nopaque/app/corpora/pj_events.py

from flask import current_app, request
from flask_login import current_user
from .cqi import CQiClient
from .. import db, socketio
from ..decorators import socketio_login_required
from ..events import connected_sessions
from ..models import Corpus, User
import math


'''
' A dictionary containing lists of, with corpus ids associated, Socket.IO
' session ids (sid). {<corpus_id>: [<sid>, ...], ...}
'''
pj_corpus_analysis_sessions = {}
'''
' A dictionary containing Socket.IO session id - CQi client pairs.
' {<sid>: CQiClient, ...}
'''
pj_corpus_analysis_clients = {}


@socketio.on('pj_corpus_analysis_init')
@socketio_login_required
def pj_init_corpus_analysis(corpus_id):
    socketio.start_background_task(pj_corpus_analysis_session_handler,
                                   current_app._get_current_object(),
                                   corpus_id, current_user.id, request.sid)


@socketio.on('pj_corpus_analysis_query')
@socketio_login_required
def pj_corpus_analysis_query(query):
    client = pj_corpus_analysis_clients.get(request.sid)
    if client is None:
        response = {'code': 404, 'msg': 'Failed Dependency'}
        socketio.emit('pj_corpus_analysis_query', response, room=request.sid)
        return
    corpus = client.corpora.get('CORPUS')
    try:
        results = corpus.query(query)
    except Exception as e:
        response = {'code': 1, 'msg': str(e)}
        socketio.emit('pj_corpus_analysis_query', response, room=request.sid)
    else:
        response = {'code': 200, 'msg': 'OK',
                    'data': {'num_matches': results.size}}
        socketio.emit('pj_corpus_analysis_query', response, room=request.sid)
        chunk_size = 100
        chunk_start = 0
        context = 100
        progress = 0
        while chunk_start <= results.size:
            chunk = results.export(context=context, offset=chunk_start,
                                   cutoff=chunk_size)
            if (results.size == 0):
                progress = 100
            else:
                progress = ((chunk_start + chunk_size) / results.size) * 100
                progress = min(100, int(math.ceil(progress)))
            socketio.emit('pj_corpus_analysis_query_results',
                          {'chunk': chunk,
                           'progress': progress},
                          room=request.sid)
            chunk_start += chunk_size
            chunk_size = 250


def pj_corpus_analysis_session_handler(app, corpus_id, user_id, session_id):
    with app.app_context():
        ''' Setup analysis session '''
        corpus = Corpus.query.get(corpus_id)
        user = User.query.get(user_id)
        if corpus is None:
            response = {'code': 404, 'msg': 'Not Found'}
            socketio.emit('pj_corpus_analysis_init', response, room=session_id)
            return
        elif not (corpus.creator == user or user.is_administrator()):
            response = {'code': 403, 'msg': 'Forbidden'}
            socketio.emit('pj_corpus_analysis_init', response, room=session_id)
            return
        while corpus.status != 'analysing':
            db.session.refresh(corpus)
            socketio.sleep(3)
        client = CQiClient('pj_corpus_{}_analysis'.format(corpus_id))
        try:
            client.connect()
        except Exception:
            response = {'code': 500, 'msg': 'Internal Server Error'}
            socketio.emit('pj_corpus_analysis_init', response, room=session_id)
            return
        pj_corpus_analysis_clients[session_id] = client
        if corpus_id not in pj_corpus_analysis_sessions:
            pj_corpus_analysis_sessions[corpus_id] = [session_id]
        else:
            pj_corpus_analysis_sessions[corpus_id].append(session_id)
        response = {'code': 200, 'msg': 'OK'}
        socketio.emit('pj_corpus_analysis_init', response, room=session_id)
        ''' Observe analysis session '''
        while session_id in connected_sessions:
            socketio.sleep(3)
        ''' Teardown analysis session '''
        try:
            client.disconnect()
        except Exception:
            pass
        pj_corpus_analysis_clients.pop(session_id, None)
        pj_corpus_analysis_sessions[corpus_id].remove(session_id)
        if not pj_corpus_analysis_sessions[corpus_id]:
            pj_corpus_analysis_sessions.pop(corpus_id, None)
            corpus.status = 'stop analysis'
            db.session.commit()