|
1
|
|
|
""" |
|
2
|
|
|
main class to support Extract script |
|
3
|
|
|
""" |
|
4
|
|
|
# useful methods to measure time performance by small pieces of code |
|
5
|
|
|
from codetiming import Timer |
|
6
|
|
|
# package to facilitate operating system operations |
|
7
|
|
|
import os |
|
8
|
|
|
# package to add support for multi-language (i18n) |
|
9
|
|
|
import gettext |
|
10
|
|
|
# custom classes specific to this project |
|
11
|
|
|
from common.BasicNeeds import BasicNeeds |
|
12
|
|
|
from common.CommandLineArgumentsManagement import CommandLineArgumentsManagement |
|
13
|
|
|
from common.DataManipulator import DataManipulator |
|
14
|
|
|
from common.FileOperations import FileOperations |
|
15
|
|
|
from common.LoggingNeeds import LoggingNeeds |
|
16
|
|
|
from db_extractor.BasicNeedsForExtractor import BasicNeedsForExtractor |
|
17
|
|
|
from db_extractor.DatabaseTalker import DatabaseTalker |
|
18
|
|
|
from db_extractor.ParameterHandling import ParameterHandling |
|
19
|
|
|
|
|
20
|
|
|
|
|
21
|
|
|
class ExtractNeeds: |
|
22
|
|
|
class_bn = None |
|
23
|
|
|
class_bnfe = None |
|
24
|
|
|
class_clam = None |
|
25
|
|
|
class_dbt = None |
|
26
|
|
|
class_dm = None |
|
27
|
|
|
class_fo = None |
|
28
|
|
|
class_ln = None |
|
29
|
|
|
class_ph = None |
|
30
|
|
|
config = None |
|
31
|
|
|
file_extract_sequence = None |
|
32
|
|
|
locale = None |
|
33
|
|
|
parameters = None |
|
34
|
|
|
script = None |
|
35
|
|
|
source_systems = None |
|
36
|
|
|
timer = None |
|
37
|
|
|
user_credentials = None |
|
38
|
|
|
|
|
39
|
|
|
def __init__(self, destination_script, default_language='en_US'): |
|
40
|
|
|
self.script = destination_script |
|
41
|
|
|
current_file_basename = os.path.basename(__file__).replace('.py', '') |
|
42
|
|
|
lang_folder = os.path.join(os.path.dirname(__file__), current_file_basename + '_Locale') |
|
43
|
|
|
self.locale = gettext.translation(current_file_basename, lang_folder, |
|
44
|
|
|
languages=[default_language]) |
|
45
|
|
|
# instantiate Basic Needs class |
|
46
|
|
|
self.class_bn = BasicNeeds(default_language) |
|
47
|
|
|
# instantiate Extractor Specific Needs class |
|
48
|
|
|
self.class_bnfe = BasicNeedsForExtractor(default_language) |
|
49
|
|
|
# instantiate File Operations class |
|
50
|
|
|
self.class_fo = FileOperations(default_language) |
|
51
|
|
|
# instantiate File Operations class |
|
52
|
|
|
self.class_dbt = DatabaseTalker(default_language) |
|
53
|
|
|
# instantiate Data Manipulator class, useful to manipulate data frames |
|
54
|
|
|
self.class_dm = DataManipulator(default_language) |
|
55
|
|
|
# instantiate Command Line Arguments class |
|
56
|
|
|
self.class_clam = CommandLineArgumentsManagement(default_language) |
|
57
|
|
|
# instantiate Logger class |
|
58
|
|
|
self.class_ln = LoggingNeeds() |
|
59
|
|
|
# instantiate Parameter Handling class |
|
60
|
|
|
self.class_ph = ParameterHandling(default_language) |
|
61
|
|
|
|
|
62
|
|
|
def close_connection(self, local_logger): |
|
63
|
|
|
self.timer.start() |
|
64
|
|
|
local_logger.info(self.locale.gettext('Closing DB connection')) |
|
65
|
|
|
self.class_dbt.conn.close() |
|
66
|
|
|
local_logger.info(self.locale.gettext('Closing DB completed')) |
|
67
|
|
|
self.timer.stop() |
|
68
|
|
|
|
|
69
|
|
|
def close_cursor(self, local_logger, in_cursor): |
|
70
|
|
|
self.timer.start() |
|
71
|
|
|
local_logger.info(self.locale.gettext('Free DB result-set started')) |
|
72
|
|
|
in_cursor.close() |
|
73
|
|
|
local_logger.info(self.locale.gettext('Free DB result-set completed')) |
|
74
|
|
|
self.timer.stop() |
|
75
|
|
|
|
|
76
|
|
|
def extract_query_to_result_set(self, local_logger, in_cursor, in_dictionary): |
|
77
|
|
|
this_session = in_dictionary['session'] |
|
78
|
|
|
this_query = in_dictionary['query'] |
|
79
|
|
|
# get query parameters into a tuple |
|
80
|
|
|
tuple_parameters = self.class_ph.handle_query_parameters(local_logger, this_session, |
|
81
|
|
|
this_session['start-isoweekday']) |
|
82
|
|
|
# measure expected number of parameters |
|
83
|
|
|
expected_number_of_parameters = str(this_query).count('%s') |
|
84
|
|
|
# simulate final query to log (useful for debugging purposes) |
|
85
|
|
|
simulated_query = self.class_ph.simulate_final_query(local_logger, self.timer, this_query, |
|
86
|
|
|
expected_number_of_parameters, |
|
87
|
|
|
tuple_parameters) |
|
88
|
|
|
local_logger.info(self.locale.gettext('Query with parameters interpreted is: %s') \ |
|
89
|
|
|
.replace('%s', |
|
90
|
|
|
self.class_bn.fn_multi_line_string_to_single_line(\ |
|
91
|
|
|
simulated_query))) |
|
92
|
|
|
# actual execution of the query |
|
93
|
|
|
in_cursor = self.class_dbt.execute_query(local_logger, self.timer, in_cursor, this_query, |
|
94
|
|
|
expected_number_of_parameters, tuple_parameters) |
|
95
|
|
|
# bringing the information from server (data transfer) |
|
96
|
|
|
return { |
|
97
|
|
|
'columns' : self.class_dbt.get_column_names(local_logger, self.timer, in_cursor), |
|
98
|
|
|
'result_set': self.class_dbt.fetch_executed_query(local_logger, self.timer, in_cursor), |
|
99
|
|
|
'rows_counted': in_cursor.rowcount, |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
def initiate_logger_and_timer(self): |
|
103
|
|
|
# initiate logger |
|
104
|
|
|
self.class_ln.initiate_logger(self.parameters.output_log_file, self.script) |
|
105
|
|
|
# initiate localization specific for this script |
|
106
|
|
|
# define global timer to use |
|
107
|
|
|
self.timer = Timer(self.script, |
|
108
|
|
|
text = self.locale.gettext('Time spent is {seconds}'), |
|
109
|
|
|
logger = self.class_ln.logger.debug) |
|
110
|
|
|
|
|
111
|
|
|
def load_configuration(self): |
|
112
|
|
|
# load application configuration (inputs are defined into a json file) |
|
113
|
|
|
ref_folder = os.path.dirname(__file__).replace('db_extractor', 'config') |
|
114
|
|
|
config_file = os.path.join(ref_folder, 'db-extractor.json').replace('\\', '/') |
|
115
|
|
|
self.config = self.class_fo.fn_open_file_and_get_content(config_file) |
|
116
|
|
|
# get command line parameter values |
|
117
|
|
|
self.parameters = self.class_clam.parse_arguments(self.config['input_options'][self.script]) |
|
118
|
|
|
# checking inputs, if anything is invalid an exit(1) will take place |
|
119
|
|
|
self.class_bn.fn_check_inputs(self.parameters) |
|
120
|
|
|
# checking inputs, if anything is invalid an exit(1) will take place |
|
121
|
|
|
self.class_bnfe.fn_check_inputs_specific(self.parameters) |
|
122
|
|
|
|
|
123
|
|
|
def load_extraction_sequence_and_dependencies(self): |
|
124
|
|
|
self.timer.start() |
|
125
|
|
|
self.file_extract_sequence = self.class_fo.fn_open_file_and_get_content( |
|
126
|
|
|
self.parameters.input_extracting_sequence_file, 'json') |
|
127
|
|
|
self.class_ln.logger.info(self.locale.gettext( \ |
|
128
|
|
|
'Configuration file name with extracting sequence(es) has been loaded')) |
|
129
|
|
|
self.timer.stop() |
|
130
|
|
|
# store file statistics |
|
131
|
|
|
self.class_fo.fn_store_file_statistics(self.class_ln.logger, self.timer, |
|
132
|
|
|
self.parameters.input_extracting_sequence_file, |
|
133
|
|
|
self.locale.gettext('Configuration file name with ' |
|
134
|
|
|
+ 'extracting sequence(es)')) |
|
135
|
|
|
# get the source system details from provided file |
|
136
|
|
|
self.timer.start() |
|
137
|
|
|
self.source_systems = self.class_fo.fn_open_file_and_get_content( \ |
|
138
|
|
|
self.parameters.input_source_system_file, 'json')['Systems'] |
|
139
|
|
|
self.class_ln.logger.info(self.locale.gettext('Source Systems file name has been loaded')) |
|
140
|
|
|
self.timer.stop() |
|
141
|
|
|
self.class_fo.fn_store_file_statistics(self.class_ln.logger, self.timer, |
|
142
|
|
|
self.parameters.input_source_system_file, |
|
143
|
|
|
self.locale.gettext('Source Systems file name')) |
|
144
|
|
|
# get the source system details from provided file |
|
145
|
|
|
self.timer.start() |
|
146
|
|
|
self.user_credentials = self.class_fo.fn_open_file_and_get_content( |
|
147
|
|
|
self.parameters.input_credentials_file, 'json')['Credentials'] |
|
148
|
|
|
self.class_ln.logger.info(self.locale.gettext( \ |
|
149
|
|
|
'Configuration file name with credentials has been loaded')) |
|
150
|
|
|
self.timer.stop() |
|
151
|
|
|
self.class_fo.fn_store_file_statistics(self.class_ln.logger, self.timer, |
|
152
|
|
|
self.parameters.input_credentials_file, |
|
153
|
|
|
self.locale.gettext( \ |
|
154
|
|
|
'Configuration file name with credentials')) |
|
155
|
|
|
|
|
156
|
|
|
def result_set_into_data_frame(self, local_logger, stats, crt_session): |
|
157
|
|
|
result_df = self.class_dbt.result_set_to_data_frame(local_logger, self.timer, |
|
158
|
|
|
stats['columns'], stats['result_set']) |
|
159
|
|
|
rdf = self.class_dbt.append_additional_columns_to_df(local_logger, self.timer, |
|
160
|
|
|
result_df, crt_session) |
|
161
|
|
|
return rdf |
|
162
|
|
|
|
|
163
|
|
|
def set_default_starting_weekday(self, crt_session): |
|
164
|
|
|
if 'start-isoweekday' not in crt_session: |
|
165
|
|
|
week_starts_with_isoweekday = 1 |
|
166
|
|
|
if 'start_isoweekday' in crt_session: |
|
167
|
|
|
week_starts_with_isoweekday = crt_session['start_isoweekday'] |
|
168
|
|
|
return week_starts_with_isoweekday |
|
|
|
|
|
|
169
|
|
|
|
|
170
|
|
|
def set_default_parameter_rules(self, crt_session): |
|
171
|
|
|
if 'parameters-handling-rules' in crt_session: |
|
172
|
|
|
dictionary_to_return = crt_session['parameters-handling-rules'] |
|
173
|
|
|
else: |
|
174
|
|
|
# assumption is for either DICT or LIST values are numeric |
|
175
|
|
|
# in case text is given different rules have to be specified |
|
176
|
|
|
dictionary_to_return = { |
|
177
|
|
|
"dict-values-glue" : ", ", |
|
178
|
|
|
"dict-values-prefix": "IN (", |
|
179
|
|
|
"dict-values-suffix": ")", |
|
180
|
|
|
"list-values-glue" : ", ", |
|
181
|
|
|
"list-values-prefix": "", |
|
182
|
|
|
"list-values-suffix": "" |
|
183
|
|
|
} |
|
184
|
|
|
return dictionary_to_return |
|
185
|
|
|
|
|
186
|
|
|
def store_result_set_to_disk(self, local_logger, in_data_frame, crt_session): |
|
187
|
|
|
output_file_setting_type = type(crt_session['output-file']) |
|
188
|
|
|
if output_file_setting_type == dict: |
|
189
|
|
|
self.class_dm.fn_store_data_frame_to_file(local_logger, self.timer, in_data_frame, |
|
190
|
|
|
crt_session['output-file']) |
|
191
|
|
|
self.class_fo.fn_store_file_statistics(local_logger, self.timer, |
|
192
|
|
|
crt_session['output-file']['name'], |
|
193
|
|
|
self.locale.gettext('Output file name')) |
|
194
|
|
|
elif output_file_setting_type == list: |
|
195
|
|
|
for crt_output in crt_session['output-file']: |
|
196
|
|
|
self.class_dm.fn_store_data_frame_to_file(local_logger, self.timer, |
|
197
|
|
|
in_data_frame, crt_output) |
|
198
|
|
|
self.class_fo.fn_store_file_statistics(local_logger, self.timer, crt_output['name'], |
|
199
|
|
|
self.locale.gettext('Output file name')) |
danielgp /
db-extractor
Loading history...