Kueri Data Semi-Terstruktur

Otomatisasi Data Pipeline di Snowflake

Emily Melhuish

Technical Curriculum Developer, Snowflake

Pemuatan Data Semi-Terstruktur

pemrosesan data.png

¹ * Materi Pembelajaran Snowflake

Mengelola Bentuk Data Berbeda

Carrier A JSON Carrier A

Carrier B

JSON Carrier B

Contoh Transformasi

Tangkapan layar 2026-05-11 pukul 5.56.34 pm.png

Tipe Data Snowflake

Kategori	Tipe	Contoh
Numerik	NUMBER, INT, FLOAT	`credits_used NUMBER(10,2)`
Teks	VARCHAR, STRING, TEXT	`carrier_name VARCHAR`
Tanggal/Waktu	DATE, TIME, TIMESTAMP	`dispatch_date DATE`
Boolean	BOOLEAN	`is_active BOOLEAN`
Semi-terstruktur	VARIANT, OBJECT, ARRAY	`event_payload VARIANT`

Tipe VARIANT

VARIANT menyimpan data semi-terstruktur apa adanya: tanpa skema
Muat JSON, Parquet, Avro, ORC, atau XML langsung
Struktur data ditentukan saat kueri, bukan saat muat (schema-on-read)

COPY INTO shipment_events
FROM @harbr_stage/events/
FILE_FORMAT = (TYPE = 'JSON');

Awan kata VARIANT yang menampilkan semua tipe data semi-terstruktur yang didukung: JSON, ARRAY, OBJECT, AVRO, PARQUET, SEMI-STRUCTURED

Menavigasi JSON

Contoh JSON

{
  "shipment_id": "SHP-001",
  "carrier": { "name": "FastFreight" },
  "stops": [
    { "location": "Rotterdam Port" },
    { "location": "Amsterdam Hub" }
  ]
}

Menavigasi JSON

Menelusuri struktur:

<kolom>:<elemen_level1>.<elemen_level2>

SELECT 
event:id::STRING AS shipment_id
, event:carrier.name::STRING AS carrier
, event:stops[0]:location::STRING 
AS first_stop 
FROM shipment_events;

Mengubah Array jadi Baris dengan FLATTEN

Menggunakan LATERAL FLATTEN

SELECT 
  s.event_data:shipment_id::STRING AS shipment_id,
  f.value:location::STRING AS stop,
  f.value:arrived_at::TIMESTAMP AS arrived_at 
FROM shipment_events s,
  LATERAL FLATTEN(input => s.event_data:stops) f;

Ayo berlatih!

Otomatisasi Data Pipeline di Snowflake