cdgranadillo/summarizer_preprocessing

1import re2 3def preprocessing_text(text):4 5    list_separators=re.findall(r"Comprador*.?\(.*?\)|Mediador*.?\(.*?\)|Vendedor*.?\(.*?\)", text)6    list_splits=re.split(r"Comprador*.?\(.*?\)|Mediador*.?\(.*?\)|Vendedor*.?\(.*?\)", text)[1:]7 8    conversation=list(map(lambda a, b: a+b, list_separators, list_splits))9 10    #remove text sent to seller11    [conversation.remove(conv) for conv in conversation if "para Vendedor" in conv.split("\n",2)[0]]12 13    final_text=""14    for text in conversation:15        text=re.sub(r".*?\.jpeg|.*?\.jpg|.*?\.png|.*?\.pdf","",text)16        new_text=text.split(" ")[0]+":"+text.split("\n",2)[2]17        final_text=final_text+new_text18 19    return final_text