def __init__(self,
hbase_meta_path,
pro_path,
see_path,
ehost='localhost',
eport=8080,
hhost='localhost',
hport=9090):
self.jso = handEl()
self.hmeta = self.jso.loadJson(hbase_meta_path, ex=True)
self.smeta = self.jso.loadJson(os.path.join(see_path,
"seeConfig.json"),
ex=True)
self.pmeta = self.jso.loadJson(os.path.join(pro_path,
"proConfig.json"),
ex=True)
self.he = handEl(ehost, eport)
self.hh = handEl(hhost, hport)
self.ps = proSweeper(ehost, eport, None, pro_path)
self.tabref = self.hmeta['TABLE']['REF']
self.fields = [
self.hmeta['TABLE']['COLUMN']['FAMILIES'][raw_cf]
for raw_cf in self.hmeta['TABLE']['COLUMN']['RAW_FAM']
]
self.hbdict = {
self.hmeta['TABLE']['COLUMN']['FAMILIES'][raw_cf]: raw_cf
for raw_cf in self.hmeta['TABLE']['COLUMN']['RAW_FAM']
}
def main() :
he = handEl(
host = ELASTICSEARCH_HOST,
port = ELASTICSEARCH_PORT,
index = ELASTICSEARCH_INDEX)
# https://search.shopping.naver.com/detail/lite.nhn?nvMid=19318896974&NaPm=ct%3Dkdy2mauo%7Cci%3D987c95070ec54eb8d3d777d1331f6c14d51d4e82%7Ctr%3Dslsl%7Csn%3D95694%7Chk%3Deaecc241a834a012c2f8a88267c0fb6804fd640d
texts = [
'오늘 노란색 옷을 입었는 데 빨간색 여성 상의를 보여줘',
'어플로 찍은거라 실제 색감과 다르게 나왔으니 핏만 봐주세요',
'사진에서는 라벤더컬러가 약간 더 보랏빛돌고 생기있어보이는데 실제로는 차분하고 아주 은은한 라벤더컬러라 린넨바지랑 같이 입기 좋을것 같아요',
'소재가 엄청 까끌한건 아닌데 누워있으면 자잘하게 자국나는 그런 니트소재에요!',
'두께가 얇아서 상의 속옷 반쯤 비침있는데 확 드러나는건 아니라 그냥 입고 다녀도 될듯해요'
]
user_dict = {
'어플':'앱'
}
try :
tinx = 0
text = texts[tinx]
for udkey in user_dict :
if udkey in text : text = text.replace(udkey, user_dict[udkey])
results = analyze(he, text)
for result in results :
if 'E(Verbal endings)' in result['rightPOS'] :
print("{0}({1}) ".format(result['token'], 'E'))
else :
print("{0}({1}) ".format(result['token'], result['rightPOS'][:result['rightPOS'].index("(")]), end="")
except KeyboardInterrupt as ki :
print("Detect ctrl+c, Bye!")
def __init__(self, host, port, cpath, CONF_PATH):
self.he = handEl(host, port)
self.cpath = cpath
self.table = self.loadJson(cpath) if cpath else None
self.config = self.loadJson(os.path.join(CONF_PATH,
'proConfig.json'))['SWEEPER']
self.dirty_indices = list()
self.clean_items = list()
self.statime = datetime.now()
self.endtime = datetime.now()
#%%
from handEl import handEl
from pprint import pprint as pp
import numpy as np
ELASTICSEARCH_HOST = "192.168.1.132"
ELASTICSEARCH_PORT = "8080"
ELASTICSEARCH_INDEX = "nori_analyzer"
he = handEl(host=ELASTICSEARCH_HOST,
port=ELASTICSEARCH_PORT,
index=ELASTICSEARCH_INDEX)
# 품사 태그 설명
# https://docs.google.com/spreadsheets/d/1-9blXKjtjeKZqsf4NzHeYJCrr49-nXeRF6D80udfcwY/edit#gid=589544265
#%%
def get_text_analyzed(text):
global he
options = ["rightPOS", "-f"]
return he.analyze(text, options)
def get_texts_split_by_E(results):
texts = [[]]
tags = [[]]
for result in results:
text = result['token']
tag = result['rightPOS'][:result['rightPOS'].index("(")]
from handEl import handEl
from pprint import pprint as pp
if __name__ == "__main__" :
HOST = 'localhost'
PORT = 8080
INDEX = "test_index"
DATA1 = {'alias':['hello', '메롱']}
DATA2 = {'alias':['world', '메렁']}
he = handEl(host=HOST, port=PORT)
# Index from the ES
he.indexing(INDEX)
# Add documents
he.prope('A0', DATA1, True)
he.prope("A1", DATA2, False)
he.doc('A0')
pp(he.result)
he.search("메롱 좀 하지마!")
pp(he.result)
he.search("그럼 메렁 해야지~")
pp(he.result)
print(he.tokenize("i might love with you."))
def __init__(self, host, port, index="default"):
self.handel = handEl(host, port, index)
self.index = index
self.table = None
from handEl import handEl
he = handEl('localhost', '8080')