Google bigquery 如果记录重复过多，则进行大查询（带展平）

这是关于以下问题的解决方案：我尝试创建了一个测试表并尝试了给定的查询，但它实际上并没有选择纽约和芝加哥的居民。测试数据如下：

{"fullname": "John Smith", "citiesLived": [{"place": "newyork"}, {"place": "chicago"}, {"place": "seattle"}]}
{"fullname": "Adam Smith", "citiesLived": [{"place": "newyork"}, {"place": "chicago"}, {"place": "phil"}]}
{"fullname": "Adam Jefferson", "citiesLived": [{"place": "boston"}, {"place": "chicago"}, {"place": "seattle"}]}

SELECT
  *
FROM (
  SELECT
    fullname,
    IF (citiesLived.place == 'newyork', 1, 0) AS ny,
    IF (citiesLived.place == 'chicago', 1, 0) AS chi
  FROM (FLATTEN(tester.citiesLived, citiesLived))
  OMIT
    RECORD IF citiesLived.place = 'seattle')
WHERE
  ny == 1
  AND chi == 1

查询如下：

{"fullname": "John Smith", "citiesLived": [{"place": "newyork"}, {"place": "chicago"}, {"place": "seattle"}]}
{"fullname": "Adam Smith", "citiesLived": [{"place": "newyork"}, {"place": "chicago"}, {"place": "phil"}]}
{"fullname": "Adam Jefferson", "citiesLived": [{"place": "boston"}, {"place": "chicago"}, {"place": "seattle"}]}

SELECT
  *
FROM (
  SELECT
    fullname,
    IF (citiesLived.place == 'newyork', 1, 0) AS ny,
    IF (citiesLived.place == 'chicago', 1, 0) AS chi
  FROM (FLATTEN(tester.citiesLived, citiesLived))
  OMIT
    RECORD IF citiesLived.place = 'seattle')
WHERE
  ny == 1
  AND chi == 1

---

您不需要进行展平（一般来说，在BigQuery查询中很少需要展平），只要忽略以下内容即可：

SELECT fullname FROM tester.citiesLived
OMIT RECORD IF NOT (
  SOME(citiesLived.place = "newyork") AND
  SOME(citiesLived.place = "chicago"))

---

省略IF的条件表示，如果一些居住的城市是纽约，一些是芝加哥，那么它符合您的标准。但是如果两者都不正确，则应省略记录（因此not谓词）。

我认为这将是对原始预期查询的更完整重写：

SELECT
  *
FROM (
  SELECT
    fullname,
    SOME(citiesLived.place == 'newyork') WITHIN RECORD AS ny,
    SOME(citiesLived.place == 'chicago') WITHIN RECORD AS chi
  FROM tester.citiesLived
  OMIT
    RECORD IF SOME(citiesLived.place = 'seattle'))
WHERE
  ny == true
  AND chi == true