Translate a YAML file into Chinese (Simplified)

What changes when a YAML file goes into Chinese

Simplified Chinese has no spaces between words and uses full-width punctuation: ，。：！？. For YAML this is mostly good news: only the ASCII colon followed by a space is syntax, so the Chinese colon ： breaks nothing. A value such as notice: 保存成功：已更新 3 项 stays valid.

The text is two to three times shorter in characters, but each character is wider, so the space it takes on screen changes little. Any limit set in characters, however (a maxlength, truncation at 40 characters), will let through far more meaning than in English, or cut a two-character word in half.

Chinese has no grammatical plural: its only CLDR category is other. Instead, a number is followed by a measure word: %{count} 条消息 (messages), %{count} 个文件 (files), %{count} 张图片 (images). The variable often moves: “You have %{count} new messages” becomes 您有 %{count} 条新消息.

Finally, a Latin font has no Chinese characters. If your stylesheet only declares a Latin web font, the browser will fall back to a system font that may not match; declare a Chinese fallback font.

Rails, Symfony, Hugo: root key, plurals and block scalars

In a Rails or i18n-js file, everything sits under a locale key such as en:. Manuce handles this root key; just check that the resulting code is the one your project expects (zh-CN, zh or zh-Hans, depending on your conventions) and name the file to match, for instance messages.zh_CN.yaml in Symfony, or after the language code declared in your Hugo configuration.

Plurals are adapted to the target language: a one: / other: pair becomes a single other form. In Rails, make sure your app knows the Chinese plural rule (the rails-i18n gem provides it, for example); without it, the default backend looks for a one key when count is 1.

Anchors (&defaults, *defaults) and comments are preserved. Watch folded blocks >: YAML turns each line break into a space. A Chinese sentence spread over two lines will therefore render with a stray space between two characters, something nobody notices in English but every Chinese reader sees. The literal block | keeps line breaks exactly as written.

What Manuce returns, and what it leaves alone

You get back the same file: indentation, key order, quotes, comments, line endings, encoding and BOM preserved. Only the values change: title: "Save" becomes title: "保存", quotes included.

Every translated value is compared with the source: the %{name} and {name} variables and any tags must all be there. If a Chinese value drops a variable and the mismatch persists, the source value is kept and the report says so; a visible source-language string is better than broken interpolation.

Not translated: keys, empty values, URLs, email addresses, numbers, colours, identifiers, date formats and anything marked as non-translatable. In practice, a format such as long: "%B %d, %Y" stays as it is. The usual Chinese order, %Y年%m月%d日, is yours to enter under date.formats.

Manuce does not take glossaries or style instructions. The register (formal 您 or familiar 你) and how a product name is handled are therefore best judged on the free sample. Traditional Chinese (Taiwan, Hong Kong) is a separate language and is ordered separately. The quote counts the words in values only, keys and tags excluded, with the ×1.6 coefficient that applies to Chinese.

Preparing the file and checking the result

• Encoding: it is kept as it is, so start from a UTF-8 file. A YAML file still in Latin-1 cannot hold Chinese characters.
• Validity: load the source file in your project or a YAML validator before uploading; a syntax error would come back unchanged in the Chinese version.
• Split sentences: if one sentence is built from two keys (prefix then suffix), Chinese word order will not follow. A single key with a variable reads far more naturally.
• After delivery: read the report, look for values left in the source language, test a screen with count at 0, 1 and 5, and check folded > blocks for stray spaces.
• Display: confirm your fallback font renders the hanzi and that no button splits a word in two.

In practice

1. Make sure your source YAML is UTF-8 and loads without errors.
2. Upload it: the quote counts words in values, keys excluded.
3. Read the free sample: register, product names, measure words.
4. Download the Chinese file and its string report.
5. Name the file with the expected locale code, adjust date.formats and test.