def back_to_start():
logging.info('getting back to start point')
site = Site()
#browser = Browser()
#browser.get()
site.back_to_start()
return jsonify(gexf=etree.tostring(site.get_gexf()).decode('utf-8'), current_page=site.current)
def follow_existing_connections():
infos = request.get_json()
logging.info('%s' % (infos['target']))
site = Site()
actions = site.get_actions_to(infos['target'])
for action in actions:
action.do()
return jsonify(gexf=etree.tostring(site.get_gexf()).decode('utf-8'), current_page=site.current)
def add_connection_and_go():
connection = request.get_json()
logging.info('%s - %s' % (connection['css'], connection['nb']))
site = Site()
connection_id = site.add_connection_to_current_page(Action.ActionType.CLICK, connection['css'], connection['nb'])
action = site.get_action_from_id(connection_id)
action.do()
return jsonify(gexf=etree.tostring(site.get_gexf()).decode('utf-8'), current_page=site.current)
def get_from_start():
starter = request.get_json()
site = Site()
if site.current != 'start':
abort(500)
browser = Browser()
browser.get(starter['url'])
return jsonify(gexf=etree.tostring(site.get_gexf()).decode('utf-8'), current_page=site.current)
def follow(connection):
config = Configuration()
site = Site()
if not connection in site._connections:
abort(404)
if site._connections[connection]['from'] != site.current:
abort(500)
action = site.get_action_from_id(connection)
action.do()
return jsonify(gexf=etree.tostring(site.get_gexf()).decode('utf-8'), current_page=site.current)
def start():
starter = request.get_json()
Configuration().browser = starter['browser']
if starter['proxy'] == 'no proxy':
Configuration().proxy_path = None
else:
Configuration().proxy_path = starter['proxy_path']
site = Site(starter['name'])
browser = Browser()
browser.start(site)
return jsonify(gexf=etree.tostring(site.get_gexf()).decode('utf-8'), current_page=site.current)
def test_on_path_finding(self):
from Site.Site import Site
from Site.Page import Page
from Main.Action import Action
site_url = "http://%s.url/for/page" % uuid.uuid4()
site = Site(site_url)
node_start = str(uuid.uuid4())
node_1 = str(uuid.uuid4())
node_2 = str(uuid.uuid4())
node_end = str(uuid.uuid4())
connection_0_id = str(uuid.uuid4())
connection_0_content = {'from': 'start', 'to': node_start, 'explored': True, 'type': site.ConnectionTypes.START, 'data': {'url': site_url}}
connection_1_id = str(uuid.uuid4())
connection_1_data = {'css': str(uuid.uuid4()), 'nb': 0}
connection_1_content = {'from': node_start, 'to': node_1, 'explored': True, 'type': site.ConnectionTypes.LINK, 'data': connection_1_data}
connection_2_id = str(uuid.uuid4())
connection_2_data = {'css': str(uuid.uuid4()), 'nb': 0}
connection_2_content = {'from': node_1, 'to': node_2, 'explored': True, 'type': site.ConnectionTypes.LINK, 'data': connection_2_data}
connection_3_id = str(uuid.uuid4())
connection_3_data = {'css': str(uuid.uuid4()), 'nb': 0}
connection_3_content = {'from': node_2, 'to': node_end, 'explored': True, 'type': site.ConnectionTypes.LINK, 'data': connection_3_data}
connection_4_id = str(uuid.uuid4())
connection_4_data = {'css': str(uuid.uuid4()), 'nb': 0}
connection_4_content = {'from': node_2, 'to': node_1, 'explored': True, 'type': site.ConnectionTypes.LINK, 'data': connection_4_data}
connection_5_id = str(uuid.uuid4())
connection_5_data = {'css': str(uuid.uuid4()), 'nb': 0}
connection_5_content = {'from': node_start, 'to': node_2, 'explored': True, 'type': site.ConnectionTypes.LINK, 'data': connection_5_data}
connection_6_id = str(uuid.uuid4())
connection_6_data = {'css': str(uuid.uuid4()), 'nb': 0}
connection_6_content = {'from': node_end, 'to': None, 'explored': False, 'type': site.ConnectionTypes.LINK, 'data': connection_6_data}
site._connections[connection_0_id] = connection_0_content
site._connections[connection_1_id] = connection_1_content
site._connections[connection_2_id] = connection_2_content
site._connections[connection_3_id] = connection_3_content
site._connections[connection_4_id] = connection_4_content
site._connections[connection_5_id] = connection_5_content
site._connections[connection_6_id] = connection_6_content
html_start = '<html><body>%s</body></html>' % uuid.uuid4()
site._pages[node_start] = Page('node_start', html_start)
html_1 = '<html><body>%s</body></html>' % uuid.uuid4()
site._pages[node_1] = Page('node_1', html_1)
html_2 = '<html><body>%s</body></html>' % uuid.uuid4()
site._pages[node_2] = Page('node_2', html_2)
html_end = '<html><body>%s</body></html>' % uuid.uuid4()
site._pages[node_end] = Page('node_end', html_end)
site._current = node_start
path = site.find_shortest_path(node_start, node_end)
assert path[0] == {'connection': {'from': node_start, 'to': node_2, 'explored': True, 'type': site.ConnectionTypes.LINK, 'data': connection_5_data}, 'id': connection_5_id}
assert path[1] == {'connection': {'from': node_2, 'to': node_end, 'explored': True, 'type': site.ConnectionTypes.LINK, 'data': connection_3_data}, 'id': connection_3_id}
actions = site.get_actions_to(node_end)
assert len(actions) == 2
assert actions[0]._type == Action.ActionType.CLICK
assert actions[0].connection == connection_5_id
assert actions[1]._type == Action.ActionType.CLICK
assert actions[1].connection == connection_3_id
assert site.get_distance_to(connection_6_id) == 2
actions = site.get_first_connection_unexplored()
assert len(actions) == 3
assert actions[0]._type == Action.ActionType.CLICK
assert actions[0].connection == connection_5_id
assert actions[1]._type == Action.ActionType.CLICK
assert actions[1].connection == connection_3_id
assert actions[2]._type == Action.ActionType.CLICK
assert actions[2].connection == connection_6_id
site._current = node_end
actions = site.get_first_connection_unexplored()
assert len(actions) == 1
assert actions[0]._type == Action.ActionType.CLICK
assert actions[0].connection == connection_6_id
html = str('<html><body><div>new_page</div></body></html>')
page = site.current_page(html, '', connection_6_id)
assert site._pages[site.get_uniq_id(html, '')] == page
assert site._connections[connection_6_id]['explored'] == True
assert site._connections[connection_6_id]['to'] == site.get_uniq_id(html, '')
actions = site.get_first_connection_unexplored()
assert actions == None
gexf = etree.Element('test')
gexf_site = site.get_gexf()
assert gexf_site.xpath('//meta/creator')[0].text == "Pallas"
assert gexf_site.xpath('//meta/description')[0].text == site_url
assert len(gexf_site.xpath('//nodes/node')) == 6
assert len(gexf_site.xpath('//edges/edge')) == 7
def go_to_url():
call = request.get_json()
browser = Browser()
site = Site()
browser.get(call['url'])
return jsonify(gexf=etree.tostring(site.get_gexf()).decode('utf-8'), current_page=site.current)