def isUsingManyExternalResources(url: str) -> int:
noOfExtRes = 0
MANY = 30
try:
source = utils.getLoadedHtmlFromUrl(url)
soup = BeautifulSoup(source, 'html.parser')
domain = utils.getDomainFromUrl(url)
imageSources = compileList(soup, 'img', 'src')
scriptSources = compileList(soup, 'script', 'src')
styleSources = compileList(soup, 'link', 'href')
videoSources = compileList(soup, 'video', 'src')
audioSources = compileList(soup, 'audio', 'src')
iframeSources = compileList(soup, 'iframe', 'src')
embedSources = compileList(soup, 'embded', 'src')
objectSources = compileList(soup, 'object', 'data')
sourceSources = compileList(soup, 'source', 'src')
sources = imageSources + scriptSources + styleSources + videoSources + audioSources + iframeSources + embedSources + objectSources + sourceSources
for s in sources:
if str(s).startswith('http'):
if utils.getDomainFromUrl(str(s)) != domain:
noOfExtRes += 1
if noOfExtRes >= MANY:
return 1
else:
return 0
except:
return -1
def isOpenningNewWindow(url: str) -> int:
hasNewWindow = False
try:
source = utils.getLoadedHtmlFromUrl(url)
soup = BeautifulSoup(source, 'html.parser')
scriptSources = compileList(soup, 'script', 'src')
scriptInners = compileList(soup, 'script', 'innerHtml')
anchorTargets = compileList(soup, 'a', 'target')
for s in scriptSources:
try:
script = utils.getHttpResponse(s)
if re.match('window.open', script):
hasNewWindow = True
except:
pass
for i in scriptInners:
if re.match('window.open', str(i)):
hasNewWindow = True
for a in anchorTargets:
if a == '_blank':
hasNewWindow = True
if hasNewWindow:
return 1
return 0
except:
return -1
def isStringInPage(url: str, searchTarget: str) -> int:
"""Return 1 if the the searchTarget exists in the page of the given URL"""
try:
source = utils.getLoadedHtmlFromUrl(url)
except:
return -1
return 1 if searchTarget in source else 0
def isPageContainingDownload(url: str) -> int:
"""Return 1 if the page contains 'download' in the page of the given URL"""
try:
source = utils.getLoadedHtmlFromUrl(url)
result1 = 'download' in source
result2 = 'Download' in source
except:
return -1
return 1 if (result1 or result2) else 0
def isPageContainingActionRequired(url: str) -> int:
"""Return 1 if the page contains 'action required' in the page of the given URL"""
try:
source = utils.getLoadedHtmlFromUrl(url)
result1 = 'action required' in source
result2 = 'Action Required' in source
result3 = 'ACTION REQUIRED' in source
except:
return -1
return 1 if (result1 or result2 or result3) else 0
def isPageContainingUrgent(url: str) -> int:
"""Return 1 if the page contains 'urgent' in the page of the given URL"""
try:
source = utils.getLoadedHtmlFromUrl(url)
result1 = 'urgent' in source
result2 = 'Urgent' in source
result3 = 'URGENT' in source
except:
return -1
return 1 if (result1 or result2 or result3) else 0
def isPageContainingCreditCard(url: str) -> int:
"""Return 1 if the page contains 'credit card' in the page of the given URL"""
try:
source = utils.getLoadedHtmlFromUrl(url)
result1 = 'credit card' in source
result2 = 'credit-card' in source
result3 = 'Credit Card' in source
except:
return -1
return 1 if (result1 or result2 or result3) else 0
def isBlockingRightClick(url: str) -> int:
isBlocking = False
try:
source = utils.getLoadedHtmlFromUrl(url)
soup = BeautifulSoup(source, 'html.parser')
scriptSources = compileList(soup, 'script', 'src')
scriptInners = compileList(soup, 'script')
# For external scripts
for s in scriptSources:
try:
# Download the script content
script = utils.getHttpResponse(s)
# For document.addEventListener('contextmenu', event => event.preventDefault());
if re.search('contextmenu', script) and re.search(
'preventDefault', script):
isBlocking = True
except:
pass
# For checking <script>CONTENT</script>
for s in scriptInners:
if re.search('contextmenu', s) and re.search('preventDefault', s):
isBlocking = True
# For checking body <body oncontextmenu="return false;"
try:
if soup.body['oncontextmenu'] and re.search(
'return\sfalse', soup.body['oncontextmenu']):
isBlocking = True
except KeyError:
pass
if isBlocking:
return 1
return 0
except:
return -1
def testHelloWorldCodepen(self):
self.assertTrue("Hello World" in utils.getLoadedHtmlFromUrl(
'https://cdpn.io/rayyue300/fullpage/RwNLBpa'))
