Update app.py

app.py

@@ -26,33 +26,33 @@ def remove_non_printable(value):
     if isinstance(value, str):
         print("Leire")
         # Eliminar \n, \t, \xa0 por un espacio en blanco
-        value = re.sub(r'[\n\t\xa0]', ' ', value) 
+        #value = re.sub(r'[\n\t\xa0]', ' ', value) 
         # Eliminar \n entre palabras
-        value = re.sub(r'(?<=[a-zA-Z])\n(?=[a-zA-Z])', ' ', value)
+        #value = re.sub(r'(?<=[a-zA-Z])\n(?=[a-zA-Z])', ' ', value)
         # Eliminar \t entre letras
-        value = re.sub(r'(?<=[a-zA-Z])\t(?=[a-zA-Z])', ' ', value)
+        #value = re.sub(r'(?<=[a-zA-Z])\t(?=[a-zA-Z])', ' ', value)
         # Eliminar \xa0 entre letras
-        value = re.sub(r'(?<=[a-zA-Z])\xa0(?=[a-zA-Z])', ' ', value)
+        #value = re.sub(r'(?<=[a-zA-Z])\xa0(?=[a-zA-Z])', ' ', value)
         # Eliminar \n antes o después de ciertos caracteres
-        value = re.sub(r'\n(?=[.,;:!?¿)\d])|\n(?<=[.,;:!?¿)\d])', ' ', value)
+        #value = re.sub(r'\n(?=[.,;:!?¿)\d])|\n(?<=[.,;:!?¿)\d])', ' ', value)
         # Eliminar \t antes o después de ciertos caracteres
-        value = re.sub(r'\t(?=[.,;:!?¿)\d])|\t(?<=[.,;:!?¿)\d])', ' ', value)
+        #value = re.sub(r'\t(?=[.,;:!?¿)\d])|\t(?<=[.,;:!?¿)\d])', ' ', value)
         # Eliminar \xa0 antes o después de ciertos caracteres
-        value = re.sub(r'\xa0(?=[.,;:!?¿)\d])|\xa0(?<=[.,;:!?¿)\d])', ' ', value)
+        #value = re.sub(r'\xa0(?=[.,;:!?¿)\d])|\xa0(?<=[.,;:!?¿)\d])', ' ', value)
         # Eliminar \n específico después de un punto y antes \
-        value = re.sub(r'\.\n(?=\\)', '. ', value)
+        #value = re.sub(r'\.\n(?=\\)', '. ', value)
         # Eliminar \t específico después de un punto y antes \
-        value = re.sub(r'\.\t(?=\\)', '. ', value)
+        #value = re.sub(r'\.\t(?=\\)', '. ', value)
         # Eliminar \xa0 específico después de un punto y antes \
-        value = re.sub(r'\.\xa0(?=\\)', '. ', value)
-        return value
+        #value = re.sub(r'\.\xa0(?=\\)', '. ', value)
+        #return value
         #return re.sub(r'[\n[\n\t\xa0]*\t[\n\t\xa0]*\xa0[\n\t\xa0]*]',' ',value)
         #return re.sub(r'[^\x20-\x7E\n\t\xa0]|(?<=\w)\n|(?<=\w)\t', '', value)
         #return re.sub(r'\n', 'WW', value)
         #return re.sub(r'[\xa0\n\t\s]+|(?<=[.,;:!?)])\\n|\n(?<=\w)\\n|\n(?<=\w)|(?<=[.,;:!?)])\\t|(?<=\w)\\t|\t(?<=\w)\\t|\t(?<=\w)|(?<=[.,;:!?)])\\xa0|(?<=\w)\\xa0|\xa0(?<=\w)\\xa0|\xa0(?<=\w)', ' ', value)
         #return re.sub(r'(?<=\w)[\n\t\xa0]+|[\n\t\xa0]+(?=\w)|^\s*[\n\t\xa0]+|[\n\t\xa0]+$|(?<=[.,:?)¿!])[\n\t\xa0]+|\b[\n\t\xa0]+\b', ' ', value)
         #return re.sub(r'\s*[\n\t\xa0]+\s*', ' ', value)
-        #return re.sub(r'[\xa0\n\t\s]|(?<=\w)\\n|(?<=[.,;:!?¿])\\n|\n(?<=\w)\\n|\n(?<=\w)|\n(?<=[.,;:!?¿])|\t(?<=[.,;:!?¿])|(?<=[.,;:!?¿])\\t|(?<=\w)\\t|\t(?<=\w)\\t|\t(?<=\w)|(?<=[.,;:!?¿])\\xa0|(?<=\w)\\xa0|\xa0(?<=\w)\\xa0|\xa0(?<=\w)\xa0(?<=[.,;:!?¿])|', ' ', value)
+        return re.sub(r'[\xa0\n\t\s]|(?<=\w)\\n|(?<=[.,;:!?¿])\\n|\n(?<=\w)\\n|\n(?<=\w)|\n(?<=[.,;:!?¿])|\t(?<=[.,;:!?¿])|(?<=[.,;:!?¿])\\t|(?<=\w)\\t|\t(?<=\w)\\t|\t(?<=\w)|(?<=[.,;:!?¿])\\xa0|(?<=\w)\\xa0|\xa0(?<=\w)\\xa0|\xa0(?<=\w)\xa0(?<=[.,;:!?¿])|', 'ww', value)
         #return re.sub(r'\s*[\n\t\xa0]+\s*', ' ', value).strip()
         #value = value.replace("\xa0", "").replace("\n", "").replace("\t", "")
     return value