Poesie

Poesie is the POEditor String Internationalization Extractor.

("poésie" also happens to be the French word for "poetry")

This repository contains a script to generate iOS' and Android's localized strings files extracted from a poeditor.com project:

Using poeditor.com, you will typically enter all your terms to translate, then all the translations for each language for those terms, using the web interface of the tool.

This tool can then automate the extraction / export of those translations to generate Localizable.strings and Localizable.stringsdict files for iOS, and strings.xml files for Android.

Even though POEditor's web interface allows you to export the strings in those format already, the script has the following advantages:

• It automates the process, which can be very welcome if you update your localized strings quite often (instead of having to download the file for each language and each platform, then move the downloaded files from your Downloads folder to the correct location in your project, etc)
• It post-processes the exported files, including:
  • Sorting the terms alphabetically
  • Filtering out terms ending with _ios when exporting for Android, and filtering out terms ending with _android when exporting for iOS
  • Normalize the string placeholder so you can use %s everywhere in your POEditor strings in the web interface — the script will automatically replace %s and %n$s with %@ and %n$@ when exporting to iOS, so no need to have different translations for those for each platform
  • Allowing you to provide text substitutions, like replacing "..." with "…" and similar for all your translations (useful when your translator isn't always consistent with those typographical rules for example)
• Allows you to extract the context strings in a JSON file if you want to use them for whatever use case in your app

The easiest solution to install the latest version of poesie is to simply run:

gem install poesie

Alternatively, the following solutions will allow you to install the latest version from master, so that you can use the newest features even if they haven't been released to RubyGems yet:

• git clone the project on you computer
• If you don't have it already, install bundler using gem install bundler
• Install poesie's dependencies by running bundle install from the directory where you cloned the repository
• Invoke the tool using its full path <path/where/you/cloned/Poesie>/bin/poesie.

You could also add the <path/where/you/cloned/Poesie>/bin/ path to your PATH environment variable if you prefer.

• git clone the project on you computer
• Run gem build poesie.gemspec to build the gem
• Run gem install poesie-*.gem to install the gem you just built (where * is the version of the gem)
• Now that it's installed in your system, you can invoke the tool using poesie from anywhere

This solution has the drawback of being potentially easily confused between versions of poesie that you installed yourself vs. official versions though.

• If you don't have a project in POEditor, start by creating one.
• Then add some terms to translate. You can also indicate for each term if it will have a plural form or not (using the "P" button in POEditor web interface)
• Then add at least a language, and translate your terms for that language in the POEditor web interface

Of course, only list in POEditor the strings that needs a translation (user-facing strings), not JSON keys or constants, etc!

💡 Tip: When adding a string to translate, be sure it doesn't already exist and hasn't already been added by another teammate (maybe using a different key) for example, by searching the text to translate in the POEditor web interface. This is especially worth checking because POEditor is typically limiting the maximum number of terms by a quota depending on your plan, so better not duplicate those terms!

• Find a name for your term that is consistent with existing keys
• As a convention, for Niji projects, we structure the name of terms in a reverse-dns hierarchical text, using _ as a separator. For example home_banner_text and home_weather_temperature.
• For Android, if you have . in term names, they will be replaced with _ (as R.string.foo.bar won't work in Android but R.string.foo_bar will)
• If a key should only be exported for Android, use the _android suffix. If a key should only be exported for iOS, use the _ios suffix.

• For keys that use string placeholders, use %s and not %@. Android doesn't know about %@ (which is an iOS/macOS-only placeholder) and %s placeholders will be translated to %@ automatically by this script
• Use positional placeholders like %n$s, instead of just %s, whenever possible.

Note

The %n$s syntax allows you to indicate the index of the parameter to use. This allows you to invert the order of the parameters in the translation, which can sometimes be needed for some languages (e.g. In English you'll use %1$s's %2$d phones to have "John's 3 phones", but in French you'll use Les %2$d téléphones de %1$s to have "Les 3 téléphones de John", where the number of phones comes before the person's name).

If you don't specify the n$ position number (but only e.g. %s), the parameters will be consumed in order. It's recommended to use the n$ positional placeholders though, even if they are in order, so that the order is explict and not implicit.

There's an oddity in POEditor when handling plurals: for a term to be marked to support plurals, you must activate the "P" round button next to the term… which will then display a field to enter the name of the term… when used for the plural variants.

I haven't seen any point of having a different term for the pluralized term and the singular term. Using a different name for the term and the plural term doesn't make sense to me.

Therefore, and given how the script interprets those terms, for terms with plurals, you should always use the same name for the "term" and the "plural term" when declaring them in the POEditor web interface:

• You can add newlines, either using actual newlines or using \n, in the translations in the web interface. Both (literal or escaped) should be interpreted and translated correctly by the script

Multiple options can be used when invoking the tool from the commandline:

$ poesie -h
Usage: poesie [options]
    -t, --token API_TOKEN            Your POEditor API token
    -p, --project PROJECT_ID         Your POEditor project identifier
    -l, --lang LANGUAGE              The POEditor project language code to extract
    -i, --ios PATH                   Path of the iOS Localizable.strings[dict] file to generate
    -a, --android PATH               Path of the Android strings.xml file path to generate
    -c, --context PATH               Path of the *.json file to generate for contexts
    -d, --date                       Generate the current date in file headers
    -s, --subst FILE                 Path to a YAML file listing all text substitutions
    -h, --help                       Show this message
    -v, --version                    Show version

• You'll typically always need to provide a token (--token) and a project ID (--project). Those can be found here on the POEditor web interface

• You'll also always need to provide the language (--lang) you wish to extract and for which you want to generate the files. You can invoke the poesie script multiple times, one for each language you need to extract, if needed.

• Depending if you want to generate the localization files for Android (strings.xml) or iOS (Localizable.strings & Localizable.stringsdict), you'll use either --ios PATH or --android PATH. (Note: for iOS, you give the path and name of the Localizable.strings file to generate, and the script will deduce itself the path for the Localizable.stringsdict to generate next to it)

Exemples :

$ poesie -t abcd1234efab5678abcd1234efab5678 -p 12345 -l fr -a /Users/me/Documents/Dev/MyApp/app/src/main/res/values/strings.xml

$ poesie -t abcd1234efab5678abcd1234efab5678 -p 12345 -l fr -i /Users/me/Documents/Dev/MyApp/Resources/Localizable.strings

When running the poesie script using the --context FILE option, it will generate a JSON file at the provided path, containing all the terms for which you provided a "Context" (the "C" round button) in the POEditor web interface.

$ poesie --token "..." --project "..." --lang fr --ios .../localizable.strings --context .../context.json

This can be useful:

• Either to use that JSON file directly in your project to do whatever you want with the contexts (e.g. parsing the JSON file at runtime using JSONSerialization, and use it as you please)
• Or use that JSON file with a template engine (like Liquid) to generate code specific to your needs. See the example script in examples/gen-context-with-liquid.rb.

In case you need substitutions to be applied to your translations, you can use the --subst flag (-s for short) to provide a YAML file listing all substitutions to be applied.

$ poesie --token "..." --project "..." --lang fr --ios .../localizable.strings -s substitutions.yaml

This can be useful:

• To replace standard spaces with non-breaking spaces before punctuation characters
• To replace "..." with "…" and similar
• To replace smiley strings like ";-)" with actual emoji characters like 😉
• To ensure proper capitalization and orthography of your brand name every time it's spelled in the translations
• etc.

The YAML file provided must be of the form of a single Hash of String pairs, or an Array listing Hashes of String pairs.

• Substitutions will be performed in the given order if listed in an Array.
• Order of substitutions isn't guaranteed if they are listed in a Hash. That's why it's sometimes preferable to use an Array of key/value pairs rather than a Hash (because an Array is ordered)
• If a key is surrounded with slashes, it will be interpreted as a regular expression.

Example:

- " :": "\u00A0:"
  " ;": "\u00A0;"
  " !": "\u00A0!"
  " ?": "\u00A0?"
- "...": "…"
- /^\s+/: ""
  /\s+$/: ""

In this example:

• The first substitutions to be applied will be the 4 first ones about the :;!? punctuation. The order of the substitutions between those 4 is undetermined (as Hashes are unordered).
• Then the substitution of ... to … will be applied
• The last two substitutions are interpreted as Regular Expressions, trimming the beginning and end of each text. They will be applied last (but those two will be applied in any order w/r/t each other).

[
  {
    " :": " :",
    " ;": " ;",
    " !": " !",
    " ?": " ?"
  },
  {
    "...": "…"
  },
  {
    "/^\\s+/": "",
    "/\\s+$/": ""
  }
]

You can find an example of a YAML file in examples/substitutions.yaml in this repository.