def clean_filetext(filetext):
lines = []
for line in filetext.split("\n"):
line = line.strip()
line = StringUtils.removeShapesSymbols(line)
line = StringUtils.removeGeneralPunctuation(line)
if len(line) < 2:
continue
else:
lines.append(line)
return "\n".join(lines)
def _clean_text(text):
text = StringUtils.removeShapesSymbols(text)
text = StringUtils.removeGeneralPunctuation(text)
text = StringUtils.removeChSpace(text)
text = text.lower()
return text