如何为数组定义 JSON 模式？答案

【问题标题】：how to define JSON schema for array?如何为数组定义 JSON 模式？
【发布时间】：2018-06-11 20:44:21
【问题描述】：

我有这个 JSON 文件：

    {  
        "tcCensusTractBlockFacesStatefp":"01",
        "tcCensusTractBlockFacesCountyfp":"001",
        "tcCensusTractBlockFacesTractce":"020200",
        "tcCensusTractBlockFacesBlockce10":"2022",
        "tcCensusTractBlockFacesGeoid":"010010202002022",
        "cellIDs":[  
        9839958675010879488,
        9839958675082706944,
        9839958677655912448,
        9839958677556297728,
        9839958676975910912,
        9839958677063991296,
        9839958677105934336,
        9839958679922409472,
        9839958679922933760,
        9839958679975886848,
        9839958679979032576,
        9839958679461036032,
        9839958679450550272,
        9839958678956670976,
        9839958678926262272,
        9839958667678187520,
        9839958667562844160,
        9839958675010879488
        ]
    }

如何定义架构以将其导入 BigQuery 中？

bq mk --table $DATASET:$TABLE tl_2017_schema.json
bq load --source_format=NEWLINE_DELIMITED_JSON $DATASET:$TABLE $WNAME

这是我尝试做的：

tl_2017_schema.json

...
---
     {
        "name": "cellIDs",
        "type": "RECORD",
        "mode": "REPEATED",
        "fields": [
            {
                "name": "cellID",
                "type": "INT64",
                "mode": "NULLABLE"
            }
        ]
    }
...

但是负责ARRAY的这部分不能匹配我在JSON文件中的数组结构：

“cellID”：[
9839958675010879488, 9839958675082706944, 9839958677655912448, 9839958677556297728, 9839958676975910912, 9839958677063991296, 9839958677105934336, 9839958679922409472, 9839958679922933760, 9839958679975886848, 9839958679979032576, 9839958679461036032, 9839958679450550272, 9839958678956670976, 9839958678926262272, 9839958667678187520, 9839958667562844160, 9839958675010879488 ]

怎么做？

【问题讨论】：

标签： arrays json google-bigquery

【解决方案1】：

运行CREATE TABLE statement 可能更容易：

CREATE TABLE dataset.tablename
(
  ...
  cellIDs ARRAY<INT64>
)

如果您确实想将架构指定为 JSON，则示例中的解决方法是将 cellIDs 设为 REPEATED INTEGER 类型：

{
    "name": "cellIDs",
    "type": "INTEGER",
    "mode": "REPEATED"
}

【讨论】：

谢谢，效果很好：{ "name": "cellIDs", "type": "NUMERIC", "mode": "REPEATED" } ]