literal substitution

[Dave Walden]

Hi,

Can someone tell me how I can obtain functionality like
$str =~ s/$a/$b/g;
where $a and $b contain strings and where instance of the the literal
string contained in $a are replaced by the literal string in $b
*without* regular expression interpretation. In other words, in the
following
my $str = 'moo \foo.* zoo';
my $a = ' \foo.*';
my $b = ' $';
$str =~ s/$a/$b/g;
I would like the value in $str to end up being 'moo $ zoo' while
allowing me to avoid having to escape in all the regex control
characters in the arguments to s///. Is there some qualifier I can use
to turn off regular expression interpretation in the arguments to s///
or some other way I can easily accomplish this without coding a new
subtroutine which takes the three arguments $str, $a, and $b and does
what I want?

Thanks, Dave

 

[kens]

Hi,

Can someone tell me how I can obtain functionality like
$str =~ s/$a/$b/g;
where $a and $b contain strings and where instance of the the literal
string contained in $a are replaced by the literal string in $b
*without* regular expression interpretation. In other words, in the
following
my $str = 'moo \foo.* zoo';
my $a = ' \foo.*';
my $b = ' $';
$str =~ s/$a/$b/g;
I would like the value in $str to end up being 'moo $ zoo' while
allowing me to avoid having to escape in all the regex control
characters in the arguments to s///. Is there some qualifier I can use
to turn off regular expression interpretation in the arguments to s///
or some other way I can easily accomplish this without coding a new
subtroutine which takes the three arguments $str, $a, and $b and does
what I want?

Thanks, Dave

You may want to look at the quotemeta function.

perldoc -f quotemeta

HTH, Ken

 

[Jürgen Exner]

Hi,

Can someone tell me how I can obtain functionality like
$str =~ s/$a/$b/g;
where $a and $b contain strings and where instance of the the literal
string contained in $a are replaced by the literal string in $b
*without* regular expression interpretation.

Sometimes I wonder why people are so obsessed with REs. Yes, they are great
and powerful and you you can do a lot with them.
However, why bend over backwards to make them fit a scenario they were not
meant for when there are much easier and straightforward solutions:
- use index() to find the location of string $a in $str
- use substr() to replace the text

jue

 

[Dave Walden]

- use index() to find the location of string $a in $str
- use substr() to replace the text

Thanks. That's the memory jog I needed (although not as ideal as
having an additional qualifier to s/// that turned off regex control
character interpretation).

 

[Brian McCauley]

Thanks. That's the memory jog I needed (although not as ideal as
having an additional qualifier to s/// that turned off regex control
character interpretation).

Did you miss kens' reply?

s/\Q$a/$b/;

 

[Ted Zlatanov]

Sometimes I wonder why people are so obsessed with REs. Yes, they are great
and powerful and you you can do a lot with them.
However, why bend over backwards to make them fit a scenario they were not
meant for when there are much easier and straightforward solutions:
- use index() to find the location of string $a in $str
- use substr() to replace the text

The first part of the problem, finding a constant string, is
definitely better with index(), but I would argue REs are exactly
right for this task as a whole. You'd have to write at least 5 lines
of code vs. one simple line (quotemeta() makes it two lines); not to
mention the complexity and potential for bugs are higher. Yes, you
can jam index()/substr() into one or two lines, I'm assuming sane
maintainable code here.

If it was just one substitution, maybe it's a tie, but with the g flag
you have to loop too. Unless you were optimizing, why would you do
this to yourself and to whoever has to maintain your code later?

Ted

 

[Paul Lalli]

The first part of the problem, finding a constant string, is
definitely better with index(), but I would argue REs are exactly
right for this task as a whole. You'd have to write at least 5 lines
of code vs. one simple line (quotemeta() makes it two lines);

The \Q..\E representation of quotemeta() keeps it one line.

s/\Q$a\E/$b;

Paul Lalli