创建列类型为 RECORD 的表

Question

我正在使用大查询，我想创建一个使用“记录”类型列填充表的作业。 数据将由查询填充 - 那么我如何编写返回“记录”类型列的查询。

谢谢！

Answer 1

Pentium10 提出的某种选项在 GBQ UI 或 API Explorer 中从未对我有用。
我可能遗漏了什么

同时，我发现的解决方法如下例所示

SELECT location.state, location.city FROM JS(
  (      // input table
  SELECT NEST(CONCAT(state, ',', city)) AS locations
  FROM (
    SELECT state, city FROM 
    (SELECT 'florida' AS state, 'miami' AS city),
    (SELECT 'california' AS state, 'la' AS city),
    (SELECT 'romania' AS state, 'transylvania' AS city)
    ) 
  ),
  locations,     // input columns
  "[    // output schema
    {'name': 'location', 'type': 'RECORD',
     'mode': 'REPEATED',
     'fields': [
       {'name': 'state', 'type': 'STRING'},
       {'name': 'city', 'type': 'STRING'}
     ]    
    }
  ]",
  "function(row, emit){    // function 
    for (var i = 0; i < row.locations.length; i++) {
      var c = [];
      x = row.locations[i].split(',');
      t = {state:x[0], city:x[1]}
      c.push(t);
      emit({location: c});  
    };
  }"
)  

请注意：
您应该使用Allow Large Results 设置目标表并取消选中Flatten Results

输出表的结果是（在 JSON 模式下）

[
  {
    "location": [
      {
        "state": "california",
        "city": "la"
      }
    ]
  },
  {
    "location": [
      {
        "state": "florida",
        "city": "miami"
      }
    ]
  },
  {
    "location": [
      {
        "state": "romania",
        "city": "transylvania"
      }
    ]
  }
]

添加以解决@AdiCohen 在他的真实示例中遇到的一些问题 他在最近的 cmets 中展示了：

问：我的查询除了记录列之外还有其他列，但是当我运行时 查询，它们返回为空。我怎样才能创建一个表 类型？

SELECT amount, currency, location.state, location.city FROM JS( 
  ( // input table 
    SELECT NEST(CONCAT(state, ',', city)) AS locations, 
      SUM(amount) AS amount, MAX(currency) as currency 
    FROM ( 
      SELECT state, city, amount, currency, ROW_NUMBER() OVER() as grp FROM 
        (SELECT 'florida' AS state, 'miami' AS city, 'coins' AS currency, 40 AS amount), 
        (SELECT 'california' AS state, 'la' AS city, 'coins' AS currency, 40 AS amount), 
        (SELECT 'romania' AS state, 'transylvania' AS city,'coins' AS currency, 40 AS amount) 
    ) GROUP BY grp
  ), 
  amount, currency, locations, // input columns 
  "[ // output schema 
    {'name': 'location', 'type': 'RECORD', 'mode': 'REPEATED', 
    'fields': [ 
      {'name': 'state', 'type': 'STRING'}, 
      {'name': 'city', 'type': 'STRING'} 
    ] }, 
    { 'name': 'amount', 'type': 'INTEGER'}, 
    { 'name': 'currency', 'type': 'STRING'} 
  ]", 
  "function(row, emit) { // function 
    for (var i = 0; i < row.locations.length; i++) { 
      var c = []; 
      x = row.locations[i].split(','); 
      t = {state:x[0], city:x[1]} 
      c.push(t); 
      emit({amount: row.amount, currency: row.currency, location: c}); 
    }; 
  }"
) 

这里的输出是：

[
  {
    "amount": "40",
    "currency": "coins",
    "location_state": "romania",
    "location_city": "transylvania"
  },
  {
    "amount": "40",
    "currency": "coins",
    "location_state": "florida",
    "location_city": "miami"
  },
  {
    "amount": "40",
    "currency": "coins",
    "location_state": "california",
    "location_city": "la"
  }
]

Answer 2

您需要使用dot 表示法将输出反映为RECORD 示例查询：

select 
  'florida' as country.state, 
  'SFO' as country.city;

在此示例中，country 是记录，state|city 是记录中的字段。