Maybe you could try OFFSET and FETCH NEXT on the righthand side of OUTER APPLY

declare
  @nth_from_top       int=2,
  @nth_from_bot ...

The problem in Steve's solution is that is doesn't account for possible CR/LF data in the last field.  In your data - there is a CR\LF after the comma...

Bump for visibility

Here's the code I'm running and the results I'm seeing.  The @string value was copy/pasted.  Thanks btw to Jeff for the splitter code.

drop FUNCTION if exists [dbo].[DelimitedSplitN4K];
go
CREATE...

I did originally post this question under my personal account and this is from my work account.

I did think about using a second replace statement but could not devise...

One way could be to use the same replacement technique as with the line breaks.  Replace the "," and ", and ," with |.  Then replace any remaining double quotes...

In the JSON Standards docs they call the forward slash symbol a 'solidus' and a...

• This reply was modified 1 years, 9 months ago by Steve Collins. Reason: Removed mistake

Yes removing the escape characters would invalidate the JSON...

Is this really SQL Server 2012?  Extended support for SQL Server 2012 ended last year.  Why are the AMOUNT and PRICE of data type FLOAT?  Generally, columns with those names...

The code posted seems to have issues imo.  Can't be certain tho. ...

We're not just a forum, though. This is a whole publishing engine that gets article submissions, runs them through a queue, publishes, scheduled, generates a newsletter, has the daily...

Imo the issue with assigning value(s) to variable(s) and then using variable(s) to resolve flow of control is it can create race conditions.  Then database serialization is attempted to hold...

Volumes are better, especially across time. Getting a consistent set of emails across time is good.

Your posts here (forwarded) did get a meeting set and some movement on upgrading...

It's tl;dr with the data.  The real questions are: 1) what is your data access pattern,  and 2) how are you handling identity and access management?

Maybe you're looking for rows WHERE the SkidStatus and the LEAD SkidStatus are unequal within a common row key partition.  In this case left(SKIDNUMBER, 10) effectively truncates the _NOTAPPLICABLE suffixes