def search_portal(case_number):
"""
Performs a search of the portal from its home page, including selecting the case number input, solving the captcha
and pressing Search. Also handles the captcha being solved incorrectly
:param case_number: Case to search
:return: A set of case number(s).
"""
# Load portal search page
load_page(f"{settings['portal-base']}/Home.aspx/Search", 'Search', settings['verbose'])
# Give some time for the captcha to load, as it does not load instantly.
time.sleep(0.8)
# Select Case Number textbox and enter case number
select_case_input()
case_input = driver.find_element_by_id('caseNumber')
case_input.click()
case_input.send_keys(case_number)
if settings['solve-captchas']:
# Solve captcha if it is required
try:
# Get Captcha
captcha_image_elem = driver.find_element_by_xpath(
'//*/img[@alt="Captcha"]')
captcha_buffer = captcha_image_elem.screenshot_as_png
captcha_answer = captcha_solver.solve_captcha(captcha_buffer)
captcha_textbox = driver.find_element_by_xpath(
'//*/input[@name="captcha"]')
captcha_textbox.click()
captcha_textbox.send_keys(captcha_answer)
except NoSuchElementException:
# No captcha on the page, continue.
pass
# Do search
search_button = driver.find_element_by_id('searchButton')
search_button.click()
else:
raise Exception("Automated captcha solving is disabled by default. Please seek advice before using this feature.")
# If the title stays as 'Search': Captcha solving failed
# If the title contains the case number or 'Search Results': Captcha solving succeeded
# If a timeout occurs, retry 'connect-thresh' times.
for i in range(settings['connect-thresh']):
try:
# Wait for page to load
WebDriverWait(driver, 5).until(
lambda x: 'Search' in driver.title or case_number in driver.title or 'Search Results:' in driver.title)
# Page loaded
if driver.title == 'Search':
# Clicking search did not change the page. This could be because of a failed captcha attempt.
try:
# Check if 'Invalid Captcha' dialog is showing
driver.find_element_by_xpath(
'//div[@class="alert alert-error"]')
print("Captcha was solved incorrectly")
captcha_solver.notify_last_captcha_fail()
except NoSuchElementException:
pass
# Clear cookies so a new captcha is presented upon refresh
driver.delete_all_cookies()
# Try solving the captcha again.
search_portal(case_number)
elif 'Search Results: CaseNumber:' in driver.title:
# Captcha solved correctly
captcha_solver.notify_last_captcha_success()
# Figure out the numer of cases returned
case_detail_tbl = driver.find_element_by_tag_name('table').text.split('\n')
case_count_idx = case_detail_tbl.index('CASES FOUND') + 1
case_count = int(case_detail_tbl[case_count_idx])
# Case number search found multiple cases.
if case_count > 1:
return ScraperUtils.get_associated_cases(driver)
# Case number search found no cases
else:
return set()
elif case_number in driver.title:
# Captcha solved correctly
captcha_solver.notify_last_captcha_success()
# Case number search did find a single court case.
return {case_number}
except TimeoutException:
if i == settings['connect-thresh'] - 1:
raise RuntimeError('Case page could not be loaded after {} attempts, or unexpected page title: {}'.format(settings['connect-thresh'], driver.title))
else:
search_portal(case_number)
def search_portal(case_number):
"""
Performs a search of the portal from its home page, including selecting the case number input, solving the captcha
and pressing Search. Also handles the captcha being solved incorrectly
:param case_number: Case to search
:return: A set of case number(s).
"""
# Load portal search page
load_page(f"{settings['portal-base']}/Home.aspx/Search", 'Search',
settings['verbose'])
# Give some time for the captcha to load, as it does not load instantly.
time.sleep(0.8)
# Select Case Number textbox and enter case number
select_case_input()
case_input = driver.find_element_by_id('caseNumber')
case_input.click()
case_input.send_keys(case_number)
# Solve captcha if it is required
try:
# Get Captcha. This is kinda nasty, but if there's no Captcha, then
# this will throw (which is a good thing in this case) and we can
# move on with processing.
captcha_image_elem = driver.find_element_by_xpath(
'//*/img[@alt="Captcha"]')
captcha_buffer = captcha_image_elem.screenshot_as_png
if settings['solve-captchas']:
solved_captcha = captcha_solver.solve_captcha(captcha_buffer)
captcha_textbox = driver.find_element_by_xpath(
'//*/input[@name="captcha"]')
captcha_textbox.click()
captcha_textbox.send_keys(solved_captcha.answer)
# Do search
search_button = driver.find_element_by_id('searchButton')
search_button.click()
else:
print(f"Captcha encountered trying to view case ID {case_number}.")
print(
"Please solve the captcha and click the search button to proceed."
)
while True:
try:
WebDriverWait(
driver, 6 * 60 *
60).until(lambda x: case_number in driver.title)
print("continuing...")
break
except TimeoutException:
print("still waiting for user to solve the captcha...")
except NoSuchElementException:
# No captcha on the page, continue.
solved_captcha = None
# Do search
search_button = driver.find_element_by_id('searchButton')
search_button.click()
# If the title stays as 'Search': Captcha solving failed
# If the title contains the case number or 'Search Results': Captcha solving succeeded
# If a timeout occurs, retry 'connect-thresh' times.
for i in range(settings['connect-thresh']):
try:
# Wait for page to load
WebDriverWait(
driver,
5).until(lambda x: 'Search' in driver.title or case_number in
driver.title or 'Search Results:' in driver.title)
# Page loaded
if driver.title == 'Search':
# Clicking search did not change the page. This could be because of a failed captcha attempt.
try:
# Check if 'Invalid Captcha' dialog is showing
driver.find_element_by_xpath(
'//div[@class="alert alert-error"]')
print("Captcha was solved incorrectly")
if settings['solve-captchas'] and solved_captcha:
solved_captcha.notify_incorrect()
except NoSuchElementException:
pass
# Clear cookies so a new captcha is presented upon refresh
driver.delete_all_cookies()
# Try solving the captcha again.
search_portal(case_number)
elif 'Search Results: CaseNumber:' in driver.title:
# Captcha solved correctly
if settings['solve-captchas'] and solved_captcha:
solved_captcha.notify_correct()
case_count = ScraperUtils.get_search_case_count(
driver, settings['county'])
# Case number search found multiple cases.
if case_count > 1:
return ScraperUtils.get_associated_cases(driver)
# Case number search found no cases
else:
return set()
elif case_number in driver.title:
# Captcha solved correctly
if settings['solve-captchas'] and solved_captcha:
solved_captcha.notify_correct()
# Case number search did find a single court case.
return {case_number}
except TimeoutException:
if i == settings['connect-thresh'] - 1:
raise RuntimeError(
'Case page could not be loaded after {} attempts, or unexpected page title: {}'
.format(settings['connect-thresh'], driver.title))
else:
search_portal(case_number)