Translate a PO file (gettext) into English

What changes when a gettext catalogue moves into English

Most projects already write their msgids in English. This page is for the other case: a catalogue whose source strings are in French, German, Spanish, Portuguese or any other language, and which now needs an English locale.

English has two CLDR plural categories, one and other. The usual header is Plural-Forms: nplurals=2; plural=(n != 1);. Be careful if your source is French or Brazilian Portuguese. Those languages put zero in the singular and use plural=(n > 1);. If that formula is copied across, your app will show “0 file” instead of “0 files”, and nothing will warn you.

Punctuation and casing need attention too:

• French spacing before colons and question marks disappears: Nom : becomes Name:;
• « » or „ “ quotation marks become English quotes;
• titles and buttons follow either sentence case (Account settings) or title case (Account Settings), depending on your UI conventions.

English text is usually 10 to 15% shorter than French, so truncation is less of a worry.

Then there is the variant. British (en_GB) uses colour, organise and cancelled, with dates written DD/MM. American (en_US) uses color, organize and canceled, with dates written MM/DD. Date format strings such as %d/%m/%Y are not translated, so switch them to %m/%d/%Y yourself if you target the US.

Context, comments and printf variables in English

One source word often maps to several English ones. Your msgctxt is preserved and used to choose the right one:

• French Commande under msgctxt "shop" becomes Order, but under msgctxt "terminal" it becomes Command;
• German Speichern unter becomes Save as in a menu context.

Extracted comments (#.), references (#:) and flags (#, such as c-format) stay in place. The context they give guides the translation. msgid_plural is kept as it is.

printf variables are checked. For example, %d fichiers supprimés becomes %d files deleted, with the same %d. English word order sometimes forces a swap. French Le %2$s de %1$s becomes %1$s's %2$s, and positional variables make that possible without breaking the call. Every translated value is compared with its source for variables and tags. If a mismatch persists, the source value is kept and the report says so.

What Manuce delivers, and what it does not

You get back the same file, ready to drop into your project. Indentation, entry order, quoting, comments, line endings, encoding and BOM are preserved; only the values change. A .pot template comes back as an English .po, with the Language and Plural-Forms headers set. The number of msgstr[n] forms matches English: two.

Left untranslated:

• keys (your msgid lines are unchanged);
• URLs and email addresses;
• numbers, colours, date formats and identifiers;
• anything marked as non-translatable.

The report lists translated strings and those left unchanged, with the reason for each. Pricing counts the words in values only, excluding keys and tags. English is charged at the base rate (×1 coefficient).

Manuce does not compile the .mo file. It does not convert your date formats between variants, and it does not take glossaries or terminology instructions. It also leaves your #, fuzzy flags in place, because comments are preserved.

Preparing the file and checking the result

Before uploading

• Make sure the header declares the encoding, for example Content-Type: text/plain; charset=UTF-8.
• Give ambiguous strings a msgctxt or a developer comment before extraction. That context is what steers the English word choice.
• Mark product names and codes that must stay as they are as non-translatable.

After delivery

• Read the free sample first. Check British or American spelling, title casing and how your brand names were handled.
• Check the Language header and set it to en_GB or en_US to match your target.
• Run msgfmt -c, which checks printf formats and plural consistency.
• Test the display with n = 0, 1 and 2.
• Entries still flagged fuzzy are ignored at compile time, so remove the flag once you have reviewed them.
• Scan the report for strings left in the source language.

In practice

1. Check the encoding header and add msgctxt to ambiguous strings.
2. Upload the .po or .pot file to get a quote and a free sample.
3. Check British or American spelling and casing in the sample.
4. Download the file and report, then adjust the Language header if needed.
5. Compile with msgfmt -c and test plurals with 0, 1 and 2.