def test_range_filter():
"""Test range filter."""
f = range_filter('test', start_date_math='startmath',
end_date_math='endmath')
assert f(['1821--1940']) == Range(test={
'gte': '1821||startmath', 'lte': '1940||endmath',
})
assert f(['>1821--']) == Range(test={'gt': '1821||startmath'})
assert f(['1821--<1940']) == Range(test={'gte': '1821||startmath',
'lt': '1940||endmath'})
assert pytest.raises(RESTValidationError, f, ['2016'])
assert pytest.raises(RESTValidationError, f, ['--'])
def seminars_start_date_range_filter(timezone):
date_range_filter = range_filter("start_datetime", time_zone=timezone)
def inner(values):
value = values and values[0]
if value == "upcoming":
now = datetime.utcnow().isoformat()
return Range(**{"start_datetime": {"gte": now}})
if value == "all":
return Q()
return date_range_filter(values)
return inner
def conferences_start_date_range_filter():
date_range_filter = range_filter("opening_date")
def inner(values):
value = values and values[0]
if value == "upcoming":
today = datetime.utcnow().strftime("%Y-%m-%d")
return Range(**{"opening_date": {"gte": today}})
if value == "all":
return Q()
return date_range_filter(values)
return inner
def test_range_filter():
"""Test range filter."""
f = range_filter('test',
start_date_math='startmath',
end_date_math='endmath')
assert f(['1821--1940']) == Range(test={
'gte': '1821||startmath',
'lte': '1940||endmath',
})
assert f(['>1821--']) == Range(test={'gt': '1821||startmath'})
assert f(['1821--<1940']) == Range(test={
'gte': '1821||startmath',
'lt': '1940||endmath'
})
assert pytest.raises(RESTValidationError, f, ['2016'])
assert pytest.raises(RESTValidationError, f, ['--'])
and_term_filter('organisation.pid'),
'language':
and_term_filter('language.value'),
'subject':
and_term_filter('facet_subjects'),
'specific_collection':
and_term_filter('specificCollections'),
'document_type':
and_term_filter('documentType'),
'controlled_affiliation':
and_term_filter('contribution.controlledAffiliation.raw'),
'author':
and_term_filter('contribution.agent.preferred_name.raw'),
'year':
range_filter('provisionActivity.startDate',
format='yyyy',
end_date_math='/y')
}),
'deposits':
dict(aggs=dict(
status=dict(terms=dict(field='status', size=DEFAULT_AGGREGATION_SIZE)),
user=dict(terms=dict(field='user.full_name.keyword',
size=DEFAULT_AGGREGATION_SIZE)),
contributor=dict(terms=dict(field='facet_contributors',
size=DEFAULT_AGGREGATION_SIZE))),
filters={
_('pid'): and_term_filter('pid'),
_('status'): and_term_filter('status'),
_('user'): and_term_filter('user.full_name.keyword'),
_('contributor'): and_term_filter('facet_contributors'),
}),
"embargoed access", "restricted access",
"metadata only access"
],
})),
_('resourceType'):
nested_filter('resourceType',
language_aware_text_terms_filter('resourceType.title')),
_('keywords'):
language_aware_text_terms_filter('keywords'),
_('subject'):
language_aware_text_terms_filter('subjectAll'),
_('language'):
nested_filter('language',
language_aware_text_terms_filter('language.title')),
_('date'):
range_filter('dateAll.date'),
_('dateIssued'):
range_filter('dateIssued.date'),
_('dateDefended'):
range_filter('dateDefended.date'),
_('dateModified'):
range_filter('dateModified.date'),
}
CURATOR_FILTERS = {
_('accessRightsCurator'):
nested_filter('accessRights',
language_aware_text_terms_filter('accessRights.title')),
_('rights'):
nested_filter('rights', language_aware_text_terms_filter('rights.title')),
_('provider'):
experiment=terms_filter('experiment.keyword'),
type=terms_filter('type.primary.keyword'),
subtype=terms_filter('type.secondary.keyword'),
year=terms_filter('date_created.keyword'),
tags=terms_filter('tags.keyword'),
keywords=terms_filter('keywords.keyword'),
collision_type=terms_filter('collision_information.type.keyword'),
collision_energy=terms_filter('collision_information.energy'
'.keyword'),
category=terms_filter('categories.primary.keyword'),
subcategory=terms_filter('categories.secondary.keyword'),
file_type=terms_filter('distribution.formats.keyword'),
collections=terms_filter('collections.keyword'),
availability=terms_filter('distribution.availability.keyword'),
signature=terms_filter('signature.keyword'),
event_number=range_filter('distribution.number_events')
)
}
}
"""Facets per index for the default facets factory."""
# Generated by scripts/get_facet_hierarchy.py
FACET_HIERARCHY = {
"availability": {},
"category": {
"B physics and Quarkonia": {"subcategory": set()},
"Exotica": {"subcategory": {"Miscellaneous", "Gravitons"}},
"Higgs Physics": {
"subcategory": {
"Beyond Standard Model",
'noquery': '-record_creation_date'
},
}
RECORDS_REST_FACETS = {
"records-record": {
"filters": {
"journal":
terms_filter("publication_info.journal_title"),
"country":
terms_filter_with_must("authors.affiliations.country"),
"collaboration":
terms_filter("facet_collaboration"),
"record_creation_date":
range_filter('record_creation_date',
format='yyyy',
end_date_math='/y')
},
"aggs": {
"journal": {
"terms": {
"field": "publication_info.journal_title",
"size": 20,
"order": {
"_term": "asc"
}
}
},
"country": {
"terms": {
"field": "authors.affiliations.country",
def inner(values):
range_values = [f'{v}-01--{v}-12' for v in values]
return range_filter(field, start_date_math, end_date_math, **kwargs)(range_values)
)
),
post_filters=dict(
identifier=terms_filter(
'other_standard_identifier.source_of_number_or_code'
),
language=terms_filter(
('language_code.language_code_of_text_sound_track_or'
'_separate_title')
),
affiliation=terms_filter(
'main_entry_personal_name.affiliation'
),
years=range_filter(
'_created',
format='yyyy',
end_date_math='/y'
),
)
)
)
# TODO: Remove once schema validation issue is fixed.
INDEXER_DEFAULT_INDEX = 'marc21-bibliographic-bd-v1.0.0'
INDEXER_DEFAULT_DOCTYPE = 'bd-v1.0.0'
# TODO: Remove me once the problem with email is solved in flask-security:
# https://github.com/mattupstate/flask-security/issues/685
SECURITY_EMAIL_SENDER = 'no-reply@localhost'
# OAI-PMH Server
document_type=dict(terms=dict(field='type', size=1000)),
author=dict(terms=dict(field='facet_authors', size=5)),
years=dict(date_histogram=dict(
field='publicationYear', interval='year', format='yyyy')),
),
# can be also post_filter
filters={
_('status'): terms_filter('itemslist._circulation.status'),
_('location'): terms_filter('itemslist.location_name'),
_('language'): terms_filter('languages.language'),
_('document_type'): terms_filter('type'),
_('author'): terms_filter('facet_authors')
},
post_filters={
_('years'):
range_filter('publicationYear', format='yyyy', end_date_math='/y')
}),
'patrons':
dict(aggs=dict(roles=dict(terms=dict(field='roles', size=100))),
filters={_('roles'): terms_filter('roles')})
}
# sort
RECORDS_REST_SORT_OPTIONS = {
'documents':
dict(
bestmatch=dict(fields=['_score'],
title='Best match',
default_order='asc'),
mostrecent=dict(fields=['-_created'],
title='Most recent',
order=2,
),
album=dict(
fields=['sort_album'],
title='Album',
default_order='asc',
order=3,
),
artist=dict(
fields=['facet_artist'],
title='Artist',
default_order='asc',
order=4,
)))
RECORDS_REST_DEFAULT_SORT = dict(records=dict(query='bestmatch',
noquery='album'), )
RECORDS_REST_FACETS = dict(records=dict(
aggs={
'mime':
dict(terms=dict(field='mime')),
'genre':
dict(terms=dict(field='genre', size=30)),
'years':
dict(date_histogram=dict(field='year', interval='year', format='yyyy'))
},
filters=dict(mime=terms_filter('mime'), genre=terms_filter('genre')),
post_filters=dict(years=range_filter(
'year', format='yyyy', end_date_math='/y'), )))
def hep_author_publications():
exclude_value = request.values.get('exclude_author_value', '', type=str)
return {
"filters": {
"author": must_match_all_filter('facet_author_name'),
"author_count": range_author_count_filter('author_count'),
"doc_type": must_match_all_filter('facet_inspire_doc_type'),
"earliest_date": range_filter(
'earliest_date',
format='yyyy',
end_date_math='/y')
},
"aggs": {
"earliest_date": {
"date_histogram": {
"field": "earliest_date",
"interval": "year",
"format": "yyyy",
"min_doc_count": 1,
},
"meta": {
"title": "Date",
"order": 1,
},
},
"author_count": {
"range": {
"field": "author_count",
"ranges": [
{
"key": "10 authors or less",
"from": 1,
"to": 11,
},
],
},
"meta": {
"title": "Number of authors",
"order": 2,
},
},
"author": {
"terms": {
"field": "facet_author_name",
"size": 20,
"exclude": exclude_value,
},
"meta": {
"title": "Collaborators",
"order": 3,
"split": True,
},
},
"doc_type": {
"terms": {
"field": "facet_inspire_doc_type",
"size": 20
},
"meta": {
"title": "Document Type",
"order": 4,
},
},
},
}
def hep_author_publications():
exclude_value = request.values.get('exclude_author_value', '', type=str)
return {
"filters": {
"author":
must_match_all_filter('facet_author_name'),
"author_count":
range_author_count_filter('author_count'),
"doc_type":
must_match_all_filter('facet_inspire_doc_type'),
"earliest_date":
range_filter('earliest_date', format='yyyy', end_date_math='/y')
},
"aggs": {
"earliest_date": {
"date_histogram": {
"field": "earliest_date",
"interval": "year",
"format": "yyyy",
"min_doc_count": 1,
},
"meta": {
"title": "Date",
"order": 1,
},
},
"author_count": {
"range": {
"field":
"author_count",
"ranges": [
{
"key": "10 authors or less",
"from": 1,
"to": 11,
},
],
},
"meta": {
"title": "Number of authors",
"order": 2,
},
},
"author": {
"terms": {
"field": "facet_author_name",
"size": 20,
"exclude": exclude_value,
},
"meta": {
"title": "Collaborators",
"order": 3,
"split": True,
},
},
"doc_type": {
"terms": {
"field": "facet_inspire_doc_type",
"size": 20
},
"meta": {
"title": "Document Type",
"order": 4,
},
},
},
}
#: Default sort for records REST API.
RECORDS_REST_DEFAULT_SORT = {
"records-record": {
'query': '-record_creation_date',
'noquery': '-record_creation_date'
},
}
RECORDS_REST_FACETS = {
"records-record": {
"filters": {
"journal": terms_filter("publication_info.journal_title"),
"country": terms_filter_with_must("authors.affiliations.country"),
"collaboration": terms_filter("facet_collaboration"),
"record_creation_date": range_filter(
'record_creation_date',
format='yyyy',
end_date_math='/y')
},
"aggs": {
"journal": {
"terms": {
"field": "publication_info.journal_title",
"size": 20,
"order": {"_term": "asc"}
}
},
"country": {
"terms": {
"field": "authors.affiliations.country",
"size": 150,
"order": {"_term": "asc"}
field='subject_added_entry_topical_term.'
'topical_term_or_geographic_name_entry_element')),
years=dict(date_histogram=dict(field='imprint.complete_date',
interval='year',
format='yyyy')),
),
post_filters=dict(
authors=terms_filter(
'added_entry_personal_name.personal_name'),
languages=terms_filter('language_code.language_code_of_text_'
'sound_track_or_separate_title'),
topic=terms_filter(
'subject_added_entry_topical_term.'
'topical_term_or_geographic_name_entry_element'),
years=range_filter('imprint.complete_date',
format='yyyy',
end_date_math='/y'),
))),
RECORDS_REST_SORT_OPTIONS=dict(
testrecords=dict(bestmatch=dict(
title='Best match',
fields=['-_score'],
default_order='asc',
order=1,
),
controlnumber=dict(
title='Control number',
fields=['control_number'],
default_order='desc',
order=2,
))),
years=dict(date_histogram=dict(
field='imprint.complete_date',
interval='year',
format='yyyy')),
),
post_filters=dict(
authors=terms_filter(
'added_entry_personal_name.personal_name'),
languages=terms_filter(
'language_code.language_code_of_text_'
'sound_track_or_separate_title'),
topic=terms_filter(
'subject_added_entry_topical_term.'
'topical_term_or_geographic_name_entry_element'),
years=range_filter(
'imprint.complete_date',
format='yyyy',
end_date_math='/y'),
)
)
),
RECORDS_REST_SORT_OPTIONS=dict(
testrecords=dict(
bestmatch=dict(
title='Best match',
fields=['-_score'],
default_order='asc',
order=1,
),
controlnumber=dict(
title='Control number',
fields=['control_number'],
#: Default sort for records REST API.
RECORDS_REST_DEFAULT_SORT = {
"scoap3-records-record": {
'query': '-date',
'noquery': '-date'
},
}
RECORDS_REST_FACETS = {
"scoap3-records-record": {
"filters": {
"journal": terms_filter("publication_info.journal_title"),
"country": terms_filter_with_must("authors.affiliations.country"),
"collaboration": terms_filter("facet_collaboration"),
"year": range_filter(
'year',
format='yyyy',
end_date_math='/y')
},
"aggs": {
"journal": {
"terms": {
"field": "publication_info.journal_title",
"size": 20,
"order": {"_term": "asc"}
}
},
"country": {
"terms": {
"field": "authors.affiliations.country",
"size": 150,
"order": {"_term": "asc"}
},
},
'years': {
'date_histogram': {
'field': 'date',
'interval': 'year',
'format': 'yyyy'
}
}
},
'filters': {
'keyword': terms_filter('keywords.name'),
'category': terms_filter('category.untouched'),
'type': terms_filter('type.untouched'),
'language': terms_filter('language'),
'years': range_filter('date', format='yyyy', end_date_math='/y'),
},
}
}
# Deposit search index.
DEPOSIT_UI_SEARCH_INDEX = 'deposits-records-videos-project'
# Options for sorting deposits.
DEPOSIT_REST_SORT_OPTIONS = {
'deposits-records-videos-project':
dict(bestmatch=dict(fields=['-_score'],
title='Best match',
default_order='asc',
order=1),
mostrecent=dict(
},
},
},
},
'post_filters': {
'type':
regex_filter('_type'),
'type_version':
terms_filter('_type'),
'cms_working_group':
prefix_filter('basic_info.cadi_id'),
'cadi_status':
terms_filter('cadi_info.status'),
'next_deadline_date':
range_filter('analysis_context.next_deadline_date',
format='yyyy',
end_date_math='/y'),
'collision_system':
terms_filter('basic_info.analysis_keywords.collision_system.keyword'),
'accelerator_parameters':
terms_filter(
'basic_info.analysis_keywords.accelerator_parameters.keyword'),
'physics_theme':
terms_filter('basic_info.analysis_keywords.physics_theme.keyword'),
'final_states':
terms_filter('basic_info.analysis_keywords.final_states.keyword'),
'sm_analysis_characteristics':
terms_filter(
'basic_info.analysis_keywords.sm_analysis_characteristics.keyword'
), # noqa
'interpretation':
"jobs_facets": JOBS_FACETS,
"journals": JOURNALS,
"experiments": EXPERIMENTS,
"conferences": CONFERENCES,
"conferences_facets": CONFERENCES_FACETS,
"data": DATA,
"institutions": INSTITUTIONS,
"seminars": SEMINARS,
"seminars_facets": SEMINARS_FACETS,
}
HEP_FILTERS = {
"author": must_match_all_filter("facet_author_name"),
"author_count": range_author_count_filter("author_count"),
"doc_type": must_match_all_filter("facet_inspire_doc_type"),
"earliest_date": range_filter("earliest_date", format="yyyy", end_date_math="/y"),
"citation_count": range_filter("citation_count"),
"citation_count_without_self_citations": range_filter(
"citation_count_without_self_citations"
),
"collaboration": must_match_all_filter("facet_collaborations"),
"refereed": must_match_all_filter("refereed"),
"citeable": must_match_all_filter("citeable"),
"collection": must_match_all_filter("_collections"),
"subject": must_match_all_filter("facet_inspire_categories"),
"arxiv_categories": must_match_all_filter("facet_arxiv_categories"),
"rpp": filter_from_filters_aggregation(hep_rpp(order=1)),
}
JOBS_FILTERS = {
"field_of_interest": terms_filter("arxiv_categories"),
op='lte'),
relation=enum_term_filter(label='relation',
field='RelationshipType',
choices={
'isCitedBy': 'Cites',
'isSupplementedBy': 'IsSupplementTo',
'isRelatedTo': 'IsRelatedTo'
}),
keyword=simple_query_string_filter('Source.Keywords_all'),
journal=nested_terms_filter('Source.Publisher.Name',
'Source.Publisher'),
),
post_filters=dict(
type=terms_filter('Source.Type.Name'),
publication_year=range_filter('Source.PublicationDate',
format='yyyy',
start_date_math='/y',
end_date_math='/y'),
)),
# The topHits agg can't be addded here due to limitations in elasticsearch_dsl aggs function so that is added in query.py
metadata=dict(
aggs=dict(
NumberOfTargets=dict(cardinality=dict(field='Target.ID')),
publication_year=dict(date_histogram=dict(
field='Source.PublicationDate',
interval='year',
format='yyyy',
), ),
),
filters=dict(
group_by=enum_term_filter(label='group_by',
"journals": JOURNALS,
"experiments": EXPERIMENTS,
"conferences": CONFERENCES,
"data": DATA,
"institutions": INSTITUTIONS,
}
HEP_COMMON_FILTERS = {
"author":
must_match_all_filter("facet_author_name"),
"author_count":
range_author_count_filter("author_count"),
"doc_type":
must_match_all_filter("facet_inspire_doc_type"),
"earliest_date":
range_filter("earliest_date", format="yyyy", end_date_math="/y"),
"citation_count":
range_filter("citation_count"),
"collaboration":
must_match_all_filter("facet_collaborations"),
"refereed":
must_match_all_filter("refereed"),
"citeable":
must_match_all_filter("citeable"),
}
HEP_FILTERS = {
"subject": must_match_all_filter("facet_inspire_categories"),
"arxiv_categories": must_match_all_filter("facet_arxiv_categories"),
}
#: Default sort for records REST API.
RECORDS_REST_DEFAULT_SORT = {
"scoap3-records-record": {
"query": "-date",
"noquery": "-date"
},
}
RECORDS_REST_FACETS = {
"scoap3-records-record": {
"filters": {
"journal": terms_filter("publication_info.journal_title"),
"country": terms_filter_with_must("authors.affiliations.country"),
"collaboration": terms_filter("facet_collaboration"),
"year": range_filter("year", format="yyyy", end_date_math="/y"),
},
"aggs": {
"journal": {
"terms": {
"field": "publication_info.journal_title",
"size": 20,
"order": {
"_term": "asc"
},
}
},
"country": {
"terms": {
"field": "authors.affiliations.country",
"size": 150,
