I have a table named data with a column pid & display_data. Pid is the serial number of all rows.

The value of display_data column is as below :

For each row display_Data has values in the format =>

For PID= 1:

The labels are constant for all rows which would become column headers. Values are changing with respect to labels. There are total 400 such rows hence, the query needs to be dynamic.

I want to parse this column value,strip html tags and into a new view split it using sql query(Regexp_Replace maybe) such that: All label values become columns, i.e.

The actual data is much more with many labels and values however this is just sample to get help.

  • its pl/sql. my mistake. removed sql server
    – Velocity
    Commented Jul 17, 2020 at 8:50
  • Do you need a stored procedure/function or plain SQL?
    – user1822
    Commented Jul 17, 2020 at 8:57
  • a sql statement using regexp/pivot would be very helpful. Just that it needs to be dynamic as in example i have given 2 rows, actual data has around 500 rows with more columns. The only thing constant is column labels across all rows and values would be placed accordingly.
    – Velocity
    Commented Jul 17, 2020 at 9:26
  • Yes, there are total 12 labels which would become column headers. I cannot share exact data so gave example of A,B. So we could assume them as A....L .
    – Velocity
    Commented Jul 17, 2020 at 9:51
  • Are the labels ever anything other than A or B? +1 for an interesting tricky question - I'll have a bash - Oracle isn't really my thing, but I could do with practicing my regexes!
    – Vérace
    Commented Jul 17, 2020 at 9:52

1 Answer 1


TL;DR Answer

Convert HTML to XML and parse with XMLTABLE()

Long Answer

HTML and XML are both descendants of SGML.

From the sample data you have posted, it looks like you just need to remove the <br> tags and wrap desc_data in a root element (eg <row>) for the conversion into XML.

Once in XML format, you can parse the data with XMLTABLE() to give you:

  • instance of <span> ( via for ordinality, column name RN )
  • value of @label
  • value of <span>

This is done within the CTE named xml_data in my example.

Assuming value always follows label, you can convert the instance of <span> ( RN ) to have the same value for each pair. (this is also done in xml_data)

From there, you can PIVOT the data into a table that has:

  • your row identifier (RID)
  • value for label (LABEL is a key word so I used LABEL_TXT )
  • value for value (VALUE is a key word so I used VALUE_TXT )

This is done within the CTE labeled key_val in my example.

For the final query, the final columns names MUST BE KNOWN at compile time. Once you know the columns, you can just do another PIVOT for a proper table.


As you can see, HTML is not a good data format to use for parsing data and/or using the data within a computer (eg a database).

Unless desc_data is used for display, I recommend that the data is stored in XML or JSON format. (I'm assuming you can change the source)

Once data is in a parseable format, PL/SQL is no longer needed.

Example Code

with data as (
    select q'[<span class ="label">Project Name</span><br> <span class ="value">Digital</span> <span class ="label">Dept Name</span><br> <span class ="value">Finance</span><span class ="label">Level</span><br> <span class ="value">Level 5</span>]'
        as txt
        ,1 AS RID
    from dual
),xml_data as (
select RID, a.rn - mod(a.rn+1,2) rn,a.label_txt, a.val
from data d, xmltable( '/row/span'
    passing xmltype( '<row>' || replace(d.txt,'<br>','') || '</row>' ) 
        rn for ordinality,
        label_txt varchar2(20)  path '/span/@class',
        val       varchar2(20)  path '/span'
    ) a
), key_val as (
    select rid, label_txt, value_txt
    from xml_data
        pivot (
            min( val )
            for label_txt in ( 'label' as LABEL_TXT, 'value' as VALUE_TXT)
from key_val
    pivot (
        min( VALUE_TXT )
        for LABEL_TXT in ('Project Name' as PNAME, 'Level' as LVL, 'Dept Name' as DNAME )
  • As the real data is bigger, can i use select desc_data from data instead of select q'[<span class ="label">Project Name</span><br> <span class ="value">Digital</span> <span class ="label">Dept Name</span><br> <span class ="value">Finance</span><span class ="label">Level</span><br> <span class ="value">Level 5</span>]' as txt ?
    – Velocity
    Commented Jul 17, 2020 at 11:44
  • The CTE data is used only to simulate the data in a real table. so, yes. Commented Jul 17, 2020 at 13:10
  • I am getting this error: ORA-31011: XML parsing failed ORA-19213: error occurred in XML processing at lines 17 LPX-00118: Warning: undefined entity "nbsp" ORA-06512: at "SYS.XMLTYPE", line 272 ORA-06512: at line 1 31011. 00000 - "XML parsing failed"
    – Velocity
    Commented Jul 19, 2020 at 16:39
  • OP updated with original data
    – Velocity
    Commented Jul 23, 2020 at 8:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.