patch: perl syntax folding independent of coding style

172 views
Skip to first unread message

Benjamin Fritz

unread,
Sep 26, 2008, 11:32:53 PM9/26/08
to vim_dev, ni...@van-laarhoven.org
I've created a patch against the latest perl.vim syntax file available
at the location given in the syntax file:

http://www.van-laarhoven.org/vim/syntax/perl.vim

This patch corrects syntax folding so that it does not depend on
indentation, brace placement, or other coding style issues. I have
tried to keep functionality as close as possible to the original,
testing on the test file provided on the same website.

Please give it a shot and respond with any comments.

perl.vim.diff

Ben Fritz

unread,
Nov 3, 2008, 3:36:14 PM11/3/08
to vim_dev, ni...@van-laarhoven.org
>  perl.vim.diff
> 4KViewDownload

Two bugs that I've found while using this patch with real perl code:

1) Commented-out lines can end a fold region
2) The closing brace in code like $abc{def} ends fold regions

Bug 1 is fixed by adding a skip group to the perlSubFold and
perlBlockFold definitions: skip="#.*"
Bug 2 is something I will try to fix later unless someone beats me to
it.

I have yet to see any feedback or developments on this other than an
"I'll look at it" from the script maintainer. Has anyone tried this?

Frew

unread,
Nov 5, 2008, 2:18:12 PM11/5/08
to vim_dev
I just found this and patched the file. Maybe I patched it
incorrectly (I don't have patch handy atm) but it seems to fold
anything with { and }, that is, if I do this:

sub foo {
$blah->bar($self->{test});

more..
}

just the first two lines get folded.

Hope that helps some.

-fREW

Benjamin Fritz

unread,
Nov 5, 2008, 3:20:19 PM11/5/08
to vim...@googlegroups.com
On Wed, Nov 5, 2008 at 1:18 PM, Frew <fri...@gmail.com> wrote:
>>
>> Two bugs that I've found while using this patch with real perl code:
>>
>> 1) Commented-out lines can end a fold region
>> 2) The closing brace in code like $abc{def} ends fold regions
>>
>> Bug 1 is fixed by adding a skip group to the perlSubFold and
>> perlBlockFold definitions: skip="#.*"
>> Bug 2 is something I will try to fix later unless someone beats me to
>> it.
>>
>> I have yet to see any feedback or developments on this other than an
>> "I'll look at it" from the script maintainer. Has anyone tried this?
>
> I just found this and patched the file. Maybe I patched it
> incorrectly (I don't have patch handy atm) but it seems to fold
> anything with { and }, that is, if I do this:
>
> sub foo {
> $blah->bar($self->{test});
>
> more..
> }
>
> just the first two lines get folded.
>
> Hope that helps some.
>
> -fREW
>

That does help some, it falls into bug (2) that I listed, which I
suspect is very common.

Unfortunately, I am very much a beginner at perl, and am not very
familiar with the syntax, so it may take me a little while. I suspect
the best way to proceed might be to just make another group that folds
all {...} blocks that aren't the special cases like if, sub, etc. I'll
try to get to it sometime this month, unless someone else fixes it
first of course :-)

Erik Falor

unread,
Nov 5, 2008, 6:48:59 PM11/5/08
to vim...@googlegroups.com

So far, so good.

Not to be uber-nitpicky, but what do you folks think about changing all of the variables that control the script's behavior (ie. perl_include_pod, perl_fold, etc.) from globals to buffer-locals?  

I like to set parameters such as these from from my after/ftplugin/*.vim files so as to keep my .vimrc neat and trim.  Similarly, I think it is better (at least aesthetically) to not clutter the global namespace with variables that only apply to particular buffers.

Food for thought,

Erik


--
Erik Falor
Registered Linux User #445632 http://counter.li.org

Tony Mechelynck

unread,
Nov 5, 2008, 7:08:59 PM11/5/08
to vim...@googlegroups.com

Maybe buffer-local, and if undefined at plugin load, default to the
global variable (if any) of the same name?

Best regards,
Tony.
--
I walked on toward Ploughwright, thinking about feces. What a lot we
had found out about the prehistoric past from the study of fossilized
dung of long-vanished animals. A miraculous thing, really; a recovery
from the past from what was carelessly rejected. And in the Middle
Ages, how concerned people who lived close to the world of nature were
with the feces of animals. And what a variety of names they had for
them: the Crotels of a Hare, the Friants of a Boar, the Spraints of
an Otter, the Werderobe of a Badger, the Waggying of a Fox, the Fumets
of a Deer. Surely there might be some words for the material so near
to the heart of Ozy Froats [an academic studying feces] than shit?
What about the Problems of a President, the Backward Passes of a
Footballer, the Deferrals of a Dean, the Odd Volumes of a Librarian,
the Footnotes of a Ph.D., the Low Grades of a Freshman, the Anxieties
of an Untenured Professor?
-- Robertson Davies, "The Rebel Angels"

Benjamin Fritz

unread,
Nov 7, 2008, 11:57:23 AM11/7/08
to vim...@googlegroups.com, ni...@van-laarhoven.org
On Nov 5, 2:20 pm, "Benjamin Fritz" <fritzophre...@gmail.com> wrote:

> On Wed, Nov 5, 2008 at 1:18 PM, Frew <fri...@gmail.com> wrote:
>
> >> Two bugs that I've found while using this patch with realperlcode:
>
> >> 1) Commented-out lines can end a fold region
> >> 2) The closing brace in code like $abc{def} ends fold regions
>
> >> Bug 1 is fixed by adding a skip group to the perlSubFold and
> >> perlBlockFold definitions: skip="#.*"
> >> Bug 2 is something I will try to fix later unless someone beats me to
> >> it.
>
> >> I have yet to see any feedback or developments on this other than an
> >> "I'll look at it" from the script maintainer. Has anyone tried this?
>
> > I just found this and patched the file. Maybe I patched it
> > incorrectly (I don't have patch handy atm) but it seems to fold
> > anything with { and }, that is, if I do this:
>
> > sub foo {
> > $blah->bar($self->{test});
>
> > more..
> > }
>
> > just the first two lines get folded.
>
> > Hope that helps some.
>
> > -fREW
>
> That does help some, it falls into bug (2) that I listed, which I
> suspect is very common.
>
> Unfortunately, I am very much a beginner atperl, and am not very
> familiar with thesyntax, so it may take me a little while. I suspect

> the best way to proceed might be to just make another group that folds
> all {...} blocks that aren't the special cases like if, sub, etc. I'll
> try to get to it sometime this month, unless someone else fixes it
> first of course :-)

Actually the fix was fairly easy. Most of the code was already there.
I fixed bug 2 as follows:

* Force behavior as if perl_extended_vars was set if perl_fold is set.
* Add an "extend" keyword to quote/match/string/substitution groups

I also fixed another bug where some character was required to occur on
the same line as a BEGIN, END, etc. or sub block after the opening
keyword for the group to match.

See attached patch, which is (so far as I know) bug free (or at least,
doesn't introduce any bugs). I may have missed some
string/quote/match/etc. groups, though...there are quite a few.

perl.vim.patch

Benjamin Fritz

unread,
Nov 7, 2008, 12:53:00 PM11/7/08
to vim...@googlegroups.com
On Fri, Nov 7, 2008 at 10:57 AM, Benjamin Fritz <fritzo...@gmail.com> wrote:
> See attached patch, which is (so far as I know) bug free (or at least,
> doesn't introduce any bugs). I may have missed some
> string/quote/match/etc. groups, though...there are quite a few.
>

One more bugfix, for this test case (note this was also present in the
unpatched file):

if (1)
{
my %arr=('a'=>"hello",'b'=>"goodbye");
print $arr{'a'};
print "\n";
print $arr {'b'};
print "\n";
print ${arr}{'a'};
print "\n";
print ${arr} {'b'};
print "\n";
}

perl.vim.patch
Reply all
Reply to author
Forward
0 new messages