Originally from @NorthboundTrain in microsoft/vscode#131322

Issue Type: Bug

Perl's regex operators can use any pair of matching characters as delimiters (see https://perldoc.perl.org/perlop#Regexp-Quote-Like-Operators). When using a bracketing character as the delimiter, e.g. ({[, then the matching closing bracket can be used to close the operator, e.g. m<foo> or s{foo}{bar}. The bug only occurs when using the substitution operator (s///), and only when using bracketing delimiters, and only with a statement modifier (a trailing if, while, etc).

s/foo/bar/; # OK
s/foo/bar/ unless $baz; # OK
s/foo/bar/ if $baz; # OK

s{foo}{bar}; # OK
s{foo}{bar} if $baz; # BROKEN
# lasts until a scalar variable is given; neither an @array nor a %hash works,
# but the $scalar does (but the remainder of the line is still FUBAR)
# next line is OK, though

s(foo)(bar) if $baz; # BROKEN - $clear;
s[foo][bar] if $baz; # BROKEN - $clear;
s<foo><bar> if $baz; # BROKEN - $clear;
s:foo:bar: if $baz; # OK
s#foo#bar# if $baz; # OK

# other statement modifiers
s(foo)(bar) for $baz; # BROKEN differently
s(foo)(bar) foreach $baz; # BROKEN - $clear
s(foo)(bar) unless $baz; # OK
s(foo)(bar) until $baz; # BROKEN differently
s(foo)(bar) when $baz; # BROKEN differently
s(foo)(bar) while $baz; # BROKEN differently

# other regex/quote-like operators are OK:
m/foo/ if $baz; # OK
m{foo} if $baz; # OK
q/foo/ if $baz; # OK
q{foo} if $baz; # OK
qq/foo/ if $baz; # OK
qq{foo} if $baz; # OK
qw/foo/ if $baz; # OK
qw{foo} if $baz; # OK

Example: <foo/bar/baz.*>

Screenshot: http://i.imgur.com/hOpWSd9.png

With the exception of the predefined variables $_ and @_ (as well as %_), it appears that any variables beginning with an underscore _ (a convention used for "private" variables) do not get highlighted as such:

I imagine the issue might have to do with how $_, @_, and %_ appear to be treated specially.

The Perl syntax highlighting breaks on regular expressions that implicitly use the default input variable and contain the '#' character.
The syntax highlighting breaks if the following conditions are met:

• The regex implicitly uses the default $_ variable
• The regex is not the first item in the conditional
• The regex follows an 'and' or 'or' operator

The rest of the line following the # is highlighted as a comment.

if (/^#/ and $something) { moo(); } # Syntax highlighting is good
if ($something && /^#/) { moo(); } # Syntax highlighting is good
if ($something and /^#/) { moo(); } # Syntax highlighting breaks!

The above isn't really meaningful or valid Perl, but it illustrates what I mean. The sigils $, @, and %, when used in reference indirection, are highlighted correctly, but & is not.

Attached patch lets perlcheckmate.pl use the TPERL env variable instead of 'perl'. Defaults to 'perl' if TPERL is not set.

From @schuan01 on December 29, 2017 14:58

Steps to Reproduce:

1. Copy the code after this steps in a new file.
2. See the bad coloring(all red) after "my @linea = split /|/, $LINEA;"

---- EXAMPLE CODE ---

my $LINEA = $_;
	chomp( $LINEA );
	my @LINEA = split /\|/, $LINEA;
	
	$HOJATNO++;

	# creamos el lote
	if ( not ref $LOTE_REF ) {
		#Supongo que 
		my ( $EMISION, $ORDEN, $LOTE) = ( $LINEA[$POS_EMISION], $LINEA[$POS_ORDEN], $LINEA[$POS_LOTE] );
		
		$LOTE_REF = CE::Lote->new(-lote => $LOTE, -ambiente => $AMBIENTE, -cliente => $CLIENTE, -trabajo => $TRABAJO,-emision => $EMISION, -orden => $ORDEN );
	}
	
	#Leo datos que necesito
	my ( $SPOOL, $INTERNO, $HOJA ) = ( $LINEA[$POS_SPOOL], $LINEA[$POS_INTERNO], $LINEA[$POS_HOJA] );
	my ( $IMPRIME, $GENERA, $VISUALIZA ) = ($LINEA[$POS_IMPRIME], $LINEA[$POS_GENERA], $LINEA[$POS_VISUALIZA] );
	my ( $VALOR_CRITERIO1A, $VALOR_CRITERIO1B ) = ( $LINEA[$POS_CEDULA], $LINEA[$POS_TIPORECIBO] );
	my ( $VALOR_CRITERIO2A, $VALOR_CRITERIO2B ) = ( $LINEA[$POS_NROEMPLEADO], $LINEA[$POS_TIPORECIBO] );
	my $VALOR_CRITERIO3A = $LINEA[$POS_GENERACION];

Copied from original issue: microsoft/vscode#40944

Hi,
Is it possible to combine Perls (5) syntax with https://github.com/Blaizer/ModernPerl-sublime?

I'm just switching from Textmate regular and Textmate 2 is creating weird indentation with if/else/elsif conditionals. It has the same behavior in the Java bundle but NOT in the C bundle. The current behavior seems abnormal/highly non-standard to me. Using underscores to represent a single tab below:

Original Textmate Perl as well as Textmate 2 C bundle (correct):
if (blah == blah) {
_print "blah";
} elsif (blah == blah2) {
_print "blah";
} else {
_return 2;
}

Textmate 2 Perl and Textmate 2 Java bundle (incorrect/unusual indentation):
if (blah == blah) {
_print "blah";
_} elsif (blah == blah2) {
__print "blah";
__} else {
___return 2;
__}

Attached an image of the default TM2 indentation:

Observed in Visual Studio Code 1.50.0.

A match operator will not be highlighted as regex if it uses / delimiters without the m prefix, is matching $_ implicitly, and is used in an assignment statement directly after the = (e.g. is not in parentheses).

Example code:

$_ = 'Hello world';
my ($match) = /Hello\s+(\S*)/;

One workaround is to use the m prefix:

In this example (from https://github.com/gisle/tkx/blob/2c96533/lib/Tkx.pm#L338), $) is highlighted as if it were an interpolated variable:

last if $pkg !~ /^Tkx(::|$)/;

Likewise $| would be highlighted in this equivalent statement:

last if $pkg !~ /^Tkx($|::)/;

However $ is highlighted correctly when using m// instead of //:

last if $pkg !~ m/^Tkx(::|$)/;
last if $pkg !~ m/^Tkx($|::)/;

First off, thanks very much for your work. :) Given how this is also used in VSCode and by GitHub for syntax highlighting, it has a lot of value to the Perl community.

Would you be open to adding one (or more?) competent Perl developers to help manage this specific repository? I'm fairly sure I could get some time at my $work scheduled to help get through the backlog as this would allow us to address at least one pain point (indented heredocs) that affects us at @maxmind.

In the case

$a = $b << ($c + $d);

The syntax will interpret this as a heredoc.

The regex on line 1981 of Syntaxes/Perl.plist currently reads as:

<string>((&lt;&lt;) *((?![=\d\$ ])[^;,'"\s)])).\n?`

Where it should be something like

<string>((&lt;&lt;) *((?![=\d\$\( ])[^;,'"\s)])).\n?`

to fix this issue.

From @gilsanx in microsoft/vscode#83321

• VSCode Version: 1.39.2
• OS Version: Fedora 20

Steps to Reproduce:

1. Have a perl function that has an underscore in its name. Example: s_expand()
2. Pass @_ as parameter. s_expand(@_)
3. The syntax is broken until another character breaks the highlight, for instance, another underscore.

sub myExample()
{
  s_expand(@_);
}

sub fixHighlight()
{
  #perl comment_
}

Regular expression breaks syntax highlighting for remainder of the file. The offending line (perl) is here:

if ((($var, $val) = /^\s*([_a-zA-Z0-9]+)\s*=\s*(.*)/) != 2) {

The *'s after each \s seem to be relevant; deleting one fixes the formatting (but deleting two breaks it again).

Steps to Reproduce:

1. Do above
2. See error

Originally from @zaharazod in microsoft/vscode#154706

From gisle/tcl.pm#20:

Affects March 2018 TextMate Perl bundle and recent Atom and VS Code.

If transliteration operators have curly braces as the delimiter, i.e. tr{}{} or y{}, and either the pattern or replacement contains exactly one /, then everything up until the next / is highlighted as a regex. See screenshot:

(Example from https://github.com/gisle/tcl.pm/blob/8935cbb97e5d74210cd4397e6c6ccffa792fc054/Makefile.PL#L269)

One workaround is to use a different delimiter such as square brackets [].

Multiline comments in the #`( comment ) style only have their first line marked as an actual comment, and the rest is incorrectly detected as code. Multine comments using =begin comment and =end comment work better but seem to have some odd behaviour as well (the comment in =end comment is in white while the rest is green with my theme).

Strings in certain places are not detected properly.

Issues as seen in this screenshot:

Atom on the left, VS Code on the right. Atom has the same issues except when using the language-perl6 package (which is being used in the above screenshot). VS Code on the right is only using the syntax files from this repo.

I don't know what langauge-perl6 for Atom does to get it correct, but perhaps someone more familiar with syntax files/grammar can investigate that.

Interestingly, if the space in .split(', ') is removed, the syntax highlight seems to work better, except the string in .split(',') is highlighted in white with my VS Code theme. Regular strings in that theme are highlighted in red.

Additionally, is this repo even maintained anymore? VS Code relies solely on this repo for syntax highlighting and it's been broken for Perl 6 for a long time now.

From @WildYorkies on February 13, 2017 21:22

Steps to Reproduce:

1. Make a file called test.pl6 or test.pm6
2. Type the following in:

sub print-board() # Wrong

sub thing-for-that() # Wrong

3. See that the words print and for are highlighted as if they are keywords.

4. Change it to snake case and notice that it is like a normal variable again:

sub print_board() # Correct

sub thing_for_that() # Correct

Solution:

Update this file with the kebab case addition.

Copied from original issue: microsoft/vscode#20563

From @atif089 on July 14, 2017 11:56

VSCode Version: Code 1.14.0
OS Version: Mac OSX
Configured Language: Perl

The the following code

if ( $product->{url} and $memcached_search ) {
    $product->{url}.=$product->{url}=~m{\?}?";":"?";
    $product->{url}.='s='.$memcached_search.'X'.++$pos;
}

Near ~m{\?}? if you add space between closing } and ? then the syntax looks fine otherwise it looks broken. I'm using the material theme but this appears on any theme.

I think the fix has something to do with the Perl tmLanguage file but I don't know where to find it and what rule I should be looking for :(

Demo : http://i.imgur.com/bjhypxh.gif

Copied from original issue: microsoft/vscode#30699

From @AlexTalker in microsoft/vscode#81309

• VSCode Version: 1.38.1
• OS Version: Windows_NT x64 10.0.17763

Steps to Reproduce:

1. Create a perl module like this:

use v5.10;
use strict;
use warning;

use constant FOO => 42;

my $x = 33;
$x = $x << FOO;

say($x);

2. Notice that after utilization of FOO, color highlight breaks:
   

Does this issue occur when all extensions are disabled?: Yes

The problem seems to be that this regular expression https://github.com/textmate/perl.tmbundle/blob/master/Syntaxes/Perl.plist#L2080 breaks rules of Perl syntax:

The terminating string may be either an identifier (a word), or some quoted text. An unquoted identifier works like double quotes. There may not be a space between the << and the identifier, unless the identifier is explicitly quoted. The terminating string must appear by itself (unquoted and with no surrounding white-space) on the terminating line.

https://perldoc.perl.org/perlop.html#Quote-and-Quote-like-Operators

As it consumes a white-spaces successfully.
I don't know who's to blame here, so I make the issue here and you can if necessary duplicate it to TextMate project.

On huge projects, it is really annoying to break syntax of a large file by simple usage of a constant.

When in a Heredoc there are html tags that are not closed in the same Heredoc the syntax highlight breaks.

Screenshots

Originally reported by @darkliquid at atom/language-perl#7:

In the following example, the portion of the code from q; until the next ; is erroneously highlighted as a quoted string - specifically the 2nd print keyword is highlighted as a string instead of like the other prints.

sub q() {}
my $test = {};
bless $test;
print $test->q;
# this is some code that should not be string highlighted
print "end";

(This TextMate bundle is used by GitHub to highlight Perl code in files and snippets.)

Consider the example below. When viewed in VSCode, the first "my $a" line is colored correctly while "my $b" is incorrectly colored.

# Example showing that CSS heredoc is treated differently
my $foo = <<"END";
Heredoc ending in end.
END

# Colored correctly:
my $a = "a";

my $foo = <<"CSS";
Heredoc ending in end.
CSS

# Colored incorrectly:
my $b = "b";

From @chriscapaci on November 9, 2017 17:18

• VSCode Version: 1.17.2
• OS Version: Windows 10

Steps to Reproduce:

1. Download attached file and remove .txt extension so extension is .pl (Perl code)
2. Open file syntax_highlighting_issues.pl in VSCode
3. Observe syntax highlighting (comments explaining what's going on)

NOTE: All usages of regex's in the code are valid as the project I work on contains all of them. Some may be uncommon, but they are valid and work, so highlighting should work with them.

syntax_highliting_issues.pl.txt

Reproduces without extensions: Yes

Copied from original issue: microsoft/vscode#37966

Originally from @mebouck in microsoft/vscode#127467:

Issue Type: Bug

A comment after a subroutine definition isn't detected, for example:

sub getPi  # calculate and return the value of pi

VS Code version: Code 1.57.1 (507ce72a4466fbb27b715c3722558bb15afa9f48, 2021-06-17T13:28:07.755Z)
OS version: Windows_NT x64 10.0.19042
Restricted Mode: No

vsliv368cf:30146710
vsreu685:30147344
python383cf:30185419
pythonvspyt602:30300191
vspor879:30202332
vspor708:30202333
vspor363:30204092
vstes627:30244334
pythonvspyt639:30300192
pythontb:30283811
vspre833:30321513
pythonptprofiler:30281270
vsdfh931:30280409
vshan820:30294714
pythondataviewer:30285071
vscus158:30321503
pythonvsuse255cf:30323309
vscorehov:30309549
vscod805cf:30301675
binariesv615:30325510

atom/language-perl@fb6b7c6

From @pcherkasova on June 6, 2016 21:53

• VSCode Version: 1.1.1
• OS Version: mac os

Steps to Reproduce:

Copied from original issue: microsoft/vscode#7270

From @leonid-s-usov in microsoft/vscode#113985

Issue Type: Bug

Thils line of PERL code

my @list = split /"/, $string;

will color all following lines as if withing a string because of the unbalanced quote.

This, however, is a valid PERL code that will split the $string by quotation mark symbols.

VS Code version: Code 1.52.1 (ea3859d4ba2f3e577a159bc91e3074c5d85c0523, 2020-12-16T16:30:02.420Z)
OS version: Darwin x64 19.6.0

From @NorthboundTrain on February 22, 2018 14:13

Issue Type

Bug

Description

1. open new editor window
2. switch to Perl mode
3. paste in the following syntax:

$foo =~ m/ff/i;
print "$foo is the name of the game\n";

$foo =~ m/ff/m;
print "$foo is the name of the game\n";

$foo =~ m{ff}i;
print "$foo is the name of the game\n";

$foo =~ m{ff}m;
print "$foo is the name of the game\n";

4. the code is formatted incorrectly; the trailing 'm' modifier in the regex match is not recognized and is therefor not end the regex formatting.

VS Code Info

VS Code version: Code 1.20.1 (f88bbf9137d24d36d968ea6b2911786bfe103002, 2018-02-13T15:23:28.607Z)
OS version: Darwin x64 16.7.0

Copied from original issue: microsoft/vscode#44204

From @skaurus on October 20, 2017 14:34

I'm talking about this: http://perldoc.perl.org/perlop.html#Indented-Here-docs

Steps to Reproduce:

1. write perl code like that:

if ($cond) {
    my $string = <<~EOF;
        indented
        text
        will be
        indented
        just by 4 spaces
        and totally not 8
    EOF
    my $res = $self->method;
}
...

2. code after closing EOF will be colored as if it's still inside the string.

Currently it works only if opening here-doc marker does not have a tilde; AND closing marker is not indented (at all).

Reproduces without extensions: Yes

Copied from original issue: microsoft/vscode#36640

The syntax highlighting for a regular expression doesn't always end correctly - causing the grammar parser in text mate to think everything after the particular start of the regex to be interpreted and colored, etc as part of the regular expression.

I only see this then using a multi-line regex with a the preceding 'm':

$var =~ m/regex/m

There may be other cases.

Sample where it breaks:

if (/ Check \s (Mate)/x)

@infininight atom/language-perl#32

as documented in https://perldoc.perl.org/perlop#EOF, it is possible to have multiple adjacent here-documents:

   print <<"foo", <<"bar"; # you can stack them
I said foo.
foo
I said bar.
bar

this isn't highlighted properly - the second here-doc isn't treated as such, which of course leads to all kinds of "interesting" followup effects when it contains quotes, etc.

go to https://github.com/git/git/blob/v2.39.2/git-send-email.perl#L835 for a real-life demo.

From @TalhaObaid on January 23, 2018 3:30

• VSCode Version: Code 1.19.2 (490ef761b76b3f3b3832eff7a588aac891e5fe80, 2018-01-10T15:49:04.682Z)
• OS Version: Darwin x64 16.7.0
• Extensions:

Steps to Reproduce:

1. Create/open a file with Perl code in editor, having .pm/.pl extension.
2. Write //= (); on a new line in the middle of the code.
3. The rest of the code below will turn leaving the identifiers only

Reproduces without extensions: Yes

Copied from original issue: microsoft/vscode#42008

$foo->x->y; # OK

$foo->m()->y; # OK

$foo->m->y; # Not OK

$foo->s->y; # Not OK
$foo->s()->y; # Not OK

Just s or m will break syntax highlight. Other chars are fine.