def __init__(self, file_name): DataFile.__init__(self, file_name) result_file_info = self._get_file_info() self.topic = Topic.create_or_update(result_file_info['topic_id']) self.user = User.create_or_update(result_file_info['user_id']) self.query = Query.create_or_update(result_file_info['query_id'], topic=self.topic, user=self.user) self.__parse()
def __parse( self ):
with open( self.file_name, 'r' ) as result_file:
result_reader = csv.DictReader( result_file, delimiter=',')
for row in result_reader:
topic = Topic.create_or_update( row['topic'] )
user = User.create_or_update( row['userid'] )
condition = Condition.create_or_update( row['condition'] )
autocomplete = row['autocomplete_used'] == 1
query = Query.create_or_update( row['queryid'], topic = topic, user = user, condition = condition, autocomplete = autocomplete, query_text = row['terms'], precision = self.__build_precision_dict( row ) )
def __init__(self, file_name): DataFile.__init__(self, file_name) result_file_info = self._get_file_info() self.topic = Topic.create_or_update( result_file_info['topic_id'] ) self.user = User.create_or_update( result_file_info['user_id'] ) self.condition = Condition.create_or_update( result_file_info['condition'] ) self.__create_or_update_session() self.query = Query.create_or_update( result_file_info['query_id'], topic = self.topic, user = self.user, session = self.session ) self.actions = self.__parse() self.topic.add_actions( self.actions ) self.user.add_actions( self.actions ) self.query.add_actions( self.actions ) self.session.add_actions( self.actions ) self.session.add_query( self.query )
def __parse(self):
with open(self.file_name, 'r') as result_file:
result_reader = csv.DictReader(result_file, delimiter=',')
for row in result_reader:
topic = Topic.create_or_update(row['topic'])
user = User.create_or_update(row['userid'])
condition = Condition.create_or_update(row['condition'])
autocomplete = row['autocomplete_used'] == 1
query = Query.create_or_update(
row['queryid'],
topic=topic,
user=user,
condition=condition,
autocomplete=autocomplete,
query_text=row['terms'],
precision=self.__build_precision_dict(row))
def __parse(self):
actions = []
query_text = None
serp_page_num = None
with open(self.file_name, 'r') as log_file:
for line in log_file:
parsed_line = _parse_line(line)
# these occur in weird places and have weird data, so ignore them to avoid trouble
if parsed_line['action'] in [
'PERFORMANCE', 'DEMOGRAPHICS_SURVEY_STARTED',
'DEMOGRAPHICS_SURVEY_COMPLETED',
'SELF_SEARCH_EFFICACY_SURVEY_STARTED',
'SELF_SEARCH_EFFICACY_SURVEY_COMPLETED',
'PRE_TASK_SURVEY_COMPLETED',
'POST_TASK_SURVEY_COMPLETED', 'SEARCH_TASK_VIEWED'
]:
continue
if parsed_line['action'] == 'QUERY_ISSUED' or parsed_line[
'action'] == 'QUERY_SUGGESTION_ISSUED':
self.query_counter += 1
query_text = parsed_line.get('action_parameters', None)
serp_page_num = None
#print ("%s - %s - X - %s: %s" % (self.query_counter, parsed_line['user_id'], parsed_line['topic_id'], query_text))
topic = Topic.create_or_update(parsed_line['topic_id'])
user = User.create_or_update(parsed_line['user_id'])
condition = Condition.create_or_update(
parsed_line['condition'])
session = self.__create_or_update_session(
user, topic, condition)
query_id = str(self.query_counter)
# These actions belong with the next query
if parsed_line['action'] in [
'SEARCH_TASK_COMMENCED', 'VIEW_SEARCH_BOX'
]:
query_id = str(self.query_counter + 1)
query = Query.create_or_update(query_id,
topic=topic,
user=user,
session=session,
query_text=query_text)
timestamp = _parse_datetime(parsed_line['date'],
parsed_line['time'])
action = Action(timestamp=timestamp,
session=session,
condition=condition,
action_type=parsed_line['action'],
query=query,
serp_page_num=serp_page_num,
action_parameters=parsed_line.get(
'action_parameters', None))
if action.is_serp_switch_event():
serp_page_num = int(action.result_page)
actions.append(action)
topic.add_actions([action])
user.add_actions([action])
query.add_actions([action])
session.add_actions([action])
session.add_query(query)
return sorted(actions, key=lambda action: action.timestamp)
