only use en as default if it is allowed

This commit is contained in:
Joshix 2023-01-18 19:50:48 +01:00
parent 4a769e0e12
commit e2ea101aa1

View file

@ -47,7 +47,13 @@ def detect_languages(text: "str | list[str]", allowed_languages: "list[str] | No
# this happens if no language could be detected
if not candidate_langs:
# use language "en" by default but with zero confidence
return [{"confidence": 0.0, "language": "en"}]
lang = (
"en"
if allowed_languages is None or "en" in allowed_languages
else allowed_languages[0]
)
return [{"confidence": 0.0, "language": lang}]
# for multiple occurrences of the same language (can happen on batch detection)
# calculate the average confidence for each language