def es_delete():
req_json = request.get_json()
print 'post req params : %s' % str(req_json)
index_name = req_json.get('index_name') or 'product_name_index_smart'
sku = req_json.get("sku")
not_required = req_params_required(['index_name', 'sku'], req_json)
if not_required:
return res_failure('es delete failure, because %s is necessary' % not_required)
else:
data = delete_one(index_name, 'fulltext', sku)
return res_success(data, 'es delete success')
def es_search():
req_json = request.get_json()
print 'post req params : %s' % str(req_json)
index_name = req_json.get('index_name') or 'product_name_index_smart'
keyword = req_json.get('keyword')
page_no = int(req_json.get('page_no')) or 0
page_size = int(req_json.get('page_size')) or 10000
sort_by = req_json.get('sort_by')
desc = int(req_json.get('desc')) or 1
not_required = req_params_required(['index_name', 'keyword'], req_json)
if not_required:
return res_failure('es search failure, because %s is necessary' % not_required)
else:
data = collapse_search(index_name, "fulltext", keyword, page_size, page_no, sort_by, desc)
return res_success(data, 'es search success')
def es_update():
req_json = request.get_json()
logger.info('Params:' + json.dumps(req_json))
print 'post req params : %s' % str(req_json)
index_name = req_json.get('index_name') or 'product_name_index_smart'
sku = req_json.get('sku')
spu = req_json.get('spu') or "TmpSKU" + sku
product_name = req_json.get("product_name")
product_no = req_json.get("product_no")
ts_score = req_json.get("ts_score")
price = req_json.get("price")
sales_volume = req_json.get("sales_volume")
data_dict = {"sku": sku, "spu": spu, "product_name": product_name, "product_no": product_no, "ts_score": ts_score}
not_required = req_params_required(['index_name', 'sku', 'product_name', 'product_no', 'price', 'sales_volume'],
req_json)
if not_required:
return res_failure('es update failure, because %s is necessary' % not_required)
else:
data = update_one(index_name, 'fulltext', sku, data_dict)
return res_success(data, 'es update success')
def es_update_index():
req_json = request.get_json()
logger.info('Params:' + json.dumps(req_json))
print 'post req params : %s' % str(req_json)
index_name = req_json.get('index_name') or 'product_name_index_smart'
not_required = req_params_required(['index_name'], req_json)
if not_required:
return res_failure('es update index failure, because %s is necessary' % not_required)
else:
es_domain = 'http://localhost:9200/'
es = Elasticsearch(hosts=[es_domain], timeout=5000)
db_name = index_name
table_name = 'fulltext'
create_table(db_name, table_name, [
['product_name', 'text'],
['product_no', 'keyword'],
['ts_score', 'keyword'],
['sku', 'keyword'],
['spu', 'keyword'],
])
count = batch_insert_es(es, db_name, table_name)
return res_success(count, 'es update index success')
def collapse_search(index_name, type_name, keyword, result_max_size, result_from, sort_by, desc):
data_dict = {
"query": {"match": {"product_name": keyword}},
"size": result_max_size,
"highlight": {
"pre_tags": ["<tag1>", "<tag2>"],
"post_tags": ["</tag1>", "</tag2>"],
"fields": {
"product_name": {},
"sku": {},
"spu": {},
"product_no": {},
"ts_score": {},
"sales_volume": {},
"price": {}
}
},
"collapse": {
"field": "spu" # 折叠去重操作 ，根据spu ,每一种spu只查询评分最高的一条数据 参考资料：https://elasticsearch.cn/article/132
},
"from": result_from
}
response = requests.post("%s%s/%s/_search" % (es_domain, index_name, type_name), json=data_dict)
spu_dict = set()
if response.status_code == 200 and 'error' not in response.json():
res_obj = response.json()
if res_obj['hits'] and res_obj['hits']['hits']:
for item in res_obj['hits']['hits']:
if sort_by not in item['_source'] or item['_source'][sort_by] is None:
item['_source'][sort_by] = 0.0
else:
try:
item['_source'][sort_by] = float(item['_source'][sort_by])
except Exception as e:
raise e
def _get_cmp_func(_sort_by, _desc):
def _cmp(x, y):
if x['_source'][_sort_by] < y['_source'][_sort_by]:
return 1 if _desc else -1
elif x['_source'][_sort_by] == y['_source'][_sort_by]:
return 0
else:
return -1 if _desc else 1
return _cmp
res_obj['hits']['hits'].sort(_get_cmp_func(_sort_by=sort_by, _desc=int(desc)))
ans_data = list()
for item in res_obj['hits']['hits']:
spu = item['_source']['spu']
if spu not in spu_dict:
spu_dict.add(spu)
tmp_dict = item['_source']
tmp_dict['_score'] = item['_score']
ans_data.append(tmp_dict)
page_no = int(result_from)
page_size = int(result_max_size)
return json.dumps(res_success(ans_data[page_no * page_size:(page_no + 1) * page_size], "es query success"))
return json.dumps(res_failure("es query failure"))