cief/data-management
1
0
Fork 0
mirror of https://gitlab.com/cief-data/data-management.git synced 2026-08-19 04:13:57 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add CRISP & ExchangeLog Data

Browse Source
This commit is contained in:
Yam ZhengLim
2022-12-19 15:19:37 +08:00
parent 5b3406e816
commit cc7cea07a5
6 changed files with 159 additions and 76 deletions
Download Patch File Download Diff File Expand all files Collapse all files
EXCHANGE_EXTRACT_SQL_OLD.json
+45
View File
@@ -40,6 +40,51 @@
"wallets" : "SELECT * FROM wallets;" "wallets" : "SELECT * FROM wallets;"
} }
{
"activity_log" : "SELECT ID, LOG_NAME, DESCRIPTION, SUBJECT_ID, SUBJECT_TYPE, CAUSER_ID, CAUSER_TYPE, '**MASKING_PROPERTIES**' AS PROPERTIES, CREATED_AT, UPDATED_AT FROM activity_log;",
"addresses" : "SELECT ID, COMPANY_ID, COUNTRY_ID, STATE_ID, DISTRICT_ID, POSTCODE, '**MASKING_STREET_ONE**'AS STREET_ONE, '**MASKING_STREET_TWO**'AS STREET_TWO, BILLING, DELETED_AT, CREATED_AT, UPDATED_AT FROM addresses;",
"announcements" : "SELECT * FROM announcements;",
"announcement_segment" : "SELECT * FROM announcement_segment;",
"banks" : "SELECT ID, COMPANY_ID, '**MASKING_REFERENCE**' AS REFERENCE, BANK_NAME, MD5(CONCAT(HOLDER_NAME,LENGTH(HOLDER_NAME)*2022*2)) AS HOLDER_NAME, MD5(CONCAT(ACCOUNT_NO,LENGTH(ACCOUNT_NO)*2022*2)) AS ACCOUNT_NO, BANK_BRANCH, MD5(CONCAT(SWIFT,LENGTH(SWIFT)*2022*2)) AS SWIFT, MD5(CONCAT(SNAP,LENGTH(SNAP)*2022*2)) AS SNAP, TYPE, `DEFAULT`, STATUS, COUNTRY_ID, DELETED_AT, CREATED_AT, UPDATED_AT FROM banks;",
"bank_logs" : "SELECT ID, BANK_ID, COMPANY_ID, '**MASKING_REFERENCE**' AS REFERENCE, BANK_NAME, MD5(CONCAT(HOLDER_NAME,LENGTH(HOLDER_NAME)*2022*2)) AS HOLDER_NAME, MD5(CONCAT(ACCOUNT_NO,LENGTH(ACCOUNT_NO)*2022*2)) AS ACCOUNT_NO, BANK_BRANCH, MD5(CONCAT(SWIFT,LENGTH(SWIFT)*2022*2)) AS SWIFT, MD5(CONCAT(SNAP,LENGTH(SNAP)*2022*2)) AS SNAP, TYPE, `DEFAULT`, STATUS, COUNTRY_ID, DELETED_AT, CREATED_AT, UPDATED_AT FROM bank_logs;",
"bookings" : "SELECT * FROM bookings;",
"companies" : "SELECT ID, '**MASKING_NAME**' AS NAME, '**MASKING_REFERENCE**' AS REFERENCE, DEBTOR, TYPE, BUSINESS_TYPE, STATUS, DELETED_AT, CREATED_AT, UPDATED_AT FROM companies;",
"contacts" : "SELECT ID, COMPANY_ID, COUNTRY_ID, '**MASKING_REFERENCE**' AS REFERENCE, MD5(CONCAT(PHONE,LENGTH(PHONE)*2022*2)) AS PHONE, MD5(CONCAT(EMAIL,LENGTH(EMAIL)*2022*2)) AS EMAIL, MD5(CONCAT(WECHAT_ID,LENGTH(WECHAT_ID)*2022*2)) AS WECHAT_ID, DELETED_AT, CREATED_AT, UPDATED_AT FROM contacts;",
"countries" : "SELECT * FROM countries;",
"currencies" : "SELECT * FROM currencies;",
"currency_logs" : "SELECT * FROM currency_logs;",
"currency_rates" : "SELECT * FROM currency_rates;",
"currency_rate_logs" : "SELECT * FROM currency_rate_logs;",
"districts" : "SELECT * FROM districts;",
"documents" : "SELECT ID, OWNER_TYPE, OWNER_ID, DOCUMENT_TYPE, '**MASKING_REFERENCE**' AS REFERENCE, STATUS, ISSUED_DATE, EXPIRED_DATE, APPROVER, APPROVAL_DATE, DELETED_AT, CREATED_AT, UPDATED_AT FROM documents;",
"employees" : "SELECT * FROM employees;",
"failed_jobs" : "SELECT ID, CONNECTION, QUEUE, '**MASKING_PAYLOAD' AS PAYLOAD, '**MASKING_EXCEPTION**' AS EXCEPTION, FAILED_AT FROM failed_jobs;",
"files" : "SELECT id, document_id, '**MASKING_FILE**', file_type, deleted_at, created_at, updated_at FROM files;",
"groups" : "SELECT ID, '**MASKING_REFERENCE**' AS REFERENCE, ISSUER, RECEIVER, AMOUNT, ORIGINAL_AMOUNT, CURRENCY_ID, ORIGINAL_CURRENCY_ID, CURRENCY_RATE, TAX, SERVICE_CHARGE, STATUS, CREATED_AT, UPDATED_AT FROM groups;",
"group_transactions" : "SELECT * FROM group_transactions;",
"jobs" : "SELECT ID, QUEUE, '**MASKING_PAYLOAD**' AS PAYLOAD, ATTEMPTS, RESERVED_AT, AVAILABLE_AT, CREATED_AT FROM jobs;",
"migrations" : "SELECT * FROM migrations;",
"model_has_roles" : "SELECT * FROM model_has_roles;",
"notifications" : "SELECT * FROM notifications;",
"password_resets" : "SELECT ID, USER_ID, '**MASKING_TOKEN**' AS TOKEN, IS_EXPIRED, IS_COMPLETE, CREATED_AT, UPDATED_AT FROM password_resets;",
"permissions" : "SELECT * FROM permissions;",
"roles" : "SELECT * FROM roles;",
"role_has_permissions" : "SELECT * FROM role_has_permissions;",
"segments" : "SELECT * FROM segments;",
"segment_companies" : "SELECT * FROM segment_companies;",
"segment_constants" : "SELECT ID, SEGMENT_ID, NAME, REFERENCE, '**MASKING_DETAIL**' AS DETAIL, DELETED_AT, CREATED_AT, UPDATED_AT FROM segment_constants;",
"service_types" : "SELECT * FROM service_types;",
"states" : "SELECT * FROM states;",
"transactions" : "SELECT ID, OWNER_TYPE, OWNER_ID, TYPE, ISSUER, RECEIVER, RECIPIENT_BANK_ACCOUNT_ID, PAYMENT_METHOD, '**PAYMENT_REFERENCE**' AS PAYMENT_REFERENCE, MD5(CONCAT(BILL_NO,LENGTH(BILL_NO)*2022*2)) AS BILL_NO, AMOUNT, ORIGINAL_AMOUNT, CURRENCY_ID, ORIGINAL_CURRENCY_ID, CURRENCY_RATE, TAX, SERVICE_CHARGE, EXPIRES_ON, STATUS, DELETED_AT, CREATED_AT, UPDATED_AT FROM transactions;",
"transaction_detail" : "SELECT id, transaction_id, REPLACE(REPLACE(REPLACE(product_code, '\n', '<br>'),'|', '&#124'),'\\\\','&#92&#92') AS product_code, REPLACE(REPLACE(REPLACE(product_name, '\n', '<br>'),'|', '&#124'),'\\\\','&#92&#92') AS product_name, quantity, price, amount, deleted_at, created_at, updated_at FROM `transaction_detail`;",
"users" : "SELECT id, '**MASKING_NAME**' AS name, MD5(CONCAT(email,LENGTH(email)*2022*2)) AS email, type, status, remember_token, active_at, deleted_at, created_at, updated_at FROM users;",
"user_email_verifications" : "SELECT ID, MD5(CONCAT(email,LENGTH(email)*2022*2)) AS email, '**MASKING_TOKEN**' AS TOKEN, IS_COMPLETE, IS_ACTIVE, IS_SENT, CREATED_AT, UPDATED_AT FROM user_email_verifications;",
"wallets" : "SELECT ID, OWNER_TYPE, OWNER_ID, MD5(CONCAT(CODE,LENGTH(CODE)*2022*2)) AS CODE, CURRENCY_ID, AMOUNT, DELETED_AT, CREATED_AT, UPDATED_AT FROM wallets;"
}
README.md
+1
View File
@@ -23,6 +23,7 @@
###### Region: US ###### Region: US
###### Directory: /home/cief_data/business_intelligence/data-management ###### Directory: /home/cief_data/business_intelligence/data-management
###### Cronjob run at: Everyday 12:30AM ###### Cronjob run at: Everyday 12:30AM
###### Snowflake task run at: Everyday 2:00AM
###### Crontab config: crontab -e ###### Crontab config: crontab -e
###### Crontab log: grep CRON /var/log/syslog ###### Crontab log: grep CRON /var/log/syslog
###### Python log: grep 'mycmd' /var/log/syslog ###### Python log: grep 'mycmd' /var/log/syslog
config/EXCHANGE_EXTRACT_SQL.json
+21 -20
View File
@@ -1,43 +1,44 @@
{ {
"activity_log" : "SELECT ID, LOG_NAME, DESCRIPTION, SUBJECT_ID, SUBJECT_TYPE, CAUSER_ID, CAUSER_TYPE, '**MASKING_PROPERTIES**' AS PROPERTIES, CREATED_AT, UPDATED_AT FROM activity_log;", "activity_log" : "SELECT * FROM activity_log;",
"addresses" : "SELECT ID, COMPANY_ID, COUNTRY_ID, STATE_ID, DISTRICT_ID, POSTCODE, '**MASKING_STREET_ONE**'AS STREET_ONE, '**MASKING_STREET_TWO**'AS STREET_TWO, BILLING, DELETED_AT, CREATED_AT, UPDATED_AT FROM addresses;", "addresses" : "SELECT * FROM addresses;",
"announcements" : "SELECT * FROM announcements;", "announcements" : "SELECT * FROM announcements;",
"announcement_segment" : "SELECT * FROM announcement_segment;", "announcement_segment" : "SELECT * FROM announcement_segment;",
"banks" : "SELECT ID, COMPANY_ID, '**MASKING_REFERENCE**' AS REFERENCE, BANK_NAME, MD5(CONCAT(HOLDER_NAME,LENGTH(HOLDER_NAME)*2022*2)) AS HOLDER_NAME, MD5(CONCAT(ACCOUNT_NO,LENGTH(ACCOUNT_NO)*2022*2)) AS ACCOUNT_NO, BANK_BRANCH, MD5(CONCAT(SWIFT,LENGTH(SWIFT)*2022*2)) AS SWIFT, MD5(CONCAT(SNAP,LENGTH(SNAP)*2022*2)) AS SNAP, TYPE, `DEFAULT`, STATUS, COUNTRY_ID, DELETED_AT, CREATED_AT, UPDATED_AT FROM banks;", "banks" : "SELECT * FROM banks;",
"bank_logs" : "SELECT ID, BANK_ID, COMPANY_ID, '**MASKING_REFERENCE**' AS REFERENCE, BANK_NAME, MD5(CONCAT(HOLDER_NAME,LENGTH(HOLDER_NAME)*2022*2)) AS HOLDER_NAME, MD5(CONCAT(ACCOUNT_NO,LENGTH(ACCOUNT_NO)*2022*2)) AS ACCOUNT_NO, BANK_BRANCH, MD5(CONCAT(SWIFT,LENGTH(SWIFT)*2022*2)) AS SWIFT, MD5(CONCAT(SNAP,LENGTH(SNAP)*2022*2)) AS SNAP, TYPE, `DEFAULT`, STATUS, COUNTRY_ID, DELETED_AT, CREATED_AT, UPDATED_AT FROM bank_logs;", "bank_logs" : "SELECT * FROM bank_logs;",
"bookings" : "SELECT * FROM bookings;", "bookings" : "SELECT * FROM bookings;",
"companies" : "SELECT ID, '**MASKING_NAME**' AS NAME, '**MASKING_REFERENCE**' AS REFERENCE, DEBTOR, TYPE, BUSINESS_TYPE, STATUS, DELETED_AT, CREATED_AT, UPDATED_AT FROM companies;", "booking_logs" : "SELECT * FROM booking_logs",
"contacts" : "SELECT ID, COMPANY_ID, COUNTRY_ID, '**MASKING_REFERENCE**' AS REFERENCE, MD5(CONCAT(PHONE,LENGTH(PHONE)*2022*2)) AS PHONE, MD5(CONCAT(EMAIL,LENGTH(EMAIL)*2022*2)) AS EMAIL, MD5(CONCAT(WECHAT_ID,LENGTH(WECHAT_ID)*2022*2)) AS WECHAT_ID, DELETED_AT, CREATED_AT, UPDATED_AT FROM contacts;", "companies" : "SELECT * FROM companies;",
"contacts" : "SELECT * FROM contacts;",
"countries" : "SELECT * FROM countries;", "countries" : "SELECT * FROM countries;",
"currencies" : "SELECT * FROM currencies;", "currencies" : "SELECT * FROM currencies;",
"currency_logs" : "SELECT * FROM currency_logs;", "currency_logs" : "SELECT * FROM currency_logs;",
"currency_rates" : "SELECT * FROM currency_rates;", "currency_rates" : "SELECT * FROM currency_rates;",
"currency_rate_logs" : "SELECT * FROM currency_rate_logs;", "currency_rate_logs" : "SELECT * FROM currency_rate_logs;",
"districts" : "SELECT * FROM districts;", "districts" : "SELECT * FROM districts;",
"documents" : "SELECT ID, OWNER_TYPE, OWNER_ID, DOCUMENT_TYPE, '**MASKING_REFERENCE**' AS REFERENCE, STATUS, ISSUED_DATE, EXPIRED_DATE, APPROVER, APPROVAL_DATE, DELETED_AT, CREATED_AT, UPDATED_AT FROM documents;", "documents" : "SELECT * FROM documents;",
"employees" : "SELECT * FROM employees;", "employees" : "SELECT * FROM employees;",
"failed_jobs" : "SELECT ID, CONNECTION, QUEUE, '**MASKING_PAYLOAD' AS PAYLOAD, '**MASKING_EXCEPTION**' AS EXCEPTION, FAILED_AT FROM failed_jobs;", "failed_jobs" : "SELECT * FROM failed_jobs;",
"files" : "SELECT id, document_id, '**MASKING_FILE**', file_type, deleted_at, created_at, updated_at FROM files;", "files" : "SELECT id, document_id, REPLACE(REPLACE(REPLACE(file, '\n', '<br>'),'|', '&#124'),'\\\\','&#92&#92') AS 'file', file_type, deleted_at, created_at, updated_at FROM files;",
"groups" : "SELECT ID, '**MASKING_REFERENCE**' AS REFERENCE, ISSUER, RECEIVER, AMOUNT, ORIGINAL_AMOUNT, CURRENCY_ID, ORIGINAL_CURRENCY_ID, CURRENCY_RATE, TAX, SERVICE_CHARGE, STATUS, CREATED_AT, UPDATED_AT FROM groups;", "groups" : "SELECT * FROM groups;",
"group_transactions" : "SELECT * FROM group_transactions;", "group_transactions" : "SELECT * FROM group_transactions;",
"jobs" : "SELECT ID, QUEUE, '**MASKING_PAYLOAD**' AS PAYLOAD, ATTEMPTS, RESERVED_AT, AVAILABLE_AT, CREATED_AT FROM jobs;", "jobs" : "SELECT * FROM jobs;",
"migrations" : "SELECT * FROM migrations;", "migrations" : "SELECT * FROM migrations;",
"model_has_roles" : "SELECT * FROM model_has_roles;", "model_has_roles" : "SELECT * FROM model_has_roles;",
"notifications" : "SELECT * FROM notifications;", "notifications" : "SELECT * FROM notifications;",
"password_resets" : "SELECT ID, USER_ID, '**MASKING_TOKEN**' AS TOKEN, IS_EXPIRED, IS_COMPLETE, CREATED_AT, UPDATED_AT FROM password_resets;", "password_resets" : "SELECT * FROM password_resets;",
"permissions" : "SELECT * FROM permissions;", "permissions" : "SELECT * FROM permissions;",
"roles" : "SELECT * FROM roles;", "roles" : "SELECT * FROM roles;",
"role_has_permissions" : "SELECT * FROM role_has_permissions;", "role_has_permissions" : "SELECT * FROM role_has_permissions;",
"segments" : "SELECT * FROM segments;", "segments" : "SELECT * FROM segments;",
"segment_companies" : "SELECT * FROM segment_companies;", "segment_companies" : "SELECT * FROM segment_companies;",
"segment_constants" : "SELECT ID, SEGMENT_ID, NAME, REFERENCE, '**MASKING_DETAIL**' AS DETAIL, DELETED_AT, CREATED_AT, UPDATED_AT FROM segment_constants;", "segment_constants" : "SELECT * FROM segment_constants;",
"service_types" : "SELECT * FROM service_types;", "service_types" : "SELECT * FROM service_types;",
"states" : "SELECT * FROM states;", "states" : "SELECT * FROM states;",
"transactions" : "SELECT ID, OWNER_TYPE, OWNER_ID, TYPE, ISSUER, RECEIVER, RECIPIENT_BANK_ACCOUNT_ID, PAYMENT_METHOD, '**PAYMENT_REFERENCE**' AS PAYMENT_REFERENCE, MD5(CONCAT(BILL_NO,LENGTH(BILL_NO)*2022*2)) AS BILL_NO, AMOUNT, ORIGINAL_AMOUNT, CURRENCY_ID, ORIGINAL_CURRENCY_ID, CURRENCY_RATE, TAX, SERVICE_CHARGE, EXPIRES_ON, STATUS, DELETED_AT, CREATED_AT, UPDATED_AT FROM transactions;", "transactions" : "SELECT * FROM transactions;",
"transaction_detail" : "SELECT id, transaction_id, REPLACE(REPLACE(REPLACE(product_code, '\n', '<br>'),'|', '&#124'),'\\\\','&#92&#92') AS product_code, REPLACE(REPLACE(REPLACE(product_name, '\n', '<br>'),'|', '&#124'),'\\\\','&#92&#92') AS product_name, quantity, price, amount, deleted_at, created_at, updated_at FROM `transaction_detail`;", "transaction_detail" : "SELECT id, transaction_id, REPLACE(REPLACE(REPLACE(product_code, '\n', '<br>'),'|', '&#124'),'\\\\','&#92&#92') AS product_code, REPLACE(REPLACE(REPLACE(product_name, '\n', '<br>'),'|', '&#124'),'\\\\','&#92&#92') AS product_name, quantity, price, amount, deleted_at, created_at, updated_at FROM `transaction_detail`;",
"users" : "SELECT id, '**MASKING_NAME**' AS name, MD5(CONCAT(email,LENGTH(email)*2022*2)) AS email, type, status, remember_token, active_at, deleted_at, created_at, updated_at FROM users;", "transaction_logs" : "SELECT * FROM transaction_logs",
"user_email_verifications" : "SELECT ID, MD5(CONCAT(email,LENGTH(email)*2022*2)) AS email, '**MASKING_TOKEN**' AS TOKEN, IS_COMPLETE, IS_ACTIVE, IS_SENT, CREATED_AT, UPDATED_AT FROM user_email_verifications;", "users" : "SELECT id, name, email, type, status, remember_token, active_at, deleted_at, created_at, updated_at FROM users;",
"wallets" : "SELECT ID, OWNER_TYPE, OWNER_ID, MD5(CONCAT(CODE,LENGTH(CODE)*2022*2)) AS CODE, CURRENCY_ID, AMOUNT, DELETED_AT, CREATED_AT, UPDATED_AT FROM wallets;" "user_email_verifications" : "SELECT * FROM user_email_verifications;",
} "wallets" : "SELECT * FROM wallets;",
"wallet_logs" : "SELECT * FROM wallet_logs"
}
crisp/extract/__pycache__/crisp_extract_function.cpython-39.pyc
BIN
View File
Binary file not shown.
crisp/extract/crisp_extract_function.py
+64 -34
View File
@@ -13,59 +13,89 @@ def connect_crisp(config):
print("[FUNCTION_ERROR]-connect_crisp") print("[FUNCTION_ERROR]-connect_crisp")
return False return False
def get_operators_csv(website_id, client, path, filename): def get_list_website_operators_csv(website_id, client, path, filename):
try: try:
output_location = path + filename output_location = path + filename
operators_list_json = client.website.list_website_operators(website_id) list_website_operators_json = client.website.list_website_operators(website_id)
operators_list_df = general_function.json_normalize_to_df(operators_list_json) list_website_operators_df = general_function.json_normalize_to_df(list_website_operators_json)
operators_list_df.to_csv(output_location, sep='|', encoding='utf-8-sig', header='true', quotechar= '"', index=False) list_website_operators_df.columns = list_website_operators_df.columns.str.replace(".", "-")
list_website_operators_df.to_csv(output_location, sep='|', encoding='utf-8-sig', header='true', quotechar= '"', index=False)
print("[Download Complete]",filename) print("[Download Complete]",filename)
except Exception as e: except Exception as e:
print(e) print(e)
print("[FUNCTION_ERROR]-get_operators_csv") print("[FUNCTION_ERROR]-get_list_website_operators_csv")
return False return False
def get_peoples_csv(website_id, client, path, filename): def get_list_people_profiles_csv(website_id, client, path, filename):
try: try:
output_location = path + filename output_location = path + filename
peoples_list_df = pd.DataFrame() list_people_profiles_df = pd.DataFrame()
peoples_data_df = pd.DataFrame()
people_subscription_status_df = pd.DataFrame()
loop_flag = True loop_flag = True
page_number = 1 page_number = 1
while loop_flag == True: while loop_flag == True:
peoples_list_json = client.website.list_people_profiles(website_id, page_number) list_people_profiles_json = client.website.list_people_profiles(website_id, page_number)
temp_df = general_function.json_normalize_to_df(peoples_list_json) temp_df = general_function.json_normalize_to_df(list_people_profiles_json)
peoples_list_df = pd.concat([temp_df,peoples_list_df], ignore_index=True) temp_df['api_page_number'] = page_number
if not peoples_list_json: list_people_profiles_df = pd.concat([temp_df,list_people_profiles_df], ignore_index=True)
if not list_people_profiles_json:
loop_flag = False loop_flag = False
else: else:
page_number += 1 page_number += 1
list_people_profiles_df.columns = list_people_profiles_df.columns.str.replace(".", "-")
# Get marking by CustomerService Operator list_people_profiles_df.to_csv(output_location, sep='|', encoding='utf-8-sig', header='true', quotechar= '"', index=False)
for people_id in peoples_list_df.people_id:
people_data_json = client.website.get_people_data(website_id, people_id)
temp_df = general_function.json_normalize_to_df(people_data_json)
temp_df['people_id'] = people_id
peoples_data_df = pd.concat([temp_df,peoples_data_df], ignore_index=True)
peoples_list_df = peoples_list_df.merge(peoples_data_df, on='people_id', how='left')
# Get email subscription status
for people_id in peoples_list_df.people_id:
people_subscription_status_json = client.website.get_people_subscription_status(website_id, people_id)
temp_df = general_function.json_normalize_to_df(people_subscription_status_json)
temp_df['people_id'] = people_id
people_subscription_status_df = pd.concat([temp_df,people_subscription_status_df], ignore_index=True)
peoples_list_df = peoples_list_df.merge(people_subscription_status_df, on='people_id', how='left')
peoples_list_df.to_csv(output_location, sep='|', encoding='utf-8-sig', header='true', quotechar= '"', index=False)
print("[Download Complete]",filename) print("[Download Complete]",filename)
except Exception as e: except Exception as e:
print(e) print(e)
print("[FUNCTION_ERROR]-get_peoples_csv") print("[FUNCTION_ERROR]-get_list_people_profiles_csv")
return False
def get_people_data_csv(website_id, client, path, filename, list_people_profiles_csv_path):
try:
output_location = path + filename
get_people_data_df = pd.DataFrame()
person_ids = pd.read_csv(list_people_profiles_csv_path, sep='|', encoding='utf-8-sig', quotechar= '"')['people_id']
# Get marking by CustomerService Operator
for people_id in person_ids:
get_people_data_json = client.website.get_people_data(website_id, people_id)
temp_df = general_function.json_normalize_to_df(get_people_data_json)
temp_df['people_id'] = people_id
get_people_data_df = pd.concat([temp_df,get_people_data_df], ignore_index=True)
get_people_data_df.to_csv(output_location, sep='|', encoding='utf-8-sig', header='true', quotechar= '"', index=False)
print("[Download Complete]",filename)
except Exception as e:
print(e)
print("[FUNCTION_ERROR]-get_people_data_csv")
return False
def get_people_subscription_status_csv(website_id, client, path, filename, list_people_profiles_csv_path):
try:
output_location = path + filename
get_people_subscription_status_df = pd.DataFrame()
person_ids = pd.read_csv(list_people_profiles_csv_path, sep='|', encoding='utf-8-sig', quotechar= '"')['people_id']
# Get email subscription status
for people_id in person_ids:
pget_people_subscription_status_df_json = client.website.get_people_subscription_status(website_id, people_id)
temp_df = general_function.json_normalize_to_df(pget_people_subscription_status_df_json)
temp_df['people_id'] = people_id
get_people_subscription_status_df = pd.concat([temp_df,get_people_subscription_status_df], ignore_index=True)
get_people_subscription_status_df.to_csv(output_location, sep='|', encoding='utf-8-sig', header='true', quotechar= '"', index=False)
print("[Download Complete]",filename)
except Exception as e:
print(e)
print("[FUNCTION_ERROR]-get_people_subscription_status_csv")
return False return False
main.py
+28 -22
View File
@@ -79,11 +79,11 @@ def shipping_load_to_cloudstorage(SHIPPING_CSV_TEMP_STORAGE_PATH):
def crisp_extract(CRISP_CSV_TEMP_STORAGE_PATH): def crisp_extract(CRISP_CSV_TEMP_STORAGE_PATH):
#Config file name #Config file name
CRISP_API_CONFIG_FILE = "CRISP_CONFIG.json" CRISP_API_CONFIG_FILE = 'CRISP_CONFIG.json'
CRISP_CSV_TEMP_STORAGE_PATH = CRISP_CSV_TEMP_STORAGE_PATH CRISP_CSV_TEMP_STORAGE_PATH = CRISP_CSV_TEMP_STORAGE_PATH
#Read config file #Read config file
CRISP_API_CONFIG = general_function.read_json_file("./config/",CRISP_API_CONFIG_FILE) CRISP_API_CONFIG = general_function.read_json_file('./config/',CRISP_API_CONFIG_FILE)
#Get website_id #Get website_id
CRISP_WEBSITE_ID = CRISP_API_CONFIG['website_id'] CRISP_WEBSITE_ID = CRISP_API_CONFIG['website_id']
@@ -92,10 +92,16 @@ def crisp_extract(CRISP_CSV_TEMP_STORAGE_PATH):
CLIENT = crisp_extract_func.connect_crisp(CRISP_API_CONFIG) CLIENT = crisp_extract_func.connect_crisp(CRISP_API_CONFIG)
#Get operators csv #Get operators csv
crisp_extract_func.get_operators_csv(CRISP_WEBSITE_ID, CLIENT, CRISP_CSV_TEMP_STORAGE_PATH, 'crisp_operators.csv') crisp_extract_func.get_list_website_operators_csv(CRISP_WEBSITE_ID, CLIENT, CRISP_CSV_TEMP_STORAGE_PATH, 'crisp_get_list_website_operators.csv')
#Get users(people) csv #Get users(people) csv
crisp_extract_func.get_peoples_csv(CRISP_WEBSITE_ID, CLIENT, CRISP_CSV_TEMP_STORAGE_PATH, 'crisp_peoples.csv') crisp_extract_func.get_list_people_profiles_csv(CRISP_WEBSITE_ID, CLIENT, CRISP_CSV_TEMP_STORAGE_PATH, 'crisp_get_list_people_profiles.csv')
# Get peoples manually marking by operator
crisp_extract_func.get_people_data_csv(CRISP_WEBSITE_ID, CLIENT, CRISP_CSV_TEMP_STORAGE_PATH, 'crisp_get_people_data.csv', './crisp/extract/crisp_temp_extract_csv/crisp_get_list_people_profiles.csv')
# Get peoples subscription status
crisp_extract_func.get_people_subscription_status_csv(CRISP_WEBSITE_ID, CLIENT, CRISP_CSV_TEMP_STORAGE_PATH, 'crisp_get_people_subscription_status.csv', './crisp/extract/crisp_temp_extract_csv/crisp_get_list_people_profiles.csv')
def crisp_load_to_cloudstorage(CRISP_CSV_TEMP_STORAGE_PATH): def crisp_load_to_cloudstorage(CRISP_CSV_TEMP_STORAGE_PATH):
#Config file name #Config file name
@@ -117,27 +123,27 @@ def crisp_load_to_cloudstorage(CRISP_CSV_TEMP_STORAGE_PATH):
def main(): def main():
#Make sure cron in the file directory # #Make sure cron in the file directory
general_function.change_to_current_directory() # general_function.change_to_current_directory()
#Extract Load Exchange Data # #Extract Load Exchange Data
EXCHANGE_CSV_TEMP_STORAGE_PATH = "./exchange/extract/exchange_temp_extract_csv/" # EXCHANGE_CSV_TEMP_STORAGE_PATH = "./exchange/extract/exchange_temp_extract_csv/"
exchange_extract(EXCHANGE_CSV_TEMP_STORAGE_PATH) # exchange_extract(EXCHANGE_CSV_TEMP_STORAGE_PATH)
exchange_load_to_cloudstorage(EXCHANGE_CSV_TEMP_STORAGE_PATH) # exchange_load_to_cloudstorage(EXCHANGE_CSV_TEMP_STORAGE_PATH)
#Delete Temp CSV # #Delete Temp CSV
general_function.delete_csv_in_path(EXCHANGE_CSV_TEMP_STORAGE_PATH) # general_function.delete_csv_in_path(EXCHANGE_CSV_TEMP_STORAGE_PATH)
#Make sure cron in the file directory # #Make sure cron in the file directory
general_function.change_to_current_directory() # general_function.change_to_current_directory()
#Extract Load SHIPPING Data # #Extract Load SHIPPING Data
SHIPPING_CSV_TEMP_STORAGE_PATH = "./shipping/extract/shipping_temp_extract_csv/" # SHIPPING_CSV_TEMP_STORAGE_PATH = "./shipping/extract/shipping_temp_extract_csv/"
shipping_extract(SHIPPING_CSV_TEMP_STORAGE_PATH) # shipping_extract(SHIPPING_CSV_TEMP_STORAGE_PATH)
shipping_load_to_cloudstorage(SHIPPING_CSV_TEMP_STORAGE_PATH) # shipping_load_to_cloudstorage(SHIPPING_CSV_TEMP_STORAGE_PATH)
#Delete Temp CSV # #Delete Temp CSV
general_function.delete_csv_in_path(SHIPPING_CSV_TEMP_STORAGE_PATH) # general_function.delete_csv_in_path(SHIPPING_CSV_TEMP_STORAGE_PATH)
#Make sure cron in the file directory #Make sure cron in the file directory
@@ -148,8 +154,8 @@ def main():
crisp_extract(CRISP_CSV_TEMP_STORAGE_PATH) crisp_extract(CRISP_CSV_TEMP_STORAGE_PATH)
crisp_load_to_cloudstorage(CRISP_CSV_TEMP_STORAGE_PATH) crisp_load_to_cloudstorage(CRISP_CSV_TEMP_STORAGE_PATH)
#Delete Temp CSV # #Delete Temp CSV
general_function.delete_csv_in_path(CRISP_CSV_TEMP_STORAGE_PATH) # general_function.delete_csv_in_path(CRISP_CSV_TEMP_STORAGE_PATH)
if True: if True:
main() main()