|
@@ -155,7 +155,7 @@ def process_web_request(db, driver, fid):
|
|
|
jstext = resp.decode('utf-8')
|
|
|
result = parsing_js(jstext)
|
|
|
|
|
|
- save_js_to_db(resultobj, fid)
|
|
|
+ save_js_to_db(result, fid)
|
|
|
time.sleep(1)
|
|
|
|
|
|
|
|
@@ -225,22 +225,26 @@ def main():
|
|
|
job = get_next_job(db)
|
|
|
|
|
|
for row, group in job.iterrows():
|
|
|
- item_url = group['item_url']
|
|
|
- reviews_cnt = group['reviews_cnt']
|
|
|
- fid = group['fid']
|
|
|
-
|
|
|
- print(reviews_cnt, item_url)
|
|
|
- driver.get(item_url)
|
|
|
- time.sleep(0.5)
|
|
|
- shop_soup = BeautifulSoup(driver.page_source, 'html.parser')
|
|
|
- tmp_value = shop_soup.find('span', {'jsaction':'pane.rating.moreReviews'})
|
|
|
-
|
|
|
- if tmp_value:
|
|
|
- get_reviews(driver, reviews_cnt)
|
|
|
- process_web_request(db, driver, fid)
|
|
|
- print(driver.current_url)
|
|
|
-
|
|
|
- db['review_process'].insert({'fid':fid, 'dt':datetime.now()})
|
|
|
+ try:
|
|
|
+ item_url = group['item_url']
|
|
|
+ reviews_cnt = group['reviews_cnt']
|
|
|
+ fid = group['fid']
|
|
|
+
|
|
|
+ print(reviews_cnt, item_url)
|
|
|
+ driver.get(item_url)
|
|
|
+ time.sleep(0.5)
|
|
|
+ shop_soup = BeautifulSoup(driver.page_source, 'html.parser')
|
|
|
+ tmp_value = shop_soup.find('span', {'jsaction':'pane.rating.moreReviews'})
|
|
|
+
|
|
|
+ if tmp_value:
|
|
|
+ get_reviews(driver, reviews_cnt)
|
|
|
+ process_web_request(db, driver, fid)
|
|
|
+ print(driver.current_url)
|
|
|
+
|
|
|
+ db['review_process'].insert({'fid':fid, 'dt':datetime.now()})
|
|
|
+
|
|
|
+ except:
|
|
|
+ traceback.print_exc()
|
|
|
|
|
|
if __name__ == '__main__':
|
|
|
main()
|