|
1
|
|
|
""" |
|
2
|
|
|
Data Input Output class |
|
3
|
|
|
""" |
|
4
|
|
|
# package to add support for multi-language (i18n) |
|
5
|
|
|
import gettext |
|
6
|
|
|
# package to handle files/folders and related metadata/operations |
|
7
|
|
|
import os |
|
8
|
|
|
# package facilitating Data Frames manipulation |
|
9
|
|
|
import pandas as pd |
|
10
|
|
|
|
|
11
|
|
|
|
|
12
|
|
|
class DataInputOutput: |
|
13
|
|
|
lcl = None |
|
14
|
|
|
|
|
15
|
|
|
def __init__(self, default_language='en_US'): |
|
16
|
|
|
current_script = os.path.basename(__file__).replace('.py', '') |
|
17
|
|
|
lang_folder = os.path.join(os.path.dirname(__file__), current_script + '_Locale') |
|
18
|
|
|
self.lcl = gettext.translation(current_script, lang_folder, languages=[default_language]) |
|
19
|
|
|
|
|
20
|
|
|
def fn_build_feedback_for_logger(self, operation_details): |
|
21
|
|
|
messages = {} |
|
22
|
|
|
if operation_details['operation'] == 'load': |
|
23
|
|
|
messages = { |
|
24
|
|
|
'failed': self.lcl.gettext( |
|
25
|
|
|
'Error encountered on loading Pandas Data Frame ' |
|
26
|
|
|
+ 'from {file_type} file type (see below)') |
|
27
|
|
|
.replace('{file_type}', operation_details['format'].upper()), |
|
28
|
|
|
'success': self.lcl.gettext( |
|
29
|
|
|
'All {files_counted} files of type {file_type} ' |
|
30
|
|
|
+ 'successfully added to a Pandas Data Frame') |
|
31
|
|
|
.replace('{files_counted}', str(operation_details['files counted'])) |
|
32
|
|
|
.replace('{file_type}', str(operation_details['format'])) |
|
33
|
|
|
} |
|
34
|
|
|
elif operation_details['operation'] == 'save': |
|
35
|
|
|
messages = { |
|
36
|
|
|
'failed': self.lcl.gettext( |
|
37
|
|
|
'Error encountered on saving Pandas Data Frame ' |
|
38
|
|
|
+ 'into a {file_type} file type (see below)') |
|
39
|
|
|
.replace('{file_type}', operation_details['format'].upper()), |
|
40
|
|
|
'success': self.lcl.gettext( |
|
41
|
|
|
'Pandas Data Frame has just been saved to file "{file_name}", ' |
|
42
|
|
|
+ 'considering {file_type} as file type') |
|
43
|
|
|
.replace('{file_name}', operation_details['name']) |
|
44
|
|
|
.replace('{file_type}', operation_details['format']), |
|
45
|
|
|
} |
|
46
|
|
|
return messages |
|
47
|
|
|
|
|
48
|
|
|
def fn_default_load_dict_message(self, in_file_list, in_format): |
|
49
|
|
|
return { |
|
50
|
|
|
'error details': None, |
|
51
|
|
|
'files counted': len(in_file_list), |
|
52
|
|
|
'format': in_format, |
|
53
|
|
|
'operation': 'load', |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
def fn_file_operation_logger(self, local_logger, in_logger_dict): |
|
57
|
|
|
messages = self.fn_build_feedback_for_logger(in_logger_dict) |
|
58
|
|
|
if in_logger_dict['error details'] is None: |
|
59
|
|
|
local_logger.info(messages['success']) |
|
60
|
|
|
else: |
|
61
|
|
|
local_logger.error(messages['failed']) |
|
62
|
|
|
local_logger.error(in_logger_dict['error details']) |
|
63
|
|
|
|
|
64
|
|
|
def fn_load_file_type_csv_into_data_frame(self, local_logger, in_file_list, csv_delimiter): |
|
65
|
|
|
details_for_logger = self.fn_default_load_dict_message(in_file_list, 'CSV') |
|
66
|
|
|
try: |
|
67
|
|
|
out_data_frame = pd.concat([pd.read_csv(filepath_or_buffer=current_file, |
|
68
|
|
|
delimiter=csv_delimiter, |
|
69
|
|
|
cache_dates=True, |
|
70
|
|
|
index_col=None, |
|
71
|
|
|
memory_map=True, |
|
72
|
|
|
low_memory=False, |
|
73
|
|
|
encoding='utf-8', |
|
74
|
|
|
) for current_file in in_file_list]) |
|
75
|
|
|
except Exception as err: |
|
76
|
|
|
details_for_logger['error details'] = err |
|
77
|
|
|
self.fn_file_operation_logger(local_logger, details_for_logger) |
|
78
|
|
|
return out_data_frame |
|
79
|
|
|
|
|
80
|
|
|
def fn_load_file_type_excel_into_data_frame(self, local_logger, in_file_list): |
|
81
|
|
|
out_data_frame = None |
|
82
|
|
|
details_for_logger = self.fn_default_load_dict_message(in_file_list, 'Excel') |
|
83
|
|
|
try: |
|
84
|
|
|
out_data_frame = pd.concat([pd.read_excel(io=current_file, |
|
85
|
|
|
verbose=True, |
|
86
|
|
|
) for current_file in in_file_list]) |
|
87
|
|
|
except Exception as err: |
|
88
|
|
|
details_for_logger['error details'] = err |
|
89
|
|
|
self.fn_file_operation_logger(local_logger, details_for_logger) |
|
90
|
|
|
return out_data_frame |
|
91
|
|
|
|
|
92
|
|
|
def fn_load_file_type_pickle_into_data_frame(self, local_logger, in_file_list, |
|
93
|
|
|
in_compression='infer'): |
|
94
|
|
|
out_data_frame = None |
|
95
|
|
|
details_for_logger = self.fn_default_load_dict_message(in_file_list, 'Pickle') |
|
96
|
|
|
try: |
|
97
|
|
|
out_data_frame = pd.concat([pd.read_pickle(filepath_or_buffer=current_file, |
|
98
|
|
|
compression=in_compression, |
|
99
|
|
|
) for current_file in in_file_list]) |
|
100
|
|
|
except Exception as err: |
|
101
|
|
|
details_for_logger['error details'] = err |
|
102
|
|
|
self.fn_file_operation_logger(local_logger, details_for_logger) |
|
103
|
|
|
return out_data_frame |
|
104
|
|
|
|
|
105
|
|
|
def fn_save_data_frame_to_csv(self, local_logger, in_data_frame, in_file_details, logger_dict): |
|
106
|
|
|
if in_file_details['format'].lower() == 'csv': |
|
107
|
|
|
if 'field-delimiter' not in in_file_details: |
|
108
|
|
|
in_file_details['field-delimiter'] = os.pathsep |
|
109
|
|
|
try: |
|
110
|
|
|
in_data_frame.to_csv(path_or_buf=in_file_details['name'], |
|
111
|
|
|
sep=in_file_details['field-delimiter'], |
|
112
|
|
|
header=True, |
|
113
|
|
|
index=False, |
|
114
|
|
|
encoding='utf-8') |
|
115
|
|
|
except Exception as err: |
|
116
|
|
|
logger_dict['error details'] = err |
|
117
|
|
|
self.fn_file_operation_logger(local_logger, logger_dict) |
|
118
|
|
|
|
|
119
|
|
|
def fn_save_data_frame_to_excel(self, local_logger, in_data_frame, in_file_details, |
|
120
|
|
|
logger_dict): |
|
121
|
|
|
if in_file_details['format'].lower() == 'excel': |
|
122
|
|
|
try: |
|
123
|
|
|
in_data_frame.to_excel(excel_writer=in_file_details['name'], |
|
124
|
|
|
engine='xlsxwriter', |
|
125
|
|
|
freeze_panes=(1, 1), |
|
126
|
|
|
verbose=True) |
|
127
|
|
|
except Exception as err: |
|
128
|
|
|
logger_dict['error details'] = err |
|
129
|
|
|
self.fn_file_operation_logger(local_logger, logger_dict) |
|
130
|
|
|
|
|
131
|
|
|
def fn_save_data_frame_to_pickle(self, local_logger, in_data_frame, in_file_details, |
|
132
|
|
|
logger_dict): |
|
133
|
|
|
if in_file_details['format'].lower() == 'pickle': |
|
134
|
|
|
if 'compression' not in in_file_details: |
|
135
|
|
|
in_file_details['compression'] = 'gzip' |
|
136
|
|
|
try: |
|
137
|
|
|
in_data_frame.to_pickle(path=in_file_details['name'], |
|
138
|
|
|
compression=in_file_details['compression']) |
|
139
|
|
|
except Exception as err: |
|
140
|
|
|
logger_dict['error details'] = err |
|
141
|
|
|
self.fn_file_operation_logger(local_logger, logger_dict) |
|
142
|
|
|
|
|
143
|
|
|
def fn_store_data_frame_to_file(self, local_logger, timmer, in_data_frame, in_file_details): |
|
144
|
|
|
timmer.start() |
|
145
|
|
|
self.fn_store_data_frame_to_file_validation(local_logger, in_file_details) |
|
146
|
|
|
if 'format' in in_file_details: |
|
147
|
|
|
details_for_logger = { |
|
148
|
|
|
'error details': None, |
|
149
|
|
|
'file name': in_file_details['name'], |
|
150
|
|
|
'format': in_file_details['format'], |
|
151
|
|
|
'operation': 'save', |
|
152
|
|
|
} |
|
153
|
|
|
self.fn_save_data_frame_to_csv(local_logger, in_data_frame, |
|
154
|
|
|
in_file_details, details_for_logger) |
|
155
|
|
|
self.fn_save_data_frame_to_excel(local_logger, in_data_frame, |
|
156
|
|
|
in_file_details, details_for_logger) |
|
157
|
|
|
self.fn_save_data_frame_to_pickle(local_logger, in_data_frame, |
|
158
|
|
|
in_file_details, details_for_logger) |
|
159
|
|
|
timmer.stop() |
|
160
|
|
|
|
|
161
|
|
|
def fn_store_data_frame_to_file_validation(self, local_logger, in_file_details): |
|
162
|
|
|
if 'format' in in_file_details: |
|
163
|
|
|
implemented_file_formats = ['csv', 'excel', 'pickle'] |
|
164
|
|
|
given_format = in_file_details['format'].lower() |
|
165
|
|
|
if given_format not in implemented_file_formats: |
|
166
|
|
|
local_logger.error(self.lcl.gettext( |
|
167
|
|
|
'File "format" attribute has a value of "{format_value}" ' |
|
168
|
|
|
+ 'which is not among currently implemented values: ' |
|
169
|
|
|
+ '"{implemented_file_formats}", therefore file saving is not possible') |
|
170
|
|
|
.replace('{format_value}', given_format) |
|
171
|
|
|
.replace('{implemented_file_formats}', |
|
172
|
|
|
'", "'.join(implemented_file_formats))) |
|
173
|
|
|
else: |
|
174
|
|
|
local_logger.error(self.lcl.gettext('File "format" attribute is mandatory ' |
|
175
|
|
|
+ 'in the file setting, but missing, ' |
|
176
|
|
|
+ 'therefore file saving is not possible')) |
|
177
|
|
|
|
danielgp /
db-extractor
