regexp help

[Janek Schleicher]

sjaak wrote at Thu, 26 Jun 2003 12:35:08 +0200:

> Can anybody set me a little on the wright direction how to replace
> route.htm"></a> to generate
> route.htm">route</a>
>
> so
> .htm"></a>
> .html"></a>
>
> must replaced to
>
> .htm">route</a>
> .html">route</a>

Simple way:

s/(\.html?">)(</a>)/$1route$2/g;

If you do a lot of such stuff,
a HTML parser would be more useful.

> I don't know where to begin this with regexp.

Have you read
perldoc perlre
?

> To avoid futher questions in this does anyone knows a good howto with many
> examples.

You might also read the execellent
"Mastering Regular Expressions" book of J. Friedl
(read perldoc -q book for all details)

> Like this problem is a combination of regexp's and I don't understand them
> all so it's heard to do.

What combination?

> I just need them ones a year that's why i don't stay in touch with regexp.

IHMO, regular expressions must only be understood one times deeply. The
exact syntax is easy if the principle is understood. In fact, the exact
syntax differs very programs (vim, sed, grep, egrep, java, Perl).

Also, I believe using Perl would also indicate using regexps more often
than once a year.


Greetings,
Janek

[fatted]

I'd like to try and combine what I currently have in 4 regexps into
(maybe) one regexp.

I'm trying to parse data from a string, where the string is in the
format:
s_left *or* s_left = s_right.
rules:
s_right may have one or more equals in it.
In a
s_left = s_right
white space between the last/first non-white space char and the '='
char should be ignored (not part of s_left or s_right).
If s_left has a equals in it, s_left should be wrapped in double
quotes ("s_le=ft" = s_right)
Otherwise s_left is taken as beginning at the start of line, and
continuing up to the '=' char. (excluding white space between the last
non-white space from the sentence_left and the '=')

So, I have:

use warnings;
use strict;

# some test values
my @array = (
'"this = that"',
'log.log',
'start = "1"',
'sum=5+3=21',
'sql=SELECT * FROM table WHERE col17 = 2',
'"eq=als"=5=7',
'my name is = fatted',
'"1=1" = chicken'
);

foreach (@array)
{
unless( /=/ )
{
print "[zero] ".$_."\n";
}
elsif( /^"(.+)"$/ )
{
print "[one] ".$1."\n";
}
elsif( /^"([^"]*)[^=]*=\s*(.+)/ )
{
print "[two] ".$1."[]".$2."\n";
}
elsif( /^([^=]*)=\s*(.+)/ )
{
my $one = $1;
my $two = $2;
$one =~ s/\s+$//;
print "[three] ".$one."[]".$two."\n";
}
}

which gives:
[one] this = that
[zero] log.log
[three] start[]"1"
[three] sum[]5+3=21
[three] sql[]SELECT * FROM table WHERE col17 = 2
[two] eq=als[]5=7
[three] my name is[]fatted
[two] 1=1[]chicken

Which is what I require. But I'd now like to see if I could write the
regexp in one go, but I'm not having much luck, for instance:

one attempt to combine the last 2 regexps:

elsif( /^("|)([^"]*|[^=]*)[^=]*=\s*(.+)/)
{
# same processing as in last elsif block above
}

Doesn't work :( (not same results as above)

[Brian McCauley]

[email]obeseted@yahoo.com[/email] (fatted) writes:

> I'd like to try and combine what I currently have in 4 regexps into
> (maybe) one regexp.
>
> I'm trying to parse data from a string, where the string is in the
> format:
> s_left *or* s_left = s_right.
> rules:
> s_right may have one or more equals in it.
> In a
> s_left = s_right
> white space between the last/first non-white space char and the '='
> char should be ignored (not part of s_left or s_right).
> If s_left has a equals in it, s_left should be wrapped in double
> quotes ("s_le=ft" = s_right)
> Otherwise s_left is taken as beginning at the start of line, and
> continuing up to the '=' char. (excluding white space between the last
> non-white space from the sentence_left and the '=')
>
> So, I have:
>
> use warnings;
> use strict;
>
> # some test values
> my @array = (
> '"this = that"',
> 'log.log',
> 'start = "1"',
> 'sum=5+3=21',
> 'sql=SELECT * FROM table WHERE col17 = 2',
> '"eq=als"=5=7',
> 'my name is = fatted',
> '"1=1" = chicken'
> );
>
> foreach (@array)
> {
> unless( /=/ )
> {
> print "[zero] ".$_."\n";
> }
> elsif( /^"(.+)"$/ )
> {
> print "[one] ".$1."\n";
> }
> elsif( /^"([^"]*)[^=]*=\s*(.+)/ )
> {
> print "[two] ".$1."[]".$2."\n";
> }
> elsif( /^([^=]*)=\s*(.+)/ )
> {
> my $one = $1;
> my $two = $2;
> $one =~ s/\s+$//;
> print "[three] ".$one."[]".$two."\n";
> }
> }
>
> which gives:
> [one] this = that
> [zero] log.log
> [three] start[]"1"
> [three] sum[]5+3=21
> [three] sql[]SELECT * FROM table WHERE col17 = 2
> [two] eq=als[]5=7
> [three] my name is[]fatted
> [two] 1=1[]chicken
>
> Which is what I require. But I'd now like to see if I could write the
> regexp in one go,

foreach (@array)
{
if ( my ( $left,$right) = /^(\"[^\"]+\"|[^\"][^=]+?)\s*(?:=\s*(.*)\s*)?$/ ) {
$left =~ s/^"(.*)"$/$1/;
if ( defined $right ) {
print "[two] $left\[]$right\n";
} else {
print "[one] $left\n";
}
}else {
# Invalid input!
print "[zero] $_\n";
}
}

Produeces:

[one] this = that
[one] log.log
[two] start[]"1"
[two] sum[]5+3=21
[two] sql[]SELECT * FROM table WHERE col17 = 2
[two] eq=als[]5=7
[two] my name is[]fatted
[two] 1=1[]chicken

Note: double quote does not need to be backslashed in // but it's a
courtesy to do so for the beniefit of people using broken
auto-indenters.

--
\\ ( )
. _\\__[oo
.__/ \\ /\@
. l___\\
# ll l\\
###LL LL\\

[Anno Siegel]

Brian McCauley <nobull@mail.com> wrote in comp.lang.perl.misc:

> Note: double quote does not need to be backslashed in // but it's a
> courtesy to do so for the beniefit of people using broken
> auto-indenters.

I object.

All auto-indenters, syntax-hilighters and their ilk are broken in some
respect. ("Only perl can parse Perl") If we start accommodating their
glitches, we'll end up with coding conventions that are completely
irrational from a Perl point of view.

Anno

[raven]

The output of lpstat -p on hpux returns this:

printer sacprn05 is idle. enabled since Jul 30 12:23
fence priority : 0

I am attempting to grab the printer name and whether or not it is
idle. The code to do this is:

if (/^printer (\w+).*(enabled|disabled)/)

Is there a more efficient way to obtain the desired information using
a Perl regexp?

[John Bokma]

raven wrote:

> The output of lpstat -p on hpux returns this:
>
> printer sacprn05 is idle. enabled since Jul 30 12:23
> fence priority : 0
>
> I am attempting to grab the printer name and whether or not it is
> idle. The code to do this is:

^^^^

idle...

> if (/^printer (\w+).*(enabled|disabled)/)

enabled/disabled... I am confused

> Is there a more efficient way to obtain the desired information using
> a Perl regexp?

Don't know if changing .* to .*? speeds things up.

--
Kind regards, feel free to mail: mail(at)johnbokma.com (or reply)
virtual home: [url]http://johnbokma.com/[/url] ICQ: 218175426
John web site hints: [url]http://johnbokma.com/websitedesign/[/url]

[John Bokma]

David Oswald wrote:

> "raven" <raven_riverwind@yahoo.com> wrote in message
>

>>The output of lpstat -p on hpux returns this:

>
>

>>printer sacprn05 is idle. enabled since Jul 30 12:23
>> fence priority : 0

[snip]

> be too broad. For example, you might actually mean [a-zA-Z], or you might

nope, 05 seems to suggest at least digits can appear in the name :-)

Thanks Dave, this was very clear and educational!

--
Kind regards, feel free to mail: mail(at)johnbokma.com (or reply)
virtual home: [url]http://johnbokma.com/[/url] ICQ: 218175426
John web site hints: [url]http://johnbokma.com/websitedesign/[/url]

[Lao Coon]

[email]raven_riverwind@yahoo.com[/email] (raven) wrote in
news:7270d1f8.0309050952.3154538e@posting.google.c om:

> The output of lpstat -p on hpux returns this:
>
> printer sacprn05 is idle. enabled since Jul 30 12:23
> fence priority : 0
>
> I am attempting to grab the printer name and whether or not it is
> idle. The code to do this is:
>
> if (/^printer (\w+).*(enabled|disabled)/)
>
> Is there a more efficient way to obtain the desired information using
> a Perl regexp?

E.g. /^printer\s(\S+)/. Shorter is usually better, if you have little or
no variation in the input.

perl -MBenchmark -e "$a = 'printer sacprn05 is idle. enabled since Jul 30
12:23'; timethese(500000, {

'First' => '$a =~ /^printer (\w+).*(enabled|disabled)/;',

'Second' => '$a =~ /^printer\s(\S+)/;'

});"

Benchmark: timing 500000 iterations of First, Second...
First: 8 wallclock secs ( 7.19 usr + 0.02 sys = 7.21 CPU) @
69338.51/s (
n=500000)
Second: 1 wallclock secs ( 1.76 usr + 0.00 sys = 1.76 CPU) @
283607.49/s
(n=500000)

[Jay Tilton]

[email]raven_riverwind@yahoo.com[/email] (raven) wrote:

: The output of lpstat -p on hpux returns this:
:
: printer sacprn05 is idle. enabled since Jul 30 12:23
: fence priority : 0
:
: I am attempting to grab the printer name and whether or not it is
: idle. The code to do this is:
:
: if (/^printer (\w+).*(enabled|disabled)/)
:
: Is there a more efficient way to obtain the desired information using
: a Perl regexp?

What is your gauge of efficiency? I have to guess you want to
minimize runtime.

Why are you worrying about a match operation that takes microseconds
to perform? Unless it's going to happen a quadrillion times a day,
fuggetaboutit.

Anyway, the real time-killer will be invoking the external 'lpstat'
command.

[raven]

Thanks for the enlightening reply. Sorry for being vague in my
specification. I am simply tracking the printers, their IPs and the
current disabled/enabled status. In the future I'm sure I'll add more
functionality. Below is the the output of the various commands and the
hack.

lpstat -vsacprn05
device for sacprn05: /dev/null
remote to: RNP6C8BF9 on 192.168.25.73


#!/usr/contrib/bin/perl -w

# Printer interfaces directory
my $interface = "/etc/lp/interface";
# Printer Array
my @printers = ();

# Get a list of printers
open PRINTER, "lpstat -p |";
while (<PRINTER>) {
if (/^printer (\w+).*?(enabled|disabled)/) {
# Push an anonymous hash REFERENCE onto the array
push @printers, { "name" => "$1", "ip" => "none", "status" =>
"$2" };
}
}
close PRINTER;

# Gather the IP address of each printer
foreach $printer (@printers) {
open PRINTER, "lpstat -v$$printer{name} |";
while (<PRINTER>) {
next if /^device.*/;
if (/^\s+remote.*?(\d{1,3}.\d{1,3}.\d{1,3}.\d{1,3})/) {
$$printer{ip} = $1;
last;
}
}
close PRINTER;
# We hope that the printers with 'none' for IP are JetDirect
if ($$printer{ip} =~ /none/) {
open INTERFACE, "$interfaces/$$printer{name}" || die
"Cannot open file!\n";
while (<INTERFACE>) {
if (/^PERIPH=(.*)/) {
$$printer{ip} = $1;
last;
}
}
close INTERFACE;
}
}

# Display the results
foreach $printer (@printers) {
print "$$printer{name} / $$printer{ip} / $$printer{status}\n";
}

[John W. Krahn]

raven wrote:

>
> Thanks for the enlightening reply. Sorry for being vague in my
> specification. I am simply tracking the printers, their IPs and the
> current disabled/enabled status. In the future I'm sure I'll add more
> functionality. Below is the the output of the various commands and the
> hack.
>
> lpstat -vsacprn05
> device for sacprn05: /dev/null
> remote to: RNP6C8BF9 on 192.168.25.73
>
> #!/usr/contrib/bin/perl -w
>
> # Printer interfaces directory
> my $interface = "/etc/lp/interface";
> # Printer Array
> my @printers = ();
>
> # Get a list of printers
> open PRINTER, "lpstat -p |";

You should _ALWAYS_ verify that the pipe opened correctly.

> while (<PRINTER>) {
> if (/^printer (\w+).*?(enabled|disabled)/) {
> # Push an anonymous hash REFERENCE onto the array
> push @printers, { "name" => "$1", "ip" => "none", "status" =>
> "$2" };

Excessive quoting. The only thing that needs to be quoted is the string
'none'.

> }
> }
> close PRINTER;

You should _ALWAYS_ verify that the pipe closed correctly.

> # Gather the IP address of each printer
> foreach $printer (@printers) {
> open PRINTER, "lpstat -v$$printer{name} |";

You should _ALWAYS_ verify that the pipe opened correctly.

> while (<PRINTER>) {
> next if /^device.*/;
> if (/^\s+remote.*?(\d{1,3}.\d{1,3}.\d{1,3}.\d{1,3})/) {

^ ^ ^
^ ^ ^
Dots are special in regular expressions. You need to escape them to
match a literal dot character.

> $$printer{ip} = $1;
> last;
> }
> }
> close PRINTER;

You should _ALWAYS_ verify that the pipe closed correctly.

> # We hope that the printers with 'none' for IP are JetDirect
> if ($$printer{ip} =~ /none/) {
> open INTERFACE, "$interfaces/$$printer{name}" || die
> "Cannot open file!\n";

Using the || operator there won't do what you expect. You need to
either use the 'or' operator or parenthesize the open function.

> while (<INTERFACE>) {
> if (/^PERIPH=(.*)/) {
> $$printer{ip} = $1;
> last;
> }
> }
> close INTERFACE;
> }
> }
>
> # Display the results
> foreach $printer (@printers) {
> print "$$printer{name} / $$printer{ip} / $$printer{status}\n";
> }

It looks like you just need one loop for all that:

#!/usr/contrib/bin/perl -w
use strict;

# Printer interfaces directory
my $interface = '/etc/lp/interface';
# Printer Array
my @printers;

# Get a list of printers
open PRINTER, 'lpstat -p |' or die "Cannot open pipe from lpstat: $!";
while ( <PRINTER> ) {

next unless /^printer (\w+).*?((?:dis|en)abled)/;
# Push an anonymous hash REFERENCE onto the array
push @printers, { name => $1, ip => 'none', status => $2 };

if ( `lpstat -v $printers[-1]{name}` =~
/^\s+remote.*?(\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3})/m ) {
$printers[ -1 ]{ ip } = $1;
}
else {
local $/;
open INTERFACE, "$interfaces/$printers[-1]{name}"
or die "Cannot open $interfaces/$printers[-1]{name}: $!";
$printers[ -1 ]{ ip } = $1 if <INTERFACE> =~ /^PERIPH=(.*)/m;
close INTERFACE;
}
}
close PRINTER or die "Cannot close pipe from lpstat: $!";

# Display the results
foreach my $printer ( @printers ) {
print join( ' / ', @{$printer}{ qw/name ip status/ } ), "\n";
}



John
--
use Perl;
program
fulfillment

[Marcelo]

Which regular expression would you use to remove the <title> and </title> from a line like this one:

<title>Here goes a webpage's title</title>

Thanks a lot in advance.

[Andrew Gaffney]

Marcelo wrote:

> Which regular expression would you use to remove the <title> and </title> from a line like this one:
>
> <title>Here goes a webpage's title</title>
>
> Thanks a lot in advance.

Try something like:

s/<\/?title>//

Although, I can remember if < and > are special regex characters so you might need to
escape them (\< and \>).

--
Andrew Gaffney
System Administrator
Skyline Aeronautics, LLC.
776 North Bell Avenue
Chesterfield, MO 63005
636-357-1548

[John McKown]

On Sat, 24 Jan 2004, Marcelo wrote:

> Which regular expression would you use to remove the <title> and
> </title> from a line like this one:
>
> <title>Here goes a webpage's title</title>
>
> Thanks a lot in advance.
>

Did you what that _exact_ input? I.e. always <title>...</title>? If so,
that's rather easy.

$line =~ s/<title>(.*)<\/title>/$1/

Now, if you want the more general form of <any_tag>...</any_tag>, that is
removing paired HTML tags, that's more difficult. Luckily, it is an
example in "Programming PERL, 3rd Edition" on page 184 which is close.

line =~ s/(<.*?>)(.*?)(?:</\1>)/$2/

In sort-of English. This says:

Match starting with a < and ending with the next >, calling it $1 (or \1).
Now, match everything up to the next < and call it $2. Now match a <
followed by a /, followed by what you matched first (in $1 or \1),
followed by a >. Now, replace all of that with $2.

A problem with this pattern is that it would not work as you would
like want it to with input such as:

<title><B>Title</B></title>

You'd end up removing the <B> and </B>, but leaving the <title> and
</title>. Of course, if your desire is to remove all paired HTML tags,
then put this in a loop until it no longer matches.

HTH,

--
Maranatha!
John McKown

[Jan Eden]

John McKown wrote:

>On Sat, 24 Jan 2004, Marcelo wrote:
>

>> Which regular expression would you use to remove the <title> and
>> </title> from a line like this one:
>>
>> <title>Here goes a webpage's title</title>
>>
>> Thanks a lot in advance.
>>

>
>Did you what that _exact_ input? I.e. always <title>...</title>? If so,
>that's rather easy.
>
>$line =~ s/<title>(.*)<\/title>/$1/
>
>Now, if you want the more general form of <any_tag>...</any_tag>, that is
>removing paired HTML tags, that's more difficult. Luckily, it is an
>example in "Programming PERL, 3rd Edition" on page 184 which is close.
>
>line =~ s/(<.*?>)(.*?)(?:</\1>)/$2/

I remember reading that using regex to parse HTML is not reliable. You should use HTML::Parse from CPAN.

HTH,

Jan
--
Either this man is dead or my watch has stopped. - Groucho Marx