odsfiles2col.py

#!/usr/bin/python3
# https://github.com/pyexcel/pyexcel-ods/issues/41
#  pyexcel-ods issue: use "<text:p>" instead of "<text:line-break/>" for newlines (line breaks) in cells

import argparse, re, sys, json
from pyexcel_ods3 import save_data
from warnings import warn

parser = argparse.ArgumentParser(
    description='read many text files into two columns of an ods file, with filenames as the key column and contents as the value column',
    formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('-o', '--outfn', type=str, default='out.ods',
    help='output file name')
parser.add_argument('-s', '--sheet', type=str, default='main',
    help='sheet name')
parser.add_argument('-x', '--keepext', type=bool, default=False,
    help='keep extension?')
parser.add_argument('textfile', nargs='*', help='input files')
args = parser.parse_args()

sheet = []
for fn in args.textfile:
    with open(fn) as F:
        val = F.read()
    fn = re.search(r'([^/]+$)', fn).group(1)
    if not args.keepext:
        fn = re.sub(r'\.\w+$', '', fn)
    sheet.append([fn, val])
save_data(args.outfn, { args.sheet: sheet })
#args.keycol
#print(json.dumps(sheet, ensure_ascii=False))

pineapple.txt

spine
vitamin C
antioxidants

raspberry.txt

temperate
fiber
summer