2021-07-10 11:33:39 +02:00
|
|
|
# File: wiktionary.py
|
|
|
|
#
|
|
|
|
# Purpose: Parses Wiktionary pages into JSON and dumps it to STDOUT.
|
|
|
|
#
|
|
|
|
# This was written for wiktionary.pl since Wiktionary::Parser in CPAN
|
|
|
|
# seems to be broken and abandoned.
|
2022-06-28 05:35:02 +02:00
|
|
|
#
|
|
|
|
# Important: This uses a custom fork of wiktionaryparser which contains
|
|
|
|
# numerous fixes. To install it use:
|
|
|
|
#
|
|
|
|
# pip install git+https://github.com/pragma-/WiktionaryParser
|
2021-07-10 11:33:39 +02:00
|
|
|
|
2021-07-11 00:00:22 +02:00
|
|
|
# SPDX-FileCopyrightText: 2021 Pragmatic Software <pragma78@gmail.com>
|
|
|
|
# SPDX-License-Identifier: MIT
|
2021-07-10 11:33:39 +02:00
|
|
|
|
|
|
|
from wiktionaryparser import WiktionaryParser
|
|
|
|
import sys
|
|
|
|
import json
|
|
|
|
|
|
|
|
parser = WiktionaryParser()
|
2021-07-10 20:07:28 +02:00
|
|
|
entries = parser.fetch(sys.argv[1], sys.argv[2])
|
2021-07-10 11:33:39 +02:00
|
|
|
print(json.dumps(entries))
|