我认为以行而不是列的形式获取所需数据会方便得多。您可以使用xmltable解析它，只需对原始html稍加修改（删除未关闭的标记，如 <br> . 这就是为什么 <br/> 更好）：

with t as (
  -- your sample data:
  select
    q'[<span class ="label">A</span><br> <span class ="value">A-Class</span> <span class ="label">B</span><br> <span class ="value">B-Class</span>.
  ]' html_data
from dual
)
-- main query:
select xt.*
from t
    ,xmltable(
      'let $labels := /root/span[@class eq "label"]
       let $values := /root/span[@class eq "value"]
       for $label at $i in $labels
          return element label {
             attribute name {$label/text()}, 
             attribute value {$values[$i]/text()}
          }
       '
      passing
      xmltype(
       --- modify your html to make it compatible with xml:
       '<root>'
       || replace(replace(t.html_data,'<br>'),'&nbsp;')
       ||'</root>'
      )
      columns
         n for ordinality,
         label_name path '@name',
         label_value path '@value'
    ) xt;

结果：

N LABEL_NAME                     LABEL_VALUE
---------- ------------------------------ ------------------------------
         1 A                              A-Class
         2 B                              B-Class

你需要用一些模式来分割你的字符串（例如 '/span> <span' )递归地。通过使用 REGEXP_REPLACE() 函数，然后应用旋转：

WITH t(desc_data) AS
(
 SELECT '<span class ="label">A</span><br> <span class ="value">A-Class</span> <span class ="label">B</span><br> <span class ="value">B-Class</span> <span class ="label">C</span><br> <span class ="value">C-Class</span>'
   FROM dual
), t2 AS
(
SELECT SUBSTR(desc_data,1,CASE WHEN INSTR(desc_data,'/span> <span',1,level) > 0
                               THEN INSTR(desc_data,'/span> <span',1,level) + 5 
                               ELSE LENGTH(desc_data)
                           END
             ) AS desc_data2
  FROM t
 CONNECT BY level <= REGEXP_COUNT(desc_data,'/span> <span') + 1
)
SELECT *
  FROM
  (
   SELECT REGEXP_REPLACE(desc_data2,'(.*"label">)(\S+)(</span>.*)','\2') AS label,
          REGEXP_REPLACE(desc_data2,'(.*"value">)(\S+)(</span>.*)','\2') AS value
     FROM t2 )
 PIVOT ( MAX(VALUE) FOR LABEL IN ('A' AS "A", 'B' AS "B", 'C' AS "C") );

A          B          C
-------    -------    -------
A-Class    B-Class    C-Class

演示