def get_documents(params, size, index, scroll_id=None):
queryObj = Query(params)
if not scroll_id:
es_uri = "/" + index + "/doc/_search?scroll=1d"
request = {"size": size, "query": {"bool": {}}}
else:
es_uri = "/_search/scroll"
request = {"scroll": "1d", "scroll_id": scroll_id}
if "query" in request:
filter = []
# 프로젝트 시퀀스 포함
filter.append(queryObj.get_project_seq_query())
filter.append(queryObj.get_project_filter_query(params['project_seq']))
# 대상 채널
if "channels" in params and params[
"channels"] and params["channels"] != 'all':
filter.append(queryObj.get_channel_query())
# 대상 기간
if "start_date" in params and "end_date" in params:
filter.append(queryObj.get_period_query())
request["query"]["bool"]["filter"] = filter
# 데이터셋의 포함 키워드
if "datasets" in params and params["datasets"]:
request["query"]["bool"]["must"] = queryObj.get_dataset_query(
params['project_seq'], params["datasets"])
logger.debug("[get_documents] Query >>> %s " % json.dumps(request))
es_conn = hc.HTTPConnection(es_ip, es_port, timeout=60)
es_conn.request("POST", es_uri, json.dumps(request),
{"Content-type": "application/json"})
result = es_conn.getresponse().read()
if 'hits' in json.loads(result):
logger.debug("[get_documents] result['hits']['total'] >>> %d" %
int(json.loads(result)['hits']['total']))
else:
logger.debug("[get_documents] result ::: " + str(result))
return json.loads(result)
def get_documents(params, size, index, scroll_id=None):
queryObj = Query(params)
if not scroll_id:
es_uri = "/" + index + "/doc/_search?scroll=1d"
request = {"size": size, "query": {"bool": {"must": []}}}
else:
es_uri = "/_search/scroll"
request = {"scroll": "1d", "scroll_id": scroll_id}
must = []
# 프로젝트 시퀀스 포함
must.append(get_project_seq_query(params))
# 대상 채널
if "channels" in params and params[
"channels"] and params["channels"] != 'all':
must.append(get_channel_query(params))
# 대상 기간
if "start_date" in params and "end_date" in params:
must.append(get_period_query(params))
# 데이터셋의 포함 키워드
if "datasets" in params and params["datasets"]: # 신라면,삼양라면,안성탕면
if len(params["datasets"].split("^")) > 1:
should = []
for dataset in params["datasets"].split("^"):
should.append(
queryObj.get_dataset_query(params['project_seq'], dataset))
must.append({"bool": {"should": should}})
else:
must.append(
queryObj.get_dataset_query(params['project_seq'],
params["datasets"]))
# elif params["type_cd"] == "CCT002": # 소셜모니터링-문서통계
# elif params["type_cd"] == "CCT003": # 소셜모니터링-감성분석
# .....
# 코드별로 request 필요한 형태로 변경해서 추가
if "query" in request:
request["query"]["bool"]["must"] = must
logger.debug("get_documents() ==> request : ")
for k, v in request.items():
logger.debug("\t{} : {}".format(k, v))
es_conn = hc.HTTPConnection(es_ip, es_port, timeout=60)
es_conn.request("POST", es_uri, json.dumps(request),
{"Content-type": "application/json"})
result = es_conn.getresponse().read()
if 'hits' in json.loads(result):
logger.debug("[get_documents] result['hits']['total'] >>> %d" %
int(json.loads(result)['hits']['total']))
else:
logger.debug("[get_documents] result ::: " + str(result))
return json.loads(result)