recursive replace command line

[George Georgalis]

Hi -

giving my perl a retry, I found some hints on a website to recursively
replace text

perl -p -i -e 's/old\(.\)atext/new\1btext/g;' $( find ./ -name '*.html' -o -name '*.txt' )

this should replace "old-atext" with "new-btext" and "old+atext" with
"new+btext" and similar occurrences in all the files from find.

but from what I can tell, perl doesn't support the \1 for \(*\) symbols
like sed does. What is the work around?

// George


--
GEORGE GEORGALIS, System Admin/Architect cell: 646-331-2027 <IXOYE><
Security Services, Web, Mail, mailto:george@galis.org
Multimedia, DB, DNS and Metrics. [url]http://www.galis.org/george[/url]

[Jeff 'Japhy' Pinyan]

On Dec 10, George Georgalis said:

>giving my perl a retry, I found some hints on a website to recursively
>replace text
>
>perl -p -i -e 's/old\(.\)atext/new\1btext/g;' $( find ./ -name '*.html' -o -name '*.txt' )

This isn't recursively replacing text; it's recursively going through a
directory tree.

>but from what I can tell, perl doesn't support the \1 for \(*\) symbols
>like sed does. What is the work around?

Because Perl is not sed. Perl uses (...), not \(...\) for its memory
capturing. In Perl's regexes, all non-alphanumeric metacharacters don't
use backslashes. That means [...] for character classes, not \[...\], and
+ for " 1 or more", not \+, and so on.

--
Jeff "japhy" Pinyan [email]japhy@pobox.com[/email] [url]http://www.pobox.com/~japhy/[/url]
RPI Acacia brother #734 [url]http://www.perlmonks.org/[/url] [url]http://www.cpan.org/[/url]
<stu> what does y/// stand for? <tenderpuss> why, yansliterate of course.
[ I'm looking for programming work. If you like my work, let me know. ]

[George Georgalis]

On Thu, Dec 11, 2003 at 09:05:20AM -0500, Jeff 'japhy' Pinyan wrote:

>On Dec 10, George Georgalis said:
>

>>giving my perl a retry, I found some hints on a website to recursively
>>replace text
>>
>>perl -p -i -e 's/old\(.\)atext/new\1btext/g;' $( find ./ -name '*.html' -o -name '*.txt' )

>
>This isn't recursively replacing text; it's recursively going through a
>directory tree.
>

>>but from what I can tell, perl doesn't support the \1 for \(*\) symbols
>>like sed does. What is the work around?

>
>Because Perl is not sed. Perl uses (...), not \(...\) for its memory
>capturing. In Perl's regexes, all non-alphanumeric metacharacters don't
>use backslashes. That means [...] for character classes, not \[...\], and
>+ for " 1 or more", not \+, and so on.

that's what I needed to hear... however replacing text (with memory
capturing) is still a problem:

perl -p -i -e 's/451(.)8229/331\12027/g;' $( find ./ -type f -name '*.html' -o -name '*.txt' )

here I'm trying to replace instances of of a phone number, that might
use "-" " " or "." to separate the first and second parts.
451-8229 to 331-2027
451 8229 to 331 2027
etc

the result "331P27" hinted 3 digits may be used for the call back

perl -p -i -e 's/451(.)8229/331\0012027/g;' $( find ./ -type f -name '*.html' -o -name '*.txt' )

that puts the characters where they go but the dash has changed to some
other character (an ascII 001 I suppose).

// George


--
GEORGE GEORGALIS, System Admin/Architect cell: 646-331-2027 <IXOYE><
Security Services, Web, Mail, mailto:george@galis.org
Multimedia, DB, DNS and Metrics. [url]http://www.galis.org/george[/url]

[Jeff 'Japhy' Pinyan]

On Dec 11, George Georgalis said:

>On Thu, Dec 11, 2003 at 09:05:20AM -0500, Jeff 'japhy' Pinyan wrote:

>>On Dec 10, George Georgalis said:
>>

>>>giving my perl a retry, I found some hints on a website to recursively
>>>replace text
>>>
>>>perl -p -i -e 's/old\(.\)atext/new\1btext/g;' $( find ./ -name '*.html' -o -name '*.txt' )

>>
>>This isn't recursively replacing text; it's recursively going through a
>>directory tree.
>>

>>>but from what I can tell, perl doesn't support the \1 for \(*\) symbols
>>>like sed does. What is the work around?

>>
>>Because Perl is not sed. Perl uses (...), not \(...\) for its memory
>>capturing. In Perl's regexes, all non-alphanumeric metacharacters don't
>>use backslashes. That means [...] for character classes, not \[...\], and
>>+ for " 1 or more", not \+, and so on.

>
>that's what I needed to hear... however replacing text (with memory
>capturing) is still a problem:
>
>perl -p -i -e 's/451(.)8229/331\12027/g;' $( find ./ -type f -name '*.html' -o -name '*.txt' )

Right; first of all, \1 should only be used IN the regex ITSELF. Outside
of the regex, you should use $1. However, "331$12027" still isn't
appropriate; you'll need "331${1}2027". The reason "331\12027" gave you
"331P27" is because octal character 120 is "P".

--
Jeff "japhy" Pinyan [email]japhy@pobox.com[/email] [url]http://www.pobox.com/~japhy/[/url]
RPI Acacia brother #734 [url]http://www.perlmonks.org/[/url] [url]http://www.cpan.org/[/url]
<stu> what does y/// stand for? <tenderpuss> why, yansliterate of course.
[ I'm looking for programming work. If you like my work, let me know. ]

[George Georgalis]

On Thu, Dec 11, 2003 at 11:21:39AM -0500, Jeff 'japhy' Pinyan wrote:

>On Dec 11, George Georgalis said:
>

>>On Thu, Dec 11, 2003 at 09:05:20AM -0500, Jeff 'japhy' Pinyan wrote:

>>>On Dec 10, George Georgalis said:
>>>
>>>>giving my perl a retry, I found some hints on a website to recursively
>>>>replace text
>>>>
>>>>perl -p -i -e 's/old\(.\)atext/new\1btext/g;' $( find ./ -name '*.html' -o -name '*.txt' )
>>>
>>>This isn't recursively replacing text; it's recursively going through a
>>>directory tree.
>>>
>>>>but from what I can tell, perl doesn't support the \1 for \(*\) symbols
>>>>like sed does. What is the work around?
>>>
>>>Because Perl is not sed. Perl uses (...), not \(...\) for its memory
>>>capturing. In Perl's regexes, all non-alphanumeric metacharacters don't
>>>use backslashes. That means [...] for character classes, not \[...\], and
>>>+ for " 1 or more", not \+, and so on.

>>
>>that's what I needed to hear... however replacing text (with memory
>>capturing) is still a problem:
>>
>>perl -p -i -e 's/451(.)8229/331\12027/g;' $( find ./ -type f -name '*.html' -o -name '*.txt' )

>
>Right; first of all, \1 should only be used IN the regex ITSELF. Outside
>of the regex, you should use $1. However, "331$12027" still isn't
>appropriate; you'll need "331${1}2027". The reason "331\12027" gave you
>"331P27" is because octal character 120 is "P".

Thanks,

perl -p -i -e 's/347(.)451(.)8229/646${1}331${2}2027/g;' $( find ./ -type f -name '*.html' -o -name '*.txt' )
perl -p -i -e 's/347(..)451(.)8229/646${1}331${2}2027/g;' $( find ./ -type f -name '*.html' -o -name '*.txt' )

worked perfect to update my web pages... :)
btw - what's the best manpage for the perl command line options?

TAI,

// George


--
GEORGE GEORGALIS, System Admin/Architect cell: 646-331-2027 <IXOYE><
Security Services, Web, Mail, mailto:george@galis.org
Multimedia, DB, DNS and Metrics. [url]http://www.galis.org/george[/url]

[R. Joseph Newton]

George Georgalis wrote:

>
> that's what I needed to hear... however replacing text (with memory
> capturing) is still a problem:
>
> perl -p -i -e 's/451(.)8229/331\12027/g;' $( find ./ -type f -name '*.html' -o -name '*.txt' )

Hi George,

If you are going to do your regex work--especially at a high level--in Perl, it is imperative that
you read the Perl documentation. The problem in the above sam[ple, or the one that jumps out at me,
is that you are using the backslash form of back reference in the replacement string. In the Perl
implementation, that is not appropriate. Use the built-in scalars ($1, $2 ...) instead. Backslash
backreferences are used within the scanning portion of the regex, but not in the replqacement
string.

Please read:
perldoc perlre

Joseph

[George Georgalis]

On Thu, Dec 11, 2003 at 11:07:53AM -0800, R. Joseph Newton wrote:

>If you are going to do your regex work--especially at a high level--in Perl, it is imperative that
>you read the Perl documentation. The problem in the above sam[ple, or the one that jumps out at me,
>is that you are using the backslash form of back reference in the replacement string. In the Perl
>implementation, that is not appropriate. Use the built-in scalars ($1, $2 ...) instead. Backslash
>backreferences are used within the scanning portion of the regex, but not in the replqacement
>string.
>
>Please read:
>perldoc perlre

Thanks for the pointer, Joseph. Indeed, I'm not yet up to speed on
accessing perl-doc, there is quite a lot. ;-)

Bye,
// George

--
GEORGE GEORGALIS, System Admin/Architect cell: 646-331-2027 <IXOYE><
Security Services, Web, Mail, mailto:george@galis.org
Multimedia, DB, DNS and Metrics. [url]http://www.galis.org/george[/url]

[Paul Johnson]

On Thu, Dec 11, 2003 at 12:17:03PM -0500, George Georgalis wrote:

> btw - what's the best manpage for the perl command line options?

perlrun

--
Paul Johnson - [email]paul@pjcj.net[/email]
[url]http://www.pjcj.net[/url]