如何通过 HTTP api 获取 BigQuery 查询的元数据

Question

我正在为 F# 开发一个类型提供程序，为了实现它，我们需要获取查询编译时的架构，这意味着它必须很快。截至目前，我们运行的是这样的：

bq query --format=json --dry_run=true --use_legacy_sql=false 'SELECT @a IS TRUE AS x, @b + 1 AS y, "foo" = @c AS z, ["tomas", "jansson"] as w, STRUCT("wat" as t, 69 as u) as v, [STRUCT(3, "allo" as g), STRUCT(5 as a, "yolo")] as u, STRUCT(["a"] as h) as t;'

这正是我们想要的，但它使用了bq 工具。我想知道在代码库中找不到的底层 http 调用是什么。

我想了解底层代码库的原因是因为我想尽可能多地删除第三方依赖项，这些依赖项可能在构建服务器上不可用或需要很长时间设置。

Answer 1

或者你可以使用

Jobs: insert 与 query 配置属性和 dryRun 设置为 true

Answer 2

编辑：米哈伊尔是对的；只有jobs.insert API 提供架构信息。我将使用这个答案来记录当前的行为。

使用jobs.insert API，这是一个示例请求和响应：

PROJECT="YOUR_PROJECT_NAME"
QUERY="\"SELECT 1 AS x, 'foo' AS y, @b IS FALSE AS z;\""
REQUEST="{\"configuration\":{\"dryRun\":true,\"query\":{\"useLegacySql\":false,\"query\":${QUERY}}}}"
echo $REQUEST | \
curl -X POST -d @- -H "Content-Type: application/json" \
    -H "Authorization: Bearer $(gcloud auth print-access-token)" \
    https://www.googleapis.com/bigquery/v2/projects/$PROJECT/jobs

(some parts omitted)
"statistics": {
 "query": {
  "schema": {
   "fields": [
    {
     "name": "x",
     "type": "INTEGER",
     "mode": "NULLABLE"
    },
    {
     "name": "y",
     "type": "STRING",
     "mode": "NULLABLE"
    },
    {
     "name": "z",
     "type": "BOOLEAN",
     "mode": "NULLABLE"
    }
   ]
  },
  "undeclaredQueryParameters": [
   {
    "name": "b",
    "parameterType": {
     "type": "BOOL"
    }
   }
  ],
  "statementType": "SELECT"
 }

这里是一个使用jobs.query的示例请求和响应，为了比较：

PROJECT="YOUR_PROJECT_NAME"
QUERY="\"SELECT 1 AS x, 'foo' AS y, @b IS FALSE AS z;\""
REQUEST="{\"kind\":\"bigquery#queryRequest\",\"dryRun\":true,\"useLegacySql\":false,\"query\":$QUERY}"
echo $REQUEST | \
  curl -X POST -d @- -H "Content-Type: application/json" \
    -H "Authorization: Bearer $(gcloud auth print-access-token)" \
    https://www.googleapis.com/bigquery/v2/projects/$PROJECT/queries

{
 "kind": "bigquery#queryResponse",
 "jobReference": {
  "projectId": "<omitted>"
 },
 "totalBytesProcessed": "0",
 "jobComplete": true,
 "cacheHit": false
}