def crawling_tourspot_visitor(district,
start_year,
end_year,
fetch=True,
result_directory='',
service_key=''):
results = []
filename = '%s/%s_tourspot_%s_%s.json' % (result_directory, district,
start_year, end_year)
if fetch:
for year in range(start_year, end_year + 1):
for month in range(1, 13):
for items in api.pd_fetch_tourspot_visitor(
district1=district,
year=year,
month=month,
service_key=service_key):
for item in items:
preprocess_tourspot_visitor(item)
results += items # 전처리 된 data가 쌓임
# save items to file
with open(filename, 'w', encoding='utf-8') as outfile:
json_string = json.dumps(results,
indent=4,
sort_keys=True,
ensure_ascii=False)
# json str으로 덤프하는 과정 텝을 4정도 주고 솔팅을 해라 모두 아스키코드로 해라
outfile.write(json_string)
return filename
def crawling(district1, start_year, end_year):
results = []
print("results type", results)
filename = '%s/%s_%s_%s.json' % (RESULT_DIRECTORY, district1, start_year,
end_year)
#for posts in api.pd_fetch_tourspot_visitor(district1, year, month):
# for post in posts:
# preprocess_post(post)
for j in range(start_year, end_year):
#print(j)
for i in range(1, 3):
#print(i)
for items in pdapi.pd_fetch_tourspot_visitor(district1,
year=j,
month=i):
print("items===", type(items), items)
if type(items) is dict:
items = [items]
#print(type(items['addrCd']))
for item in items:
print("item========!!!!", item)
print()
preprocess_post(item)
results += items
print("type results", type(results), results)
#print("tmp의type=====",type(tmp),tmp)# 이건 딕셔
#print("result의type=====",type(results), results)#이건 리스트 인데 , 여기다 딕셔를 넣음 , 키값만 보임
#print(results)
print("results====", results)
# results += posts
#save results to file (저장, 적재)
with open(filename, 'w', encoding='utf-8') as outfile:
json_string = json.dumps(results,
indent=4,
sort_keys=True,
ensure_ascii=False)
outfile.write(json_string)
from analysis_pd.collect.api import api
# test for pd_gen_url
url = api.pd_gen_url(
'http://openapi.tour.go.kr/openapi/service/TourismResourceStatsService/getPchrgTrrsrtVisitorList',
YM='{0:04d}{1:02d}'.format(2012, 7),
SIDO='서울특별시',
GUNGU='',
RES_NM='',
numOfRows=10,
_type='json',
pageNo=1)
print(url)
# test for pd_tourspot_visitor
for items in api.pd_fetch_tourspot_visitor(district1='서울특별시',
year=2017,
month=7):
print(items)
# test for pd_fetch_foreign_visitor
item = api.pd_fetch_foreign_visitor(112, 2017, 7)
print(item)
request.get_method = lambda: 'GET'
response_body = urlopen(request).read()
print(response_body)
'''
'''
#url 어떻게 만드는지 테스트
url = pdapi.pd_gen_url(
'http://openapi.tour.go.kr/openapi/service/TourismResourceStatsService/getPchrgTrrsrtVisitorList',
YM='{0:04d}{1:02d}'.format(2017, 1),
SIDO='부산광역시',
GUNGU='해운대구',
RES_NM='부산시립미술관',
numOfRows=10,
_type='json',
pageNo=1
)
print(url)
'''
for items in pdapi.pd_fetch_tourspot_visitor(district1='부산광역시',
year=2012,
month=1):
print(items)
#tesst=ttt.html_request(url='http://openapi.tour.go.kr/openapi/service/TourismResourceStatsService/getPchrgTrrsrtVisitorList')
#print(tesst)
#for items in pdapi.pd_fetch_tourspot_visitor(district1='서울특별시', year=2012, month=7):
#print(items)
import analysis_pd.collect.api.api as pdapi
# test for pd_gen_url
# url = pdapi.pd_gen_url(
# 'http://openapi.tour.go.kr/openapi/service/TourismResourceStatsService/getPchrgTrrsrtVisitorList',
# YM='{0:04d}{1:02d}'.format(2017, 1),
# SIDO='서울특별시',
# GUNGU='',
# RES_NM='',
# numOfRows=10,
# _type='json',
# pageNo=1)
#
# print(url)
# test for .pd_fetch_tourspot_visitor
for items in pdapi.pd_fetch_tourspot_visitor(district1="서울특별시", year=2012, month=7):
for item in items:
print(item)
# item = pdapi.pd_fetch_tourspot_visitor(district1="서울특별시", year=2012, month=7)
# print(item)
# for i in item.get('response').get('body').get('items').get('item'):
# print("item : ", i)
# test for pd_fetch_foreign_visitor
# item = pdapi.pd_fetch_foreign_visitor(112, 2012, 7)
# print(item)