srinuksv commited on
Commit
adb6ea7
·
verified ·
1 Parent(s): 8d488aa

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +7 -13
app.py CHANGED
@@ -19,26 +19,20 @@ app.mount("/static", StaticFiles(directory="static"), name="static")
19
 
20
  def scrape_upwork_data(search_query, num_jobs, page):
21
  # Setup Chrome options for remote WebDriver
22
- chrome_options = webdriver.ChromeOptions()
23
- chrome_options.add_argument("--headless")
24
- chrome_options.add_argument("--window-size=1920x1080")
25
- chrome_options.add_argument("--disable-gpu")
26
- chrome_options.add_argument("--disable-extensions")
27
- chrome_options.add_argument("--no-sandbox")
28
- chrome_options.add_argument("--disable-dev-shm-usage")
29
 
30
  # Setup the remote WebDriver
31
- driver = webdriver.Remote(
32
- command_executor='http://localhost:4444/wd/hub',
33
- options=chrome_options
34
- )
35
-
36
  job_listings = []
37
  try:
38
  url = f'https://www.upwork.com/nx/search/jobs?amount=500-&hourly_rate=25-&location=Americas,Europe,Australia%20and%20New%20Zealand,Canada,India,Switzerland,United%20States&per_page=50&q={search_query}&sort=recency&t=0,1&page={page}'
39
  driver.get(url)
40
-
41
  time.sleep(5) # Wait for the page to load
 
42
  jobs = driver.find_elements(By.CSS_SELECTOR, 'article[data-test="JobTile"]')
43
 
44
  for job in jobs:
 
19
 
20
  def scrape_upwork_data(search_query, num_jobs, page):
21
  # Setup Chrome options for remote WebDriver
22
+ options = webdriver.ChromeOptions()
23
+ options.add_argument('--headless')
24
+ options.add_argument('--no-sandbox')
25
+ options.add_argument('--disable-dev-shm-usage')
 
 
 
26
 
27
  # Setup the remote WebDriver
28
+ driver=webdriver.Chrome(options=options)
 
 
 
 
29
  job_listings = []
30
  try:
31
  url = f'https://www.upwork.com/nx/search/jobs?amount=500-&hourly_rate=25-&location=Americas,Europe,Australia%20and%20New%20Zealand,Canada,India,Switzerland,United%20States&per_page=50&q={search_query}&sort=recency&t=0,1&page={page}'
32
  driver.get(url)
33
+ driver.set_window_size(1080, 720)
34
  time.sleep(5) # Wait for the page to load
35
+ page_title = driver.title
36
  jobs = driver.find_elements(By.CSS_SELECTOR, 'article[data-test="JobTile"]')
37
 
38
  for job in jobs: