Perl language syntax highlighting troubles (bug or limitation ?)

[Gilles Maisonneuve]

J’ai tellement l’habitude d’utiliser $1, $2, …, qui, eux, ne fonctionnent pas dans un simple ‘match’ mais uniquement dans un ‘substitute’, que je ne connaissais pas cette façon de répéter les ‘patterns’ de ‘matching’. J’ai appris quelque chose.
Dont acte.

[Ekopalypse]

@Gilles-Maisonneuve

Lunch break :-)

First, I’m sorry not to telling you that the single quote has to be escaped as it was
used to denote a python string - good, you figured it already out.

Let me break down the parts of that python code

regexes = OrderedDict()
regexes[(3, (255,0,0))] = (r'(?s)(\s*(<<)\s*("{0,1}.+"{0,1})\s*;.*?\3)', [0])

regexes is variable, containing an OrderedDict class instance.
OrderedDict is more or less the same as a perl associative array or hash

regexes[] is the python way to access a key in that hash, like in perl regexes{}
regexes[()] the round bracket denotes a python tuple, in perl a list I guess (immutable)
the python tuple contains the items 3 and (255,0,0) <- this is again a tuple
The number 3 is here to create an unique key - has nothing to do with the regex itself.
So, regexes[(3, (255,0,0))] means, get me the value for key (3, (255,0,0)) from dict(hash) regexes

The value is (r’(?s)(\s*(<<)\s*(“{0,1}.+”{0,1})\s*;.*?\3)‘, [0])
Again, a python tuple containing the items r’…’ (raw string) and a list [] (in perl an array = mutable)
Everything within the raw string is the regex to be searched for and the list contains the information
which match group should be used for coloring
[0] is always the overall match of the complete regex and [1] would be the result from group 1,
[2] from group 2 and [1,2] from group 1 and group 2

So, in terms of regular expressions only the value part of the regexes hash/dict is of interest.
For searching only the raw string and for coloring which part was defined in the list [].

Does this makes sense to you?

The reason why this regex

regexes[(4, (0,0,224))] = (r'(?s-i)((<<)\h+([\'"])(\w+?)\2\h*;.*?\3)', [1,3])

doesn’t do what you want is that you use 4 groups now whereas @guy038 has
removed the outer matching group brackets.

(?s-i)(<<)(['"]?)(\w+?)\2\h*;.*?\3

In order to make it work either use

regexes[(4, (0,0,224))] = (r'(?s-i)(<<)\h+([\'"])(\w+?)\2\h*;.*?\3', [1,3])
or
regexes[(4, (0,0,224))] = (r'(?s-i)((<<)\h+([\'"])(\w+?)\3\h*;.*?\4)', [1,3])

[Alan Kilborn]

No idea what the “chcp 1250…” posting was supposed to be saying to me. :)

This thread gets my vote for the biggest jumbled mess in the history of the community. :)

[Meta Chuh]

maybe @Ekopalypse will write a resuming manual, once this is over … i refuse :)

[Ekopalypse]

You mean a short manager summary I guess :-D

[Meta Chuh]

if a short manager summary is, in your eyes, a fully featured guide, covering all eventualities, based on all caveats of the whole topic … then yes 😉

[Ekopalypse]

LOL - back to business

[Gilles Maisonneuve]

@Ekopalypse
Replying at the message “Lunch break…” with all the explainations.

1. THANK YOU ! I start enjoying Python since I read you.
   Well, I’ll never be a disciple, because not fan of OO and have difficulty to accept a language where the tabulations and spaces define the code syntax… reminds me too much of my youth with the punch cards and the punched paper roll (and yes, I’m that old), but it’s kind of fun to read when one understand it better.
   Your analogies with Perl made it very comprehensible, very kind of you.

2. ** Y Y Y Y E E E E S S S S ! ! ! ! ! ! **
   It works. I just changed a ‘+’ into a ‘*’ after the first ‘\h’ to allow for no horizontal space between the preceding keyword / Perl-separator and the here doc starter (‘<<’). So my (yours with my ridiculous pinch of salt) regexp is now:

   regexes[(4, (0,0,224))] = (r’(?s-i)((<<)\h*([‘"])(\w+?)\3\h*;.*?\4)’, [1,3])
   ^^

Thank to you and your patience (and for the readers, pissed at my garbage, this includes you Alan, and BTW ignore my perl joke, I was upset by the tone you used and perhaps, even further by the fact that you were right [chcp needed to get the accents on the French vowels on a Windows Perl console]).

I was without-a-clue and you saved my day. And now…

Already several adventures have begun to take shape which can be solved by no-one else. Right, Eko ?
Right you are, Meta Chuh.
And so, without further ado… …I hereby declare this case… …closed.

{to sum up the solution provided by Ekopalypse in this thread of discussion}:

regexes[(1, (255,0,128))] = (r'\bq[rwqx]{0,1}\b([^\h]).*?\1|(\bq[rwqx]{0,1}\b\h+(\w).*?\3)', [0])
regexes[(2, (255,0,128))] = (r'\bq[rwqx]{0,1}\b\h*(\(.+?\)|\[.+?\]|\{.+?\})', [0])
regexes[(4, (0,112,112))] = (r'(?s-i)((<<)\h*([\'"])(\w+?)\3\h*;.*?\4)', [1,3])

Allow you to colorize your Perl ‘q*’ keywords and args with color RGB 255,0,128 and your here-docs with color RGB 0,112,112 in Notepad-plus-plus, using “Python Script” plugin. By so, you correct the coloring limitation of Scintilla Library for those keywords in Perl.

Hope this summary will be satisfying enough for “Meta Chuh al.”.

Have nice week (end of) all of you.

Gilles

[Alan Kilborn]

@Gilles-Maisonneuve said:

pissed at my garbage, this includes you Alan

Oh, not at all…at least after 98 messages a positive outcome!

[Ekopalypse]

@Gilles-Maisonneuve

Thank you for your kind words.
I have enjoyed developing the script with you and the
discussion afterwards was also helpful and intersting
as it pointed out that the script comments aren’t 100% bulletproof.
And please don’t hesitate to post here if you find something that
doesn’t work the way you expected it - just let us know, in the end
we can all benefit from it.

Regarding the off topic comments, these were not meant to be insulting.
Sometimes there is a sound in it that is only understood if you read through
several other answers - they are mostly nice but sometimes they do provoke
but they are still meant to be nice or at least helpful.
The one with the summaries, for example, was from another thread where
I caused a confusion, because of my recklessness, that was only cleared up
a few posts later and then I gave a “manager summary” to among other things,
to make life easier for future readers. Seems to become a running gag now :-)

I agree with you, Python is in the beginning odd, especially
when coming from a different language.
It took some time getting used to it and I had my difficulties too,
but now I find the language super - especially Python3, which is
by the way not supported by the PythonScript plugin, :-(
has syntax constructs and language extensions that I really like.

So, have a nice weekend too.

[guy038]

Hello @gilles-maisonneuve, @eko-palypse, @meta-chuh, @alan-kilborn, @peterjones and All,

Ah ! So I’m going to do the 101th post ;-)) Don’t worry, I won’t be [too] long ! I will :

• Explain why my previous regexes did not work ( almost obvious )

• Give you a new version of all the regexes, used in the EkoPalypse script, which :

  • Matches the case qq|qr|qw|qx|q with the < and > delimiters

  • Matches the case of here-docs, containing an escaped delimiter ( \' or \" ), inside the starting and ending blocks ( legal syntax )

For information, refer to :

https://perldoc.perl.org/perlop.html

And particularly :

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

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

---

So, Gilles, in my previous regexes, I used the single quote symbol as it ! At that moment I just tested the regular expressions, without using the Python script :-((

Once I decided to use the Eko’s script, I quickly understood that the single quote symbol is, first, interpreted by the Python engine. So, when I changed any ' single quote with the syntax \x27, in my previous regexes, everything went OK ;-))

---

Then, I decided to test the m Perl instruction, first, with all possible delimiters ( See my .pl test file, below )

For instance :

m bPATTERNb
m ZPATTERNZ
m 0PATTERN0
m _PATTERN_
m(PATTERN)
m<PATTERN>
m [PATTERN]
m {PATTERN}
m!PATTERN!
m"PATTERN"
m/PATTERN/
m $PATTERN$
m %PATTERN%
m &PATTERN&
....
....

On the same way, I tried all syntaxes of the PERL instruction qrDelimiterPATTERNDelimiter So, in regex #1, relative to the q PERL instructions, I enumerated all possible delimiters, different from a word char and from the four sets () [] {} <>, in the character class [!"#$%&\x27*+,./:;=?@``|~\\^-]

Note that in that regex #1, I used the special syntax (?|........|.......|... ....|....), which forces the renumbering of the groups, located inside the group, for each alternative ! ( See an example, at the end of that post )

Then, I tried to enumerate all the variations of the here-docs syntax, including special cases as, for instance

$x=<< "TE\"XT";
Plain text here
TE"XT

And I succeeded to manage this case in my new regexes #3 and #4 ;-))

However, note that the highlighting of any here-document is effective ONLY IF the ending text is visible, in the current editor window !

---

So, here are my new regexes :

# Color every instruction word qq|qr|qw|qx|q with PERL Style 5 ( r = 0 v = 0 b = 255 => Blue )

regexes[(1, 5)] = (r'(?s-i)\bq[qrwx]?(?|\h*([!"#$%&\x27*+,./:;=?@`|~\\^-])|\h+(\w)).*?\1', [0])
regexes[(2, 5)] = (r'(?s-i)\bq[qrwx]?\h*(\(.+?\)|\[.+?\]|\{.+?\}|<.+?>)', [0])

# Color every here-document with the USER color r = 255 g = 0 b = 255 ( => Magenta )

regexes[(3, (255,0,255))] = (r'(?s-i)(<<)([\x27"]?)(?|(\w+)\\([\x27"]\w+)|(\w+)())\2\h*;.*?\3\4', [1])
regexes[(4, (255,0,255))] = (r'(?s-i)(<<)\h+(\x27|")(?|(\w+)\\([\x27"]\w+)|(\w+)())\2\h*;.*?\3\4', [1,3,4])

# If, on addition, you want to highlight the END of here-docs :

#regexes[(3, (255,0,255))] = (r'(?s-i)(<<)([\x27"]?)(?|(\w+)\\([\x27"]\w+)|(\w+)())\2\h*;.*?(\3)(\4)', [1,5,6])
#regexes[(4, (255,0,255))] = (r'(?s-i)(<<)\h+(\x27|")(?|(\w+)\\([\x27"]\w+)|(\w+)())\2\h*;.*?(\3)(\4)', [1,3,4,5,6])

If you want to know how these regexes work, I could give you some hints, next time. Just too lazy to do it, right now ;-))

Note also, that I added, in comments, regexes #3 and #4 if you want, also, highlight the end of here-docs, by placing the back-references \3 and \4, inside parentheses => Two new groups 5 and 6

Remark : Do not delete the empty group () in regexes #3 and #4 : it represents an empty group 4, re-used by the back-reference \4

---

And, of course, here is, below, the Test_Gilles.pl file, used to test these 4 new regexes :

#--------------------------------------------------------------------------------------

# Various examples of INSTRUCTION WORDS q, qq, qr, qw, qx, highlighted with the PYTHON script

q/ok/error    q(ok);
qrw/ok/error    q(ok);
qq/ok/error;   qq{ok};
qr/ok/error;   qr(ok);
qw/ok/error;   qw[ok];
qq/ok/error;   qx(ok);

q        xokxerror      q (ok);
qq     hokherror;     qq {ok};
qr                      rokerror;    qr  (ok);
qw aokaerror;   qw [ok];
qx    zokzerror;   qx (ok);

#--------------------------------------------------------------------------------------

my var1 = q xfoobarx;
my var2 = q getservbyname g;
my var3 = q getservbyname getservbyent;
my var4 = qx{ verify > NUL: };
my var5 = qr/$singer.*grand chanteur/;

#--------------------------------------------------------------------------------------

$bar = q(\n);              # or   $bar = '\n'

$foo = qq(\n);             # or   $bar = "\n" ( Interpolation )

$abc = qx(echo .);         # or   $abc = `echo .`

$perl_info  = qx(ps $$);   # That's Perl's $$
$shell_info = qx'ps $$';   # That's the new shell's $$

use POSIX qw( setlocale localeconv )
@EXPORT = qw( foo bar baz );

qr/PATTERN/msixpodualn     # Interpolation occurs unless delimiter is a SINGLE quote '

#--------------------------------------------------------------------------------------

$r = qr//;
$rex = qr/my.STRING/is;
$re = qr/$pattern/;
qr/$_/i

next if qr#^/usr/spool/uucp# ;

#--------------------------------------------------------------------------------------

 my $sentence_rx = qr&
   (?: (?<=^) | (?<=\s) )  # after start-of-string or # whitespace
   \p{Lu}                  # capital letter
   .*?                     # a bunch of anything
   [.?!]                   # followed by a sentence ender
   (?= $ | \s )            # in front of end-of-string or whitespace
  &sx;

#--------------------------------------------------------------------------------------

The cases where WHITESPACE must be used are when the QUOTING character is a WORD character :

q XfooX                    # Means the string 'foo'
qx XfooX                   # Means the string 'foo', too

qXfooX                     # WRONG !
qxXfooX                    # WRONG !

#--------------------------------------------------------------------------------------

# There can (and in some cases, must) be WHITESPACE between the operator and
# the quoting characters, EXCEPT when # is being used as the quoting character :
#
# q#foo# is parsed as the string foo , while q #foo# is the operator q followed by a
# comment. So, its argument will be taken from the next line.

q#foo#

q #foo#   #  => ONLY q SHOULD be colored ( Exception )

#--------------------------------------------------------------------------------------

# Instruction Word m, already CORRECTLY highlighted, by DEFAULT, by Scintilla :

mBPATTERNB   # KO ( normal )
mZPATTERNZ   # KO ( normal )
mbPATTERNb   # KO ( normal )
mzPATTERNz   # KO ( normal )
m0PATTERN0   # KO ( normal )
m9PATTERN9   # KO ( normal )
m_PATTERN_   # KO ( normal )

m BPATTERNB
m ZPATTERNZ
m zPATTERNz
m 0PATTERN0
m 9PATTERN9
m _PATTERN_

m(PATTERN)
m<PATTERN>
m[PATTERN]
m{PATTERN}

m (PATTERN)
m <PATTERN>
m [PATTERN]
m {PATTERN}

m!PATTERN!
m"PATTERN"
m#PATTERN#
m$PATTERN$
m%PATTERN%
m&PATTERN&
m'PATTERN'
m*PATTERN*
m+PATTERN+
m,PATTERN,
m-PATTERN-
m.PATTERN.
m/PATTERN/
m:PATTERN:
m;PATTERN;
m=PATTERN=
m?PATTERN?
m@PATTERN@
m\PATTERN\
m^PATTERN^
m`PATTERN`
m|PATTERN|
m~PATTERN~

m !PATTERN!
m "PATTERN"
m #PATTERN#
m $PATTERN$
m %PATTERN%
m &PATTERN&
m 'PATTERN'
m *PATTERN*
m +PATTERN+
m ,PATTERN,
m -PATTERN-
m .PATTERN.
m /PATTERN/
m :PATTERN:
m ;PATTERN;
m =PATTERN=
m ?PATTERN?
m @PATTERN@
m \PATTERN\
m ^PATTERN^
m `PATTERN`
m |PATTERN|
m ~PATTERN~

#--------------------------------------------------------------------------------------

# Instruction Word qr, CORRECTLY highlighted, with the PYTHON script :

qrBPATTERNB    # KO ( normal )
qrZPATTERNZ    # KO ( normal )
qrbPATTERNb    # KO ( normal )
qrzPATTERNz    # KO ( normal )
qr0PATTERN0    # KO ( normal )
qr9PATTERN9    # KO ( normal )
qr_PATTERN_    # KO ( normal )

qr BPATTERNB
qr ZPATTERNZ
qr bPATTERNb
qr zPATTERNz
qr 0PATTERN0
qr 9PATTERN9
qr _PATTERN_

qr(PATTERN)
qr<PATTERN>
qr[PATTERN]
qr{PATTERN}

qr (PATTERN)
qr <PATTERN>
qr [PATTERN]
qr {PATTERN}

qr!PATTERN!
qr"PATTERN"
qr#PATTERN#
qr$PATTERN$
qr%PATTERN%
qr&PATTERN&
qr'PATTERN'
qr*PATTERN*
qr+PATTERN+
qr,PATTERN,
qr-PATTERN-
qr.PATTERN.
qr/PATTERN/
qr:PATTERN:
qr;PATTERN;
qr=PATTERN=
qr?PATTERN?
qr@PATTERN@
qr\PATTERN\
qr^PATTERN^
qr`PATTERN`
qr|PATTERN|
qr~PATTERN~

qr !PATTERN!
qr "PATTERN"
qr #PATTERN#
qr $PATTERN$
qr %PATTERN%
qr &PATTERN&
qr 'PATTERN'
qr *PATTERN*
qr +PATTERN+
qr ,PATTERN,
qr -PATTERN-
qr .PATTERN.
qr /PATTERN/
qr :PATTERN:
qr ;PATTERN;
qr =PATTERN=
qr ?PATTERN?
qr @PATTERN@
qr \PATTERN\
qr ^PATTERN^
qr `PATTERN`
qr |PATTERN|
qr ~PATTERN~

#--------------------------------------------------------------------------------------

# For completeness, << as shift operator

$b = (1 << 5);

#  Here-documents, CORRECTLY highlighted, with the PYTHON script :

$x=<<TEXT;
Plain text here
TEXT

$x=<<'TEXT';
Plain text here
TEXT

$x=<<"TEXT";
Plain text here
TEXT

$x=<<'TE"XT';
Plain text here
TE"XT

$x=<<"TE'XT";
Plain text here
TE'XT

# Here-documents, with a SPACE char, before the SEMI-COLON

$x=<<TEXT ;
Plain text here
TEXT

$x=<<'TEXT' ;
Plain text here
TEXT

$x=<<"TEXT" ;
Plain text here
TEXT

$x=<<'TE"XT' ;
Plain text here
TE"XT

$x=<<"TE'XT" ;
Plain text here
TE'XT


#  Here-documents, with the ESCAPED delimiter in the TEXT, CORRECTLY highlighted, too !

$x=<<'TE\'XT';
Plain text here
TE'XT

$x=<<"TE\"XT";
Plain text here
TE"XT

# The SAME + a SPACE char, before the SEMI-COLON

$x=<<'TE\'XT' ;
Plain text here
TE'XT

$x=<<"TE\"XT" ;
Plain text here
TE"XT

# Here-documents with SPACE highlights as operator, in Notepad++

$x=<< 'TEXT';
Plain text here
TEXT

$x=<< "TEXT";
Plain text here
TEXT

$x=<< 'TE"XT';
Plain text here
TE"XT

$x=<< "TE'XT";
Plain text here
TE'XT

# The SAME + a SPACE char, before the SEMI-COLON

$x=<< 'TEXT' ;
Plain text here
TEXT

$x=<< "TEXT" ;
Plain text here
TEXT

$x=<< 'TE"XT' ;
Plain text here
TE"XT

$x=<< "TE'XT" ;
Plain text here
TE'XT

#  Here-docs with SPACE highlights as operator, and the ESCAPED delimiter in TEXT, CORRECTLY highlighted !

$x=<< 'TE\'XT';
Plain text here
TE'XT

$x=<< "TE\"XT";
Plain text here
TE"XT

# The SAME + a SPACE char, before the SEMI-COLON

$x=<< 'TE\'XT' ;
Plain text here
TE'XT

$x=<< "TE\"XT" ;
Plain text here
TE"XT


#-----  Note that MULTIPLE Here-docs are NOT managed, yet -:(( -----

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

myfunc(<< "THIS", 23, <<'THAT');
Here's a line
or two.
THIS
and here's another.
THAT

#------------------ END ----------------------------

Cheers,

guy038

P.S :

Here a simple example of the (?|......|.......|.......)

Let’s suppose that you want to match these two expressions :

foo12345fooABCDE
bar12345barABCDE

A classic syntax should be (foo)12345\1ABCDE|(bar)12345\2ABCDE, where group 1 = foo and group 2 = bar

But you can use this second shorter regex (?|(foo)|(bar))12345\1ABCDE, where group 1 represents, either, foo or bar, depending of the part of the alternative has matched

For a more complete example, refer to :

https://www.boost.org/doc/libs/1_55_0/libs/regex/doc/html/boost_regex/syntax/perl_syntax.html#boost_regex.syntax.perl_syntax.branch_reset

This (?|PATTERN) syntax is, commonly, called a branch-reset !

[Gilles Maisonneuve]

@guy038
Thanks.
I picked your 4th regex with END of here doc coloring included.
Just changed once again the \h+ in \h* to allow for no space between previous keyword or variable and the ‘<<’ operator.

I think I’m going to use the regexp and python scripts provided in this thread to enhanced other syntax hilighting (I think about CMD and Yori for example. It’ll make me progress in Python, now that I have tested and tasted it.

Thanks to all again.

Gilles

[Gilles Maisonneuve]

@Ekopalypse

Hello Eko,

I met the following shortcoming with the q* coloring:
use subs qw(divide_by_hand resultline is_fine_resultline resultline_len fine_resultline_len);
works fine, but if I want to split my qw() in 2 lines (because is becoming to be too long on the right side of screen), then I do:

use subs qw(divide_by_hand resultline is_fine_resultline
            resultline_len fine_resultline_len);

and I loose the coloring.

Could it be possible with python to include CRLF/NL (\n) in the pattern so that I can fold my qw (and even my other q*) statements ?

[Ekopalypse]

@Gilles-Maisonneuve

Hello Gilles,

how about using the single-line modifier (?s) in front of the regex?
r'(?s)\bq[rwqx]{0,1}\b\h*(\(.+?\)|\[.+?\]|\{.+?\})'

This should do the job, I assume.

Concerning the idea of using the script for multiple languages at the same time.
Currently the issue is that a second script would overwrite the variables
like regexes, BUILTIN_LEXER, EnhanceBuiltinLexer etc… which would break
the logic of the first script.
My first thought would be ensuring namespace integrity, which means the script
needs to get modified in a way that no code gets be executed while importing the script. Which then would result in something like
import EnhancePerlLexer to import the script and an additional
EnhancePerlLexer.start() to activate it. Other scripts would then be configured like

import EnhanceCmdLexer and EnhanceCmdLexer.start() etc…

Another way would be to make a base class and overwrite it … hmm… need to think about it. Will try to find a way which is reasonable and easy to adapt.
Will keep you informed.

Have a nice Sunday and greetings
Eren

[Gilles Maisonneuve]

@Ekopalypse

Hello Eren,

Worked like a charm (the (?s) syntax), thank you.

About the multiple language script… well I did not get your point exactly beside the fact that it is a lot more tricky than I thought. So, lazy as I am to learn Python, I’ll wait for you to find a solution to put in your EnhancePerlLexer.py (which then would become EnhanceAnyLexer.py).

Whenever you have time…

Thanks for all.

Gilles

[Gilles Maisonneuve]

Changed the regexp from:
regexes[(1, (128,0,128))] = (r'(?s)\bq[rwqx]{0,1}\b([^\h]).*?\1|(\bq[rwqx]{0,1}\b\h+(\w).*?\3)', [0])
to:
..........................................................vv
regexes[(1, (128,0,128))] = (r'(?s)\bq[rwqx]{0,1}\b([^\h])\b.*?\1|(\bq[rwqx]{0,1}\b\h+(\w).*?\3)', [0])

so that

qw( my words here
       are colored but not the remaing text after
       the closing parenthese, until it finds another one...);

Without the word boundary when I used "qw(" instead of "qw (" I got the coloring extending past the closing ) (respectively {...}, [..], etc.)

I know that this regex does not fit the full Perl syntax capability (any char separator like in @array = qw xfoo bar quux; -> @array = ('foo','bar','quu');) but I only use common separators like /{[()]}/. I leave it to someone else to figure this out…

[guy038]

Hello, @gilles-maisonneuve,

If you decide to use \h*, instead of \h+, in regex #4, then, some configurations can be found, either, with the regexes #3 and #4 ! So, to be rigorous, you could use, other regexes, which define two disjoint sets :

• Here-doc documents, with TEXT, right after the << operator ( regex #3 )

• Here-doc documents, with TEXT, after the << operator and possible space chars and a mandatory delimiter ' or ' ( regex #4 )

So, in the Python script :

# Color every "here-document" with the USER color r = 255 g = 0 b = 255 ( => Magenta )
regexes[(3, (255,0,255))] = (r'(?s-i)(<<)(\w+)\h*;.*?\2', [1])
regexes[(4, (255,0,255))] = (r'(?s-i)(<<)\h*(\x27|")(?|(\w+)\\?([\x27"]\w+)|(\w+)())\2\h*;.*?\3\4', [1,3,4])

Tested, with success, against the Perl file “Test_2_Gilles.pl”, recapitulating all cases, below :

#----- NO SPACE nor " nor ', AFTER << ----- Colored with Regex #3 -----

$x=<<TEXT;
Plain text here
TEXT

$x=<<TEXT;
Plain text here
TEXT

#----- DELIMITER " or ', AFTER << ----- Colored with Regex #4 -----

$x=<<"TEXT";
Plain text here
TEXT

$x=<<'TE"XT' ;
Plain text here
TE"XT

$x=<<"TE'XT" ;
Plain text here
TE'XT

$x=<<'TE\'XT';
Plain text here
TE'XT

$x=<<"TE\"XT";
Plain text here
TE"XT

#----- SPACE(S) + DELIMITER " or ', AFTER << ----- Colored with Regex #4 -----

$x=<< 'TEXT';
Plain text here
TEXT

$x=<< "TEXT";
Plain text here
TEXT

$x=<< 'TE"XT';
Plain text here
TE"XT

$x=<< "TE'XT";
Plain text here
TE'XT

$x=<< 'TE\'XT';
Plain text here
TE'XT

$x=<< "TE\"XT";
Plain text here
TE"XT

#------------------------- END --------------------------

Cheers,

guy038

[Ekopalypse]

I’m testing the EnhanceAnyBuiltinLexer right now and I noticed a situation where I don’t know exactly how to solve it better.
The issue is the following.
Assume that the text, which should get colored, are multiple lines. When the start or the end of that text is not within the visible area then the whole text is not colored.
I assume that is what @guy038 was mentioning.
:-(
To avoid this I use a sliding window approach (term borrowed from networks :-), which means as long as the first visible line is not greater than 15 the algorithm starts from line 0 and afterwards it is current line minus 15. Similar the end line gets added 15 unless the maximum lines would be reached.

Not nice, but works as long as there is no text, which should be colored, contains more than 15 lines.

Any ideas how I could solve it differently?

Here the current code - maybe this makes it easier to understand what I’m trying to solve.

        start_line = editor.docLineFromVisible(editor.getFirstVisibleLine())
        end_line = editor.docLineFromVisible(start_line + editor.linesOnScreen())

        start_line -= 15 if start_line > 15 else 0

        max_line = editor.getLineCount()
        if editor.getWrapMode():
            end_line = sum([editor.wrapCount(x) for x in range(end_line)])
        end_line += 15 if max_line - 15 > end_line else max_line

        start_position = editor.positionFromLine(start_line)
        end_position = editor.getLineEndPosition(end_line)

        editor.setIndicatorCurrent(self.INDICATOR_ID)
        editor.indicatorClearRange(0, editor.getTextLength())
        for color, regex in self.regexes.items():
            editor.research(regex[0],
                            lambda match: self.paint_it(color[1],
                                                        regex[1],
                                                        match),
                            0,
                            start_position,
                            end_position)

Thank you.

[Alan Kilborn]

@Ekopalypse

You have a good approach, I think. The only really great solution, that is great in one way and bad in another, is to lex over the entire document constantly. Obviously the problem with that is that it could take a great deal of time.

In keeping with your current approach, have you considered what happens when lines are folded or hidden? It seems that this would alter how far you need to “reach” beyond the current line (in both directions).

Pythonscript/Scintilla has editor.docLineFromVisible() which may be of some use for this situation. Notepad++ itself uses it as well, see https://github.com/notepad-plus-plus/notepad-plus-plus/blob/master/PowerEditor/src/ScitillaComponent/SmartHighlighter.cpp#L83-L104

[Ekopalypse]

@Alan-Kilborn

thank you very much for your insight, very much appreciated.

Obviously the problem with that is that it could take a great deal of time.

Right, I tested with a 5000 lines of code script and there was a noticeable delay.

In keeping with your current approach, have you considered what happens when lines are folded or hidden?

Yes, I’ve tested different scenarios when upper part has folded/hidden text or
within current visible area and bottom but neither of those are re-calculated,
so, you are right - this can become another issue.
What if hundreds of lines are folded but within the calculated range?
Need to test this more thoroughly.

Pythonscript/Scintilla has editor.docLineFromVisible()

:-D yes, - looks like the first two lines of the example :-)

Again, thank you very much.