continue
# Except for % character.
if '%' in search_query:
continue
# Tuning, Add "" to both side on query.
original_search_query = search_query
search_query = '"%s"' % search_query
print('search_query:', search_query)
# Hgnc response
elapsed_millis = get_current_millis()
while True:
try:
response = hgnc_http.request('/search/' + quote(search_query),
'GET', '', '')['response']
except:
print('HGNC request failed, so retry after 5 seconds')
time.sleep(5)
continue
else:
print('HGNC request time:',
get_elapsed_seconds(get_current_millis(), elapsed_millis))
break
# Ignore empty docs, score less than current saved gene.
if not response['docs']:
# Cache gene result information.
query_result_map[original_search_query] = {
'hgnc_id': None,
'symbol': None,
from urllib.parse import quote
import sys
START_S_ID = None
if len(sys.argv) > 1:
START_S_ID = sys.argv[1]
hgnc_search_uri = 'http://rest.genenames.org'
db_uri = 'http://e3cb0988.ngrok.io'
# Get all preprocessed MeSH terms in DB.
db_http = HttpWrapper(db_uri)
all_processed = db_http.request(
'/sql',
'GET',
{
'query': 'SELECT * FROM LUNG_PROCESSED WHERE S_ID = 28',
},
'',
)
# Search using P_NAME from processed rows.
# S_ID and HGNC Map.
# Key: S_ID
# Value: HGNC doc max score
sid_hgnc_map = dict()
hgnc_http = HttpWrapper(hgnc_search_uri)
def get_max_score_doc(result_docs):
get_elapsed_seconds(get_current_millis(), elapsed_millis))
# Get MeSH query
mesh_name = None
max_doc = None
if len(processeds) == 1:
mesh_name = processeds[0]['P_NAME']
else:
for processed in processeds:
elapsed_millis = get_current_millis()
response = None
while True:
try:
# Hgnc response
response = hgnc_http.request(
'/search/' + quote(processed['P_NAME']), 'GET', '',
'')['response']
except:
print('HGNC request failed, so retry after 5 seconds')
time.sleep(5)
continue
else:
print(
'HGNC response time:',
get_elapsed_seconds(get_current_millis(),
elapsed_millis))
break
if not response['docs']:
continue
if not max_doc or max_doc['score'] < response['maxScore']:
print('Found max doc')