thalovant-languages

Every word a Thalovant component's rule turns on, for every language the fleet supports, in one place and out of every codebase.

A voice satellite waits on "turn off the" because "the" is a continuation word. An intent listing closes "quelle heure est-il" with a question mark because "quelle" opens a question, and reads "volume {level} percent" aloud as "volume cinquante pour cent" in French. A French synthesiser is told to say "onze minutes" where it would swallow a consonant. Each of those is a fact about a language, not about the program that needs it, and each used to be a constant in whichever program needed it first. Here they are data: one directory per language, read by the SDK and by the satellite, and a new language is a directory in this repository rather than an edit to a program.

Layout

src/thalovant_languages/languages/
  scripts.yaml            what a writing system does that no word list can
  en-US/language.yaml     English
  fr-FR/language.yaml     French
  es/language.yaml        Spanish, every region (bare tag)
  de/ it/ pt/ nl/         German, Italian, Portuguese, Dutch
  tr/ az/                 Turkish, Azerbaijani (the dotless i)

A directory is named by a BCP-47 tag: bare (es) when what it says holds for every region, regional (en-US) when it does not. Every key in a language file is optional. A language that states none of them gets no rule at all, never another language's: a wrong question mark reads as a defect, a bare line does not.

scripts.yaml lists the scripts written without spaces (a partial transcript in one is a single "word", so characters are counted), the scripts where one character is a syllable (how a synthesiser weighs a sentence), and the marks that close a sentence or break a clause, spaced and unspaced.

Use

import thalovant_languages as languages

languages.language("fr-CA")["continuation_words"]   # the French file
languages.words("en", "trailing_words")             # lower-cased, as a set
languages.language("zh")                            # {} -- nothing describes it
languages.script_pattern("unspaced").search("今天") # a character of an unspaced script
languages.marks("sentence_ends", "spaced")          # ".!?…;"

A language is found by the matcher the rest of OVOS uses (ovos_spec_tools.language): fr-CA reads fr-FR, pt-BR reads pt, and a language nothing describes gets an empty mapping.

THALOVANT_LANGUAGES_DIR=/path/to/checkout/src/thalovant_languages/languages reads a checkout instead of the installed data, for trying a language before it is released; languages.refresh() forgets what was read after changing it.

Adding a language

1. Add src/thalovant_languages/languages/<tag>/language.yaml with the keys the language needs, each with a comment saying why the words are there. Quote a word YAML would read as something else ("on", "no").
2. Run thalovant-languages check (or pytest). It refuses a key no component reads, a pattern that does not compile, a plural category that is not one, and a list holding a boolean.
3. Open a pull request. Merging to main releases the package; the SDK and the satellite pick the language up when they move their pin.

Who reads it

• thalovant (the Python SDK): thalovant.listing sets a registered pattern the way a person reads it, ranks phrases, and reads slots aloud.
• thalovant-voice (the satellite): the semantic endpoint's continuation words, the lower-casing of an all-capitals transcript, the plural forms of its own counted text, and the synthesiser's pronunciation repairs.

Nothing here parses dates, numbers or colours; the OVOS parsers do that for skills on the hub. Contractions come from ovos-utterance-normalizer, which already carries them.