enable using get_language_detector with many languages + add unit test

annif/simplemma_util.py

@@ -1,5 +1,7 @@
 """Wrapper code for using Simplemma functionality in Annif"""
 
+from typing import Tuple, Union
+
 from simplemma import LanguageDetector, Lemmatizer
 from simplemma.strategies import DefaultStrategy
 from simplemma.strategies.dictionaries import DefaultDictionaryFactory
@@ -11,5 +13,5 @@
 lemmatizer = Lemmatizer(lemmatization_strategy=_lemmatization_strategy)
 
 
-def get_language_detector(lang: str) -> LanguageDetector:
+def get_language_detector(lang: Union[str, Tuple[str, ...]]) -> LanguageDetector:
     return LanguageDetector(lang, lemmatization_strategy=_lemmatization_strategy)

tests/test_simplemma_util.py

@@ -0,0 +1,17 @@
+"""Unit tests for Simplemma utility functions"""
+
+from annif.simplemma_util import get_language_detector
+
+
+def test_get_language_detector():
+    detector = get_language_detector("en")
+    text = "She said 'au revoir' and left"
+    proportion = detector.proportion_in_target_languages(text)
+    assert proportion == 0.75
+
+
+def test_get_language_detector_many():
+    detector = get_language_detector(("en", "fr"))
+    text = "She said 'au revoir' and left"
+    proportion = detector.proportion_in_target_languages(text)
+    assert proportion == 1.0