Couchbase N1QL：在平面结构输出的条件下从数组返回特定项目

Question

我想输出一个平面数据表（对于 csv），其中我使用字段 predictionModels.modelID 作为标题和 predictionModels.predictions.passengers 的相应值，当 predictionModels.predictions.daysAhead = 1 时。模型的数量是可变的但有限，所以我很乐意将它们硬编码到查询中，但不能只是 UNNEST 并输出结果。

我不愿意创建索引，因为它是一个正在生产中的大型数据存储，而且我是办公室里的初级人员，实际上是在玩游戏以更好地了解我们的数据，我担心做任何持久的事情或者可能会影响其他地方的资源可用性。我一直在尝试这样的事情-

SELECT departureDateTime, lineID, weight, 
CASE WHEN predictionModels.modelID = "model1" AND predictionModels.predictions.daysAhead=1 THEN predictionModels.predictions.passengers END AS Model1,
CASE WHEN predictionModels.modelID = "model2" AND predictionModels.predictions.daysAhead=1 THEN predictionModels.predictions.passengers END AS Model2
FROM bucket

样本数据：

    'departureDateTime': '2022-12-23T00:10:00+00:00',
    'lineID': 'f2b4d1d9',
    'weight': '630'
    'predictionModels': [
        {
        "modelID":"model1",
        "predictions": [
            {"daysAhead":1, "passengers":11},
            {"daysAhead":2, "passengers":12},
            {"daysAhead":3, "passengers":13},
            {"daysAhead":4, "passengers":14}
                ]
        },
        {
     "modelID":"model2",
     "predictions": [
            {"daysAhead":1, "passengers":21},
            {"daysAhead":2, "passengers":22},
            {"daysAhead":3, "passengers":23},
            {"daysAhead":4, "passengers":24}
                ]
        }
    ]
},
{
    'departureDateTime': '2023-01-24T00:17:00+00:00',
    'lineID': 'f2b4d1d9',
    'weight': '520'
    'predictionModels': [
    {
        "modelID":"model2",
        "predictions": [
            {"daysAhead":1, "passengers":210},
            {"daysAhead":2, "passengers":220},
            {"daysAhead":3, "passengers":230},
            {"daysAhead":4, "passengers":240}
        ]
    }
    ]
}

期望的输出：

        departureDateTime |   lineID | weight | model1 | model2
---------------------------------------------------------------
2022-12-23T00:10:29+00:00 | f2b4d1d9 |   630  |     11 |     21
2023-01-24T00:17:29+00:00 | f2b4d1d9 |   520  |    Nan |    210