sporada/nopaque
1
0
Fork 0
mirror of https://gitlab.ub.uni-bielefeld.de/sfb1288inf/nopaque.git synced 2025-06-11 16:40:40 +00:00
Code Issues Projects Releases Wiki Activity

Add export options to subcorpora

Browse Source
This commit is contained in:
Patrick Jentsch
2023-01-19 14:59:09 +01:00
parent 8b8df68781
commit 14820643ed
4 changed files with 218 additions and 68 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
app/corpora/cqi_over_socketio/cqi_corpora_corpus_subcorpora.py
View File

@ -1,13 +1,9 @@
from flask import session
import cqi
import json
import math
import os
from app import socketio
from app.decorators import socketio_login_required
from app.models import Corpus
from . import NAMESPACE as ns
from .utils import cqi_over_socketio, export_subcorpus
from .utils import cqi_over_socketio, export_subcorpus, partial_export_subcorpus
@socketio.on('cqi.corpora.corpus.subcorpora.get', namespace=ns)
@ -109,6 +105,16 @@ def cqi_corpora_corpus_subcorpora_subcorpus_paginate(cqi_client: cqi.CQiClient,
return {'code': 200, 'msg': 'OK', 'payload': payload}
@socketio.on('cqi.corpora.corpus.subcorpora.subcorpus.partial_export', namespace=ns)
@socketio_login_required
@cqi_over_socketio
def cqi_corpora_corpus_subcorpora_subcorpus_partial_export(cqi_client: cqi.CQiClient, corpus_name: str, subcorpus_name: str, match_id_list: list, context: int = 50): # noqa
cqi_corpus = cqi_client.corpora.get(corpus_name)
cqi_subcorpus = cqi_corpus.subcorpora.get(subcorpus_name)
cqi_subcorpus_partial_export = partial_export_subcorpus(cqi_subcorpus, match_id_list, context=context)
return {'code': 200, 'msg': 'OK', 'payload': cqi_subcorpus_partial_export}
@socketio.on('cqi.corpora.corpus.subcorpora.subcorpus.export', namespace=ns)
@socketio_login_required
@cqi_over_socketio
@ -116,8 +122,4 @@ def cqi_corpora_corpus_subcorpora_subcorpus_export(cqi_client: cqi.CQiClient, co
cqi_corpus = cqi_client.corpora.get(corpus_name)
cqi_subcorpus = cqi_corpus.subcorpora.get(subcorpus_name)
cqi_subcorpus_export = export_subcorpus(cqi_subcorpus, context=context)
corpus = Corpus.query.get(session['d']['corpus_id'])
file_path = os.path.join(corpus.path, f'{subcorpus_name}.json')
with open(file_path, 'w') as file:
json.dump(cqi_subcorpus_export, file)
return {'code': 200, 'msg': 'OK'}
return {'code': 200, 'msg': 'OK', 'payload': cqi_subcorpus_export}

69
app/corpora/cqi_over_socketio/utils.py
View File

@ -68,7 +68,7 @@ def lookups_by_cpos(corpus, cpos_list):
cpos_attr_values[i]
for attr in corpus.structural_attributes.list():
# We only want to iterate over non subattributes, identifiable by
# attr.attrs['has_values']==False
# attr.attrs['has_values'] == False
if attr.attrs['has_values']:
continue
cpos_attr_ids = attr.ids_by_cpos(cpos_list)
@ -93,43 +93,86 @@ def lookups_by_cpos(corpus, cpos_list):
return lookups
def partial_export_subcorpus(subcorpus, match_id_list, context=25):
if subcorpus.attrs['size'] == 0:
return {"matches": []}
match_boundaries = []
for match_id in match_id_list:
if match_id < 0 or match_id >= subcorpus.attrs['size']:
continue
match_boundaries.append(
(
match_id,
subcorpus.dump(subcorpus.attrs['fields']['match'], match_id, match_id)[0],
subcorpus.dump(subcorpus.attrs['fields']['matchend'], match_id, match_id)[0]
)
)
cpos_set = set()
matches = []
for match_boundary in match_boundaries:
match_num, match_start, match_end = match_boundary
c = (match_start, match_end)
if match_start == 0 or context == 0:
lc = None
cpos_list_lbound = match_start
else:
lc_lbound = max(0, (match_start - context))
lc_rbound = match_start - 1
lc = (lc_lbound, lc_rbound)
cpos_list_lbound = lc_lbound
if match_end == (subcorpus.collection.corpus.attrs['size'] - 1) or context == 0:
rc = None
cpos_list_rbound = match_end
else:
rc_lbound = match_end + 1
rc_rbound = min(
(match_end + context),
(subcorpus.collection.corpus.attrs['size'] - 1)
)
rc = (rc_lbound, rc_rbound)
cpos_list_rbound = rc_rbound
match = {'num': match_num, 'lc': lc, 'c': c, 'rc': rc}
matches.append(match)
cpos_set.update(range(cpos_list_lbound, cpos_list_rbound + 1))
lookups = lookups_by_cpos(subcorpus.collection.corpus, list(cpos_set))
return {'matches': matches, **lookups}
def export_subcorpus(subcorpus, context=25, cutoff=float('inf'), offset=0):
if subcorpus.attrs['size'] == 0:
return {"matches": []}
first_match = max(0, offset)
last_match = min((offset + cutoff - 1), (subcorpus.attrs['size'] - 1))
match_boundaries = zip(
subcorpus.dump(
subcorpus.attrs['fields']['match'], first_match, last_match),
subcorpus.dump(
subcorpus.attrs['fields']['matchend'], first_match, last_match)
list(range(first_match, last_match + 1)),
subcorpus.dump(subcorpus.attrs['fields']['match'], first_match, last_match),
subcorpus.dump(subcorpus.attrs['fields']['matchend'], first_match, last_match)
)
cpos_set = set()
matches = []
match_num = offset + 1
for match_start, match_end in match_boundaries:
for match_num, match_start, match_end in match_boundaries:
c = (match_start, match_end)
if match_start == 0 or context == 0:
lc = None
cpos_list_lbound = match_start
else:
lc_lbound = max(0, (match_start - 1 - context))
lc_lbound = max(0, (match_start - context))
lc_rbound = match_start - 1
lc = (lc_lbound, lc_rbound)
cpos_list_lbound = lc_lbound
if (match_end == (subcorpus.collection.corpus.attrs['size'] - 1)
or context == 0):
if match_end == (subcorpus.collection.corpus.attrs['size'] - 1) or context == 0:
rc = None
cpos_list_rbound = match_end
else:
rc_lbound = match_end + 1
rc_rbound = min(match_end + 1 + context,
subcorpus.collection.corpus.attrs['size'] - 1)
rc_rbound = min(
(match_end + context),
(subcorpus.collection.corpus.attrs['size'] - 1)
)
rc = (rc_lbound, rc_rbound)
cpos_list_rbound = rc_rbound
match = {'num': match_num, 'lc': lc, 'c': c, 'rc': rc}
matches.append(match)
cpos_set.update(range(cpos_list_lbound, cpos_list_rbound + 1))
match_num += 1
lookups = lookups_by_cpos(subcorpus.collection.corpus, list(cpos_set))
return {'matches': matches, **lookups}