Item11556: $n newline format token doesn't seem to be expanded consistently in SEARCH formats

Priority:	CurrentState:	AppliesTo:	Component:	WaitingFor:
Low	New	Engine

The docco says Use $n() if followed by alphanumeric character.

$n$formfield(Summary)$n= is detectable, given that the following char is not alphanumeric, its a $.

Similarly, $n$n$n$web is rendered as \n\n$nBlog for some inane reason.

Mind you, It can't be that hard to detect that there's an unexpanded $ in there and work backwards until we get it.

%STARTSECTION{"query"}%
%SEARCH{
web="Blog"
type="query"
"form.name ~ 'BlogPostForm'"
date="P1w/$today"
excludetopic="*Template"
order="formfield(PublishingDate)"
reverse="on"
NNnonoise="on"
format="---+++ [[$web.$topic][$formfield(Title)]] $nby $formfield(Author) $n$formfield(Summary)$n $n$n$n$web"
}%
%ENDSECTION{"query"}%

renders as

I think the right approach is to either make the detection cleverer, by sorting the list of tokens by length, then replacing the longest first, matching greedily, but that has a small chance of tripping up someone that relies on the quirk produced by the code we have.

To work out if that is a reasonable worry, it'll be worth testing on a number of older releases - as we re-jiggered how the processing was done a few times, which may have resulted in different corner cases.

if that approach does not fly, then we should start by replacing all $n tokens that are not followed by and alphanum with $n(). that way at least the $n$ cases will not be broken.

-- SvenDowideit - 24 Feb 2012