nopaque/app/corpora/json_routes.py

from datetime import datetime
from flask import abort, current_app, request, url_for
from flask_login import current_user
from threading import Thread
from app import db
from app.decorators import content_negotiation
from app.models import Corpus, CorpusFollowerRole
from . import bp
from .decorators import corpus_follower_permission_required, corpus_owner_or_admin_required
import nltk
from string import punctuation


@bp.route('/<hashid:corpus_id>', methods=['DELETE'])
@corpus_owner_or_admin_required
@content_negotiation(produces='application/json')
def delete_corpus(corpus_id):
    def _delete_corpus(app, corpus_id):
        with app.app_context():
            corpus = Corpus.query.get(corpus_id)
            corpus.delete()
            db.session.commit()

    corpus = Corpus.query.get_or_404(corpus_id)
    thread = Thread(
        target=_delete_corpus,
        args=(current_app._get_current_object(), corpus.id)
    )
    thread.start()
    response_data = {
        'message': f'Corpus "{corpus.title}" marked for deletion',
        'category': 'corpus'
    }
    return response_data, 200


@bp.route('/<hashid:corpus_id>/build', methods=['POST'])
@corpus_follower_permission_required('MANAGE_FILES')
@content_negotiation(produces='application/json')
def build_corpus(corpus_id):
    def _build_corpus(app, corpus_id):
        with app.app_context():
            corpus = Corpus.query.get(corpus_id)
            corpus.build()
            db.session.commit()

    corpus = Corpus.query.get_or_404(corpus_id)
    if len(corpus.files.all()) == 0:
        abort(409)
    thread = Thread(
        target=_build_corpus,
        args=(current_app._get_current_object(), corpus_id)
    )
    thread.start()
    response_data = {
        'message': f'Corpus "{corpus.title}" marked for building',
        'category': 'corpus'
    }
    return response_data, 202

@bp.route('/stopwords')
@content_negotiation(produces='application/json')
def get_stopwords():
    # data = request.json
    # if not isinstance(data, dict):
    #     abort(400)
    # language = data.get('language')
    # if not isinstance(language, str):
    #     abort(400)
    nltk.download('stopwords')
    languages = ["german", "english", "catalan", "greek", "spanish", "french", "italian", "russian", "chinese"]
    stopwords = {}
    for language in languages:
        stopwords[language] = nltk.corpus.stopwords.words(language)
    stopwords['punctuation'] = list(punctuation) + ['—', '|']
    stopwords['user_stopwords'] = []
    print(stopwords)
    response_data = {
        'stopwords': stopwords
    }
    return response_data, 202

# @bp.route('/<hashid:corpus_id>/generate-share-link', methods=['POST'])
# @corpus_follower_permission_required('MANAGE_FOLLOWERS')
# @content_negotiation(consumes='application/json', produces='application/json')
# def generate_corpus_share_link(corpus_id):
#     data = request.json
#     if not isinstance(data, dict):
#         abort(400)
#     expiration = data.get('expiration')
#     if not isinstance(expiration, str):
#         abort(400)
#     role_name = data.get('role')
#     if not isinstance(role_name, str):
#         abort(400)
#     expiration_date = datetime.strptime(expiration, '%b %d, %Y')
#     cfr = CorpusFollowerRole.query.filter_by(name=role_name).first()
#     if cfr is None:
#         abort(400)
#     corpus = Corpus.query.get_or_404(corpus_id)
#     token = current_user.generate_follow_corpus_token(corpus.hashid, role_name, expiration_date)
#     corpus_share_link = url_for(
#         'corpora.follow_corpus',
#         corpus_id=corpus_id,
#         token=token,
#         _external=True
#     )
#     response_data = {
#         'message': 'Corpus share link generated',
#         'category': 'corpus',
#         'corpusShareLink': corpus_share_link
#     }
#     return response_data, 200
    


# @bp.route('/<hashid:corpus_id>/is_public', methods=['PUT'])
# @corpus_owner_or_admin_required
# @content_negotiation(consumes='application/json', produces='application/json')
# def update_corpus_is_public(corpus_id):
#     is_public = request.json
#     if not isinstance(is_public, bool):
#         abort(400)
#     corpus = Corpus.query.get_or_404(corpus_id)
#     corpus.is_public = is_public
#     db.session.commit()
#     response_data = {
#         'message': (
#             f'Corpus "{corpus.title}" is now'
#             f' {"public" if is_public else "private"}'
#         ),
#         'category': 'corpus'
#     }
#     return response_data, 200