#!/bin/env python
import argparse
import sys
import base64
from loomchild.segmenter import LoomchildSegmenter

parser = argparse.ArgumentParser()
parser.add_argument('-l', '--lang', dest="lang", required=True, type=str, help="Language of the input")
options = parser.parse_args()

segmenter = LoomchildSegmenter(options.lang)

for doc in sys.stdin:
    content = base64.b64decode(doc.strip().replace("\t", " ")).decode('utf-8')
    output_text = segmenter.get_document_segmentation(content)
    output_text = "\n".join(output_text)
    print(base64.b64encode(output_text.encode("utf-8")).decode("utf-8"))
