def AdvancedSearchForm(*args, **kwargs):
search_url, store = kwargs.pop('search_url'), kwargs.pop('store')
q = {'size': 0,
'facets': {'formalOrganisation': {'terms': {'field': 'formalOrganisation.uri'}},
'basedNear': {'terms': {'field': 'basedNear.uri'}}}}
results = json.load(urllib2.urlopen(search_url, json.dumps(q)))
formal_organisation_choices = [t['term'] for t in results['facets']['formalOrganisation']['terms']]
based_near_choices = [t['term'] for t in results['facets']['basedNear']['terms']]
labels = get_labels(map(rdflib.URIRef, formal_organisation_choices + based_near_choices),
endpoint=store.query_endpoint)
formal_organisation_choices = [('', '-'*20)]+[(uri, labels.get(rdflib.URIRef(uri), uri)) for uri in formal_organisation_choices]
based_near_choices = [('', '-'*20)]+[(uri, labels.get(rdflib.URIRef(uri), uri)) for uri in based_near_choices]
attrs = {'q': forms.CharField(label='Query'),
'filter.basedNear.uri': forms.ChoiceField(label='Location',
choices=based_near_choices,
initial='',
required=False),
'filter.formalOrganisation.uri': forms.ChoiceField(label='Institution',
choices=formal_organisation_choices,
initial='http://oxpoints.oucs.ox.ac.uk/id/00000000',
required=False)}
form = type('AdvancedSearchForm', (forms.Form,), attrs)
return form(*args, **kwargs)
def get_results(self, parameters, cleaned_data):
page = cleaned_data.get('page') or 1
page_size = cleaned_data.get('page_size') or self.page_size
start = (page - 1) * page_size
query = self.get_query(parameters, cleaned_data, start, page_size)
# If there aren't any filters defined, we don't want a filter part of
# our query.
if 'filter' in query:
if 'and' in query['filter'] and not query['filter']['and']:
del query['filter']['and']
if not query['filter']:
del query['filter']
results = self.search_endpoint.query(query)
results.update(self.get_pagination(page_size, page, start, results))
results['q'] = cleaned_data['q']
facet_labels = set()
for key in query['facets']:
meta = results['facets'][key]['meta'] = query['facets'][key]
filter_value = parameters.get(
'filter.%s' % query['facets'][key]['terms']['field'])
results['facets'][key]['filter'] = {
'present': filter_value is not None,
'value': filter_value
}
if meta['terms']['field'].endswith('.uri'):
for term in results['facets'][key]['terms']:
facet_labels.add(term['term'])
term['value'] = contract(term['term'])
else:
for term in results['facets'][key]['terms']:
term['value'] = term['term']
labels = get_labels(map(rdflib.URIRef, facet_labels),
endpoint=self.endpoint)
for key in query['facets']:
if results['facets'][key]['meta']['terms']['field'].endswith(
'.uri'):
for term in results['facets'][key]['terms']:
uri = rdflib.URIRef(term['term'])
if uri in labels:
term['label'] = unicode(labels[uri])
for hit in results['hits']['hits']:
try:
hit['_url'] = doc_forwards(hit['_source']['uri'])[None]
except KeyError:
raise
return results
def with_labels(self, triples):
subjects = set()
already_labelled = set()
for s, p, o in triples:
yield s, p, o
if p in label_predicates:
already_labelled.add(s)
if isinstance(s, rdflib.URIRef):
subjects.add(s)
if isinstance(o, rdflib.URIRef):
subjects.add(o)
for triple in get_labels(subjects - already_labelled, self.endpoint, mapping=False):
yield triple
def get_results(self, parameters, cleaned_data):
page = cleaned_data.get('page') or 1
page_size = cleaned_data.get('page_size') or self.page_size
start = (page - 1) * page_size
query = self.get_query(parameters, cleaned_data, start, page_size)
# If there aren't any filters defined, we don't want a filter part of
# our query.
if 'filter' in query:
if 'and' in query['filter'] and not query['filter']['and']:
del query['filter']['and']
if not query['filter']:
del query['filter']
results = self.search_endpoint.query(query)
results.update(self.get_pagination(page_size, page, start, results))
results['q'] = cleaned_data['q']
facet_labels = set()
for key in query['facets']:
meta = results['facets'][key]['meta'] = query['facets'][key]
filter_value = parameters.get('filter.%s' % query['facets'][key]['terms']['field'])
results['facets'][key]['filter'] = {'present': filter_value is not None,
'value': filter_value}
if meta['terms']['field'].endswith('.uri'):
for term in results['facets'][key]['terms']:
facet_labels.add(term['term'])
term['value'] = contract(term['term'])
else:
for term in results['facets'][key]['terms']:
term['value'] = term['term']
labels = get_labels(map(rdflib.URIRef, facet_labels), endpoint=self.endpoint)
for key in query['facets']:
if results['facets'][key]['meta']['terms']['field'].endswith('.uri'):
for term in results['facets'][key]['terms']:
uri = rdflib.URIRef(term['term'])
if uri in labels:
term['label'] = unicode(labels[uri])
for hit in results['hits']['hits']:
try:
hit['_url'] = doc_forwards(hit['_source']['uri'])[None]
except KeyError:
raise
return results
def get_results(self, parameters, cleaned_data):
page = cleaned_data.get('page') or 1
page_size = cleaned_data.get('page_size') or self.page_size
start = (page - 1) * page_size
query = {
'query': {'query_string': {'query': cleaned_data['q'],
'default_operator': 'AND'}},
'from': start,
'size': page_size,
# A blank conjunctive filter. We'll remove this later if necessary.
'filter': {'and': []},
}
# Parse query parameters of the form 'filter.FIELDNAME'.
for key in list(parameters):
parameter = parameters[key]
if key.startswith('filter.'):
if not parameter:
del parameters[key]
continue
elif parameter == '-':
filter = {'missing': {'field': key[7:]}}
else:
if key.endswith('.uri') and ':' in parameter:
parameter = expand(parameter)
filter = {'term': {key[7:]: parameter}}
query['filter']['and'].append(filter)
# If there aren't any filters defined, we don't want a filter part of
# our query.
if not query['filter']['and']:
del query['filter']['and']
if not query['filter']:
del query['filter']
if self.facets:
# Copy the facet definitions as we'll be playing with them shortly.
facets = copy.deepcopy(self.facets)
# Add facet filters for all active filters except any acting on this
# particular facet.
if 'filter' in query:
for facet in facets.itervalues():
for filter in query['filter']['and']:
if facet['terms']['field'] not in (filter.get('term') or filter['missing']):
if 'facet_filter' not in facet:
facet['facet_filter'] = {'and': []}
facet['facet_filter']['and'].append(filter)
query['facets'] = facets
response = urllib2.urlopen(self.search_url, json.dumps(query))
results = self.Deunderscorer(json.load(response))
results.update(self.get_pagination(page_size, page, start, results))
results['q'] = cleaned_data['q']
facet_labels = set()
for key in query['facets']:
meta = results['facets'][key]['meta'] = query['facets'][key]
filter_value = parameters.get('filter.%s' % query['facets'][key]['terms']['field'])
results['facets'][key]['filter'] = {'present': filter_value is not None,
'value': filter_value}
if meta['terms']['field'].endswith('.uri'):
for term in results['facets'][key]['terms']:
facet_labels.add(term['term'])
term['value'] = contract(term['term'])
else:
for term in results['facets'][key]['terms']:
term['value'] = term['term']
labels = get_labels(facet_labels, endpoint=self.endpoint)
for key in query['facets']:
if results['facets'][key]['meta']['terms']['field'].endswith('.uri'):
for term in results['facets'][key]['terms']:
uri = URIRef(term['term'])
if uri in labels:
term['label'] = unicode(labels[uri])
for hit in results['hits']['hits']:
try:
hit['_url'] = doc_forwards(hit['_source']['uri'])[None]
except KeyError:
raise
return results
