Teradata REGEXP_SPLIT_TO_TABLE 输入参数

Question

我希望能够在 CTE 中定义输入参数，然后在 REGEXP_SPLIT_TO_TABLE 函数中使用它，如下所示：

WITH PARAMS (INPUT_PARAMS) AS
(SELECT
    '?InputParams' AS INPUT_PARAMS
)
SELECT
    TEST_TABLE.SPLIT_PARAMS
FROM
    TABLE (REGEXP_SPLIT_TO_TABLE(1, PARAMS.INPUT_PARAMS, '\|', 'i')
        RETURNS (outkey INTEGER, TOKENNUM INTEGER, SPLIT_PARAMS VARCHAR(8192) CHARACTER SET UNICODE)) AS TEST_TABLE
    CROSS JOIN PARAMS

但是这会产生以下错误：

SELECT Failed. [3706] Syntax error: Joined table is not supported in conjuction with table operators or table function invoked with variable input argument.

把参数放在函数中就可以了，如下：

SELECT
    TEST_TABLE.SPLIT_PARAMS
FROM
    TABLE (REGEXP_SPLIT_TO_TABLE(1, '?InputParams', '\|', 'i') --
        RETURNS (outkey INTEGER, TOKENNUM INTEGER, SPLIT_PARAMS VARCHAR(8192) CHARACTER SET UNICODE)) AS TEST_TABLE

但我想知道是否有办法做一些更像上面第一个例子的事情。

Answer 1

此函数不允许 JOIN 语法，但您可以进行旧式连接：

WITH PARAMS (INPUT_PARAMS) AS
(SELECT
    '?InputParams' AS INPUT_PARAMS
)
SELECT
    TEST_TABLE.SPLIT_PARAMS
FROM
    TABLE (REGEXP_SPLIT_TO_TABLE(1, PARAMS.INPUT_PARAMS, '\|', 'i')
        RETURNS (outkey INTEGER, TOKENNUM INTEGER, SPLIT_PARAMS VARCHAR(8192) CHARACTER SET UNICODE)) AS TEST_TABLE
    ,PARAMS

事实上你根本不需要加入：

WITH PARAMS (INPUT_PARAMS) AS
(SELECT
    '?InputParams' AS INPUT_PARAMS
)
SELECT
    TEST_TABLE.SPLIT_PARAMS
FROM
    TABLE (REGEXP_SPLIT_TO_TABLE(1, PARAMS.INPUT_PARAMS, '\|', 'i')
        RETURNS (outkey INTEGER, TOKENNUM INTEGER, SPLIT_PARAMS VARCHAR(8192) CHARACTER SET UNICODE)) AS TEST_TABLE

并且由于您不使用正则表达式，您也可以切换到 STRTOK_SPLIT_TO_TABLE:

WITH PARAMS (INPUT_PARAMS) AS
(SELECT
    '?InputParams' AS INPUT_PARAMS
)
SELECT
    TEST_TABLE.SPLIT_PARAMS
FROM
    TABLE (StrTok_Split_To_Table(1, PARAMS.INPUT_PARAMS, '\|')
        RETURNS (outkey INTEGER, TOKENNUM INTEGER, SPLIT_PARAMS VARCHAR(8192) CHARACTER SET Unicode)) AS TEST_TABLE

Teradata REGEXP_SPLIT_TO_TABLE 输入参数

Teradata REGEXP_SPLIT_TO_TABLE Input Parameter

function

teradata