diff --git a/qurator/dinglehopper/substitute_equivalences.py b/qurator/dinglehopper/substitute_equivalences.py index faa873e..3556a52 100644 --- a/qurator/dinglehopper/substitute_equivalences.py +++ b/qurator/dinglehopper/substitute_equivalences.py @@ -1,3 +1,6 @@ +import unicodedata + + def substitute_equivalences(s): # These are for OCR-D GT vs Tesseract frk vs Calamari GT4HistOCR