mirror of
https://gitlab.ub.uni-bielefeld.de/sfb1288inf/nopaque.git
synced 2024-11-15 01:05:42 +00:00
119 lines
5.4 KiB
Python
119 lines
5.4 KiB
Python
from app import db, logger, socketio
|
|
from app.events import connected_sessions
|
|
from app.models import Corpus
|
|
from flask import current_app, request
|
|
from flask_login import current_user, login_required
|
|
from .CQiWrapper.CQiWrapper import CQiWrapper
|
|
import sys
|
|
import gzip
|
|
import zlib
|
|
import json
|
|
|
|
'''
|
|
' A dictionary containing lists of, with corpus ids associated, Socket.IO
|
|
' session ids (sid). {<corpus_id>: [<sid>, ...], ...}
|
|
'''
|
|
corpus_analysis_sessions = {}
|
|
'''
|
|
' A dictionary containing Socket.IO session id - CQi client pairs.
|
|
' {<sid>: CQiClient, ...}
|
|
'''
|
|
corpus_analysis_clients = {}
|
|
|
|
|
|
@socketio.on('request_corpus_analysis')
|
|
@login_required
|
|
def request_corpus_analysis(corpus_id):
|
|
corpus = Corpus.query.get(corpus_id)
|
|
if corpus is None:
|
|
socketio.emit('init_corpus_analysis', '[404]: Not Found',
|
|
room=request.sid)
|
|
elif not (corpus.creator == current_user
|
|
or current_user.is_administrator()):
|
|
socketio.emit('init_corpus_analysis', '[403]: Forbidden',
|
|
room=request.sid)
|
|
else:
|
|
socketio.start_background_task(corpus_analysis_session_handler,
|
|
current_app._get_current_object(),
|
|
corpus_id, request.sid)
|
|
|
|
|
|
@socketio.on('corpus_analysis')
|
|
@login_required
|
|
def corpus_analysis(message):
|
|
client = corpus_analysis_clients.get(request.sid)
|
|
if client is None:
|
|
socketio.emit('query', '[424]: Failed Dependency',
|
|
room=request.sid)
|
|
return
|
|
""" Prepare and execute a query """
|
|
corpus = 'CORPUS'
|
|
query = (message['query'])
|
|
query_subcorpus = 'Results'
|
|
client.cqp_query(corpus, query_subcorpus, query)
|
|
|
|
data = {'matches': [], 'cpos_lookup': {}, 'text_loopup': {}}
|
|
|
|
""" Evaluate query results """
|
|
match_corpus = '{}:{}'.format(corpus, query_subcorpus)
|
|
match_num = min(int(message['hits_per_page']), client.cqp_subcorpus_size(match_corpus))
|
|
match_boundaries = zip(client.cqp_dump_subcorpus(match_corpus,
|
|
0x10,
|
|
0, match_num - 1),
|
|
client.cqp_dump_subcorpus(match_corpus,
|
|
0x11,
|
|
0, match_num - 1))
|
|
context = 15
|
|
corpus_len = 10000
|
|
for match_start, match_end in match_boundaries:
|
|
data['matches'].append({'lc': list(range(max(0, match_start - int(message['context'])), match_start)),
|
|
'hit': list(range(match_start, match_end + 1)),
|
|
'rc': list(range(match_end + 1, min(corpus_len, match_end + 1 + int(message['context']))))})
|
|
cpos_list = []
|
|
for match in data['matches']:
|
|
cpos_list += match['lc'] + match['hit'] + match['rc']
|
|
cpos_list = list(set(cpos_list))
|
|
lemma_list = client.cl_cpos2str('{}.lemma'.format(corpus), cpos_list)
|
|
pos_list = client.cl_cpos2str('{}.pos'.format(corpus), cpos_list)
|
|
simple_pos_list = client.cl_cpos2str('{}.simple_pos'.format(corpus), cpos_list)
|
|
s_id_list = client.cl_cpos2struc('{}.s'.format(corpus), cpos_list)
|
|
text_id_list = client.cl_cpos2struc('{}.text'.format(corpus), cpos_list)
|
|
word_list = client.cl_cpos2str('{}.word'.format(corpus), cpos_list)
|
|
for cpos, lemma, pos, simple_pos, s_id, text_id, word in zip(cpos_list, lemma_list, pos_list, simple_pos_list, s_id_list, text_id_list, word_list):
|
|
data['cpos_lookup'][cpos] = {'lemma': lemma, 'pos': pos, 'simple_pos': simple_pos, 's_id': s_id, 'text_id': text_id, 'word': word}
|
|
text_author_list = client.cl_struc2str('{}.text_author'.format(corpus), text_id_list)
|
|
text_publishing_year_list = client.cl_struc2str('{}.text_publishing_year'.format(corpus), text_id_list)
|
|
text_title_list = client.cl_struc2str('{}.text_title'.format(corpus), text_id_list)
|
|
for text_id, text_author, text_publishing_year, text_title in zip(text_id_list, text_author_list, text_publishing_year_list, text_title_list):
|
|
data['text_loopup'][text_id] = {'author': text_author, 'publishing_year': text_publishing_year, 'title': text_title}
|
|
socketio.emit('corpus_analysis', data, room=request.sid)
|
|
|
|
|
|
def corpus_analysis_session_handler(app, corpus_id, session_id):
|
|
with app.app_context():
|
|
''' Setup analysis session '''
|
|
corpus = Corpus.query.get(corpus_id)
|
|
while corpus.status != 'analysing':
|
|
db.session.refresh(corpus)
|
|
socketio.sleep(3)
|
|
client = CQiWrapper(host='analyse_corpus_{}'.format(corpus_id))
|
|
client.connect()
|
|
corpus_analysis_clients[session_id] = client
|
|
if corpus_id not in corpus_analysis_sessions:
|
|
corpus_analysis_sessions[corpus_id] = [session_id]
|
|
else:
|
|
corpus_analysis_sessions[corpus_id].append(session_id)
|
|
socketio.emit('request_corpus_analysis', '[201]: Created',
|
|
room=session_id)
|
|
''' Observe analysis session '''
|
|
while session_id in connected_sessions:
|
|
socketio.sleep(3)
|
|
''' Teardown analysis session '''
|
|
client.disconnect()
|
|
corpus_analysis_clients.pop(session_id, None)
|
|
corpus_analysis_sessions[corpus_id].remove(session_id)
|
|
if not corpus_analysis_sessions[corpus_id]:
|
|
corpus_analysis_sessions.pop(corpus_id, None)
|
|
corpus.status = 'stop analysis'
|
|
db.session.commit()
|