History

Jim Derry 53f2a2da2a msgunfmt works properly with escaped hex.		2016-01-30 15:51:53 +08:00
..
translations	msgunfmt works properly with escaped hex.	2016-01-30 15:51:53 +08:00
.gitignore	Localization Support added to HTML Tidy	2016-01-30 15:51:53 +08:00
Gemfile	Localization Support added to HTML Tidy	2016-01-30 15:51:53 +08:00
language_ll_cc.h.erb	Localization Support added to HTML Tidy	2016-01-30 15:51:53 +08:00
poconvert.rb	msgunfmt works properly with escaped hex.	2016-01-30 15:51:53 +08:00
README.md	Localization Support added to HTML Tidy	2016-01-30 15:51:53 +08:00

README.md

README.md - Localize HTML Tidy

Thank you for your interest in helping us localize HTML Tidy and LibTidy. Users throughout the world will thank you.

The document describes Tidy's localization philosophy and instructs you on how you can use standard gettext tools to generate language and region localizations that will work with Tidy. Optionally instructions are included in the event that you want to build Tidy with your new language.

Introduction

PO and POT files

HTML Tidy uses PO and POT files for language translations. The file tidy.pot is the correct template to use as a basis for translations. In a typical gettext workflow a translator will use the tidy.pot file to create a language translation PO file that contains original English strings and the translated strings.

If a language has already been translated (or if the translation has begun) then PO files may already exist. These files are named language_ll.po or langage_ll_CC.po, where ll represents the language code, and optionally, CC represents the region code of the translation.

Tidy does not use MO files that gettext tools generate from PO files.

H files

Tidy does not use gettext to display strings and so gettext-generated MO files are not necessary. Instead translated PO files must be converted to Tidy's language header H file format. Translators are not required to perform this step, but we provide a tool to perform this function if desired.

Differences for translators

Experienced users and translators of PO files may note that we use the PO file's msgctxt field a bit uniquely. Rather than point to a line in the source code it contains a reference to the string's identifier. Because the PO format does not allow for arbitrary metadata this is a requirement for generating our header files.

If you're the type of translator the does dig into the source code, then this msgtext symbol is still useful to you and adds a single extra step to finding where a string is in context: a symbol or string search using the msgctxt value will reveal the context in source code.

Finally the msgid field is a throwaway; Tidy's language tools do not use this value and so it's only for the translator's convenience. This fact makes it convenient for translators to translate from languages other than English, which is fully supported by our tools.

`poconvert.rb` versus `gettext`' tools

Please don't use gettext' tools with our PO and POT files (unless you are using our strings for a different project). Instead all workflows can be accomplished with our poconvert.rb tool.

More information about this tool can be found below.

How to Contribute

Find or Create the Translation Files

If you've not already cloned the HTML Tidy source code repository that will be your first step.

In the localize\translations directory you can find existing languages, e.g.,

tidy.pot (Tidy's POT template for translations).
language_en_gb.po (British English variants for the built in language)
…and perhaps more.

If the language that you want to work on is already present:

Simply open the file in your favorite PO editor and then get to work.
Note that although you can use a text editor, we recommend that you use a dedicated PO editor so that you don't accidentally make the file illegible to our conversion utility.

If the language that you want to work on is not already present:

You can open tidy.pot in your favorite PO editor and use its functions to begin a new translation into your desired language.
Note that although you can use a text editor, we recommend that you use a dedicated PO editor so that you don't accidentally make the file illegible to our conversion utility.
To perform the work manually:
- Copy tidy.pot to language_ll.po (for a non-regional variant, or base language), or to language_ll_cc.po (for a region-specific variant), where ll indicates the two letter language code and cc indicates the two letter region or country code.
- Change the pertinent PO header section accordingly.
Use poconvert.rb to generate a PO:
- poconvert.rb msginit --locale ll, where ll indicates the language code for the language you want to translate to. The tool recognizes the same languages as gettext' msginit. If your chosen language is not supported, then please see the manual method, above.
- See also poconvert.rb help for more options.
See GNU's The Format of PO Files for more specific instructions and important information.

Issue a Pull Request to HTML Tidy

Once your translation is complete commit your entire HTML Tidy repository to GitHub and issue a pull request (PR) against the master branch. If accepted a friendly developer will convert your PO into a format useful to Tidy.

You are also welcome to perform the conversion yourself, add the language to Tidy, and issue a PR for the whole change.

Using Git appropriately

Fork the repository to your GitHub account.
Optionally create a topical branch - a branch whose name is succinct but explains what you're doing, such as "localize Portuguese".
Make your changes, committing at logical breaks.
Push your work to your personal account.
Create a pull request.
Watch for comments or acceptance.

Repository Notes

Please only commit PO files with English msgid fields. The gettext convention specifies only English msgid, and other translators may not understand the original strings.

Our poconvert.rb script can generate PO files using another language as msgid. This can be very useful if it's easier for you to translate from another language instead of English. It can also be useful for translating from a base language to a regional variant, such as from Spanish to Mexican Spanish.

If you choose to work locally with a non-English PO, you can easily convert your PO to a Tidy header file and back to an English-based PO using our poconvert.rb script. See its documentation (poconvert.rb help) for instructions.

Adding Languages to Tidy

Although we don't require you to follow these steps to contribute a language to Tidy, you may want to add the language to Tidy yourself to test the translation, or to save one of the developer team a few extra steps.

Generate the header files:
- Convert your PO file to a Tidy header file by executing poconvert.rb msgfmt <path_to_your_file.po>. Note that on Windows you will likely have to preface this line with ruby.
- The tool should generate a file named language_ll_cc.h in the working directory, where ll_cc will be replaced with the language/region of your translation.
- Copy this .h file into src\.
Modify Tidy's source:
- Edit the file src\language.c to ensure that the new .h file you added is in the #include section.
- Look for the static tidyLanguagesType tidyLanguages structure starting near line 40, and look for the comment These languages are installed.. You can add your new language to the list along with the other languages present, following the same format.
Build Tidy:
- Build Tidy per the usual instructions, and try it out using the -lang option.

Best Practices

Language Inheritance

HTML Tidy will fallback from the specified language to the base language and then finally to the default English as required. This means, for example, a programmer might set libtidy to use “es_mx”, and if it doesn’t exist Tidy will automatically use “es”. If that doesn’t exist libtidy will continue to use whatever language it is currently using.

String Inheritance

HTML Tidy will also fallback for individual strings. For example if libtidy is set to use “es_mx” and a particular string is requested and not found, the library will look for the string in “es”. If the string is not found there then the “en” string will be given.

Base Language First and Regional Variants

Because of this inheritance we hope to localize base languages first, as the only strings then required for regional variants are the strings that differ. This will help us keep HTML Tidy and libtidy small.

If you are working on a regional variation (such as “us_CA”) please only localize strings that are actually different from the base language!

Testing

We hope to develop a comprehensive test suite in the future, but in the meantime you can test localized output like this.

Command line option

Use the -lang/-language option and specify a POSIX or Windows language name. This option should be first option used because the console application parses and acts on options first-in, first-out.

Changing your locale

On Unix/Mac and Linux operating systems you can change your shell’s locale temporarily with:

export LANG=en_GB export LC_ALL=en_GB

…substituting, of course the language of your choice.

gettext

Although HTML Tidy uses gettext-compatible tools and PO files for language localization, Tidy itself does not use gettext. Tidy's build philosophy is build it anywhere and build it with anything. As gettext is not universally available on every platform under the sun, Tidy cannot count on gettext.

Instead Tidy builds all translations into its library (and command line executable if built monolithically), and can run on virtually any general purpose computer with any operating system.

While this does not pose a significant problem for storage or execution space on modern PC's, we understand that certain applications may still be space critical. As such it's trivial to build Tidy without this extra language support using the -DSUPPORT_LOCALIZATIONS=NO switch.

poconvert.rb

Tidy's source code includes a Ruby batch file called poconvert.rb which can be used to generate POT, PO, and H files, and convert them back and forth. It has been designed to work in a similar fashion as gettext's tools, and includes conveniences that let translators work in different source languages. Please use poconvert.rb help for complete information (ruby poconvert.rb help on Windows).

Note that you must install Ruby on your system, as well as install the required dependencies. These can be manually installed with [sudo] gem install xxx, where xxx represents the packages listed in Gemfile. For convenience, if you have the Bundler gem installed, you can bundle install for automated dependency installation.

Also take note of these two important characteristics:

poconvert.rb is currently dependent on its current path. You can move it from its current location, but you will have to change the values of the @@default_en and @@header_template variables within the script.
All files will be output in the current working directory. This will prevent accidental overwrites of important files while we all get used to the workflows.

Below are some sample workflows.

Create a new POT file

Although we provide tidy.pot in the source, you can generate your own.

./poconvert.rb xgettext

This will put a fresh, new copy of tidy.pot in the working directory.

Create a new POT file with non-English `msgid` strings

Although gettext officially recognizes English as the one, true source language for PO and POT files, if you're more comfortable translating from a non-English language, we can support you.

./poconvert.rb xgettext <language_cc_ll.h>

Where <language_cc_ll.h> is the path to an existing Tidy language header file. This will produce a tidy.pot using the translated strings as msgid, using English as a backup when translated strings are not present.

This can be valuable in producing regional variant translations, e.g., when translating from es to es_mx.

Convert an existing H to PO

In many cases you may want to have a fresh PO generated from a Tidy H file. This can be accomplished with:

./poconvert.rb msgunfmt <language_cc_ll.h>

Convert an existing H to PO using a different `msgid` language

If you want to generate a fresh PO file from a Tidy H file, but also want to have untranslated strings from a language other than English, try:

./poconvert.rb msgunfmt <language_cc_ll.h> --baselang=<other-language_cc_ll.h>

Create a blank PO file for a particular region

./poconvert.rb msginit or ./poconvert.rb msginit --locale=LOCALE

The first example will try to guess your current region, and the second will use a region specified.

Tidy only knows about the same regions that gettext knows; if our msginit does not recognize the region you specify, you will have to create a new PO and modify the region settings yourself.

To create the blank PO using msgid strings from a different Tidy language, you can use:

./poconvert.rb msginit <language_cc_ll.h> [--locale=LOCALE]

Create a Tidy Language Header H file

When you're ready to include the language in Tidy, you can generate its header file with:

./poconvert.rb msgfmt <language_cc_ll.po>

In the event you are creating a regional variant of a language, it's an excellent idea to have Tidy exclude strings that are already present in the parent language in order to reduce library and executable size. For example if es already includes the string "archivo" there is no reason for your translation to es_mx to include it, too. You can tell poconvert.rb to exclude strings matching another localization like so:

./poconvert.rb msgfmt <language_cc_ll.po> --baselang=<other-language_cc_ll.h>

Prepare your non-English PO for a PR

Although we have provided tools to allow you to work in languages other than English, we can only accept PO's in the repository that have English msgid fields. It's easy to convert your PO back to English:

./poconvert msgfmt <language_cc_ll.po>

./poconvert msgunfmt <language_cc_ll.h>

The first command converts your non-standard PO into a Tidy Language Header file, and the second will create a fresh, new PO file from the header that you've just created.

Update your PO to match the new POT

If Tidy's POT changes, e.g., new strings are added, new comments, etc., the simplest way to update your PO is to convert it to a header (which normalizes it to the latest Tidy standard), and then convert the header to a new PO again.