|
1
|
|
|
""" |
|
2
|
|
|
facilitates File Operations |
|
3
|
|
|
""" |
|
4
|
|
|
# package to handle date and times |
|
5
|
|
|
from datetime import datetime |
|
6
|
|
|
# package to use for checksum calculations (in this file) |
|
7
|
|
|
import hashlib |
|
8
|
|
|
# package to handle json files |
|
9
|
|
|
import json |
|
10
|
|
|
# package to facilitate operating system operations |
|
11
|
|
|
import os |
|
12
|
|
|
# package to facilitate os path manipulations |
|
13
|
|
|
import pathlib |
|
14
|
|
|
# package regular expressions |
|
15
|
|
|
import re |
|
16
|
|
|
|
|
17
|
|
|
|
|
18
|
|
|
class FileOperations: |
|
19
|
|
|
timestamp_format = '%Y-%m-%d %H:%M:%S.%f %Z' |
|
20
|
|
|
|
|
21
|
|
|
def build_file_list(self, local_logger, timmer, given_input_file): |
|
22
|
|
|
if re.search(r'(\*|\?)*', given_input_file): |
|
23
|
|
|
local_logger.debug('File pattern identified') |
|
24
|
|
|
parent_directory = os.path.dirname(given_input_file) |
|
25
|
|
|
# loading from a specific folder all files matching a given pattern into a file list |
|
26
|
|
|
relevant_files_list = self.fn_build_relevant_file_list(local_logger, timmer, |
|
27
|
|
|
parent_directory, |
|
28
|
|
|
given_input_file) |
|
29
|
|
|
else: |
|
30
|
|
|
local_logger.debug('Specific file name provided') |
|
31
|
|
|
relevant_files_list = [given_input_file] |
|
32
|
|
|
return relevant_files_list |
|
33
|
|
|
|
|
34
|
|
|
@staticmethod |
|
35
|
|
|
def fn_build_relevant_file_list(local_logger, timmer, in_folder, matching_pattern): |
|
36
|
|
|
timmer.start() |
|
37
|
|
|
local_logger.info('Listing all files within ' + in_folder |
|
38
|
|
|
+ ' folder looking for ' + matching_pattern + ' as matching pattern') |
|
39
|
|
|
list_files = [] |
|
40
|
|
|
file_counter = 0 |
|
41
|
|
|
if os.path.isdir(in_folder): |
|
42
|
|
|
working_path = pathlib.Path(in_folder) |
|
43
|
|
|
for current_file in working_path.iterdir(): |
|
44
|
|
|
if current_file.is_file() and current_file.match(matching_pattern): |
|
45
|
|
|
list_files.append(file_counter) |
|
46
|
|
|
list_files[file_counter] = str(current_file.absolute()) |
|
47
|
|
|
local_logger.info(str(current_file.absolute()) + ' file identified') |
|
48
|
|
|
file_counter = file_counter + 1 |
|
49
|
|
|
local_logger.info(str(file_counter) + ' file(s) from ' + in_folder + ' folder identified!') |
|
50
|
|
|
timmer.stop() |
|
51
|
|
|
return list_files |
|
52
|
|
|
|
|
53
|
|
|
def fn_get_file_content(self, in_file_handler, in_content_type): |
|
54
|
|
|
if in_content_type == 'json': |
|
55
|
|
|
try: |
|
56
|
|
|
json_interpreted_details = json.load(in_file_handler) |
|
57
|
|
|
print(datetime.utcnow().strftime(self.timestamp_format) + |
|
58
|
|
|
'I have interpreted JSON structure from given file') |
|
59
|
|
|
return json_interpreted_details |
|
60
|
|
|
except Exception as e: |
|
61
|
|
|
print(datetime.utcnow().strftime(self.timestamp_format) + |
|
62
|
|
|
'Error encountered when trying to interpret JSON') |
|
63
|
|
|
print(e) |
|
64
|
|
|
elif in_content_type == 'raw': |
|
65
|
|
|
raw_interpreted_file = in_file_handler.read() |
|
66
|
|
|
print(datetime.utcnow().strftime(self.timestamp_format) + |
|
67
|
|
|
'I have read file entire content') |
|
68
|
|
|
return raw_interpreted_file |
|
69
|
|
|
else: |
|
70
|
|
|
print(datetime.utcnow().strftime(self.timestamp_format) + |
|
71
|
|
|
'Unknown content type provided, expected either "json" or "raw" but got ' |
|
72
|
|
|
+ in_content_type) |
|
73
|
|
|
|
|
74
|
|
|
@staticmethod |
|
75
|
|
|
def fn_get_file_statistics(file_to_evaluate): |
|
76
|
|
|
try: |
|
77
|
|
|
file_handler = open(file=file_to_evaluate, mode='r', encoding='mbcs') |
|
78
|
|
|
except UnicodeDecodeError: |
|
79
|
|
|
file_handler = open(file=file_to_evaluate, mode='r', encoding='utf-8') |
|
80
|
|
|
file_content = file_handler.read().encode() |
|
81
|
|
|
file_handler.close() |
|
82
|
|
|
file_checksums = { |
|
83
|
|
|
'md5': hashlib.md5(file_content).hexdigest(), |
|
84
|
|
|
'sha1': hashlib.sha1(file_content).hexdigest(), |
|
85
|
|
|
'sha256': hashlib.sha256(file_content).hexdigest(), |
|
86
|
|
|
'sha512': hashlib.sha512(file_content).hexdigest(), |
|
87
|
|
|
} |
|
88
|
|
|
f_dts = { |
|
89
|
|
|
'created': datetime.fromtimestamp(os.path.getctime(file_to_evaluate)), |
|
90
|
|
|
'modified': datetime.fromtimestamp(os.path.getctime(file_to_evaluate)), |
|
91
|
|
|
} |
|
92
|
|
|
return { |
|
93
|
|
|
'date when created': datetime.strftime(f_dts['created'], '%Y-%m-%d %H:%M:%S.%f'), |
|
94
|
|
|
'date when last modified': datetime.strftime(f_dts['modified'], '%Y-%m-%d %H:%M:%S.%f'), |
|
95
|
|
|
'size [bytes]': os.path.getsize(file_to_evaluate), |
|
96
|
|
|
'MD5-Checksum': file_checksums['md5'], |
|
97
|
|
|
'SHA256-Checksum': file_checksums['sha256'], |
|
98
|
|
|
'SHA512-Checksum': file_checksums['sha512'], |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
|
|
@staticmethod |
|
102
|
|
|
def fn_move_files(local_logger, timmer, source_folder, file_names, destination_folder): |
|
103
|
|
|
timmer.start() |
|
104
|
|
|
resulted_files = [] |
|
105
|
|
|
for current_file in file_names: |
|
106
|
|
|
new_file_name = current_file.replace(source_folder, destination_folder) |
|
107
|
|
|
if new_file_name.is_file(): |
|
108
|
|
|
os.replace(current_file, new_file_name) |
|
109
|
|
|
local_logger.info('File ' + current_file |
|
110
|
|
|
+ ' has just been been overwritten as ' + new_file_name) |
|
111
|
|
|
else: |
|
112
|
|
|
os.rename(current_file, new_file_name) |
|
113
|
|
|
local_logger.info('File ' + current_file |
|
114
|
|
|
+ ' has just been renamed as ' + new_file_name) |
|
115
|
|
|
resulted_files.append(new_file_name) |
|
116
|
|
|
timmer.stop() |
|
117
|
|
|
return resulted_files |
|
118
|
|
|
|
|
119
|
|
|
def fn_open_file_and_get_content(self, input_file, content_type='json'): |
|
120
|
|
|
if os.path.isfile(input_file): |
|
121
|
|
|
with open(input_file, 'r') as file_handler: |
|
122
|
|
|
print(datetime.utcnow().strftime(self.timestamp_format) + |
|
123
|
|
|
'I have opened file: ' + input_file) |
|
124
|
|
|
return self.fn_get_file_content(file_handler, content_type) |
|
125
|
|
|
else: |
|
126
|
|
|
print(datetime.utcnow().strftime(self.timestamp_format) + |
|
127
|
|
|
'Given file ' + input_file + ' does not exist, please check your inputs!') |
|
128
|
|
|
|
|
129
|
|
|
def fn_store_file_statistics(self, local_logger, timmer, file_name, file_meaning): |
|
130
|
|
|
timmer.start() |
|
131
|
|
|
file_name_variable_type = str(type(file_name)) |
|
132
|
|
|
list_file_names = [file_name] |
|
133
|
|
|
if file_name_variable_type == "<class 'list'>": |
|
134
|
|
|
list_file_names = file_name |
|
135
|
|
|
for current_file_name in list_file_names: |
|
136
|
|
|
local_logger.info(file_meaning + ' file "' + current_file_name |
|
137
|
|
|
+ '" has the following characteristics: ' |
|
138
|
|
|
+ str(self.fn_get_file_statistics(current_file_name))) |
|
139
|
|
|
timmer.stop() |
|
140
|
|
|
|
danielgp /
db-extractor
