如何用PowerShell/Bash批量删除Azure Purview集合资产（AAD认证）

前提条件 #

• 已安装Azure PowerShell模块（版本≥5.1）或Azure CLI（版本≥2.40.0）
• 已通过AAD账户登录Azure（Connect-AzAccount 或 az login）
• 拥有目标Purview账户的集合管理员权限
• 提前获取目标集合的集合ID（可通过Purview UI的集合详情页获取，或调用API查询）

PowerShell实现方案 #

1. 定义基础变量 #

$purviewAccountName = "your-purview-account-name"  # 替换为你的Purview账户名
$collectionId = "your-collection-id"              # 替换为目标集合ID
$apiBaseUrl = "https://$purviewAccountName.purview.azure.com/catalog/api/atlas/v2"

2. 获取Purview API访问令牌 #

$accessToken = (Get-AzAccessToken -ResourceUrl "https://purview.azure.com").Token
$headers = @{
    "Authorization" = "Bearer $accessToken"
    "Content-Type"  = "application/json"
}

3. 分页获取集合内所有资产GUID #

$allAssetGuids = @()
$nextPageToken = $null

do {
    $queryParams = if ($nextPageToken) { "?filter=collectionId eq '$collectionId'&nextPageToken=$nextPageToken" } else { "?filter=collectionId eq '$collectionId'" }
    $response = Invoke-RestMethod -Method Get -Uri "$apiBaseUrl/entity/bulk$queryParams" -Headers $headers
    
    $allAssetGuids += $response.entities.guid
    $nextPageToken = $response.nextPageToken
} while ($nextPageToken)

4. 批量删除资产 #

Purview API单次支持最多删除1000个资产，因此需要分批处理：

$batchSize = 1000
for ($i=0; $i -lt $allAssetGuids.Count; $i+=$batchSize) {
    $batchGuids = $allAssetGuids[$i..($i+$batchSize-1)]
    $deleteBody = @{ guids = $batchGuids } | ConvertTo-Json
    
    try {
        Invoke-RestMethod -Method Post -Uri "$apiBaseUrl/entity/bulk/delete" -Headers $headers -Body $deleteBody
        Write-Host "成功删除第 $($i/$batchSize + 1) 批资产（共 $($batchGuids.Count) 个）"
    } catch {
        Write-Host "删除第 $($i/$batchSize + 1) 批资产失败：$($_.Exception.Message)"
    }
}

Azure Cloud Bash（Azure CLI）实现方案 #

1. 定义基础变量 #

PURVIEW_ACCOUNT_NAME="your-purview-account-name"  # 替换为你的Purview账户名
COLLECTION_ID="your-collection-id"                # 替换为目标集合ID
API_BASE_URL="https://$PURVIEW_ACCOUNT_NAME.purview.azure.com/catalog/api/atlas/v2"

2. 获取Purview API访问令牌 #

ACCESS_TOKEN=$(az account get-access-token --resource https://purview.azure.com --query accessToken -o tsv)
HEADERS=(-H "Authorization: Bearer $ACCESS_TOKEN" -H "Content-Type: application/json")

3. 分页获取集合内所有资产GUID #

ALL_ASSET_GUIDS=()
NEXT_PAGE_TOKEN=""

while true; do
    QUERY_PARAMS=""
    if [ -n "$NEXT_PAGE_TOKEN" ]; then
        QUERY_PARAMS="?filter=collectionId eq '$COLLECTION_ID'&nextPageToken=$NEXT_PAGE_TOKEN"
    else
        QUERY_PARAMS="?filter=collectionId eq '$COLLECTION_ID'"
    fi

    RESPONSE=$(curl -s "${API_BASE_URL}/entity/bulk$queryParams" "${HEADERS[@]}")
    CURRENT_GUIDS=$(echo "$RESPONSE" | jq -r '.entities[].guid')
    ALL_ASSET_GUIDS+=($CURRENT_GUIDS)
    
    NEXT_PAGE_TOKEN=$(echo "$RESPONSE" | jq -r '.nextPageToken')
    if [ "$NEXT_PAGE_TOKEN" == "null" ]; then
        break
    fi
done

4. 批量删除资产 #

同样按1000个/批处理：

BATCH_SIZE=1000
TOTAL_GUIDS=${#ALL_ASSET_GUIDS[@]}

for ((i=0; i<TOTAL_GUIDS; i+=BATCH_SIZE)); do
    BATCH_GUIDS=("${ALL_ASSET_GUIDS[@]:i:BATCH_SIZE}")
    DELETE_BODY=$(jq -n --argjson guids "$(printf '%s\n' "${BATCH_GUIDS[@]}" | jq -R . | jq -s .)" '{guids: $guids}')
    
    RESPONSE=$(curl -s -X POST "${API_BASE_URL}/entity/bulk/delete" "${HEADERS[@]}" -d "$DELETE_BODY")
    if [ $(echo "$RESPONSE" | jq -r '.guidAssignments | length') -eq ${#BATCH_GUIDS[@]} ]; then
        echo "成功删除第 $((i/BATCH_SIZE + 1)) 批资产（共 ${#BATCH_GUIDS[@]} 个）"
    else
        echo "删除第 $((i/BATCH_SIZE + 1)) 批资产失败：$(echo "$RESPONSE" | jq -r '.error.message')"
    fi
done

注意事项 #

• 执行删除前建议先验证资产列表，避免误删：可将删除步骤替换为打印GUID列表确认
• 若资产数量极大，建议在非业务高峰时段执行，避免影响Purview服务性能
• 部分特殊资产（如通过扫描生成的资产）可能需要先删除扫描规则或关联资源，否则可能重新被同步

内容的提问来源于stack exchange，提问作者mytabi