add config switch for spaCy model

src/lang_main/constants.py

@@ -72,7 +72,6 @@ SKIP_TIME_ANALYSIS: Final[bool] = CONFIG['control']['time_analysis_skip']
 
 # ** models
 # ** loading
-# MODEL_BASE_FOLDER_NAME: Final[str] = 'lang-models'
 model_folder_cfg = LIB_PATH / Path(CONFIG['paths']['models'])
 MODEL_BASE_FOLDER: Final[Path] = model_folder_cfg.resolve()
 if not MODEL_BASE_FOLDER.exists():
@@ -84,9 +83,18 @@ os.environ['SENTENCE_TRANSFORMERS_HOME'] = str(MODEL_BASE_FOLDER)
 # LANG_MAIN_STFR_MODEL : Sentence Transformer model used; if not provided, use constant value defined in library; more internal use
 # LANG_MAIN_STFR_BACKEND : STFR backend, choice between "torch" and "onnx"
 
-SPACY_MODEL_NAME: Final[str | SpacyModelTypes] = os.environ.get(
+# config option for switch between spaCy TRF und medium model
-    'LANG_MAIN_SPACY_MODEL', SpacyModelTypes.DE_DEP_NEWS_TRF
+# env variable has prio: if set, use this
-)
+SPACY_USE_LARGE_MODEL: Final[bool] = CONFIG['models']['use_large_model']
+spacy_model_name: str | SpacyModelTypes | None
+spacy_model_name = os.environ.get('LANG_MAIN_SPACY_MODEL', None)
+if spacy_model_name is None:
+    if SPACY_USE_LARGE_MODEL:
+        spacy_model_name = SpacyModelTypes.DE_DEP_NEWS_TRF
+    else:
+        spacy_model_name = SpacyModelTypes.DE_CORE_NEWS_MD
+
+SPACY_MODEL_NAME: Final[str | SpacyModelTypes] = spacy_model_name
 STFR_MODEL_NAME: Final[str | STFRModelTypes] = os.environ.get(
     'LANG_MAIN_STFR_MODEL', STFRModelTypes.E5_BASE_STS_EN_DE
 )

src/lang_main/lang_main_config.toml

@@ -5,6 +5,9 @@ inputs = '../data/in/'
 results = '../data/out/'
 models = './lang-models'
 
+[models]
+use_large_model = true
+
 [logging]
 enabled = true
 stderr = true