4 doc for "ja" translation https://drone.io/bitbucket.org/shimizukawa/sphinx-doc14/admin  Docutils doc for "ja" translation https://drone.io/bitbucket.org/sphinxjp/docutils-translation/admin 60 5. Tips, tricks, traps
  61. 61. Questions? @shimizukawa Grab me after the session :) 61
  62. 62. Thanks :) 62

Editor's Notes

  • (ここの文字サイズ大きくする、マウスをレーザーポインタにする)
    Hi everyone. Thank you for coming my session.
    //Can you hear me? OK
    //Can you see the session title?
    This session title is: Easy contributable internationalization process with Sphinx.
  • こんにちは
    Second time in this conference.
  • At first, Let me introduce myself.

    My name is Takayuki Shimizukawa came from Japan.
    I do 3 opensource works.

    1. Sphinx co-maintainer since the end of 2011.
    2. organize Sphinx-users.jp users group in Japan.
    3. member of PyCon JP Committee.

    And I'm working for BePROUD.
    We develop web applications for business customers by using Django and Pyramid.
  • Before my main presentation, I'd like to introduce "PyCon JP 2015" in Tokyo Japan.
    We will held the event in this October.
    Registration is opened. Please join us.
    Anyway,
  • This is Agenda of this session.

    1. Sphinx introduction & Setup for i18n.
    2. A easy way and Non easy ways to translate 3. Sphinx i18n feature 4. Automated translation process with several services
    5. Tips, tricks, and traps
  • Before proceeding to the subject, I want to ask you.
  • How many people do you use Open Source Software?
    -> 10, 20, ... Thanks.
    A-> Obviously, most people are using OSS.
    B-> Oh, It's surprising. Half of the people don't use OSS.
  • How many people may have contributed to the OSS?
    -> 1, 2, ... Thanks,
    A-> Well,... it's a very small number.
    B-> Oh, a lot of people have contributed. It's very good news.

    IMO, using OSS is the first contribution.
    Also we can say that introducing the OSS to other people is another contribution.
    So, almost (a lot of / a half of) people who have raised your hand to the first question is already contributing to the OSS.
  • Many of the OSS provides the document in English.
    How do you think, Does the translation into familiar language contribute to other developers? Yes?

    10, 20,... Thanks.
    A-> most people think it's a contribution, I agree it completely.
    B-> Looks not too many. I guess everyone can read English, right?

    In Japan, many programmers aren't familiar to English.
    In such situations, translated documents helps me when I introduce some library to other developers.
    Because of this, I occasionally translate documents.
  • I think, by your translation of a OSS document, more people will be able to use the OSS.
    (クリック)
    If more than one person can join such translation work easily, it would be great.
    In this session, I'll introduce how to make a mechanism which is easy to join as a translator.
  • In order to make the mechanism, we'll use such tools and services in this session.
    Tools; sphinx, docutils, sphinx-intl, transifex-client
    Services; Transifex, Drone.io.

  • First subject, I'll introduce the sphinx introduction and setup Sphinx for i18n.
  • What is Sphinx?
    Sphinx is a documentation generator.
    Sphinx generates doc as several output formats from reStructuredText markup that is an extensible.
    (ポインタでinputとoutputを指す)
  • This man, Georg Brandl is the father of Sphinx.
    (クリック)
    Until 2007, python official document was written by LaTeX.
    But, it's too hard to maintenance.
    Georg was trying to change such situation.
    (クリック)
    So then, he created the Sphinx in 2007.
    The sphinx provides ease of use and maintainability for the Python official document.
  • Sphinx before and after.
    Before
    There was no standard ways to write documents. One of example is a Python official document. It was written by LaTeX and several some python scripts jungle.
    And, Sometime, we need converting markups into other formats

    Since sphinx has been released,
    * We can generate more multiple output format from single source.
    * Integrated html themes make read docs easier.
    * API references can be integrated into narrative docs.
    * Automated doc building and hosting by ReadTheDocs service.
  • Nowadays, sphinx has been used by these libraries and tools.
    Python libraries/tools: Python, Sphinx, Flask, Jinja2, Django, Pyramid, SQLAlchemy, Numpy, SciPy, scikit-learn, pandas, fabric, ansible, awscli, …

    And Non python library/tools also using Sphinx for them docs: Chef, CakePHP(2.x), MathJax, Selenium, Varnish
  • Nowadays, sphinx has been used by these libraries and tools.
    Python libraries/tools: Python, Sphinx, Flask, Jinja2, Django, Pyramid, SQLAlchemy, Numpy, SciPy, scikit-learn, pandas, fabric, ansible, awscli, …

    And Non python library/tools also using Sphinx for them docs: Chef, CakePHP(2.x), MathJax, Selenium, Varnish
  • Sphinx extend the i18n mechanism to provide translation feature.
    In this session, let's take a look the i18n feature.
  • How to install Sphinx with i18n tools.
    You need to install sphinx to build a sphinx document.
    Also you need to install sphinx-intl and transifex-client for translation.
    IMO, newer transifex is not stable, so I'm using older one.
  • At first, you should get source code whether from github or other place to translate it.
    In this case, the source code already has a sphinx doc directory.

    So, type "make html" in doc directory to generate html output.
    You can see the output in _build/html directory.
  • Now you can see the directories/files structure, like this.
    (ポインタ)
    * Target library for doc that is cloned from existent repository.
    * Document build output that was generated by "make html"
    * Document source that was generated by "sphinx-quickstart"
    (クリック)
    This is a conf.py file in the doc directory.
    You should add these 2 lines to the conf.py
    * Line-3rd: set a language code you want to use for output doc.
    * Line-4th: set a directory relative path from conf.py location. The locale directory will have translation catalog files.
    Now setup is over.
  • Well, it is the main subjects from here.

    Easy contributable internationalization process with Sphinx.
  • What is internationalization? It's usually called I18N.
    The internationalization process is sometimes called translation or localization enablement without rewriting the source code from a technical point of view.

    "without rewriting the source code from a technical point of view" is important.
  • What is easy contributable?
    Some software users or document readers have wanted to contribute by translation.
    But, if the preparation of the translation takes long time or requires many kind of skill, they are frustrated and give up.
    "Easy contributable" is the STATE that will be accepted the contributions immediately.
  • So, for the comparison, I'll introduce the three cases of difficult to contribute.

    First one.
    The manual are provided only in the HTML files.
    Of course, you can translate it.
    But, usually the html files are frequently updated.
    How do you follow the original changes to update your translations?

    This is the first case. I have no idea how to follow the original changes.
  • 2nd example.
    When the manual are generated from reST files and docstrings in the source files,
    You can rewrite original files to translate it.

    But this approach has three problems: No1. You must be careful to maintain reStructuredText structure. No2. It's hard to divide translation tasks for a number of volunteer translators. No3. It's hard to pursue the upstream document source that is frequently updated.
  • last example.
    OK, we are engineers. We can use git!
    (上を読む)
    Additionally, in this case, the repository owner should allow write access permission for each translators.

    This is also not a easy way.
  • Well then, what is a easy way?
    My opinion is this. Left side is Not a easy way.
    Right side is A easy way.
    We need a easy way.
    The combination of the Sphinx i18n feature and some of the services can realize such a mechanism.

    That's what is the goal of this session
    I'll explain what how to create the mechanism in this session.
    OK, let's move forward.
  • 3rd Subject. Sphinx i18n feature.
    I'll explain the detail of the feature.
  • Sphinx i18n feature is composed of two functions.

    First one is the generation pot files from reST files.
    pot file is a famous translation catalog format for gettext that is used for i18n, you know.

    Second one is the generation translated html files with using po files that is translated version of catalogs.
    It's the human work to prepare the translated po files from the pot files.
  • Like this, Converting the POT files into PO files is the job of humans.

    (クリック)In this flow, Author generate pot file at first part, and then, Author ask translators to translate them.
    (クリック)Translators receive the pot files from Author in some way, then translate them.
    (クリック)Once it has been finished, translators send back the po files to Author.

    Let's look a little more detail from first part.
  • Author can use "make gettext" command to generate pot files.

    "make gettext" command extract text from reST files and python sources that referenced by autodoc.
    It means, you can translate them into any language without rewriting the original reST files and python source files.

    As you see, pot translation catalog makes easy to translate from one language to another languages. "msgid" line have a original sentence.
    "msgstr" line will have translated one.
  • Translated PO files should have as a right side structure.
    "locale" directory has "zh_cn" and it has "LC_MESSAGES" directory, and it has translated PO files.
    PO files are copies of POT files with changed file's extension.
    If you want to translate it into other languages, like 'ja', you can create other directories under "locale" directory.

    When original POT files are updated by changing of original document, you should update PO files too.
    If you had done it manually, it is very hard.
    You can automated it by sphinx-intl.
  • At first, sphinx-intl copy pot files and rename them.
    When the document is changed, translators can use sphinx-intl to update PO differences.
    By same command.

    As you see, translators can finish the boring tasks by one command.
    sphinx-intl copy pot files and rename them on behalf of you.
    sphinx-intl also update differences for each po files when the document is changed.

    It's easy to use.
    As a consequence of introducing the sphinx-intl tool, translators can concentrate for translation.
  • This is a translation work.

    Translators can use their favorite editors or helpful tools/services easily that provide translation support features like a translation memory, recommending similar translation, glossary and auto-translation.
    Once the translation is complete, finally let's run the "make html" command.
  • Finally author and translators can run "make html" and can get translated output w/o editing reST and python code.
    This is a translated HTML sample.
    Internally, Sphinx parse reST file and replace them into translated one by using po file.
  • Entire process to translate sphinx docs is this.
    If you translate some sphinx document without using the i18n feature, you need to rewrite original document source files.
    The Sphinx i18n feature provides a easy way to translate a document as you seen.
  • So far, we've found the mechanism of i18n of Sphinx.
    In 4th Subject, Automated translation process with several services, we will continue to automate this mechanism.
  • Right now, we have seen an overall picture of the process.
    Actually, I have a thing that was not pictured here.
    (クリック)
    Yes, these commands such as "make blah blah" require your hands to invoke.
    In other words, the translator must learn such steps, yet.
    Additionally, it would be difficult to translate in parallel.

    It's not a easy way, you know.
  • OK, Let's make it to be automation and parallelization.

    And then, the translator doesn't need to know the mechanism.
    And, make the translation work possible to parallelize.

    Let's take a look at from the parallelization of the translation work.
  • Once you get po files, you can use helpful tools/services easily that provide translation support features like a translation memory, recommending similar translation, glossary and auto-translation.

    In particular, the use of the online translation service is important for parallelization.
    I'd like to introduce the Transifex as a online translation service.
  • Transifex provides the following functions: Upload pot and download po files by API, glossary feature, translation memory, recommending translations, automatic translation by using Google / Microsoft API.
    Because you can do translation on the Web screen, you can translate them with other translators in parallel.
  • Translation on Transifex web console.
    1. You can select one untranslated message
    2. And Translate it by your self, or by machine translation
    3. If the message contains long module or method name, you can click the icon to copy original to translation box w/o typing.
    4. Sometimes, you are suggested translation messages from other translator, or Translation Memory. Translation Memory (called TM) pick up suggestions from similar translation in the project or its historical changes. For example, When Doc Author fix a typo, translation will be discarded. However, TM suggests previous translation that matches 99% with current original message.
    5. Once you finish the translation, you can save it.
    6. Optionally, reviewer can review it.
  • We have made translation parallelization.
    Next.
    Let's automate the central part.
  • The procedure you want to automate is made up of 6 steps.
    At first, we will review the 6 steps.
  • First step is, Clone repository to be translated.
    Step 2. make gettext to generate translation catalog.
    Step 3. Upload pot translation catalog source.
    Step 4. Download translated po files.
    Step 5. "make html" command to generate translated doc.
    Step 6. Upload html to publish it

    That's all.
    Next, we will replace these all steps on the command line.
  • This is the command line that include the steps.

    Line1: install sphinx and relates tools
    Line2: clone repository
    Line4: create a transifex account file by using sphinx-intl that reference environment variables (USERNAME and PASSWORD).
    Line5: create a transifex config file by using sphinx-intl that also reference a environment variable (PROJECT_NAME)
    Line6: generate pot files by make gettext
    Line7: update resources information in transifex config file that also reference a environment variable (POT_DIR)
    Line8: push pot files to transifex
    Line9: pull translated po files from transifex
    Line10: make html to generate translated html with using translated catalogs.

    Now we replaced 5 steps on the command line, except HTML uploading part.
    I'll explain the part later.

    As you seen above, You can automate the process by this script.

    Rest of work is preparing a environment to run the script automatically.
    As one of the environment, I'll introduce the drone.io service.
  • The drone.io is a continuous integration service.
    Drone integrates seamlessly with Github, Bitbucket and Google Code to make setup fast and simple.
    Drone also integrates with Amazon, Heroku, Google AppEngine and more. For custom deployments you can use SSH and shell scripts.
    You can use the service for public project without charge.
  • In my case, I usually use Github and S3 integrations.
    Process of that is,
    (クリック)
    1. Push notification from GitHub
    2. That invoke the script you've seen
    3. Then, deploy html files onto S3 storage. (You need a publish html setting of S3)

    If you set WebHook notification URL to transifex hook setting, updating translations also invoke the drone.io script.
  • By using drone.io service, automation can be achieved.
    Upper-Left process is replaced with Right-Down one.
  • 1 WebHook invoke the script.
    The script execute commands 2, 3, 4, 5.
    Finally drone.io deployment feature will deploy HTML to AWS S3 storage.

    As this, we got the automated mechanism by using drone.io environment.
  • In summary, let's look at the results of the automation at each point of view as author and translators.

    At first, a point of view from the doc author.

    When author push to the repository, translation source of Transifex will be updated, and author can check the translated html in the site.
    Doc Author doesn't require annoying procedure.
  • The second is, a point of view from doc translators.
    They can translate the document in parallel with;
    No git, No file, No conflict.
    Translation source are updated Automatically
    They can get a translated HTML output w/o hand-build.

    It means that the translation contributors can focus on the translation.
  • By the mechanism of automation, both of doc author and translators will provide the maximum effect with a minimum of effort.
  • In order to make the mechanism, we'll use such tools and services in this session.
    Tools; sphinx, docutils, sphinx-intl, transifex-client
    Services; Transifex, Drone.io.

  • I'd like to introduce some of the tips.
  • Drone.io limits 15mins for a build.
    If your process takes more than 15 minutes, you may be able to reduce the installation time with using wheel.
    Wheel package is a installation format.
    Installation speed is much faster than install from source distribution.

    If you interested in it, please refer to https://bitbucket.org/sphinxjp/docutils-translation/
  • Version of transifex-client.

    Looks like transifex-client 0.11b3 is not stable, for me.
    Especially for Windows users.

    If you have encountered trouble with using newer version, please try:
    $ pip install "transifex-client=0.8"
  • This is a New project page of Drone.io.
    Drone.io only list-up repositories which you have admin permission.
    Because drone.io setup webhook, admin permission is required.

    However, if you prepare a empty repository for drone.io usage, you can create a drone.io project.
    It means you can prepare the i18n automation environment, except webhook push from github.
    I think, project owner might be setup the webhook url for translation work, if you ask it.

    It's bit a hacky way. But works.
  • These 3 examples are using earlier techniques.

×