def brand():
'''return products meet the industry& brand'''
if request.method=='GET':
idsty = dget(mapping, '%s.data'%industry)
idsty_sort = dget(mapping, '%s.sort'%industry)
brand_name = request.args.get('brand', '')
products_db = db.webpages.find({'transformed.data.brand_name': brand_name})
#for easy to get, write back will be a mess .. so we need a map
products_key = list(idsty.keys())
products = [
[dget(p, idsty[k], '') for k, v in idsty.items()]
for p in products_db
]
#sort the data, to the convinent of displaying..
for fs in idsty_sort:
try:
products = sorted(products, key=lambda x: x[products_key.index(fs)])
except:
print(fs)
return render_template('row_template.html',
products = products,
titles = list(idsty.keys()))
def get_value(data):
"""Gets Python type value from WikiData response field"""
data_type = dget(data, 'type.value')
value = dget(data, 'valLabel.value')
if data_type == WikiDataAnswer.TIME_VALUE:
dt = parser.parse(value)
dt = dt.replace(tzinfo=None)
return dt
elif data_type == WikiDataAnswer.QUANTITY_VALUE:
if value.isdigit():
return float(value)
return value
else:
return value
def _get_property(self, subject, prop, prop_id=None):
"""Queries Wikidata to get property"""
self.debug('{0}, {1}', subject, prop)
subject_id = self._get_id(subject, 'item')
if not prop_id:
prop_id = self._get_id(prop, 'property')
if not prop_id or not subject_id:
return None
query = """
SELECT ?valLabel ?type
WHERE {
"""
sub_queries = []
for pid in prop_id.split(','):
sub_query = """{
wd:%s p:%s ?prop .
?prop ps:%s ?val .
OPTIONAL {
?prop psv:%s ?propVal .
?propVal rdf:type ?type .
}
}""" % (subject_id, pid, pid, pid)
sub_queries.append(sub_query)
query += ' UNION '.join(sub_queries)
query += """
SERVICE wikibase:label { bd:serviceParam wikibase:language "en"}
}
"""
result = self._query_wdsparql(query)
bindings = dget(result, 'results.bindings')
return WikiDataAnswer(sparql_query=query, bindings=bindings)
def _get_aliases(self, subject):
"""Get all aliases of an entity"""
self.debug('Get alias {0}'.format(subject))
subject_id = self._get_id(subject, 'item')
query = """
SELECT ?valLabel
WHERE {
{ wd:%s skos:altLabel ?val FILTER (LANG (?val) = "en") }
UNION
{ wd:%s rdfs:label ?val FILTER (LANG (?val) = "en") }
SERVICE wikibase:label { bd:serviceParam wikibase:language "en"}
}""" % (subject_id, subject_id)
result = self._query_wdsparql(query)
bindings = dget(result, 'results.bindings')
return WikiDataAnswer(sparql_query=query, bindings=bindings)
def _get_id(self, name, _type='item'):
"""Get WikiData ID of a name"""
item = self._search_entity(name, _type)
return dget(item, 'search.0.id')
def _get_desc(self, subject):
"""Get WikiData description of subject"""
data = self._search_entity(subject)
desc = dget(data, 'search.0.description')
return Answer(data=desc)
def _find_entity(self, qtype, inst, params):
"""Count number of things instance/subclass of inst with props"""
self.info('Get instances of {0} that are {1}'.format(inst, params))
inst_id = self._get_id(inst)
if not inst_id:
self.info('Cannot find id of: {0}'.format(inst))
return None
if qtype == 'how many':
select = '(count(*) as ?count)'
elif qtype in ['which', 'who']:
select = '?valLabel'
else:
self.warn('Qtype {0} not known'.format(qtype))
return None
query = """
SELECT %s
WHERE {
{ ?val p:P39 ?pos . # position held
?pos ps:P39 wd:%s . # pos = inst
?val wdt:P31 wd:Q5 . # as a human
} UNION
{
?val wdt:P31 wd:%s . # instance of
}
""" % (select, inst_id, inst_id)
for prop, prop_val, op in params:
if op in ['>', '<']:
prop_id = self._get_id(prop, 'property')
self.info('Count number of {0} where {1} {2} {3}'.format(
inst, prop_id, op, prop_val))
query += """
?val wdt:%s ?value FILTER(?value %s %s) . # Filter by value
""" % (prop_id, op, prop_val)
elif op in ['in', 'by', 'of', 'from']:
if op == 'in' and prop_val.isdigit():
iso_time = parser.parse(prop_val).isoformat()
query += """
?pos pq:P580 ?startDate . # pos.startDate
?pos pq:P582 ?endDate . # pos.endDate
FILTER (?startDate < "%s"^^xsd:dateTime && ?endDate > "%s"^^xsd:dateTime)
""" % (iso_time, iso_time)
elif op == 'of' and prop_val:
prop_val_id = self._get_id(prop_val)
query += """
?pos pq:P108 wd:%s . # pos.employer
""" % (prop_val_id)
else:
# Get value entity
prop_val_id = self._get_id(prop_val)
if prop:
# Get property id
if prop in ['died', 'killed'
] and op in ['from', 'by', 'of']:
# slight hack because lookup for died defaults to place of death
# cause of death
prop_id = 'P509'
else:
prop_id = self._get_id(prop, 'property')
query += '?val wdt:%s wd:%s .\n' % (prop_id,
prop_val_id)
else:
# Infer property from value (e.g. How many countries are in China?)
# e.g. infer: How many countries with continent China?
prop_id = '*'
query += """
wd:%s wdt:P31 ?instance . # Get entities that value is an instance of. Ex: ?instance = wd:Q5107 (continent)
?instance wdt:P1687 ?propEntity . # instance of Entity to property. Ex: ?propEntity = wd:P30 (continent)
?propEntity wikibase:directClaim ?prop . # wd to wdt. Ex: ?prop = wdt:P30 (continent)
?val ?prop wd:%s .
""" % (prop_val_id, prop_val_id)
self.info('Count number of {0} where {1}={2}'.format(
inst, prop_id, prop_val_id))
query += 'SERVICE wikibase:label { bd:serviceParam wikibase:language "en"} }'
result = {
'sparql_query': query,
}
try:
data = self._query_wdsparql(query)
except ValueError:
self.error('Error parsing data')
return WikiDataAnswer(**result)
if qtype == 'how many':
result['data'] = dget(data, 'results.bindings.0.count.value')
elif qtype in ['which', 'who']:
result['bindings'] = dget(data, 'results.bindings')
return WikiDataAnswer(**result)
def _get_label(self, subject):
"""Get WikiData label of subject"""
data = self._search_entity(subject)
label = dget(data, 'search.0.label')
return Answer(data=label)
