sporada/nopaque
1
0
Fork 0
mirror of https://gitlab.ub.uni-bielefeld.de/sfb1288inf/nopaque.git synced 2025-06-12 17:10:41 +00:00
Code Issues Projects Releases Wiki Activity

Add metadata object on analysis

Browse Source
This commit is contained in:
Stephan Porada
2020-04-30 14:45:54 +02:00
parent a17771a772
commit 1350e6ddd6
4 changed files with 33 additions and 29 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
app/corpora/events.py
View File

@ -8,6 +8,7 @@ from ..models import Corpus, User
import cqi
import math
from app import logger
from datetime import datetime
'''
@ -47,16 +48,18 @@ def corpus_analysis_get_meta_data(corpus_id):
metadata['corpus_properties'] = client_corpus.attrs['properties']
metadata['corpus_size_tokens'] = client_corpus.attrs['size']
texts = {}
text_attr = client_corpus.structural_attributes.get('text')
for text_id in range(0, (text_attr.attrs['size'])):
texts[text_id] = {}
logger.warning('Text ID: {}'.format(text_id))
struct_attrs = client_corpus.structural_attributes.list(filters={'part_of': text_attr})
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:
logger.warning(struct_attr)
for struct_id in list(range(0, struct_attr.attrs['size'])):
logger.warning(struct_attr.values_by_ids([struct_id]))
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_version'] = cqi.version
metadata['corpus_cqpserver_version'] = 'CQPserver v3.4.22' # TODO: make this dynamically
payload = metadata
response = {'code': 200, 'desc': 'Corpus meta data', 'msg': 'OK', 'payload': payload}