nopaque/app/namespaces/corpora.py

from datetime import datetime
from flask import current_app, Flask, url_for
from flask_login import current_user
from flask_socketio import Namespace
from string import punctuation
import nltk
from app import db, hashids, socketio
from app.decorators import socketio_login_required
from app.models import Corpus, CorpusFollowerAssociation, CorpusFollowerRole


def _delete_corpus(app: Flask, corpus_id: int):
    with app.app_context():
        corpus = Corpus.query.get(corpus_id)
        corpus.delete()
        db.session.commit()


def _build_corpus(app: Flask, corpus_id: int):
    with app.app_context():
        corpus = Corpus.query.get(corpus_id)
        corpus.build()
        db.session.commit()


class CorporaNamespace(Namespace):
    @socketio_login_required
    def on_delete(self, corpus_hashid: str) -> dict:
        if not isinstance(corpus_hashid, str):
            return {'status': 400, 'statusText': 'Bad Request'}

        corpus_id = hashids.decode(corpus_hashid)

        if not isinstance(corpus_id, int):
            return {'status': 400, 'statusText': 'Bad Request'}

        corpus = Corpus.query.get(corpus_id)

        if corpus is None:
            return {'status': 404, 'statusText': 'Not Found'}

        if not (
            corpus.user == current_user
            or current_user.is_administrator
        ):
            return {'status': 403, 'statusText': 'Forbidden'}

        socketio.start_background_task(
            _delete_corpus,
            current_app._get_current_object(),
            corpus_id
        )

        return {
            'body': f'Corpus "{corpus.title}" marked for deletion',
            'status': 202,
            'statusText': 'Accepted'
        }

    @socketio_login_required
    def on_build(self, corpus_hashid: str) -> dict:
        if not isinstance(corpus_hashid, str):
            return {'status': 400, 'statusText': 'Bad Request'}

        corpus_id = hashids.decode(corpus_hashid)

        if not isinstance(corpus_id, int):
            return {'status': 400, 'statusText': 'Bad Request'}

        corpus = Corpus.query.get(corpus_id)

        if corpus is None:
            return {'status': 404, 'statusText': 'Not Found'}

        cfa = CorpusFollowerAssociation.query.filter_by(corpus_id=corpus_id, follower_id=current_user.id).first()
        if not (
            cfa is not None and cfa.role.has_permission('MANAGE_FILES')
            or corpus.user == current_user
            or current_user.is_administrator
        ):
            return {'status': 403, 'statusText': 'Forbidden'}

        if len(corpus.files.all()) == 0:
            return {'status': 409, 'statusText': 'Conflict'}

        socketio.start_background_task(
            _build_corpus,
            current_app._get_current_object(),
            corpus_id
        )

        return {
            'body': f'Corpus "{corpus.title}" marked for building',
            'status': 202,
            'statusText': 'Accepted'
        }

    # TODO: Think about where to place this, as this does not belong here...
    @socketio_login_required
    def on_get_stopwords(self):
        languages = [
            'german',
            'english',
            'catalan',
            'greek',
            'spanish',
            'french',
            'italian',
            'russian',
            'chinese'
        ]

        nltk.download('stopwords', quiet=True)
        stopwords = {
            language: nltk.corpus.stopwords.words(language)
            for language in languages
        }
        stopwords['punctuation'] = list(punctuation)
        stopwords['punctuation'] += ['—', '|', '–', '“', '„', '--']
        stopwords['user_stopwords'] = []

        return {
            'body': stopwords,
            'status': 200,
            'statusText': 'OK'
        }

    @socketio_login_required
    def on_create_share_link(self, corpus_hashid: str, expiration_date: str, role_name: str) -> dict:
        if not isinstance(corpus_hashid, str):
            return {'status': 400, 'statusText': 'Bad Request'}

        if not isinstance(expiration_date, str):
            return {'status': 400, 'statusText': 'Bad Request'}

        if not isinstance(role_name, str):
            return {'status': 400, 'statusText': 'Bad Request'}

        print(corpus_hashid, expiration_date, role_name)

        corpus_id = hashids.decode(corpus_hashid)

        if not isinstance(corpus_id, int):
            return {'status': 400, 'statusText': 'Bad Request'}

        corpus = Corpus.query.get(corpus_id)

        if corpus is None:
            return {'status': 404, 'statusText': 'Not Found'}

        cfa = CorpusFollowerAssociation.query.filter_by(corpus_id=corpus_id, follower_id=current_user.id).first()
        if not (
            cfa is not None and cfa.role.has_permission('MANAGE_FOLLOWERS')
            or corpus.user == current_user
            or current_user.is_administrator
        ):
            return {'status': 403, 'statusText': 'Forbidden'}

        _expiration_date = datetime.strptime(expiration_date, '%b %d, %Y')

        cfr = CorpusFollowerRole.query.filter_by(name=role_name).first()
        if cfr is None:
            return {'status': 400, 'statusText': 'Bad Request'}

        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
        )

        return {
            'body': corpus_share_link,
            'status': 200,
            'statusText': 'OK'
        }

    @socketio_login_required
    def on_set_is_public(corpus_hashid: str, new_value: bool) -> dict:
        if not isinstance(corpus_id, str):
            return {'status': 400, 'statusText': 'Bad Request'}

        if not isinstance(new_value, bool):
            return {'status': 400, 'statusText': 'Bad Request'}

        corpus_id = hashids.decode(corpus_hashid)

        if not isinstance(corpus_id, int):
            return {'status': 400, 'statusText': 'Bad Request'}

        corpus = Corpus.query.get(corpus_id)

        if corpus is None:
            return {'status': 404, 'statusText': 'Not Found'}

        if not (
            corpus.user == current_user
            or current_user.is_administrator
        ):
            return {'status': 403, 'statusText': 'Forbidden'}

        corpus.is_public = new_value
        db.session.commit()

        return {
            'body': f'Corpus "{corpus.title}" is now {"public" if new_value else "private"}',
            'status': 200,
            'statusText': 'OK'
        }