正则表达式递归替换(增加缩进)

Regular expression recursive replace (increasing indentation)

我确信这应该是可能的,我想避免一个函数,因为它是页面上的一次性报告 (Oracle 11g),但我真的很想用换行符格式化以下字符串每个后续替换“-”的缩进以显示层次结构

举个例子:

This is the base level - This is level 2 - And this is the second

所以,我可以开始并用换行符替换“-”并应用 CSS pre 以确保它们都在一行上

但理想情况下,我喜欢这样:

This is the base level
> This is level 2
>> And this is the second

非常感谢

这有点丑陋,但有效(至少,我是如何理解这个问题的)。查看代码中的注释。

SQL> WITH test (col)
  2    -- Sample string
  3       AS (SELECT 'This is the base level - This is level 2 - And this is the second'
  4             FROM DUAL),
  5    -- Split it to rows (minus sign is a delimiter)
  6       temp
  7       AS (    SELECT LEVEL lvl,
  8                      TRIM (REGEXP_SUBSTR (col,
  9                                           '[^-]+',
 10                                           1,
 11                                           LEVEL))
 12                         val
 13                 FROM test
 14           CONNECT BY LEVEL <= REGEXP_COUNT (col, '-') + 1)
 15  -- join it back, using '>' along with CHR(10) as delimiter
 16  SELECT REPLACE (
 17            XMLAGG (XMLELEMENT (e, LPAD ('>', lvl - 1, '>') || val || CHR (10)) ORDER BY
 18                                                                                   lvl).EXTRACT (
 19               '//text()'),
 20            '&gt;',
 21            '>')
 22            result
 23    FROM temp;

RESULT
--------------------------------------------------------------------------------
This is the base level
>This is level 2
>>And this is the second


SQL>

如果你问“为什么是XMLAGG而不是LISTAGG”,那是因为

 16  SELECT LISTAGG (val, LPAD ('>', lvl - 1, '>')) WITHIN GROUP (ORDER BY lvl)
 17    FROM temp;
SELECT LISTAGG (val, LPAD ('>', lvl - 1, '>')) WITHIN GROUP (ORDER BY lvl)
                                *
ERROR at line 16:
ORA-30496: Argument should be a constant.

如果填充在第一个参数内完成,您可以使用 listagg

with rcte (id, value, lvl, result) as (
  select id, value, 1, regexp_substr(value, '(.*?)( - |$)', 1, 1, null, 1)
  from your_table
  union all
  select id, value, lvl + 1, regexp_substr(value, '(.*?)( - |$)', 1, lvl + 1, null, 1)
  from rcte
  where regexp_substr(value, '(.*?)( - |$)', 1, lvl + 1, null, 1) is not null
)
select id,
  listagg(case when lvl > 1 then rpad(chr(10), lvl, '>') || ' ' end || result)
    within group (order by lvl) as result
from rcte
group by id
order by id;
ID | RESULT
-: | :--------------------------------------------------------------------------------
 1 | This is the base level
     > This is level 2
     >> And this is the second           
 2 | Base only                                                                        

但你也说:

I can do a start and replace the " - " with a line break and apply a CSS pre to ensure they're all on one line

所以如果你想要一行,你可以不添加换行符:

with rcte (id, value, lvl, result) as (
  select id, value, 1, regexp_substr(value, '(.*?)( - |$)', 1, 1, null, 1)
  from your_table
  union all
  select id, value, lvl + 1, regexp_substr(value, '(.*?)( - |$)', 1, lvl + 1, null, 1)
  from rcte
  where regexp_substr(value, '(.*?)( - |$)', 1, lvl + 1, null, 1) is not null
)
select id,
  listagg(case when lvl > 1 then rpad(' ', lvl, '>') || ' ' end || result)
    within group (order by lvl) as result
from rcte
group by id
order by id;
ID | RESULT
-: | :-------------------------------------------------------------------------------
 1 | This is the base level > This is level 2 >> And this is the second              
 2 | Base only                                                                       

您还可以在递归 CTE 中添加 >,这可能更简洁:

with rcte (id, value, lvl, result) as (
  select id, value, 1, regexp_substr(value, '(.*?)( - |$)', 1, 1, null, 1)
  from your_table
  union all
  select id, value, lvl + 1,
    rpad('>', lvl, '>') || ' ' || regexp_substr(value, '(.*?)( - |$)', 1, lvl + 1, null, 1)
  from rcte
  where regexp_substr(value, '(.*?)( - |$)', 1, lvl + 1, null, 1) is not null
)
select id,
  listagg(result, ' ') within group (order by lvl) as result
from rcte
group by id
order by id;

db<>fiddle