Glossary

phil/translation-memory

A Glossary\Glossary holds the Terms of one pair of languages.

$glossary = new Glossary(LanguageTag::of('en'), LanguageTag::of('fr'), [
    new Term('hit points', ['points de vie', 'PV'], forbidden: ['vie']),
    new Term('Dungeon Master'),                                  // kept as it is
    new Term('Gnome', ['Gnome'], caseSensitive: true, note: 'The people, not the garden ornament.'),
]);

Term	Meaning
`source`	The term in the source language, white space normalized
`translations`	The allowed translations; none when the term is kept as it is — a name, a brand
`forbidden`	Translations refused
`caseSensitive`	Whether case matters, both in the source and in the target
`note`	A note for translators
`forms`	The inflected forms of the term — `hit point` for `hit points`
`translationForms`	The inflected forms of its translations — `point de vie` for `points de vie` — or of the term when it is kept

Finding terms

find($source) gives the TermHits of a source, in reading order: the term, the words as written, and their offset in characters of the plain text. Terms match as whole words — never inside a longer word — and literally, without regard to case unless the term says otherwise. Where two terms overlap, the longer one wins: hit points before hit.

Checking a target

check($source, $target) gives the TermIssues of a target:

Kind	When
`missing`	The target holds the allowed translations — or the kept term — fewer times than the source holds the term; the issue says how many of each
`forbidden`	The target holds a refused translation outside an allowed one: `vie` is refused while `points de vie` is allowed

Only the terms of the source are checked.

Inflected forms

A term and its translations change in a sentence: one hit point, two hit points. The forms of a term are found as the term itself, and the forms of its translations count as its translations:

$term = new Term('hit points', ['points de vie'], forms: ['hit point'], translationForms: ['point de vie']);

$term->variants();   // ['hit points', 'hit point'] — what find() looks for
$term->accepted();   // ['points de vie', 'point de vie'] — what check() counts

Forms equal to the term or to a translation are left out, and duplicates kept once.

An Inflector makes the forms instead: given one, the glossary adds the forms it gives for each term, in the source language, and for each translation, in the target language; $term->inflected($inflector, $source, $target) does the same for one term. RegularPlurals gives the regular plural in English and in French, regional variants included:

Language	Rule	Examples
English	The last word: `-es` after `s`, `x`, `z`, `ch`, `sh`; `-ies` after a consonant and `y`; `-s` otherwise	`hit point`, `hit points`; `box`, `boxes`; `city`, `cities`
French	Every word before the first preposition: unchanged after `s`, `x`, `z`; `-aux` for `-al`; `-x` after `au`, `eu` and seven nouns in `-ou`; `-s` otherwise, and for `bal`, `bleu`, `festival`, `pneu` and the like	`point de vie`, `points de vie`; `épée longue`, `épées longues`; `cheval`, `chevaux`

Words with a digit or an apostrophe are left alone, and other languages get no form. Irregular plurals — mouse, œil — and other inflections — genders, cases, conjugations — go into the term's forms, or into an inflector of the application.

$glossary = new Glossary(LanguageTag::of('en'), LanguageTag::of('fr-CA'), $terms, new RegularPlurals());

Exchange

A glossary leaves and arrives as TBX: see TMX and TBX.