Re: [BUG?] XMLSERIALIZE( ... INDENT) won't work with blank nodes

Hi Tom

On 06.09.24 18:34, Tom Lane wrote:
> I think it'd be quite foolish to assume that every extant and future
> version of libxml2 will share this glitch. Probably should use
> logic more like pg_strip_crlf(), although we can't use that directly.
Makes sense. I Introduced this logic in the end of
xmltotext_with_options() in case it was called with INDENT and DOCUMENT
type xml string.

SELECT xmlserialize(DOCUMENT '<foo><bar>42</bar></foo>' AS text INDENT);
  xmlserialize   
-----------------
 <foo>          +
   <bar>42</bar>+
 </foo>
(1 row)

The regression tests were updated accordingly - see patch v2-0002.
> Would it ever be the case that trailing whitespace would be valid
> data? In a bit of testing, it seems like that could be true in
> CONTENT mode but not DOCUMENT mode.
Yes, in case of CONTENT it is valid data and it will be preserved, as
CONTENT can be pretty much anything.

SELECT xmlserialize(CONTENT E'<foo><bar>42</bar></foo>\n\n\t\t\t' AS
text INDENT);
       xmlserialize       
--------------------------
 <foo>                   +
   <bar>42</bar>         +
 </foo>                  +
                         +
                         
(1 row)

With DOCUMENT it is superfluous and should be removed after indentation.
IIRC there's an xmlSaveToBuffer option called XML_SAVE_WSNONSIG that can
be used to preserve it.

Thanks

Best, Jim