Spaces:
Sleeping
Sleeping
Linh Vuu
commited on
Commit
·
3f1beca
1
Parent(s):
9335b14
updated requirements
Browse files- requirements.txt +2 -1
- scraper_shopee.py +15 -8
requirements.txt
CHANGED
|
@@ -1,4 +1,5 @@
|
|
| 1 |
selenium==4.3.0
|
| 2 |
pandas==1.2
|
| 3 |
streamlit==1.13.0
|
| 4 |
-
altair==4.2.0
|
|
|
|
|
|
| 1 |
selenium==4.3.0
|
| 2 |
pandas==1.2
|
| 3 |
streamlit==1.13.0
|
| 4 |
+
altair==4.2.0
|
| 5 |
+
webdriver-manager==3.7.1
|
scraper_shopee.py
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
from selenium import webdriver
|
| 2 |
-
from
|
|
|
|
| 3 |
from selenium.common.exceptions import NoSuchElementException
|
| 4 |
from selenium.webdriver.common.by import By
|
| 5 |
-
from selenium.webdriver.support.ui import WebDriverWait
|
| 6 |
from selenium.webdriver.support import expected_conditions as EC
|
| 7 |
import time
|
| 8 |
|
|
@@ -24,12 +24,19 @@ def start_driver(force_restart=False):
|
|
| 24 |
close_driver()
|
| 25 |
|
| 26 |
# Setting up the driver
|
| 27 |
-
options = Options()
|
| 28 |
-
options.add_argument('-headless') # we don't want a chrome browser opens, so it will run in the background
|
| 29 |
-
options.add_argument('-no-sandbox')
|
| 30 |
-
options.add_argument('-disable-dev-shm-usage')
|
| 31 |
-
|
| 32 |
-
DRIVER = webdriver.Chrome(options=options)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
|
| 34 |
### Function to extract product info from the necessary html and json tags
|
| 35 |
def get_shopee_product_info_single(product_element, extra_info):
|
|
|
|
| 1 |
from selenium import webdriver
|
| 2 |
+
from webdriver_manager.chrome import ChromeDriverManager
|
| 3 |
+
# from selenium.webdriver.chrome.options import Options
|
| 4 |
from selenium.common.exceptions import NoSuchElementException
|
| 5 |
from selenium.webdriver.common.by import By
|
|
|
|
| 6 |
from selenium.webdriver.support import expected_conditions as EC
|
| 7 |
import time
|
| 8 |
|
|
|
|
| 24 |
close_driver()
|
| 25 |
|
| 26 |
# Setting up the driver
|
| 27 |
+
# options = Options()
|
| 28 |
+
# options.add_argument('-headless') # we don't want a chrome browser opens, so it will run in the background
|
| 29 |
+
# options.add_argument('-no-sandbox')
|
| 30 |
+
# options.add_argument('-disable-dev-shm-usage')
|
| 31 |
+
|
| 32 |
+
# DRIVER = webdriver.Chrome(options=options)
|
| 33 |
+
|
| 34 |
+
gChromeOptions = webdriver.ChromeOptions()
|
| 35 |
+
gChromeOptions.add_argument("window-size=1920x1480")
|
| 36 |
+
gChromeOptions.add_argument("disable-dev-shm-usage")
|
| 37 |
+
DRIVER = webdriver.Chrome(
|
| 38 |
+
chrome_options=gChromeOptions, executable_path=ChromeDriverManager().install()
|
| 39 |
+
)
|
| 40 |
|
| 41 |
### Function to extract product info from the necessary html and json tags
|
| 42 |
def get_shopee_product_info_single(product_element, extra_info):
|