将单个 sql 列拆分为五个

Question

我试图围绕“>”定界符将一列分成最多五列，但我尝试过的东西没有成功：

我试过了

select
id, 
compoundColumn,
split(compoundColumn," > ")[1] as "first"
split(compoundColumn," > ")[2] as "second"
from table
where compoundColumn is not null

没用，

这有点像（无论如何是第一部分，而不是第 n 部分）

select
id, 
compoundColumn,
first(split(compoundColumn," > ")) as "first"
nth(compoundColumn," > ")[n] as "second"
from table

我在这里找到了很多例子，但他们似乎都在说要使用括号，但括号会引发错误：

Exception: Malformed SQL. More information: Error with SQL statement: You have an error in your SQL syntax; check the manual that corresponds to your MySQL server version for the right syntax to use near '[1] as "first" from table where compoundColumn IS NOT NULL' at line 3.

Answer 1

您的 SQL
我猜 CloudSQL 是基于某些旧版本的 MySQL 的，它只能使用 substring_index 进行拆分（请参阅下面的查询 - 是的，它冗长且笨拙，case 子句必须清理短字符串)
也许尝试使用 [offset(0)] 或 [ordinal(1)] 括号，这对我们有用，尽管我们使用 Postgres 方言，也作为 #standardSql，而不是 #legacySql

SQL 从第二点开始：(fiddle)

select id,
  case when substring_index(cc,' > ',0) = cc then null else substring_index(substring_index(cc,' > ',1),' > ',-1) end as a1,
  case when substring_index(cc,' > ',1) = cc then null else substring_index(substring_index(cc,' > ',2),' > ',-1) end as a2,
  case when substring_index(cc,' > ',2) = cc then null else substring_index(substring_index(cc,' > ',3),' > ',-1) end as a3,
  case when substring_index(cc,' > ',3) = cc then null else substring_index(substring_index(cc,' > ',4),' > ',-1) end as a4,
  case when substring_index(cc,' > ',4) = cc then null else substring_index(substring_index(cc,' > ',5),' > ',-1) end as a5
from d

Answer 2

我终于在 bigquery pull 中而不是在 appmaker 中使用正则表达式提取到了我需要去的地方：

SELECT 
  CompoundColumn,

  REGEXP_EXTRACT(CompoundColumn+">",  r'^(.*?)>') first_number,
  REGEXP_EXTRACT(CompoundColumn+">",  r'^(?:(?:.*?)>){1}(.*?)>') second_number,
  REGEXP_EXTRACT(CompoundColumn+">", r'^(?:(?:.*?)>){2}(.*?)>') third_number,
  REGEXP_EXTRACT(CompoundColumn+">",  r'^(?:(?:.*?)>){3}(.*?)>') fourth_number
FROM
  myTable
WHERE
  CompoundColumn IS NOT NULL

代码的 +">" 部分很难看，但我无法让它匹配没有以方括号结尾的字符串（">?" 破坏了整个东西）所以我把它们都做了以括号结尾。

Answer 3

所需的旧版SQL将是：

SELECT id, 
       compoundColumn,
       FIRST(SPLIT(compoundColumn, " > ")) AS "first",
       NTH(2, SPLIT(compoundColumn, " > ")) AS "second"
FROM table

有关 SPLIT、FIRST 和 NTH 函数的详细信息，请参阅 this BigQuery documentation page。

将单个 sql 列拆分为五个

Split a single sql column into five

sql

google-cloud-sql

legacy-sql