def test_get_shortest_path(self):
automata = Automata()
state0 = State('state0')
state1 = State('state1')
state2 = State('state2')
state3 = State('state3')
state4 = State('state4')
state5 = State('state5')
state6 = State('state6')
automata.add_state(state0)
automata.add_state(state1)
automata.add_state(state2)
automata.add_state(state3)
automata.add_state(state4)
automata.add_state(state5)
automata.add_state(state6)
automata.add_edge(state0, state1, Clickable('0-1'))
automata.add_edge(state0, state2, Clickable('0-2'))
automata.add_edge(state0, state3, Clickable('0-3'))
automata.add_edge(state2, state4, Clickable('2-4'))
automata.add_edge(state4, state5, Clickable('4-5'))
automata.add_edge(state3, state5, Clickable('3-5'))
automata.add_edge(state3, state5, Clickable('5-0'))
automata.add_edge(state5, state6, Clickable('5-6'))
self.assertEqual(automata.get_shortest_path(state0), [])
edges = automata.get_shortest_path(state6)
# 0-3, 3-5, 5-6
self.assertEqual([int(e[0].get_id()) for e in edges], [0, 3, 5])
def load_automata(fname):
t_start = time.time()
assert os.path.isfile(fname) and os.path.exists(fname)
automata = Automata()
with open(fname) as f:
data = json.load(f)
for state in data['state']:
with open(os.path.join(os.path.dirname(os.path.realpath(fname)), state['dom_path']), 'r') as df:
s = State(df.read())
s.set_id(state['id'])
for clickable in state['clickable']:
c = Clickable(clickable['id'], clickable['xpath'], clickable['tag'])
s.add_clickable(c)
automata.add_state(s)
for edge in data['edge']:
from_state = automata.get_state_by_id(edge['from'])
to_state = automata.get_state_by_id(edge['to'])
clickable = from_state.get_clickable_by_id(edge['clickable'])
assert from_state and to_state and clickable
automata.add_edge(from_state, to_state, clickable)
return automata
def run():
automata = Automata('q1', ['q3', 'q5', 'q7', 'q8'] +
['q{}'.format(i) for i in range(9, 41)])
primitives = {
'q3': 'INTEGER NUMBER',
'q5': 'REAL NUMBER',
'q7': 'COMMENT',
'q8': 'IDENTIFIER',
'q9': 'IDENTIFIER',
'q10': 'INT',
'q11': 'IDENTIFIER',
'q12': 'IDENTIFIER',
'q13': 'IDENTIFIER',
'q14': 'FLOAT',
'q15': 'IDENTIFIER',
'q16': 'IDENTIFIER',
'q17': 'REAL',
'q18': 'IDENTIFIER',
'q19': 'IDENTIFIER',
'q20': 'IDENTIFIER',
'q21': 'BREAK',
'q22': 'IDENTIFIER',
'q23': 'IDENTIFIER',
'q24': 'CHAR',
'q25': 'IDENTIFIER',
'q26': 'IDENTIFIER',
'q27': 'CASE',
'q28': 'IDENTIFIER',
'q29': 'IDENTIFIER',
'q30': 'IDENTIFIER',
'q31': 'IDENTIFIER',
'q32': 'DOUBLE',
'q33': 'IDENTIFIER',
'q34': 'IDENTIFIER',
'q35': 'CONST',
'q36': 'IDENTIFIER',
'q37': 'IDENTIFIER',
'q38': 'IDENTIFIER',
'q39': 'IDENTIFIER',
'q40': 'CONTINUE'
}
identifiers = ['IDENTIFIER', 'INTEGER NUMBER', 'REAL NUMBER']
automata.add_state('q1', {'+': 'q2', '-': 'q3'})
automata.add_state('q1', get_numerals('q3'))
automata.add_state('q1', {'/': 'q6'})
automata.add_state('q1', get_alphabet('q8'))
automata.add_state('q2', get_numerals('q3'))
automata.add_state('q3', get_numerals('q3'))
automata.add_state('q3', {'.': 'q4'})
automata.add_state('q4', get_numerals('q5'))
automata.add_state('q5', get_numerals('q5'))
automata.add_state('q6', {'/': 'q7'})
automata.add_state('q7', {
'+': 'q7',
'-': 'q7',
'/': 'q7',
'.': 'q7',
' ': 'q7'
})
automata.add_state('q7', get_numerals('q7'))
automata.add_state('q7', get_alphabet('q7'))
automata.add_state(
'q8', {
'n': 'q9',
'l': 'q11',
'e': 'q15',
'r': 'q18',
'h': 'q22',
'a': 'q25',
'o': 'q28'
})
automata.add_state(
'q8', {
i: 'q8'
for i in list(
set(get_alphabet('q8')) -
set(automata.get_transitions_values('q8')))
})
automata.add_state('q8', get_numerals('q8'))
automata.add_state('q9', {'t': 'q10'})
automata.add_state(
'q9', {
i: 'q8'
for i in list(
set(get_alphabet('q9')) -
set(automata.get_transitions_values('q9')))
})
automata.add_state('q10', get_numerals('q8'))
automata.add_state('q10', get_alphabet('q8'))
automata.add_state('q11', {'o': 'q12'})
automata.add_state(
'q11', {
i: 'q8'
for i in list(
set(get_alphabet('q11')) -
set(automata.get_transitions_values('q11')))
})
automata.add_state('q12', {'a': 'q13'})
automata.add_state(
'q12', {
i: 'q8'
for i in list(
set(get_alphabet('q12')) -
set(automata.get_transitions_values('q12')))
})
automata.add_state('q13', {'t': 'q14'})
automata.add_state(
'q13', {
i: 'q8'
for i in list(
set(get_alphabet('q13')) -
set(automata.get_transitions_values('q13')))
})
automata.add_state('q14', get_numerals('q8'))
automata.add_state('q14', get_alphabet('q8'))
automata.add_state('q15', {'a': 'q16'})
automata.add_state(
'q15', {
i: 'q8'
for i in list(
set(get_alphabet('q15')) -
set(automata.get_transitions_values('q15')))
})
automata.add_state('q16', {'l': 'q17'})
automata.add_state(
'q16', {
i: 'q8'
for i in list(
set(get_alphabet('q16')) -
set(automata.get_transitions_values('q16')))
})
automata.add_state('q17', get_numerals('q8'))
automata.add_state('q17', get_alphabet('q8'))
automata.add_state('q18', {'e': 'q19'})
automata.add_state(
'q18', {
i: 'q8'
for i in list(
set(get_alphabet('q18')) -
set(automata.get_transitions_values('q18')))
})
automata.add_state('q19', {'a': 'q20'})
automata.add_state(
'q19', {
i: 'q8'
for i in list(
set(get_alphabet('q19')) -
set(automata.get_transitions_values('q19')))
})
automata.add_state('q20', {'k': 'q21'})
automata.add_state(
'q20', {
i: 'q8'
for i in list(
set(get_alphabet('q20')) -
set(automata.get_transitions_values('q20')))
})
automata.add_state('q21', get_numerals('q8'))
automata.add_state('q21', get_alphabet('q8'))
automata.add_state('q22', {'a': 'q23'})
automata.add_state(
'q22', {
i: 'q8'
for i in list(
set(get_alphabet('q22')) -
set(automata.get_transitions_values('q22')))
})
automata.add_state('q23', {'r': 'q24'})
automata.add_state(
'q23', {
i: 'q8'
for i in list(
set(get_alphabet('q23')) -
set(automata.get_transitions_values('q23')))
})
automata.add_state('q24', get_numerals('q8'))
automata.add_state('q24', get_alphabet('q8'))
automata.add_state('q25', {'s': 'q26'})
automata.add_state(
'q25', {
i: 'q8'
for i in list(
set(get_alphabet('q25')) -
set(automata.get_transitions_values('q25')))
})
automata.add_state('q26', {'e': 'q27'})
automata.add_state(
'q26', {
i: 'q8'
for i in list(
set(get_alphabet('q26')) -
set(automata.get_transitions_values('q26')))
})
automata.add_state('q27', get_numerals('q8'))
automata.add_state('q27', get_alphabet('q8'))
automata.add_state('q28', {'u': 'q29', 'n': 'q33'})
automata.add_state(
'q28', {
i: 'q8'
for i in list(
set(get_alphabet('q28')) -
set(automata.get_transitions_values('q28')))
})
automata.add_state('q29', {'b': 'q30'})
automata.add_state(
'q29', {
i: 'q8'
for i in list(
set(get_alphabet('q29')) -
set(automata.get_transitions_values('q29')))
})
automata.add_state('q30', {'l': 'q31'})
automata.add_state(
'q30', {
i: 'q8'
for i in list(
set(get_alphabet('q30')) -
set(automata.get_transitions_values('q30')))
})
automata.add_state('q31', {'e': 'q32'})
automata.add_state(
'q31', {
i: 'q8'
for i in list(
set(get_alphabet('q31')) -
set(automata.get_transitions_values('q31')))
})
automata.add_state('q32', get_numerals('q8'))
automata.add_state('q32', get_alphabet('q8'))
automata.add_state('q33', {'s': 'q34', 't': 'q36'})
automata.add_state(
'q33', {
i: 'q8'
for i in list(
set(get_alphabet('q33')) -
set(automata.get_transitions_values('q33')))
})
automata.add_state('q34', {'t': 'q35'})
automata.add_state(
'q34', {
i: 'q8'
for i in list(
set(get_alphabet('q34')) -
set(automata.get_transitions_values('q34')))
})
automata.add_state('q35', get_numerals('q8'))
automata.add_state('q35', get_alphabet('q8'))
automata.add_state('q36', {'i': 'q37'})
automata.add_state(
'q36', {
i: 'q8'
for i in list(
set(get_alphabet('q36')) -
set(automata.get_transitions_values('q36')))
})
automata.add_state('q37', {'n': 'q38'})
automata.add_state(
'q37', {
i: 'q8'
for i in list(
set(get_alphabet('q37')) -
set(automata.get_transitions_values('q37')))
})
automata.add_state('q38', {'u': 'q39'})
automata.add_state(
'q38', {
i: 'q8'
for i in list(
set(get_alphabet('q38')) -
set(automata.get_transitions_values('q38')))
})
automata.add_state('q39', {'e': 'q40'})
automata.add_state(
'q39', {
i: 'q8'
for i in list(
set(get_alphabet('q39')) -
set(automata.get_transitions_values('q39')))
})
automata.add_state('q40', get_numerals('q8'))
automata.add_state('q40', get_alphabet('q8'))
words = read_text_file(sys.argv[1])
get_validation(automata, words, primitives, identifiers)
class B2gCrawler(Crawler):
def __init__(self, configuration, executor):
self.automata = Automata()
self.configuration = configuration
self.executor = executor
self.exe_stack = [] # stack of executed clickables (events)
self.invariant_violation = []
self.num_clickables = {
'unexamined': 0, # num of candidate clickables found with rules in DomAnalyzer
'true': 0, # num of clickables triggered new state (different screen dom)
'false': 0, # num of clickables not triggering new state
}
def run(self):
self.executor.restart_app()
initial_state = State(self.executor.get_source())
self.automata.add_state(initial_state)
self.save_screenshot(initial_state.get_id() + '.png', self.executor.get_screenshot(), 'state')
self.save_dom(initial_state)
self.crawl(1)
self.invariant_violation = sorted(self.invariant_violation, key=lambda k: int(k['state']))
return self.automata, self.invariant_violation, self.num_clickables
def crawl(self, depth, prev_state=None):
if depth <= self.configuration.get_max_depth():
cs = self.automata.get_current_state()
if not self.violate_invariant(cs.get_dom(), cs.get_id()):
candidate_clickables = DomAnalyzer.get_clickables(cs.get_dom(), prev_state.get_dom() if prev_state else None)
self.num_clickables['unexamined'] += len(candidate_clickables)
for clickable in candidate_clickables:
# prefetch image of the clickable
time.sleep(0.2) # time for correctly fetching image
img_name = cs.get_id() + '-' + clickable.get_id() + '.png'
img_data = self.executor.get_screenshot(clickable)
# fire the clickable
logger.debug('Fire event in state %s', cs.get_id())
self.executor.empty_form(clickable)
self.executor.fill_form(clickable)
ft = FireEventThread(self.executor, clickable)
ft.start()
ft.join(self.configuration.get_sleep_time()*2) # time out after sleep_time*2 seconds
if ft.is_alive(): # timed out
logger.error('No response while firing an event. Execution sequences:')
self.exe_stack.append(clickable) # add the clickable triggering No Response
for c in self.exe_stack:
logger.error(c)
logger.error('Total clickables found: %d (true: %d, false: %d, unexamined: %d)',
self.num_clickables['unexamined'] + self.num_clickables['true'] + self.num_clickables['false'],
self.num_clickables['true'],
self.num_clickables['false'],
self.num_clickables['unexamined']
)
logger.error('Program terminated.')
sys.exit()
time.sleep(self.configuration.get_sleep_time())
self.num_clickables['unexamined'] -= 1
new_dom = self.executor.get_source()
if DomAnalyzer.is_equal(cs.get_dom(), new_dom):
self.num_clickables['false'] += 1
else:
self.num_clickables['true'] += 1
cs.add_clickable(clickable)
self.exe_stack.append(clickable)
self.save_screenshot(img_name, img_data, 'clickable')
ns, is_newly_added = self.automata.add_state(State(new_dom))
self.automata.add_edge(cs, ns, clickable)
if is_newly_added:
self.save_screenshot(ns.get_id() + '.png', self.executor.get_screenshot(), 'state')
self.save_dom(ns)
self.automata.change_state(ns)
self.crawl(depth+1, cs)
self.exe_stack.pop(-1)
self.automata.change_state(cs)
self.backtrack(cs)
def backtrack(self, state):
logger.debug('Backtrack to state %s', state.get_id())
edges = self.automata.get_shortest_path(state)
self.executor.restart_app()
for (state_from, state_to, clickable, cost) in edges:
time.sleep(self.configuration.get_sleep_time())
self.executor.empty_form(clickable)
self.executor.fill_form(clickable)
self.executor.fire_event(clickable)
def save_screenshot(self, fname, b64data, my_type):
path = os.path.join(self.configuration.get_abs_path(my_type), fname)
imgdata = base64.b64decode(b64data)
with open(path, 'wb') as f:
f.write(imgdata)
def save_dom(self, state):
with open(os.path.join(self.configuration.get_abs_path('dom'), state.get_id() + '.txt'), 'w') as f:
f.write(state.get_dom())
def violate_invariant(self, dom, statd_id):
is_violated = False
for inv in self.configuration.get_invariants():
if inv.check(dom):
is_violated = True
violation = {
'state': statd_id,
'name': str(inv),
'sequence': list(self.exe_stack) # shallow copy of clickables
}
self.invariant_violation.append(violation)
return is_violated
class B2gCrawler(Crawler):
def __init__(self, configuration, executor):
self.automata = Automata()
self.configuration = configuration
self.executor = executor
self.exe_stack = [] # stack of executed clickables (events)
self.invariant_violation = []
self.num_clickables = {
'unexamined':
0, # num of candidate clickables found with rules in DomAnalyzer
'true':
0, # num of clickables triggered new state (different screen dom)
'false': 0, # num of clickables not triggering new state
}
def run(self):
self.executor.restart_app()
initial_state = State(self.executor.get_source())
self.automata.add_state(initial_state)
self.save_screenshot(initial_state.get_id() + '.png',
self.executor.get_screenshot(), 'state')
self.save_dom(initial_state)
self.crawl(1)
self.invariant_violation = sorted(self.invariant_violation,
key=lambda k: int(k['state']))
return self.automata, self.invariant_violation, self.num_clickables
def crawl(self, depth, prev_state=None):
if depth <= self.configuration.get_max_depth():
cs = self.automata.get_current_state()
if not self.violate_invariant(cs.get_dom(), cs.get_id()):
candidate_clickables = DomAnalyzer.get_clickables(
cs.get_dom(),
prev_state.get_dom() if prev_state else None)
self.num_clickables['unexamined'] += len(candidate_clickables)
for clickable in candidate_clickables:
# prefetch image of the clickable
time.sleep(0.2) # time for correctly fetching image
img_name = cs.get_id() + '-' + clickable.get_id() + '.png'
img_data = self.executor.get_screenshot(clickable)
# fire the clickable
logger.debug('Fire event in state %s', cs.get_id())
self.executor.empty_form(clickable)
self.executor.fill_form(clickable)
ft = FireEventThread(self.executor, clickable)
ft.start()
ft.join(self.configuration.get_sleep_time() *
2) # time out after sleep_time*2 seconds
if ft.is_alive(): # timed out
logger.error(
'No response while firing an event. Execution sequences:'
)
self.exe_stack.append(
clickable
) # add the clickable triggering No Response
for c in self.exe_stack:
logger.error(c)
logger.error(
'Total clickables found: %d (true: %d, false: %d, unexamined: %d)',
self.num_clickables['unexamined'] +
self.num_clickables['true'] +
self.num_clickables['false'],
self.num_clickables['true'],
self.num_clickables['false'],
self.num_clickables['unexamined'])
logger.error('Program terminated.')
sys.exit()
time.sleep(self.configuration.get_sleep_time())
self.num_clickables['unexamined'] -= 1
new_dom = self.executor.get_source()
if DomAnalyzer.is_equal(cs.get_dom(), new_dom):
self.num_clickables['false'] += 1
else:
self.num_clickables['true'] += 1
cs.add_clickable(clickable)
self.exe_stack.append(clickable)
self.save_screenshot(img_name, img_data, 'clickable')
ns, is_newly_added = self.automata.add_state(
State(new_dom))
self.automata.add_edge(cs, ns, clickable)
if is_newly_added:
self.save_screenshot(
ns.get_id() + '.png',
self.executor.get_screenshot(), 'state')
self.save_dom(ns)
self.automata.change_state(ns)
self.crawl(depth + 1, cs)
self.exe_stack.pop(-1)
self.automata.change_state(cs)
self.backtrack(cs)
def backtrack(self, state):
logger.debug('Backtrack to state %s', state.get_id())
edges = self.automata.get_shortest_path(state)
self.executor.restart_app()
for (state_from, state_to, clickable, cost) in edges:
time.sleep(self.configuration.get_sleep_time())
self.executor.empty_form(clickable)
self.executor.fill_form(clickable)
self.executor.fire_event(clickable)
def save_screenshot(self, fname, b64data, my_type):
path = os.path.join(self.configuration.get_abs_path(my_type), fname)
imgdata = base64.b64decode(b64data)
with open(path, 'wb') as f:
f.write(imgdata)
def save_dom(self, state):
with open(
os.path.join(self.configuration.get_abs_path('dom'),
state.get_id() + '.txt'), 'w') as f:
f.write(state.get_dom())
def violate_invariant(self, dom, statd_id):
is_violated = False
for inv in self.configuration.get_invariants():
if inv.check(dom):
is_violated = True
violation = {
'state': statd_id,
'name': str(inv),
'sequence':
list(self.exe_stack) # shallow copy of clickables
}
self.invariant_violation.append(violation)
return is_violated
class AutomataTestCase(unittest.TestCase):
def setUp(self):
dom1 = '''
<!DOCTYPE html>
<html lang="en">
<head> <meta charset="UTF-8"> <title></title> </head>
<body>dom1</body>
</html>
'''
state = State(dom1)
self.automata = Automata()
self.automata.add_state(state)
self.assertEqual(len(self.automata.get_states()), 1)
self.assertEqual(self.automata.get_initial_state().get_id(), self.automata.get_current_state().get_id())
self.assertEqual(self.automata.get_current_state().get_id(), '0')
# test adding and removing inputs and forms
form1 = FormField('form1')
form1.add_input(InputField('username', '//*[@id="username"]', 'castman'))
form1.add_input(InputField('password', '', 'p@ssw0rd'))
form1.add_input(InputField('password', '', 'p@ssw0rd'))
self.assertEqual(len(form1.get_inputs()), 2)
form1.remove_input(InputField('username', '//*[@id="username"]', 'castman'))
self.assertEqual(len(form1.get_inputs()), 1)
form2 = FormField('', '//*[@id="lst-ib"]')
clickable = Clickable('', '//*[@id="btn1"]')
clickable.add_form(form1)
clickable.add_form(FormField('form1'))
clickable.add_form(form2)
self.assertEqual(len(clickable.get_forms()), 2)
clickable.remove_form(FormField('', '//*[@id="lst-ib"]'))
self.assertEqual(len(clickable.get_forms()), 1)
# add the clickable into state 0
self.automata.get_current_state().add_clickable(clickable)
def test_automata(self):
dom1 = self.automata.get_current_state().get_dom()
dom1 += '<custom></custom>'
dom2 = dom1
state1 = State(dom1)
state2 = State(dom2)
self.automata.add_state(state1)
state3, is_newly_added = self.automata.add_state(state2)
self.assertTrue(state3 == state1)
self.assertFalse(is_newly_added)
self.assertEqual(len(self.automata.get_states()), 2)
clickable = self.automata.get_current_state().get_clickables()[0]
clickable2 = Clickable('', '//html/body/button[3]')
self.assertEqual(len(self.automata.get_current_state().get_clickables()), 1)
self.automata.get_current_state().add_clickable(clickable)
self.automata.get_current_state().add_clickable(clickable2)
self.automata.get_current_state().add_clickable(clickable2)
self.assertEqual(len(self.automata.get_current_state().get_clickables()), 2)
self.automata.add_edge(self.automata.get_current_state(), state1, self.automata.get_current_state().get_clickables()[0])
self.assertEqual(len(self.automata.get_edges()), 1)
state1.add_prev_state(self.automata.get_current_state())
self.assertEqual(self.automata.get_current_state().get_id(), '0')
self.automata.change_state(state1)
self.assertEqual(self.automata.get_initial_state().get_id(), '0')
self.assertEqual(self.automata.get_current_state().get_id(), '1')
self.assertEqual(self.automata.get_current_state().get_prev_states()[0].get_id(), '0')
'''
for s in self.automata.get_states():
print s
for c in s.get_clickables():
print c
for f in c.get_forms():
print f
for _i in f.get_inputs():
print _i
for (state_from, state_to, clickable, cost) in self.automata.get_edges():
print state_from, state_to, clickable, cost
'''
def test_get_shortest_path(self):
automata = Automata()
state0 = State('state0')
state1 = State('state1')
state2 = State('state2')
state3 = State('state3')
state4 = State('state4')
state5 = State('state5')
state6 = State('state6')
automata.add_state(state0)
automata.add_state(state1)
automata.add_state(state2)
automata.add_state(state3)
automata.add_state(state4)
automata.add_state(state5)
automata.add_state(state6)
automata.add_edge(state0, state1, Clickable('0-1'))
automata.add_edge(state0, state2, Clickable('0-2'))
automata.add_edge(state0, state3, Clickable('0-3'))
automata.add_edge(state2, state4, Clickable('2-4'))
automata.add_edge(state4, state5, Clickable('4-5'))
automata.add_edge(state3, state5, Clickable('3-5'))
automata.add_edge(state3, state5, Clickable('5-0'))
automata.add_edge(state5, state6, Clickable('5-6'))
self.assertEqual(automata.get_shortest_path(state0), [])
edges = automata.get_shortest_path(state6)
# 0-3, 3-5, 5-6
self.assertEqual([int(e[0].get_id()) for e in edges], [0, 3, 5])
#for e in edges:
# print e[0].get_id(), e[1].get_id(), e[2].get_id()
def test_load_save(self):
automata = Automata(fname='test_data/automata-example.json')
config = B2gConfiguration('test-app-name', 'test-app-id', mkdir=False)
config.set_path('root', 'test_data')
config.set_path('dom', 'test_data/dom')
config.set_path('state', 'test_data/screenshot/state')
config.set_path('clickable', 'test_data/screenshot/clickable')
saved_file_path = automata.save(config)
import filecmp
self.assertTrue(filecmp.cmp('test_data/automata-example.json', 'test_data/automata.json'))
try:
os.remove(saved_file_path)
except OSError:
pass
