Perl language syntax highlighting troubles (bug or limitation ?)

Gilles Maisonneuve

This post is deleted!

Gilles Maisonneuve

Ok
another one: in Python you must say ["|'] instead of Perl ["'] (‘either one of the set’) ? Is that what it means ?

Ekopalypse

No, afaik non-capturing group is (?:pattern)
This, (?s), just tells the engine that the dot . is matching
EOLs like \r\n - if I’m right.

Ekopalypse

Just for clarification, the python script does NOT use the python regex engine instead
it uses the one notepad++ offers, the boost::regex.
Yes, you can use the enumeration without the pipe but makes it more visible for me with
the pipe sign. Or is there a difference if used with pipe sign or without?

Ekopalypse

or maybe this one might be even better
(?s)(<<)\h+(["'])(\w+?)\2\h*;.*?\3

Gilles Maisonneuve

Can’t reply what I wanted, a robot says I’m spamming…

Ekopalypse

@Gilles-Maisonneuve

Can’t reply what I wanted, a robot says I’m spamming…

I have no idea why this happens sometimes.

By the way, now that you have installed pythonscript plugin would you mind
clicking Plugins->Python Script->Scripts->Samples->RegexTester ?

I know not everyone is recommending it but, personally, I love it.

Gilles Maisonneuve

AFAIK, at least in Perl, ["|'] means double-quote OR pipre OR simple-quote, everything between square brakets is literal. Also true in “awk” and C regexp I think.
I don’t know for Python.

Gilles Maisonneuve

@Ekopalypse

Now, if I say in Pyhton (attempt to transliterate from Perl) :

(r'(?s)(\h*(<<)\h*["|']?([^"|^']+?)["|']?\h*;.*?\3)', [2])

does it mean :

form REGEXP
do not match NL with DOT
matches any horizontal blanks (0 or more), don’t make a group
matches ‘<<’ make it a group
matches any horizontal blanks (0 or more), don’t make a group
matches 0 or 1 text quote (either double or single), no group
matches a group of any chars not " nor ’ one or more time(s) (in perl it would be [^"'])
matches 0 or 1 text quote (either double or single), no group
possible blanks until semi-colon, semi-colon, then possible chars until NL

BUT THEN, what does mean ?\3. I’m lost there.

Gilles Maisonneuve

a slash m

Ekopalypse

the r at the beginning just informs python that this is a raw string and
every char must be taken literally otherwise backslashes would be treated
as escapes under some circumstances.

The regex string is only this part

(?s)(\h*(<<)\h*["|']?([^"|^']+?)["|']?\h*;.*?\3)

and I would say, but as said - not an regex expert at all,

(?s) means Dot matches newline characters
the first matching group is

(\h*(<<)\h*["|']?([^"|^']+?)["|']?\h*;.*?\3)

the second

(<<)

and the third must be

([^"|^']+?)

if I’m right.

\3 should be the same as $3 in perl

Gilles Maisonneuve

@Ekopalypse

still confused: ([^"|^']+?) why a ‘?’ after the ‘+’ what’s for this ‘?’

and then \3 would mean the 3rd matching group (third ‘()’) but in Perl is used only in subsitutions. What is the use here ? There are only 2 groups in the regex (two blocks surrounded by parenthèses only.

Ekopalypse

@Gilles-Maisonneuve

maybe this picture makes it a little bit clearer

Ekopalypse

@Gilles-Maisonneuve

still confused: ([^"|^']+?) why a ‘?’ after the ‘+’ what’s for this ‘?’

as less as possible - non-greedy

and then \3 would mean the 3rd matching group (third ‘()’) but in Perl is used only in >subsitutions. What is the use here ? There are only 2 groups in the regex (two blocks >surrounded by parenthèses only.

placeholder for what was found in match group 3, to find the EOT at the end

and there are 3 match groups or am I missing something??

Gilles Maisonneuve

@Ekopalypse

2 sets of parenteses only, where is the third set ?
so only 2 match groups

can you make this work :

no syntax error on the python console but absolutely no result, where is my bug ?

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

Ekopalypse

@Gilles-Maisonneuve

[1] informs the python script, that only the results from sub match group 1 should be colored in white (255,255,255)
sub match group 1 is the result of (<<)

In order to make it painting all you can use [0]

I’m still confused about the 2 to 3 match groups.
Am I incorrect when saying that
(\s*(<<)\s*("{0,1}.+"{0,1})\s*;.*?\3)
(<<)
("{0,1}.+"{0,1})
are three match groups?

Maybe the confusion comes from the fact that references matches within a
regular expression starts by 1 but python starts counting match results by 0.

Sorry, but I have to stay up early tomorrow and it is already 1am but I’m really
interested in solving our (mis)understanding today later (maybe in ~16-18hours)?

Gilles Maisonneuve

ok, tomorrow is another day
‘see’ you tomorrow.
have a good night.
g

Ekopalypse

you too - see you

Gilles Maisonneuve

@Ekopalypse

OK, so the

(\s*(<<)\s*("{0,1}.+"{0,1})\s*;.*?\3)

is a regex group, not a function call surrounded by parenthèses or a logical group provided by the ‘r’ keyword. My mistake.
BUT THEN, it is possible in Python to enclose an instruction such as ?\3 which means (as far as I understood what you explained to me earlier) recursive reference to a regexp named ‘3’) ??? The ‘3’ name being given in the expression regexes[(3, (255,255,255))] is that correct ? SO you can reference an expression within itself while it has not be closed yet: the last parenthese of the expression 3 is after the \3). Is that what it means ?

Python syntax is a bit complicated to me.

Alan Kilborn

@Gilles-Maisonneuve said:

Python syntax is a bit complicated to me

It’s not Python syntax, it’s regular expression syntax. It’s just not Perl regular expression syntax. :)

And, BTW, nobody in the history of the world, especially someone coming from a Perl background, has ever uttered the phrase you typed.