def test_sota_sampler(self):
local_file = os.path.abspath(os.path.dirname(__file__) + os.path.sep + "../../") + '/pkl_example/yelp_3000.pkl'
localdata = LocalData(local_file, 'pkl', "business_id", ["name"], ["name", "full_address"])
localdata_ids, localdata_query, localdata_er = localdata.getlocalData()
initQueries = utils.queryGene(localdata_query, 2)
sampler.sota_sampler(query_pool=initQueries, api=self.yelp, match_term=localdata.getQueryList(), top_k=300,
adjustment=1, samplenum=1)
self.yelp.getSession().close()
assert True
def setUp(self):
search_term = 'q'
parameters = {'h': 1000}
self.dblp = PublApi(delay=5, search_term=search_term, **parameters)
localdata_file = os.path.abspath(
os.path.dirname(__file__) + os.path.sep +
"../../") + '/example/dblp_10000.pkl'
localdata = LocalData(localdata_file, 'pkl', "row['key']",
["row['title']"], ["row['title']"])
localdata_ids, localdata_query, localdata_er = localdata.getlocalData()
initQueries = utils.queryGene(localdata_query, 2)
self.initQueries = initQueries
class LocaldataTestCase(unittest.TestCase):
def setUp(self):
data_file = os.path.abspath(os.path.dirname(__file__) + os.path.sep + "../../") + '/yelp_example/yelp_3000_AZ.csv'
self.localdata = LocalData(data_file, 'csv', "business_id", ["name"],
["name", "full_address"])
def tearDown(self):
self.localdata = None
def test_loadLocalData(self):
self.localdata.setlocalData(None, None, None)
self.localdata.read_csv()
localdata_ids, localdata_query, localdata_er = self.localdata.getlocalData()
assert len(localdata_ids) == 3000
def setUp(self):
search_term = 'q'
parameters = {'h': 1000}
self.dblp = PublApi(top_k=1000,
delay=5,
search_term=search_term,
**parameters)
localdata_file = os.path.abspath(
os.path.dirname(__file__) + os.path.sep +
"../../") + '/csv_example/dblp_sample.csv'
localdata = LocalData(localdata_file, 'csv', "key", ["title"],
["title"])
localdata_ids, localdata_query, localdata_er = localdata.getlocalData()
initQueries = utils.queryGene(localdata_query, 2)
self.initQueries = initQueries
class LocaldataTestCase(unittest.TestCase):
def setUp(self):
data_file = os.path.abspath(
os.path.dirname(__file__) + os.path.sep +
"../../") + '/example/yelp_3000.pkl'
self.localdata = LocalData(data_file, 'pkl', "row['business_id']",
["row['name']"],
["row['name']", "row['full_address']"])
def tearDown(self):
self.localdata = None
def test_loadLocalData(self):
self.localdata.setlocalData(None, None, None)
self.localdata.read_pickle()
localdata_ids, localdata_query, localdata_er = self.localdata.getlocalData(
)
assert len(localdata_ids) == 3000
from deeperlib.api.yelp.searchapi import SearchApi
from deeperlib.core import utils
from deeperlib.data_processing.local_data import LocalData
from deeperlib.estimator import sampler
# ==== Sota-Sampler Yelp ====
client_id = "kCe2YbZePXsPnC204ZrXoQ"
client_secret = "s9KnvEEQW7jaA2wlrBi4X2fnDQ0F7asdklXVvJUidWp8i50ov24E8EjkHX2AUhoL"
search_term = 'term'
parameters = {'limit': 50, 'location': 'AZ'}
yelp = SearchApi(client_id=client_id,
client_secret=client_secret,
top_k=300,
delay=5,
search_term=search_term,
**parameters)
local_file = 'yelp_3000.pkl'
localdata = LocalData(local_file, 'pkl', "row['business_id']", ["row['name']"],
["row['name']", "row['full_address']"])
localdata_ids, localdata_query, localdata_er = localdata.getlocalData()
initQueries = utils.queryGene(localdata_query, 2)
sampler.sota_sampler(query_pool=initQueries,
api=yelp,
match_term=localdata.getQueryList(),
top_k=300,
adjustment=1)
yelp.getSession().close()