fix: [ocr] fix ocr supported languages

CIRCL · Apr 26, 2024 · 5503d81 · 5503d81
1 parent 1d1671c
commit 5503d81
Showing 1 changed file with 3 additions and 4 deletions.
diff --git a/bin/modules/OcrExtractor.py b/bin/modules/OcrExtractor.py
@@ -24,7 +24,7 @@
 
 
 # Default to eng
-def get_model_languages(obj, ocr_languages, add_en=True):
+def get_model_languages(obj, add_en=True):
     if add_en:
         model_languages = {'en'}
     else:
@@ -54,8 +54,6 @@ def get_model_languages(obj, ocr_languages, add_en=True):
             model_languages.add(lang)
             return model_languages
 
-    model_languages = Ocrs.sanityze_ocr_languages(model_languages, ocr_languages=ocr_languages)
-
     return model_languages
 
     #  TODO thread
@@ -100,7 +98,8 @@ def compute(self, message):
 
         if not ocr.exists():
             path = image.get_filepath()
-            languages = get_model_languages(image, self.ocr_languages)
+            languages = get_model_languages(image)
+            languages = Ocrs.sanityze_ocr_languages(languages, ocr_languages=self.ocr_languages)
             print(image.id, languages)
             texts = Ocrs.extract_text(path, languages)
             if texts: