从 SQL 中选择 JSON 并按 ID 连接元素

Posted 2023-03-25

【中文标题】从 SQL 中选择 JSON 并按 ID 连接元素

【英文标题】：Select JSON from SQL and JOIN elements by IDs

【发布时间】：2018-07-28 15:30:06

【问题描述】：

问题：如何为每个问题/答案对返回一行？我觉得我几乎有过它，但是答案中的嵌套 ID 让我失望了，因为我正在尝试 Regex，但我想有一种更纯粹的 JSON 方式可以在没有任何 Regex 的情况下做到这一点，对吧？

这就是我要找的东西：

i.e.
--------------------------------------------------------------------------------
Question                                      Answer
What do you think?                            Love it. So easy to use.
Which sales channels do you use?              Facebook / Instagram
Scale of 1 to 10  how much do you like this?  7

这是数据：

# setup our data
WITH survey_results -- table name
AS
(
    SELECT '''
    
      "fields": [
        
          "id": "ycq4WbK9EdNF",
          "title": "What do you think?",
          "type": "long_text"
        ,
        
          "id": "uPLaLxMKZ9lm",
          "title": "Scale of 1 to 10 how much do you like this?",
          "type": "opinion_scale"
        ,
        
          "choices": [
            
              "id": "k4227hvYbTaR",
              "label": "eBay"
            ,
            
              "id": "zjWplAMROVuP",
              "label": "Etsy"
            
          ],
          "id": "f6ois7aPxVz4",
          "title": "Which sales channels do you use?",
          "type": "multiple_choice"
        
      ],
      "id": "k56nsh",
      "title": "Amazon Survey"
    
    '''
    AS questions, --- column name
    '''
    [
      
        "field": 
          "id": "ycq4WbK9EdNF",
          "type": "long_text"
        ,
        "text": "Love it. So easy to use.",
        "type": "text"
      ,
      
        "field": 
          "id": "uPLaLxMKZ9lm",
          "type": "opinion_scale"
        ,
        "number": 7,
        "type": "number"
      ,
      
        "choice": 
          "label": "Facebook / Instagram"
        ,
        "field": 
          "id": "f6ois7aPxVz4",
          "type": "multiple_choice"
        ,
        "type": "choice"
      
    ]
    '''
    AS answers --- column name
    )

# run the query

SELECT * from survey_results

【参考方案1】：

以下是 BigQuery 标准 SQL

#standardSQL
CREATE TEMPORARY FUNCTION CUSTOM_EXTRACT_QUESTIONS(json STRING)
RETURNS ARRAY<STRUCT<id STRING, question STRING>>
LANGUAGE js AS """
  try  var parsed = JSON.parse(json).fields; var result = []; 
    for (i = 0; i < parsed.length; i++) 
      e = []; e.id = parsed[i].id; e.question = parsed[i].title; result.push(e);
    ; return result;  catch (e)  return null 
""";
CREATE TEMPORARY FUNCTION CUSTOM_EXTRACT_ANSWERS(json STRING)
RETURNS ARRAY<STRUCT<id STRING, answer STRING>>
LANGUAGE js AS """
  try  var parsed = JSON.parse(json);
    var result = []; for (i = 0; i < parsed.length; i++) 
      e = []; e.id = parsed[i].field.id; 
      e.answer = parsed[i].text || parsed[i].number || parsed[i].choice.label; result.push(e);
    ; return result;  catch (e)  return null 
""";
SELECT q.id, question, answer
FROM survey_results, 
UNNEST(CUSTOM_EXTRACT_QUESTIONS(questions)) q,
UNNEST(CUSTOM_EXTRACT_ANSWERS(answers)) a
WHERE q.id  = a.id

如果应用于您问题中的虚拟数据 - 结果是（测试）

Row id              question                                       answer    
1   ycq4WbK9EdNF    What do you think?                             Love it. So easy to use.
2   uPLaLxMKZ9lm    Scale of 1 to 10 how much do you like this?    7
3   f6ois7aPxVz4    Which sales channels do you use?               Facebook / Instagram

【讨论】：

另外，相当复杂。如果我们取消选中标准 SQL 会更简单吗？

遇到了一个拦截器：需要在 Data Studio 中公开它，所以我尝试将此查询保存为视图。但是视图不支持临时功能。你会建议我如何解决这个问题？我还尝试在数据工作室创建自定义查询，但看起来其他人遇到了与我相同的错误：en.advertisercommunity.com/t5/Data-Studio/…

如果您要发布具有新要求的新问题，以便能够保存为视图等 - 我很可能会回答。但您还需要重新接受上述答案 - 因为它完全回答了您最初的问题:o)