예제 #1
0
try:
    with open('timeBasedIndexNewTest.txt', 'rb') as f3:
        timeBasedIndex = json.load(f3)
except Exception as e:
    print(e.with_traceback)
    try:
        with open('timeBasedIndexNewFixed.txt', 'rb') as f3:
            timeBasedIndex = json.load(f3)
    except Exception:
        print('file does not exist time based')
        timeBasedIndex = {}

es = Elasticsearch()
s = Search(using=es, index='trec-news-index', doc_type='_doc')
s = s.script_fields(
)  # only get ids, otherwise `fields` takes a list of field names
reg = re.compile("[a-z]{4,}")
e1 = {}
ids = []

baseYear = 2017
#previous_count = 0
counter = 0
faultyIds = []


def getTermWeights(doc):
    termWeightsDict = {}
    global actions
    previous_count = 0
    print(f"doc here is {doc['_id']}")