def get_all_normalize_dom(self, configuration):
if not self._dom_list:
dom_list = self.get_dom_list(configuration)
dom = [ DomAnalyzer.normalize( stateDom['dom'] ) for stateDom in dom_list ]
dom = "\n".join(dom)
return dom
else:
dom = [ DomAnalyzer.normalize( stateDom['dom'] ) for stateDom in self._dom_list ]
dom = "\n".join(dom)
return dom
def get_all_normalize_dom(self, configuration):
if not self._dom_list:
dom_list = self.get_dom_list(configuration)
dom = [
DomAnalyzer.normalize(stateDom['dom']) for stateDom in dom_list
]
dom = "\n".join(dom)
return dom
else:
dom = [
DomAnalyzer.normalize(stateDom['dom'])
for stateDom in self._dom_list
]
dom = "\n".join(dom)
return dom
def save_dom(self, state):
try:
#make dir for each state
state_dir = os.path.join( self.configuration.get_abs_path('dom'), state.get_id() )
if not os.path.isdir(state_dir):
os.makedirs(state_dir)
iframe_key_dict = { 'num': 0 }
for stateDom in state.get_dom_list(self.configuration):
iframe_key = ';'.join(stateDom['iframe_path']) if stateDom['iframe_path'] else None
#make new dir for iframe
if stateDom['iframe_path']:
iframe_key_dict['num'] += 1
iframe_key_dict[ str(iframe_key_dict['num']) ] = { 'path' : stateDom['iframe_path'], 'url': stateDom['url'] }
dom_dir = os.path.join( self.configuration.get_abs_path('dom'), state.get_id(), str(iframe_key_dict['num']) )
if not os.path.isdir(dom_dir):
os.makedirs(dom_dir)
else:
iframe_key_dict['basic'] = { 'url' : stateDom['url'] }
dom_dir = os.path.join( self.configuration.get_abs_path('dom'), state.get_id() )
with codecs.open( os.path.join( dom_dir, state.get_id()+'.txt'), 'w', encoding='utf-8' ) as f:
f.write( stateDom['dom'] )
with codecs.open( os.path.join( dom_dir, state.get_id()+'_nor.txt'), 'w', encoding='utf-8' ) as f:
f.write( DomAnalyzer.normalize( stateDom['dom'] ) )
with codecs.open( os.path.join( dom_dir, state.get_id()+'_inputs.txt'), 'w', encoding='utf-8' ) as f:
json.dump(state.get_inputs_json( iframe_key ), f, indent=2, sort_keys=True, ensure_ascii=False)
with codecs.open( os.path.join( dom_dir, state.get_id()+'_selects.txt'), 'w', encoding='utf-8' ) as f:
json.dump(state.get_selects_json(iframe_key ), f, indent=2, sort_keys=True, ensure_ascii=False)
with codecs.open( os.path.join( dom_dir, state.get_id()+'_radios.txt'), 'w', encoding='utf-8' ) as f:
json.dump(state.get_radios_json(iframe_key ), f, indent=2, sort_keys=True, ensure_ascii=False)
with codecs.open( os.path.join( dom_dir, state.get_id()+'_checkboxes.txt'), 'w', encoding='utf-8' ) as f:
json.dump(state.get_checkboxes_json(iframe_key ), f, indent=2, sort_keys=True, ensure_ascii=False)
with codecs.open( os.path.join( dom_dir, state.get_id()+'_clicks.txt'), 'w', encoding='utf-8' ) as f:
json.dump(state.get_candidate_clickables_json( iframe_key ), f, indent=2, sort_keys=True, ensure_ascii=False)
with codecs.open( os.path.join( state_dir, 'iframe_list.json'), 'w', encoding='utf-8' ) as f:
json.dump( iframe_key_dict, f, indent=2, sort_keys=True, ensure_ascii=False)
"""
TODO: turn TempFile stateDom into FilePath stateDom
"""
state.clear_dom()
except Exception as e:
logging.error(' save dom : %s \t\t__from automata.py save_dom()', str(e))
def save_dom(self, state):
try:
#make dir for each state
state_dir = os.path.join(self.configuration.get_abs_path('dom'),
state.get_id())
if not os.path.isdir(state_dir):
os.makedirs(state_dir)
iframe_key_dict = {'num': 0}
for stateDom in state.get_dom_list(self.configuration):
iframe_key = ';'.join(stateDom['iframe_path']
) if stateDom['iframe_path'] else None
#make new dir for iframe
if stateDom['iframe_path']:
iframe_key_dict['num'] += 1
iframe_key_dict[str(iframe_key_dict['num'])] = {
'path': stateDom['iframe_path'],
'url': stateDom['url']
}
dom_dir = os.path.join(
self.configuration.get_abs_path('dom'), state.get_id(),
str(iframe_key_dict['num']))
if not os.path.isdir(dom_dir):
os.makedirs(dom_dir)
else:
iframe_key_dict['basic'] = {'url': stateDom['url']}
dom_dir = os.path.join(
self.configuration.get_abs_path('dom'), state.get_id())
with codecs.open(os.path.join(dom_dir,
state.get_id() + '.txt'),
'w',
encoding='utf-8') as f:
f.write(stateDom['dom'])
with codecs.open(os.path.join(dom_dir,
state.get_id() + '_nor.txt'),
'w',
encoding='utf-8') as f:
f.write(DomAnalyzer.normalize(stateDom['dom']))
with codecs.open(os.path.join(dom_dir,
state.get_id() + '_inputs.txt'),
'w',
encoding='utf-8') as f:
json.dump(state.get_inputs_json(iframe_key),
f,
indent=2,
sort_keys=True,
ensure_ascii=False)
with codecs.open(os.path.join(dom_dir,
state.get_id() + '_selects.txt'),
'w',
encoding='utf-8') as f:
json.dump(state.get_selects_json(iframe_key),
f,
indent=2,
sort_keys=True,
ensure_ascii=False)
with codecs.open(os.path.join(dom_dir,
state.get_id() + '_radios.txt'),
'w',
encoding='utf-8') as f:
json.dump(state.get_radios_json(iframe_key),
f,
indent=2,
sort_keys=True,
ensure_ascii=False)
with codecs.open(os.path.join(
dom_dir,
state.get_id() + '_checkboxes.txt'),
'w',
encoding='utf-8') as f:
json.dump(state.get_checkboxes_json(iframe_key),
f,
indent=2,
sort_keys=True,
ensure_ascii=False)
with codecs.open(os.path.join(dom_dir,
state.get_id() + '_clicks.txt'),
'w',
encoding='utf-8') as f:
json.dump(state.get_candidate_clickables_json(iframe_key),
f,
indent=2,
sort_keys=True,
ensure_ascii=False)
with codecs.open(os.path.join(state_dir, 'iframe_list.json'),
'w',
encoding='utf-8') as f:
json.dump(iframe_key_dict,
f,
indent=2,
sort_keys=True,
ensure_ascii=False)
"""
TODO: turn TempFile stateDom into FilePath stateDom
"""
state.clear_dom()
except Exception as e:
logging.error(' save dom : %s \t\t__from automata.py save_dom()',
str(e))
