def jobs_similar(id):
out = ''
es_query = RecordsSearch(index='records-jobs', doc_type='jobs')
es_query = es_query.query(
{
"more_like_this": {
"docs": [
{
"_id": id
}
],
"min_term_freq": 0,
"min_doc_freq": 0,
}
}
)[0:2]
similar_jobs = es_query.execute()
for job in similar_jobs:
out = out + (render_template_to_string(
"inspirehep_theme/similar_jobs.html",
record=job))
return out
def cleanup_indexed_deposits():
"""Delete indexed deposits that do not exist in the database.
.. note:: This task exists because of deposit REST API calls sometimes
failing after the deposit has already been sent for indexing to ES,
leaving an inconsistent state of a deposit existing in ES and not in
the database. It should be removed once a proper signal mechanism has
been implemented in the ``invenio-records-rest`` and
``invenio-deposit`` modules.
"""
search = RecordsSearch(index='deposits')
q = (search.query('term', **{
'_deposit.status': 'draft'
}).source(['_deposit.id']))
res = q.scan()
es_depids_info = [(d.to_dict().get('_deposit.id', [None])[0], d.meta.id,
d.meta.index, d.meta.doc_type) for d in res]
es_depids = {p[0] for p in es_depids_info}
db_depids_query = PersistentIdentifier.query.filter(
PersistentIdentifier.pid_type == 'depid',
PersistentIdentifier.pid_value.in_(es_depids))
db_depids = {d.pid_value for d in db_depids_query}
missing_db_depids = filter(lambda d: d[0] not in db_depids, es_depids_info)
indexer = RecordIndexer()
for _, deposit_id, index, doc_type in missing_db_depids:
indexer.client.delete(id=str(deposit_id),
index=index,
doc_type=doc_type)
def test_elasticsearch_query_min_score(app):
"""Test building a query with min_score."""
app.config.update(SEARCH_RESULTS_MIN_SCORE=0.1)
q = RecordsSearch()
q = q.query(Q('match', title='Higgs'))
search_dict = q.to_dict()
assert 'min_score' in search_dict
assert search_dict['min_score'] == app.config['SEARCH_RESULTS_MIN_SCORE']
def _build_query(id_):
result = RecordsSearch(index='records-jobs', doc_type='jobs')
return result.query({
'more_like_this': {
'docs': [
{
'_id': id_,
},
],
'min_term_freq': 0,
'min_doc_freq': 0,
}
})[0:2]