Guia de Exportação S3 por Domínio

Estrutura de particionamento (todos os domínios)

s3://seu-bucket/{domain}_events/
    organization_id={org_id}/
        brand_id={brand_id}/
            year={YYYY}/
                month={MM}/
                    day={DD}/
                        events_part_0001.parquet
                        events_part_0002.parquet

Cada arquivo Parquet deve ter no máximo 500 MB. Um arquivo por dia é aceitável; vários arquivos por dia também funcionam.

Domínio: User (register / update) [#user]

Schema obrigatório

Coluna	Tipo Parquet	Obrigatório	Descrição
`eid`	STRING	✓	ID único do evento (UUID v4 ou v7)
`organization_id`	STRING	✓	ID da organização
`brand_id`	STRING	✓	ID da brand
`user_ext_id`	STRING	✓	ID do usuário no seu sistema
`event`	STRING	✓	`register` ou `update`
`timestamp`	TIMESTAMP (UTC)	✓	Momento do evento
`status`	STRING		`ACTIVE`, `BLOCKED`, `SUSPENDED`, `BANNED`, `SELF_EXCLUDED`, `DEACTIVATED`, `PENDING`
`full_name`	STRING
`first_name`	STRING
`last_name`	STRING
`email`	STRING
`phone`	STRING
`document`	STRING		CPF/documento
`birthdate`	TIMESTAMP (UTC)
`gender`	INT32		1=M, 2=F, 9=outro (ISO/IEC 5218)
`country`	STRING (2 chars)		ISO 3166 Alpha-2, ex: `BR`
`state`	STRING (2 chars)		Sigla do estado
`city`	STRING
`post_code`	STRING
`registered_at`	TIMESTAMP (UTC)		Data de cadastro original
`updated_at`	TIMESTAMP (UTC)
`registration_platform`	STRING		`DESKTOP`, `MOBILE`, `APP`
`kyc_status`	STRING		`VERIFIED`, `PENDING`, `REJECTED`
`is_test_account`	BOOL
`ip`	STRING
`geolocation_lat`	FLOAT64
`geolocation_long`	FLOAT64

Não inclua ingestion_source ou ingested_at — são preenchidos automaticamente pela plataforma.

Exemplo Python (pandas + pyarrow)

import pandas as pd
import pyarrow as pa
import pyarrow.parquet as pq
import s3fs
from datetime import datetime, timezone

df = pd.DataFrame([{
    "eid":                  "b7c3e8a0-1234-7bcd-8901-abcdef012345",
    "organization_id":      "org_123",
    "brand_id":             "brand_456",
    "user_ext_id":          "user_789",
    "event":                "register",
    "timestamp":            datetime(2025, 10, 15, 14, 30, 0, tzinfo=timezone.utc),
    "status":               "ACTIVE",
    "full_name":            "João Silva",
    "email":                "joao@exemplo.com",
    "document":             "12345678900",
    "country":              "BR",
    "state":                "SP",
    "city":                 "São Paulo",
    "registered_at":        datetime(2025, 10, 15, 14, 30, 0, tzinfo=timezone.utc),
    "registration_platform": "MOBILE",
    "kyc_status":           "VERIFIED",
    "is_test_account":      False,
}])

schema = pa.schema([
    pa.field("eid", pa.string()),
    pa.field("organization_id", pa.string()),
    pa.field("brand_id", pa.string()),
    pa.field("user_ext_id", pa.string()),
    pa.field("event", pa.string()),
    pa.field("timestamp", pa.timestamp("ms", tz="UTC")),
    pa.field("status", pa.string(), nullable=True),
    pa.field("full_name", pa.string(), nullable=True),
    pa.field("email", pa.string(), nullable=True),
    pa.field("document", pa.string(), nullable=True),
    pa.field("country", pa.string(), nullable=True),
    pa.field("state", pa.string(), nullable=True),
    pa.field("city", pa.string(), nullable=True),
    pa.field("registered_at", pa.timestamp("ms", tz="UTC"), nullable=True),
    pa.field("registration_platform", pa.string(), nullable=True),
    pa.field("kyc_status", pa.string(), nullable=True),
    pa.field("is_test_account", pa.bool_(), nullable=True),
])

table = pa.Table.from_pandas(df, schema=schema, preserve_index=False)
pq.write_table(
    table,
    "s3://seu-bucket/user_events/organization_id=org_123/brand_id=brand_456/year=2025/month=10/day=15/events_part_0001.parquet",
    filesystem=s3fs.S3FileSystem(),
)

Domínio: Casino (open / win / lose) [#casino]

Schema obrigatório

Coluna	Tipo Parquet	Obrigatório	Descrição
`eid`	STRING	✓
`organization_id`	STRING	✓
`brand_id`	STRING	✓
`user_ext_id`	STRING	✓
`event`	STRING	✓	`open`, `win`, `lose`
`timestamp`	TIMESTAMP (UTC)	✓
`currency`	STRING	✓	`BRL`, `USD`, etc.
`bet_id`	STRING
`casino_session_id`	STRING
`bet_dt`	TIMESTAMP (UTC)
`bet_status`	STRING		`open`/`win`/`lose` (normalizado para `OPEN`/`WIN`/`LOSS` internamente)
`bet_amount`	DECIMAL(18,2)
`bet_amount_bonus`	DECIMAL(18,2)
`win_amount`	DECIMAL(18,2)
`win_amount_bonus`	DECIMAL(18,2)
`after_balance`	DECIMAL(18,2)
`before_balance`	DECIMAL(18,2)
`game_ext_id`	STRING
`game_name`	STRING
`game_provider`	STRING
`game_provider_ext_id`	STRING
`game_type`	STRING		`SLOT`, `TABLE`, `LIVE`
`is_free_bet`	BOOL
`platform`	STRING		`WEB`, `MOBILE`, `APP`

bet_amount e win_amount devem ser Decimal (não Float) para evitar erros de arredondamento.

Domínio: Transaction (deposit / withdraw) [#transaction]

Schema obrigatório

Coluna	Tipo Parquet	Obrigatório	Descrição
`eid`	STRING	✓
`organization_id`	STRING	✓
`brand_id`	STRING	✓
`user_ext_id`	STRING	✓
`event`	STRING	✓	`deposit`, `withdraw`
`timestamp`	TIMESTAMP (UTC)	✓
`transaction_id`	STRING
`transaction_dt`	TIMESTAMP (UTC)
`amount`	DECIMAL(18,2)
`status`	STRING		`RECEIVED`, `APPROVED`, `PENDING`, `DENIED`
`currency`	STRING
`payment_method`	STRING		`PIX`, `TED`, `CREDIT_CARD`
`payment_provider`	STRING
`bonus_credited`	BOOL
`bonus_code`	STRING
`kyc_verified`	BOOL
`is_first_transaction`	BOOL
`after_balance`	DECIMAL(18,2)
`before_balance`	DECIMAL(18,2)

Domínio: Sportsbook (open / win / lose / cashout / refund / cancel / pending) [#sportsbook]

Campos principais

Coluna	Tipo Parquet	Obrigatório	Descrição
`eid`	STRING	✓
`organization_id`	STRING	✓
`brand_id`	STRING	✓
`user_ext_id`	STRING	✓
`event`	STRING	✓	`open`, `win`, `lose`, `cashout`, `refund`, `cancel`, `pending`
`timestamp`	TIMESTAMP (UTC)	✓
`currency`	STRING	✓
`bet_id`	STRING
`bet_dt`	TIMESTAMP (UTC)
`ticket_status`	STRING		`open`, `win`, `lose`, `cancel`, `cashout`, `refund`, `reject`
`bet_type`	STRING		`SINGLE`, `MULTIPLE`, `SYSTEM`
`bet_amount`	DECIMAL(18,2)
`bet_amount_bonus`	DECIMAL(18,2)
`bet_odds`	FLOAT64
`bet_platform`	STRING		`WEB`, `MOBILE`, `APP`
`bet_timing`	STRING		`LIVE`, `PREMATCH`
`bet_virtual`	BOOL
`win_amount`	DECIMAL(18,2)
`after_balance`	DECIMAL(18,2)
`before_balance`	DECIMAL(18,2)

Campos de selections (formato COLUMNAR — obrigatório)

As selections devem ser exportadas em formato columnar (uma coluna por atributo, cada coluna é um array). NÃO use array de objetos JSON. Prefixo das colunas: selection_* (singular).

Coluna	Tipo Parquet	Descrição
`selection_id`	LIST(STRING)
`selection_status`	LIST(STRING)
`selection_choice`	LIST(STRING)	Ex: `OVER_2.5`
`selection_sport_type`	LIST(STRING)	Ex: `SOCCER`
`selection_league`	LIST(STRING)
`selection_market`	LIST(STRING)
`selection_odds`	LIST(FLOAT64)
`selection_is_live`	LIST(BOOL)
`selection_virtual`	LIST(BOOL)
`selection_home_team`	LIST(STRING)
`selection_away_team`	LIST(STRING)
`selection_competitors`	LIST(STRING)	Concatenação (ex: `"Flamengo,Palmeiras"`)
`selection_event_name`	LIST(STRING)
`sport_match_id`	LIST(STRING)

Exemplo Python — converter array de objetos → columnar

import pandas as pd
import pyarrow as pa
import pyarrow.parquet as pq
from datetime import datetime, timezone

raw_bets = [{
    "eid": "sport-eid-001",
    "organization_id": "org_123",
    "brand_id": "brand_456",
    "user_ext_id": "user_789",
    "event": "open",
    "timestamp": datetime(2025, 10, 15, 20, 0, 0, tzinfo=timezone.utc),
    "bet_id": "bilhete_001",
    "ticket_status": "open",
    "bet_amount": 50.00,
    "bet_odds": 2.75,
    "currency": "BRL",
    "bet_timing": "LIVE",
    "bet_platform": "MOBILE",
    "selections": [
        {
            "selection_id": "sel-001",
            "selection_status": "open",
            "selection_sport_type": "SOCCER",
            "selection_league": "Campeonato Brasileiro",
            "selection_market": "1X2",
            "selection_choice": "1",
            "selection_odds": 1.80,
            "selection_is_live": True,
            "selection_virtual": False,
            "selection_home_team": "Flamengo",
            "selection_away_team": "Palmeiras",
            "selection_competitors": "Flamengo,Palmeiras",
            "selection_event_name": "FLA x PAL",
            "sport_match_id": "match-001",
        },
    ],
}]


def to_columnar(bets):
    """Converte selections de array-de-objetos para colunar."""
    result = []
    for bet in bets:
        sels = bet.pop("selections", [])
        flat = {
            "selection_id":          [s.get("selection_id", "")          for s in sels],
            "selection_status":      [s.get("selection_status", "")      for s in sels],
            "selection_choice":      [s.get("selection_choice", "")      for s in sels],
            "selection_sport_type":  [s.get("selection_sport_type", "")  for s in sels],
            "selection_league":      [s.get("selection_league", "")      for s in sels],
            "selection_market":      [s.get("selection_market", "")      for s in sels],
            "selection_odds":        [float(s.get("selection_odds", 0))  for s in sels],
            "selection_is_live":     [bool(s.get("selection_is_live", False)) for s in sels],
            "selection_virtual":     [bool(s.get("selection_virtual", False)) for s in sels],
            "selection_home_team":   [s.get("selection_home_team", "")   for s in sels],
            "selection_away_team":   [s.get("selection_away_team", "")   for s in sels],
            "selection_competitors": [s.get("selection_competitors", "") for s in sels],
            "selection_event_name":  [s.get("selection_event_name", "")  for s in sels],
            "sport_match_id":        [s.get("sport_match_id", "")        for s in sels],
        }
        result.append({**bet, **flat})
    return result


rows = to_columnar(raw_bets)

schema = pa.schema([
    pa.field("eid", pa.string()),
    pa.field("organization_id", pa.string()),
    pa.field("brand_id", pa.string()),
    pa.field("user_ext_id", pa.string()),
    pa.field("event", pa.string()),
    pa.field("timestamp", pa.timestamp("ms", tz="UTC")),
    pa.field("bet_id", pa.string(), nullable=True),
    pa.field("ticket_status", pa.string(), nullable=True),
    pa.field("bet_amount", pa.decimal128(18, 2), nullable=True),
    pa.field("bet_odds", pa.float64(), nullable=True),
    pa.field("currency", pa.string()),
    pa.field("bet_timing", pa.string(), nullable=True),
    pa.field("bet_platform", pa.string(), nullable=True),
    pa.field("selection_id", pa.list_(pa.string())),
    pa.field("selection_status", pa.list_(pa.string())),
    pa.field("selection_choice", pa.list_(pa.string())),
    pa.field("selection_sport_type", pa.list_(pa.string())),
    pa.field("selection_league", pa.list_(pa.string())),
    pa.field("selection_market", pa.list_(pa.string())),
    pa.field("selection_odds", pa.list_(pa.float64())),
    pa.field("selection_is_live", pa.list_(pa.bool_())),
    pa.field("selection_virtual", pa.list_(pa.bool_())),
    pa.field("selection_home_team", pa.list_(pa.string())),
    pa.field("selection_away_team", pa.list_(pa.string())),
    pa.field("selection_competitors", pa.list_(pa.string())),
    pa.field("selection_event_name", pa.list_(pa.string())),
    pa.field("sport_match_id", pa.list_(pa.string())),
])

df = pd.DataFrame(rows)
table = pa.Table.from_pandas(df, schema=schema, preserve_index=False)
pq.write_table(
    table,
    "s3://seu-bucket/sportsbook_events/organization_id=org_123/brand_id=brand_456/year=2025/month=10/day=15/events_part_0001.parquet",
)

Get Started

Quickstart

Event Tracking

Backfill S3

Analytics & Reports

Risk & Scoring

Compliance

Monitoria & Alertas

Platform

Resources

Guia de Exportação S3 por Domínio

Guia de Exportação S3 por Domínio

Estrutura de particionamento (todos os domínios)

Domínio: User (register / update) [#user]

Schema obrigatório

Exemplo Python (pandas + pyarrow)

Domínio: Casino (open / win / lose) [#casino]

Schema obrigatório

Domínio: Transaction (deposit / withdraw) [#transaction]

Schema obrigatório

Domínio: Sportsbook (open / win / lose / cashout / refund / cancel / pending) [#sportsbook]

Campos principais

Campos de selections (formato COLUMNAR — obrigatório)

Exemplo Python — converter array de objetos → columnar

​Guia de Exportação S3 por Domínio

​Estrutura de particionamento (todos os domínios)

​Domínio: User (register / update) [#user]

​Schema obrigatório

​Exemplo Python (pandas + pyarrow)

​Domínio: Casino (open / win / lose) [#casino]

​Schema obrigatório

​Domínio: Transaction (deposit / withdraw) [#transaction]

​Schema obrigatório

​Domínio: Sportsbook (open / win / lose / cashout / refund / cancel / pending) [#sportsbook]

​Campos principais

​Campos de selections (formato COLUMNAR — obrigatório)

​Exemplo Python — converter array de objetos → columnar

Guia de Exportação S3 por Domínio

Estrutura de particionamento (todos os domínios)

Domínio: User (register / update) [#user]

Schema obrigatório

Exemplo Python (pandas + pyarrow)

Domínio: Casino (open / win / lose) [#casino]

Schema obrigatório

Domínio: Transaction (deposit / withdraw) [#transaction]

Schema obrigatório

Domínio: Sportsbook (open / win / lose / cashout / refund / cancel / pending) [#sportsbook]

Campos principais

Campos de selections (formato COLUMNAR — obrigatório)

Exemplo Python — converter array de objetos → columnar