From Newsgroup: comp.lang.tcl


Hello,

Simply :

set _res [regexp -inline -all {\S+} "dum dummy a b c d c4"]

gives a list of 7 items : dum dummy a b c d c4

set _res [regexp -inline -all {c+} "dum dummy a b c d c4"]

gives a list of " items : c c

I understand that {c+} acts as a specification, but I cannot understand why {\S+}
acts as a counter-specification, it should return only spaces and/or tabs ?

I found this "trick" on https://wiki.tcl-lang.org/page/Get+a+Certain+Line+From+a+File
I often deal with simulation data where items of line are homogeneously separated
either by space or by tab. So the line can be considered as a list. But in rare case
you stumble on hybrid separators : line defining units or names for instance, that is
how I came to this question. And this "trick" works as a charm !

Olivier.


--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.lang.tcl

On 10/14/26 17:05, Olivier wrote:

regexp -inline -all {\S+} "dum dummy a b c d c4"

Did you mean to do:

regexp -inline -all {\s+} "dum dummy a b c d c4"

Note the lower case character "s".

Erik Leunissen
--
elns@ nl | Merge the left part of these two lines into one,
xs4all. | respecting a character's position in a line.

--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.lang.tcl


Erik Leunissen <[email protected]> posted:

On 10/14/26 17:05, Olivier wrote:

regexp -inline -all {\S+} "dum dummy a b c d c4"

Did you mean to do:

regexp -inline -all {\s+} "dum dummy a b c d c4"

Note the lower case character "s".

Erik Leunissen

Ah ! Thank you, your message makes me look (again) at the definition of \S with an uppercase : \S <==> [^[:space:]] . I did not pay attention to the "^" wich means :
"If the list begins with “^”, it matches any single character (but see below) not from the rest of the list."

The behaviour is now obvious ! It was in the documentation, but I had in mind that "^"
was relative to the "beginning of", not that usage.

Thanks for your time,

Olivier.
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.lang.tcl

Olivier <[email protected]d> wrote:

Hello,

Simply :

set _res [regexp -inline -all {\S+} "dum dummy a b c d c4"]

\S is a special symbol, it is documented in the re_syntax man page
under the "CLASS-SHORTHAND ESCAPES" header:

\S [^[:space:]]

So using \S is the same as having entered [^[:space:]] into the regex.
And that [^[:space:]] means "any character that is not in the set
":space:" (whitespace). Or shorter: "not whitespace characters".

"+" in a regex means one or more matches of the prior atom (in this
instance the shorthand for "[^[:space:]]".

So \S+ means one or more characters that are not whitespace characters.

Which is why you get this result:

gives a list of 7 items : dum dummy a b c d c4

set _res [regexp -inline -all {c+} "dum dummy a b c d c4"]

"c" is just a regular character. So it means "c". + means one or more
of the prior atom (the "c") so it means a run of one or more c
characters. You have two runs of "c" characters, both one character
long. The lone "c" and the "c4". So this result also makes perfect
sense:

gives a list of " items : c c

I understand that {c+} acts as a specification, but I cannot understand why {\S+}
acts as a counter-specification, it should return only spaces and/or tabs ?

/S+ (the capitolized variety) is already a "not" specifier by
definition. So it is "not whitespace" which is why you get everything
but the white space. You may be thinking of \s which is shorthand for [[:space:]]
which is "all white space characters".

I found this "trick" on https://wiki.tcl-lang.org/page/Get+a+Certain+Line+From+a+File I often
deal with simulation data where items of line are homogeneously
separated either by space or by tab. So the line can be considered
as a list. But in rare case you stumble on hybrid separators : line defining units or names for instance, that is how I came to this
question. And this "trick" works as a charm !

Yes, it's a useful "trick" to normalize a set of "white space
separatated items" into a list, when you don't care how many white
space separaters there are, or which exact white space characters are
present.
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.lang.tcl

Olivier <[email protected]d> wrote:

Erik Leunissen <[email protected]> posted:

On 10/14/26 17:05, Olivier wrote:

regexp -inline -all {\S+} "dum dummy a b c d c4"

Did you mean to do:

regexp -inline -all {\s+} "dum dummy a b c d c4"

Note the lower case character "s".

Erik Leunissen

Ah ! Thank you, your message makes me look (again) at the definition
of \S with an uppercase : \S <==> [^[:space:]] . I did not pay
attention to the "^" wich means : "If the list begins with “^”, it matches any single character (but see below) not from the rest of the
list."

The behaviour is now obvious ! It was in the documentation, but I
had in mind that "^" was relative to the "beginning of", not that
usage.

Sadly, ^ does double duty in regular expressions. It is an anchor,
when used outside bracket expressions ([]) and it is the negation
operator when used inside bracket expressions, but then, only if it is
the first character in the bracket expression, otherwise it is just
itself if it occurs elsewhere in brackets.

--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.lang.tcl


Rich <[email protected]d> posted:

Yes, it's a useful "trick" to normalize a set of "white space
separatated items" into a list, when you don't care how many white
space separaters there are, or which exact white space characters are present.

Thank you, your answer came simultaneously when I finally understood
when coming back to the documentation after Erik's
answer. I would say that your explanation is somewhat exhaustive.

The devil is in the details, and in Tcl/Tk documentation there are a lot.

Olivier.

PS : I read your explanation twice, not to forget that "^" !
--- Synchronet 3.22a-Linux NewsLink 1.2