def parse_gmail(case):
file_path = case.takeout_my_activity_gmail_path
if os.path.exists(file_path) == False:
return False
with open(file_path, 'r', encoding='utf-8') as f:
file_contents = f.read()
soup = BeautifulSoup(file_contents, 'lxml')
list_gmail_logs = TakeoutHtmlParser.find_log(soup)
if list_gmail_logs != []:
for i in trange(
len(list_gmail_logs),
desc=
"[Parsing the My Activity -> Gmail data..............]",
unit="epoch"):
# print("..........................................................................")
dic_my_activity_gmail = {
'service': "",
'type': "",
'keyword_url': "",
'keyword': "",
'timestamp': ""
}
MyActivityGmail.parse_gmail_log_title(
dic_my_activity_gmail, list_gmail_logs[i])
MyActivityGmail.parse_gmail_log_body(
dic_my_activity_gmail, list_gmail_logs[i])
MyActivityGmail.insert_log_info_to_preprocess_db(
dic_my_activity_gmail, case.preprocess_db_path)
def parse_maps(case):
file_path = case.takeout_my_activity_maps_path
if os.path.exists(file_path) == False:
return False
with open(file_path, 'r', encoding='utf-8') as f:
file_contents = f.read()
soup = BeautifulSoup(file_contents, 'lxml')
list_maps_logs = TakeoutHtmlParser.find_log(soup)
if list_maps_logs != []:
for i in trange(
len(list_maps_logs),
desc=
"[Parsing the My Activity -> Maps data...............]",
unit="epoch"):
# print("..........................................................................")
dic_my_activity_maps = {'timestamp':"", 'service':"", 'type':"", 'keyword':"", 'keyword_url':"", \
'keyword_latitude':"", 'keyword_longitude':"", 'latitude':"", 'longitude':"", 'geodata_description':"", \
'used_device':""}
MyActivityMaps.parse_maps_log_title(
dic_my_activity_maps, list_maps_logs[i])
MyActivityMaps.parse_maps_log_body(dic_my_activity_maps,
list_maps_logs[i])
MyActivityMaps.parse_maps_log_caption(
dic_my_activity_maps, list_maps_logs[i])
MyActivityMaps.insert_log_info_to_preprocess_db(
dic_my_activity_maps, case.preprocess_db_path)