Feature parity with old implementation

This commit is contained in:
Stephan Porada
2020-09-08 10:42:39 +02:00
parent 521fd72d9b
commit ccfbf852b2
5 changed files with 97 additions and 147 deletions

View File

@ -35,41 +35,58 @@ def init_corpus_analysis(corpus_id):
def corpus_analysis_get_meta_data(corpus_id):
# get meta data from db
db_corpus = Corpus.query.get(corpus_id)
# TODO: Check if current user is actually the creator of the corpus?
metadata = {}
metadata['corpus_name'] = db_corpus.title
metadata['corpus_description'] = db_corpus.description
metadata['corpus_creation_date'] = db_corpus.creation_date.isoformat()
metadata['corpus_last_edited_date'] = db_corpus.last_edited_date.isoformat()
# get meta data from corpus in cqp server
client = corpus_analysis_clients.get(request.sid)
client_corpus = client.corpora.get('CORPUS')
metadata['corpus_properties'] = client_corpus.attrs['properties']
metadata['corpus_size_tokens'] = client_corpus.attrs['size']
if client is None:
response = {'code': 424, 'desc': 'No client found for this session',
'msg': 'Failed Dependency'}
socketio.emit('corpus_analysis_query', response, room=request.sid)
return
# check if client is busy or not
if client.status == 'running':
client.status = 'abort'
while client.status != 'ready':
socketio.sleep(0.1)
# get meta data from corpus in cqp server
client.status = 'running'
try:
client_corpus = client.corpora.get('CORPUS')
metadata['corpus_properties'] = client_corpus.attrs['properties']
metadata['corpus_size_tokens'] = client_corpus.attrs['size']
text_attr = client_corpus.structural_attributes.get('text')
struct_attrs = client_corpus.structural_attributes.list(filters={'part_of': text_attr})
text_ids = range(0, (text_attr.attrs['size']))
texts_metadata = {}
for text_id in text_ids:
texts_metadata[text_id] = {}
for struct_attr in struct_attrs:
texts_metadata[text_id][struct_attr.attrs['name'][(len(text_attr.attrs['name']) + 1):]] = struct_attr.values_by_ids(list(range(struct_attr.attrs['size'])))[text_id]
metadata['corpus_all_texts'] = texts_metadata
metadata['corpus_analysis_date'] = datetime.utcnow().isoformat()
metadata['corpus_cqi_py_protocol_version'] = client.api.version
metadata['corpus_cqi_py_package_version'] = cqi.__version__
metadata['corpus_cqpserver_version'] = 'CQPserver v3.4.22' # TODO: make this dynamically
text_attr = client_corpus.structural_attributes.get('text')
struct_attrs = client_corpus.structural_attributes.list(filters={'part_of': text_attr})
text_ids = range(0, (text_attr.attrs['size']))
texts_metadata = {}
for text_id in text_ids:
texts_metadata[text_id] = {}
for struct_attr in struct_attrs:
texts_metadata[text_id][struct_attr.attrs['name'][(len(text_attr.attrs['name']) + 1):]] = struct_attr.values_by_ids(list(range(struct_attr.attrs['size'])))[text_id]
metadata['corpus_all_texts'] = texts_metadata
metadata['corpus_analysis_date'] = datetime.utcnow().isoformat()
metadata['corpus_cqi_py_protocol_version'] = client.api.version
metadata['corpus_cqi_py_package_version'] = cqi.__version__
metadata['corpus_cqpserver_version'] = 'CQPserver v3.4.22' # TODO: make this dynamically
# write some metadata to the db
db_corpus.current_nr_of_tokens = metadata['corpus_size_tokens']
db.session.commit()
# write some metadata to the db
db_corpus.current_nr_of_tokens = metadata['corpus_size_tokens']
db.session.commit()
# emit data
payload = metadata
response = {'code': 200, 'desc': 'Corpus meta data', 'msg': 'OK',
'payload': payload}
socketio.emit('corpus_analysis_meta_data', response, room=request.sid)
# emit data
payload = metadata
response = {'code': 200, 'desc': 'Corpus meta data', 'msg': 'OK',
'payload': payload}
socketio.emit('corpus_analysis_meta_data', response, room=request.sid)
except cqi.errors.CQiException as e:
payload = {'code': e.code, 'desc': e.description, 'msg': e.name}
response = {'code': 500, 'desc': None, 'msg': 'Internal Server Error',
'payload': payload}
socketio.emit('corpus_analysis_meta_data', response, room=request.sid)
client.status = 'ready'
@socketio.on('corpus_analysis_query')
@ -100,7 +117,6 @@ def corpus_analysis_query(query):
'match_count': results.attrs['size']}
response = {'code': 200, 'desc': None, 'msg': 'OK', 'payload': payload}
socketio.emit('corpus_analysis_query', response, room=request.sid)
# TODO: Stop here and add a new method for transmission
chunk_size = 100
chunk_start = 0
context = 50
@ -142,6 +158,11 @@ def corpus_analysis_inspect_match(payload):
socketio.emit('corpus_analysis_inspect_match', response,
room=request.sid)
return
if client.status == 'running':
client.status = 'abort'
while client.status != 'ready':
socketio.sleep(0.1)
client.status = 'running'
try:
corpus = client.corpora.get('CORPUS')
s = corpus.structural_attributes.get('s')
@ -171,6 +192,7 @@ def corpus_analysis_inspect_match(payload):
'type': type,
'data_indexes': data_indexes}
socketio.emit('corpus_analysis_inspect_match', response, room=request.sid)
client.status = 'ready'
def corpus_analysis_session_handler(app, corpus_id, user_id, session_id):