Edouard@788: #! /usr/bin/env python3
Edouard@788: 
vb@1513: # This file is under GNU General Public License 3.0
vb@1513: # see LICENSE.txt
vb@1513: 
vb@1513: 
Edouard@788: from argparse import ArgumentParser
Edouard@788: from fileinput import FileInput, hook_encoded
Edouard@788: import re, itertools, sys
Edouard@788: 
Edouard@788: space = re.compile(r'^\s')
Edouard@788: 
Edouard@788: p = ArgumentParser(description="re-write re-order csv and strip lines with too long words")
Edouard@788: p.add_argument('--input', '-i', type=str, default="somefile.cvs",
Edouard@788:     help='input file')
Edouard@788: p.add_argument('--length', '-l', type=int, default=100,
Edouard@788:     help='min word length to stripp a line')
Edouard@788: 
Edouard@788: args = p.parse_args()
Edouard@788: 
Edouard@788: try:
Edouard@788:     from icu import UnicodeString, Locale
Edouard@788: except ImportError:
Edouard@788:     print("warning: PyICU not installed, using fallback", file=sys.stderr)
Edouard@788: else:
Edouard@788:     locale = Locale("utf-8")
Edouard@788: 
Edouard@788: _all = (
Edouard@788:         line.split(',')
Edouard@788:         for line in FileInput(
Edouard@788:                 args.input,
Edouard@788:                 openhook=hook_encoded("utf-8")
Edouard@788:             )
Edouard@788:         if not space.match(line)
Edouard@788: )
Edouard@788: 
edouard@1478: _some = (line for line in _all if len(line[2]) < args.length)
Edouard@788: 
edouard@1478: for i, w in enumerate(_some):
Edouard@788:     print("{l},{i},{w},0".format(l=w[0], i=i, w=w[2]))