إخفاء البيانات في ClickHouse

CREATE TABLE orders (
    user_id UInt32,
    name String,
    email String,
    phone String,
    total_amount Decimal(10,2),
    order_date Date,
    shipping_address String
)
ENGINE = MergeTree()
ORDER BY user_id;

INSERT INTO orders VALUES
    (1001, 'John Smith', 'john.smith@gmail.com', '555-123-4567', 299.99, '2024-01-15', '123 Main St, New York, NY 10001'),
    (1002, 'Sarah Johnson', 'sarah.johnson@outlook.com', '555-987-6543', 149.50, '2024-01-16', '456 Oak Ave, Los Angeles, CA 90210'),
    (1003, 'Michael Brown', 'mbrown@company.com', '555-456-7890', 599.00, '2024-01-17', '789 Pine Rd, Chicago, IL 60601'),
    (1004, 'Emily Rogers', 'emily.rogers@yahoo.com', '555-321-0987', 89.99, '2024-01-18', '321 Elm St, Houston, TX 77001'),
    (1005, 'David Wilson', 'dwilson@email.net', '555-654-3210', 449.75, '2024-01-19', '654 Cedar Blvd, Phoenix, AZ 85001');

CREATE ROLE masked_data_viewer;

CREATE MASKING POLICY mask_pii_data ON orders
    UPDATE
        name = replaceRegexpOne(name, '^([A-Za-z]+)\\s+(.*)$', '\\1 ****'),
        email = replaceRegexpOne(email, '^(.{2})[^@]*(@.*)$', '\\1****\\2'),
        phone = replaceRegexpOne(phone, '^(\\d{3})-(\\d{3})-(\\d{4})$', '\\1-***-\\3'),
        shipping_address = replaceRegexpOne(shipping_address, '^[^,]+,\\s*(.*)$', '*** \\1')
    TO masked_data_viewer;

CREATE MASKING POLICY mask_high_value_orders ON orders
    UPDATE
        name = replaceRegexpOne(name, '^([A-Za-z]+)\\s+(.*)$', '\\1 ****'),
        email = replaceRegexpOne(email, '^(.{2})[^@]*(@.*)$', '\\1****\\2')
    WHERE total_amount > 200
    TO masked_data_viewer;

-- Lower priority: Basic masking for all sensitive data
CREATE MASKING POLICY basic_masking ON orders
    UPDATE
        name = '****',
        email = '****@****.com'
    TO masked_data_viewer
    PRIORITY 0;

-- Higher priority: More refined masking (overrides basic_masking)
CREATE MASKING POLICY refined_masking ON orders
    UPDATE
        name = replaceRegexpOne(name, '^([A-Za-z]+)\\s+(.*)$', '\\1 ****')
    WHERE total_amount > 100
    TO masked_data_viewer
    PRIORITY 10;

CREATE MASKING POLICY hash_sensitive_data ON orders
    UPDATE
        email = concat(toString(cityHash64(email)), '@masked.com'),
        phone = concat('555-', toString(cityHash64(phone) % 10000000))
    TO masked_data_viewer;

SHOW MASKING POLICIES;

DROP MASKING POLICY mask_pii_data ON orders;

CREATE OR REPLACE MASKING POLICY mask_pii_data ON orders
    UPDATE name = '[REDACTED]'
    TO masked_data_viewer;

CREATE TABLE orders (
    user_id UInt32,
    name String,
    email String,
    phone String,
    total_amount Decimal(10,2),
    order_date Date,
    shipping_address String
)
ENGINE = MergeTree()
ORDER BY user_id;

INSERT INTO orders VALUES
    (1001, 'John Smith', 'john.smith@gmail.com', '555-123-4567', 299.99, '2024-01-15', '123 Main St, New York, NY 10001'),
    (1002, 'Sarah Johnson', 'sarah.johnson@outlook.com', '555-987-6543', 149.50, '2024-01-16', '456 Oak Ave, Los Angeles, CA 90210'),
    (1003, 'Michael Brown', 'mbrown@company.com', '555-456-7890', 599.00, '2024-01-17', '789 Pine Rd, Chicago, IL 60601'),
    (1004, 'Emily Rogers', 'emily.rogers@yahoo.com', '555-321-0987', 89.99, '2024-01-18', '321 Elm St, Houston, TX 77001'),
    (1005, 'David Wilson', 'dwilson@email.net', '555-654-3210', 449.75, '2024-01-19', '654 Cedar Blvd, Phoenix, AZ 85001');

CREATE VIEW masked_orders AS
SELECT
    user_id,
    replaceRegexpOne(name, '^([A-Za-z]+)\\s+(.*)$', '\\1 ****') AS name,
    replaceRegexpOne(email, '^(.{0})[^@]*(@.*)$', '\\1****\\2') AS email,
    replaceRegexpOne(phone, '^(\\d{3})-(\\d{3})-(\\d{4})$', '\\1-***-\\3') AS phone,
    total_amount,
    order_date,
    replaceRegexpOne(shipping_address, '^[^,]+,\\s*(.*)$', '*** \\1') AS shipping_address
FROM orders;

CREATE ROLE masked_orders_viewer;

GRANT SELECT ON masked_orders TO masked_orders_viewer;

REVOKE SELECT ON orders FROM masked_orders_viewer;

GRANT masked_orders_viewer TO your_user;

DROP TABLE IF EXISTS orders;
CREATE TABLE orders (
    user_id UInt32,
    name String,
    name_masked String MATERIALIZED replaceRegexpOne(name, '^([A-Za-z]+)\\s+(.*)$', '\\1 ****'),
    email String,
    email_masked String MATERIALIZED replaceRegexpOne(email, '^(.{0})[^@]*(@.*)$', '\\1****\\2'),
    phone String,
    phone_masked String MATERIALIZED replaceRegexpOne(phone, '^(\\d{3})-(\\d{3})-(\\d{4})$', '\\1-***-\\3'),
    total_amount Decimal(10,2),
    order_date Date,
    shipping_address String,
    shipping_address_masked String MATERIALIZED replaceRegexpOne(shipping_address, '^[^,]+,\\s*(.*)$', '*** \\1')
)
ENGINE = MergeTree()
ORDER BY user_id;

INSERT INTO orders VALUES
    (1001, 'John Smith', 'john.smith@gmail.com', '555-123-4567', 299.99, '2024-01-15', '123 Main St, New York, NY 10001'),
    (1002, 'Sarah Johnson', 'sarah.johnson@outlook.com', '555-987-6543', 149.50, '2024-01-16', '456 Oak Ave, Los Angeles, CA 90210'),
    (1003, 'Michael Brown', 'mbrown@company.com', '555-456-7890', 599.00, '2024-01-17', '789 Pine Rd, Chicago, IL 60601'),
    (1004, 'Emily Rogers', 'emily.rogers@yahoo.com', '555-321-0987', 89.99, '2024-01-18', '321 Elm St, Houston, TX 77001'),
    (1005, 'David Wilson', 'dwilson@email.net', '555-654-3210', 449.75, '2024-01-19', '654 Cedar Blvd, Phoenix, AZ 85001');

DROP ROLE IF EXISTS masked_order_viewer;
CREATE ROLE masked_order_viewer;

GRANT SELECT ON orders TO masked_data_reader;

REVOKE SELECT(name) ON orders FROM masked_data_reader;
REVOKE SELECT(email) ON orders FROM masked_data_reader;
REVOKE SELECT(phone) ON orders FROM masked_data_reader;
REVOKE SELECT(shipping_address) ON orders FROM masked_data_reader;

GRANT masked_orders_viewer TO your_user;

DROP TABLE IF EXISTS orders;
CREATE TABLE orders (
    user_id UInt32,
    name String EPHEMERAL,
    name_masked String MATERIALIZED replaceRegexpOne(name, '^([A-Za-z]+)\\s+(.*)$', '\\1 ****'),
    email String EPHEMERAL,
    email_masked String MATERIALIZED replaceRegexpOne(email, '^(.{2})[^@]*(@.*)$', '\\1****\\2'),
    phone String EPHEMERAL,
    phone_masked String MATERIALIZED replaceRegexpOne(phone, '^(\\d{3})-(\\d{3})-(\\d{4})$', '\\1-***-\\3'),
    total_amount Decimal(10,2),
    order_date Date,
    shipping_address String EPHEMERAL,
    shipping_address_masked String MATERIALIZED replaceRegexpOne(shipping_address, '^([^,]+),\\s*(.*)$', '*** \\2')
)
ENGINE = MergeTree()
ORDER BY user_id;

INSERT INTO orders (user_id, name, email, phone, total_amount, order_date, shipping_address) VALUES
    (1001, 'John Smith', 'john.smith@gmail.com', '555-123-4567', 299.99, '2024-01-15', '123 Main St, New York, NY 10001'),
    (1002, 'Sarah Johnson', 'sarah.johnson@outlook.com', '555-987-6543', 149.50, '2024-01-16', '456 Oak Ave, Los Angeles, CA 90210'),
    (1003, 'Michael Brown', 'mbrown@company.com', '555-456-7890', 599.00, '2024-01-17', '789 Pine Rd, Chicago, IL 60601'),
    (1004, 'Emily Rogers', 'emily.rogers@yahoo.com', '555-321-0987', 89.99, '2024-01-18', '321 Elm St, Houston, TX 77001'),
    (1005, 'David Wilson', 'dwilson@email.net', '555-654-3210', 449.75, '2024-01-19', '654 Cedar Blvd, Phoenix, AZ 85001');

<query_masking_rules>
    <rule>
        <name>hide SSN</name>
        <regexp>(^|\D)\d{3}-\d{2}-\d{4}($|\D)</regexp>
        <replace>000-00-0000</replace>
    </rule>
</query_masking_rules>