def _parse_clickable_from_db_to_model(self, clickable):
c = Clickable(clickable['event'], clickable['tag'], clickable['dom_address'], clickable['html_id'], clickable['html_class'], clickable['clickable_depth'], clickable['function_id'])
c.clicked = clickable['clicked']
c.clickable_type = self._num_to_clickable_type(clickable['clickable_type'])
c.links_to = clickable['links_to']
c.clickable_depth = clickable['clickable_depth']
return c
def test_clickables(self):
clickable1 = Clickable("click",
"a",
"body/div/div/a",
id="Test1",
html_class="Test2",
clickable_depth=243,
function_id="Test3")
self.database._insert_clickable_into_db(SESSION, WEBPAGE_ID,
clickable1)
clickables = self.database.get_all_clickables_to_page_id_from_db(
SESSION, WEBPAGE_ID)
self.assertEqual(len(clickables), 1)
self.assertEqual(clickable1, clickables[0])
self.database.set_clickable_clicked(
SESSION,
WEBPAGE_ID,
clickable1.dom_address,
clickable1.event,
clickable_depth=243,
clickable_type=ClickableType.CreatesNewNavigatables)
clickables = self.database.get_all_clickables_to_page_id_from_db(
SESSION, WEBPAGE_ID)
self.assertEqual(len(clickables), 1)
clickable1.clicked = True
clickable1.clickable_type = ClickableType.CreatesNewNavigatables
self.assertEqual(clickable1, clickables[0])
def get_all_clickables_to_page_id_from_db(self, current_session, page_id):
clickables = self.clickables.find({"web_page_id": page_id, "session": current_session})
result = []
for clickable in clickables:
c = Clickable(clickable['event'], clickable['tag'], clickable['dom_address'], clickable['html_id'], clickable['html_class'], clickable_depth=clickable['clickable_depth'], function_id=clickable['function_id'])
c.links_to = clickable['links_to']
c.clickable_type = self._num_to_clickable_type(clickable['clickable_type'])
c.clicked = clickable['clicked']
result.append(c)
return result
def loadData(fn):
indata = json.load(open(fn))
url = indata["url"]
initclick = Clickable.fromDict(
indata["element_to_click"]
) if indata["element_to_click"] != None else None
preclicks = [
Clickable(None, None, None).fromDict(x) if x != None else None
for x in indata["pre_clicks"]
]
return url, initclick, preclicks
def add_eventlistener_to_element(self, msg):
#logging.debug(msg)
if "id" in msg and msg['id'] != "":
id = msg['id']
else:
id = None
dom_address = msg['addr']
event = msg['event']
if event == "":
event = None
tag = msg['tag']
if "class" in msg and msg['class'] != "":
html_class = msg['class']
else:
html_class = None
function_id = msg['function_id']
if tag is not None and dom_address != "":
tmp = Clickable(event,
tag,
dom_address,
id,
html_class,
function_id=function_id)
if tmp not in self._new_clickables:
self._new_clickables.append(tmp)
def _extract_new_links_from_links(elements, requested_url):
found_links = []
new_clickables = []
if (len(elements) == 0):
#logging.debug("No links found...")
return [], []
else:
for elem in elements:
href = elem.attribute("href")
#logging.debug(str(type(elem)) + " href: " + str(href) + " Tagname: " + str(elem.tagName()))
if href == "/" or href == requested_url or href == "": #or href[0] == '#':
continue
elif "javascript:" in href: #We assume it as clickable
html_id = elem.attribute("id")
html_class = elem.attribute("class")
dom_address = elem.evaluateJavaScript("getXPath(this)")
event = href
tag = "a"
new_clickables.append(
Clickable(event, tag, dom_address, html_id, html_class,
None, None))
elif "#" in href:
html_id = elem.attribute("id")
html_class = elem.attribute("class")
dom_address = elem.evaluateJavaScript("getXPath(this)")
event = "click"
tag = "a"
new_clickables.append(
Clickable(event, tag, dom_address, html_id, html_class,
None, None))
elif len(href) > 0:
html_id = elem.attribute("id")
html_class = elem.attribute("class")
dom_address = elem.evaluateJavaScript("getXPath(this)")
url = href
link = Link(url, dom_address, html_id, html_class)
found_links.append(link)
else:
logging.debug(
"Elem has attribute href: " +
str(elem.attribute("href") + " and matches no criteria"))
return found_links, new_clickables
def get_all_clickables_to_page_id_from_db(self, current_session, page_id):
clickables = self.clickables.find({
"web_page_id": page_id,
"session": current_session
})
result = []
for clickable in clickables:
c = Clickable(clickable['event'],
clickable['tag'],
clickable['dom_address'],
clickable['html_id'],
clickable['html_class'],
clickable_depth=clickable['clickable_depth'],
function_id=clickable['function_id'])
c.links_to = clickable['links_to']
c.clickable_type = self._num_to_clickable_type(
clickable['clickable_type'])
c.clicked = clickable['clicked']
result.append(c)
return result
def _parse_clickable_from_db_to_model(self, clickable):
c = Clickable(clickable['event'], clickable['tag'],
clickable['dom_address'], clickable['html_id'],
clickable['html_class'], clickable['clickable_depth'],
clickable['function_id'])
c.clicked = clickable['clicked']
c.clickable_type = self._num_to_clickable_type(
clickable['clickable_type'])
c.links_to = clickable['links_to']
c.clickable_depth = clickable['clickable_depth']
return c
def property_helper(frame):
all_elements = frame.findAllElements("*")
result = []
properties = ['onclick', "onmouseover", "onabort", "onblur", "onchange", "onblclick", "onerror", "onfocus", "onkeydown",
"onkeypress", "onkeyup", "onmousedown", "onmousemove", "onmouseout", "onmouseup"]
for element in all_elements:
element_id = None
element_class = None
if element.hasAttribute("id"):
element_id = element.attribute("id")
if element.hasAttribute("class"):
element_class = element.attribute("class")
element_dom_address = None
for prop in properties:
if element.hasAttribute(prop):
if element_dom_address is None:
element_dom_address = element.evaluateJavaScript("getXPath(this)")
result.append(Clickable(prop, element.tagName(), element_dom_address, element_id, element_class, function_id="None"))
return result
def test_webpage(self):
clickable1 = Clickable("click",
"a",
"body/div/div/a",
id="Test1",
html_class="Test2",
clickable_depth=243,
function_id="Test3")
web_page = WebPage(1,
url=TEST_URL1,
html=TEST_HTML,
cookiesjar=None,
depth=24,
base_url=TEST_URL2)
web_page.clickables.extend([clickable1])
self.database.insert_page_into_db(SESSION, web_page)
web_page1 = self.database.get_webpage_to_id_from_db(SESSION, 1)
self.assertEqual(web_page.toString(), web_page1.toString())
web_page2 = self.database.get_webpage_to_url_from_db(
SESSION, TEST_URL1)
self.assertEqual(web_page.toString(), web_page2.toString())
def __init__(self, clickable, key_event):
Clickable.__init__(self, clickable.event, clickable.tag, clickable.dom_address, clickable.id, clickable.html_class, clickable.clickable_depth, clickable.function_id)
self.random_char = key_event #Is the key typed in for triggering the clickabel
from models.url import Url
from models.webpage import WebPage
__author__ = 'constantin'
import unittest
SESSION = 12345
WEBPAGE_ID = 99
TEST_URL1 = "http://example.com"
TEST_URL2 = "http://example.com/exmaple.php"
TEST_HTML = "<html><head></head><body></body></html>"
CLICKABLE = Clickable("click",
"a",
"body/div/div/a",
id="Test1",
html_class="Test2",
clickable_depth=243,
function_id="Test3")
WEBPAGE = WebPage(1,
url=TEST_URL1,
html=TEST_HTML,
cookiesjar=None,
depth=24,
base_url=TEST_URL2)
AJAXREQUEST = AjaxRequest("GET",
TEST_URL1,
CLICKABLE,
parameters=["test=Test"])
def __init__(self, clickable, key_event):
Clickable.__init__(self, clickable.event, clickable.tag,
clickable.dom_address, clickable.id,
clickable.html_class, clickable.clickable_depth,
clickable.function_id)
self.random_char = key_event #Is the key typed in for triggering the clickabel