Add alphabets for more languages

opuscleaner/filters/clean_common.py

@@ -5,6 +5,7 @@
     'ar': r'[\u0600-\u06FF]', # This is not entirely right, as it also includes farsi symbols and whatnot
     'bg': r'[АаБбВвГгДддЕеЖжЗзИиЙйКкkasЛлМмНнОоПпРрСсТтУуФфХхЦцЧчШшЩщЪъЬьЮюЯя]',
     'bn': r'[\u0980-\u09FF]', # bangla
+    'bs': r'[a-zćĆčČđĐšŠžŽЉљЊњЋћЏџа-я]',
     'ca': r'[a-zÀàÈèÉéÍíÒòÓóÚúÇç]',
     'cs': r'[a-zÁáČčĎďÉéěÍíŇňÓóŘřŠšŤťÚúůÝýŽž]',
     'da': r'[a-zÆæØøÅå]',
@@ -22,6 +23,7 @@
     'hr': r'[abcčČćĆdđĐefghijklmnoprsšŠtuvzžŽ]',
     'hu': r'[a-zÁáÉéÍíÓóÖöŐőŰű]',
     'hy': r'[\u0530-\u058F]',
+    'id': r'[a-z]',
     'is': r'[abdefghijklmnoprstuvxyÁáðÐÉéÍíÓóÚúÝýÞþÆæÖö]',
     'it': r'[a-zàÀèÈéÉìÌíÍîÎòÒóÓùÙúÚ]',
     'ko': r'[\uac00-\ud7af]|[\u1100-\u11ff]|[\u3130-\u318f]|[\ua960-\ua97f]|[\ud7b0-\ud7ff]',
@@ -38,7 +40,10 @@
     'ru': r'[а-я]',
     'sk': r'[a-záäÁÄčČďĎžéÉíÍĺĹľĽňŇóÓôÔŕŔšŠťŤúÚýÝžŽ]',
     'sl': r'[abcčČdđĐefghijklmnoprsšŠtuvzžŽ]',
+    'sr': r'[a-zčČćĆđĐšŠžŽЉљЊњЋћЏџа-я]',
     'sv': r'[a-zÅåÄäÖö]',
+    'tr': r'[a-zçÇğĞıİöÖşŞüÜ]',
     'uk': r'[А-ЩЬЮЯҐЄІЇа-щьюяґєії\'`’ʼ]',
+    'vi': r'[a-zàáảãạâầấẩẫậăằắẳẵặèéẻẽẹêềếểễệìíỉĩịòóỏõọôồốổỗộơờớởỡợùúủũụưừứửữựỳýỷỹỵÀÁẢÃẠÂẦẤẨẪẬĂẰẮẲẴẶÈÉẺẼẸÊỀẾỂỄỆÌÍỈĨỊÒÓỎÕỌÔỒỐỔỖỘƠỜỚỞỠỢÙÚỦŨỤƯỪỨỬỮỰỲÝỶỸỴ]',
     'zh': r'[\u3400-\u4dbf\u4e00-\u9fff\uf900-\ufaff]',
 }