Início rápido: Execute a sua primeira consulta DAX

Neste quickstart, autentica-se no Power BI, executa uma consulta DAX num modelo semântico e desserializa a resposta do Arrow numa estrutura de dados local.

Pré-requisitos

Um espaço de trabalho Power BI com pelo menos um modelo semântico.
Permissões de construção e leitura no modelo semântico.
Um registo de aplicação Microsoft Entra (ou usar autenticação interativa para testes).

Python 3.10 ou posterior.
Dependências de instalação:
```
pip install msal pyarrow pandas
```

1 - Autenticar

Adquira um token de portador com o https://analysis.windows.net/powerbi/api/.default telescópio.

from msal import PublicClientApplication

client_id = "YOUR_APP_CLIENT_ID"
authority = "https://login.microsoftonline.com/YOUR_TENANT_ID"
scopes = ["https://analysis.windows.net/powerbi/api/.default"]

app = PublicClientApplication(client_id, authority=authority)
result = app.acquire_token_interactive(scopes=scopes)

access_token = result["access_token"]

using Microsoft.Identity.Client;

var clientId = "YOUR_APP_CLIENT_ID";
var tenantId = "YOUR_TENANT_ID";
var scopes = new[] { "https://analysis.windows.net/powerbi/api/.default" };

var app = PublicClientApplicationBuilder
    .Create(clientId)
    .WithAuthority(AzureCloudInstance.AzurePublic, tenantId)
    .WithRedirectUri("http://localhost")
    .Build();

var result = await app.AcquireTokenInteractive(scopes).ExecuteAsync();
var accessToken = result.AccessToken;

$clientId = "YOUR_APP_CLIENT_ID"
$tenantId = "YOUR_TENANT_ID"
$scopes = @("https://analysis.windows.net/powerbi/api/.default")

$token = Get-MsalToken -ClientId $clientId -TenantId $tenantId -Scopes $scopes -Interactive
$accessToken = $token.AccessToken

2 - Executar uma consulta DAX

Envie um pedido POST para o endpoint Execute Queries Arrow com uma instrução simples EVALUATE .

import io
import pyarrow as pa
import requests

group_id = "YOUR_WORKSPACE_ID"
dataset_id = "YOUR_DATASET_ID"

url = (f"https://api.powerbi.com/v1.0/myorg/groups/{group_id}"
       f"/datasets/{dataset_id}/executeDaxQueries")

headers = {
    "Authorization": f"Bearer {access_token}",
    "Content-Type": "application/json",
}

body = {"query": "EVALUATE TOPN(5, 'DimProduct')"}

response = requests.post(url, headers=headers, json=body)
response.raise_for_status()

reader = pa.ipc.open_stream(io.BytesIO(response.content))
table = reader.read_all()
df = table.to_pandas()

print(df)

using System.Net.Http.Headers;
using Apache.Arrow;
using Apache.Arrow.Ipc;

var groupId = "YOUR_WORKSPACE_ID";
var datasetId = "YOUR_DATASET_ID";

var url = $"https://api.powerbi.com/v1.0/myorg/groups/{groupId}"
        + $"/datasets/{datasetId}/executeDaxQueries";

using var client = new HttpClient();
client.DefaultRequestHeaders.Authorization =
    new AuthenticationHeaderValue("Bearer", accessToken);

var body = new StringContent(
    """{"query": "EVALUATE TOPN(5, 'DimProduct')"}""",
    System.Text.Encoding.UTF8,
    "application/json");

var response = await client.PostAsync(url, body);
response.EnsureSuccessStatusCode();

using var stream = await response.Content.ReadAsStreamAsync();
using var reader = new ArrowStreamReader(stream);

var batch = await reader.ReadNextRecordBatchAsync();
Console.WriteLine($"Rows: {batch.Length}, Columns: {batch.ColumnCount}");

$groupId = "YOUR_WORKSPACE_ID"
$datasetId = "YOUR_DATASET_ID"

$url = "https://api.powerbi.com/v1.0/myorg/groups/$groupId" +
       "/datasets/$datasetId/executeDaxQueries"

$headers = @{
    "Authorization" = "Bearer $accessToken"
    "Content-Type"  = "application/json"
}

$body = @{ query = "EVALUATE TOPN(5, 'DimProduct')" } | ConvertTo-Json

$response = Invoke-WebRequest -Uri $url -Method Post -Headers $headers -Body $body

# Save the binary Arrow stream to a file for inspection
$arrowFile = "result.arrow"
[System.IO.File]::WriteAllBytes($arrowFile, $response.Content)
Write-Host "Arrow stream saved to $arrowFile ($($response.Content.Length) bytes)"

3 - Inspecionar os resultados

# Print the Arrow schema (column names and types)
print(table.schema)

# Show the first few rows as a pandas DataFrame
print(df.head())

# Access a specific column
print(table.column("ProductName").to_pylist()[:5])

// Print the schema (column names and types)
for (var i = 0; i < batch.Schema.FieldsList.Count; i++)
{
    var field = batch.Schema.FieldsList[i];
    Console.WriteLine($"  {field.Name}: {field.DataType}");
}

// Print the first row's values
for (var i = 0; i < batch.ColumnCount; i++)
{
    var column = batch.Column(i);
    Console.Write($"{batch.Schema.FieldsList[i].Name}=");
    Console.Write(column.GetType().GetMethod("GetValue")?.Invoke(column, new object[] { 0 }));
    Console.Write("  ");
}

# Use Python with pyarrow to read the saved Arrow file
python -c @"
import pyarrow as pa
reader = pa.ipc.open_stream('result.arrow')
table = reader.read_all()
print(table.schema)
print(table.to_pandas().head())
"@

Limpeza de recursos

Se criaste um registo de Microsoft Entra app apenas para testes, navega até ao portal Azure e apaga-o. Os tokens de acesso expiram automaticamente e não precisam de revogação manual.

Comentários

Esta página foi útil?

Last updated on 2026-05-07