def ingest_to_index():
"""
Ingests all products into the index. Does some preprocessing before hand.
:return:
"""
es = get_connection_from_env()
with open("../../meta_Clothing_Shoes_and_Jewelry.json") as myfile:
while next(myfile):
head = [eval(next(myfile)) for x in range(50000)]
parsed = [
dict(_preprocess_product(pr), **{"_index": INDEX})
for pr in head if pr.get("description")
]
print(bulk(es, parsed))
:return:
"""
es = get_connection_from_env()
with open("../../meta_Clothing_Shoes_and_Jewelry.json") as myfile:
while next(myfile):
head = [eval(next(myfile)) for x in range(50000)]
parsed = [
dict(_preprocess_product(pr), **{"_index": INDEX})
for pr in head if pr.get("description")
]
print(bulk(es, parsed))
if __name__ == '__main__':
es = get_connection_from_env()
delete_index_by_name(es_connection=es, index=INDEX)
ingest_to_index()
print("Done")
# def store_old_brand_value():
# """
# Creates a new field for all articles named "old_brand" containing the old value of the "brand" field.
# """
# es = get_connection_from_env()
# n_results = 1
#
# while n_results > 0:
# results = execute_query(es_connection=es,
# query=query,
# index=INDEX)