|
@@ -81,7 +81,7 @@ def brower_start(port):
|
|
|
# browser = webdriver.Chrome(options=options)
|
|
|
options.add_argument('--ignore-certificate-errors')
|
|
|
options.add_argument("--no-sandbox")
|
|
|
- options.add_argument("--headless")
|
|
|
+# options.add_argument("--headless")
|
|
|
options.add_argument("--disable-gpu")
|
|
|
options.add_argument("--disable-dev-shm-usage")
|
|
|
browser = webdriver.Chrome(options=options)
|
|
@@ -97,9 +97,17 @@ def brower_start(port):
|
|
|
|
|
|
def get_next_job(db):
|
|
|
result = {}
|
|
|
- result = db.query('select * from error_list2 where check_=0 ORDER BY RAND() limit 100')
|
|
|
+# result = db.query('select * from error_list2 where check_=0 ORDER BY RAND() limit 100')
|
|
|
+
|
|
|
+ result = db.query('SELECT * FROM swire_store_list a WHERE not exists (select 1 from error_list2 tei where tei.fid = a.fid limit 1 ) ORDER BY RAND() limit 100')
|
|
|
+
|
|
|
+
|
|
|
url_pd = pd.DataFrame([dict(i) for i in result])
|
|
|
- url_pd['fid'] = url_pd['item_url'].apply(lambda x: x.split('data=')[1].split('!')[3])
|
|
|
+
|
|
|
+ url_pd['item_url'] = url_pd['place_id'].apply(lambda x: 'https://www.google.com/maps/place/?q=place_id:{}'.format(x) )
|
|
|
+
|
|
|
+# url_pd['fid'] = url_pd['item_url'].apply(lambda x: x.split('data=')[1].split('!')[3])
|
|
|
+
|
|
|
# url_pd['item_url'] = url_pd['fid'].apply(lambda x: 'https://www.google.com.tw/maps/@24.1753633,120.6747136,15z/data=!4m5!3m4!1s{}!8m2!3d24.1760271!4d120.6705323'.format(x))
|
|
|
|
|
|
remove = db.query('select item_url from error_list3')
|
|
@@ -463,9 +471,9 @@ def main():
|
|
|
if not chrome_window:
|
|
|
print('restart docker pw{}'.format(port))
|
|
|
# os.system('sudo docker container restart p'+str(port))
|
|
|
- os.system('sudo docker container restart pw'+str(port))
|
|
|
+# os.system('sudo docker container restart pw'+str(port))
|
|
|
|
|
|
- time.sleep(10)
|
|
|
+# time.sleep(10)
|
|
|
|
|
|
print('drvier start...')
|
|
|
driver = brower_start(port)
|