Reg expression help please... 2

[nickroper]

Hi,

I have a regular expression that converts newline characters to '<BR>'. I need to modify this so that it only converts the newline characters if they are not preceeded by certain other strings ('table>', 'td>' and 'tr>')

So, examples would be as follows:

original string:

'line1\n line2\n line3'

would be converted to:

'line1<BR> line2<BR> line3'

Whereas:

'<table>\n<tr>\<td>line1\nline2\nline3</td>\n</tr>\n</table>'

would be converted to:

'<table>\n<tr>\<td>line1<BR>line2<BR>line3</td>\n</tr>\n</table>'

My existing expression is:

$foo =~ s/\n/<BR>/g;

which converts ALL occurrences of \n. If any regex gurus out there can let me know the trick to ignore \n preceeded by 'table>', 'td>' and 'tr>' (I've left the opening bracket off in order to catch both openng and closing tags) then I'd be extremely grateful.

Cheers,

Nick Roper

 

[duncdude]

s/([^>])\\n/$1<br>/g;


Kind Regards
Duncan

 

[nickroper]

Hi Duncan,

Great, thanks for the quick response. One question - will this ignore a linefeed preceeded by any closing tag ? I really just want to ignore table tags.

Thanks,

Nick

 

[duncdude]

yes! i'll try to improve this to suit your needs more exactly - but i can't work it out at the moment

it's going to be necessary to use negative lookbehind assertion with alternation ... i can't seem to crack it!?


Kind Regards
Duncan

 

[duncdude]

just had to drop the alternation...

s/(?<!table>)(?<!tr>)(?<!td>)\\n/<br>/g;


Kind Regards
Duncan

 

[duncdude]

this should do exactly what you require


Kind Regards
Duncan

 

[nickroper]

Duncan - you are a true star!

Just to make it more interesting, here's another tweak:

If the \n IS preceeded by 'table>', 'tr>' or 'td>' then, instead of just leaving it as a \n character, it would be better to replace it with a pseudo linefeed - say a string such as '{n}' If I leave the actual line feed in place then it messes up the text file.

So, 'foo\n bar' becomes 'foo<BR> bar' but 'foo <table>\n ' becomes 'foo <table>{n}'

Regards,

Nick

 

[duncdude]

probably have to be done with 2 regex's

do this after the previous regex:-

s/\\n/{n}/g;


Kind Regards
Duncan

 

[TrojanWarBlade]

Nick,
It's better to give someone a star than to just tell them they are a start. Better still do both.
I thought it was a good piece of work so Dunc, have a star from me.
;-)


Trojan.

 

[nickroper]

That makes sense.

If you can get to the George Inn at Croscombe - just outside Wells in Somerset - in the next hour or so - then the beers are on me...

If not, then I'll have one for you ;-)

Thanks again for the help.

Nick

 

[nickroper]

Trojan,

OK, I didn't realise I could give a star - but have now done so.

By the way, the offer of the beer was to Duncan, but appears just after your post. However, the more the merrier...

Nick

 

[duncdude]

Thanks again Trojan! ;-)


Kind Regards
Duncan

 

[duncdude]

Thanks for the offer of the beer Nick.... however, I think i might get a few points on my licence getting there in that time!!! I hope your are sorted now.


Kind Regards
Duncan

 

[TrojanWarBlade]

Surely it's worth a try though Duncs?
Good excuse for a sunday blast on the bike!
;-)


Trojan.

 

[duncdude]

a pic of the bike...

Kind Regards
Duncan

 

[nickroper]

Hi Dunc,

Nice looking bike - how far away are you ?

Actually I just tried the regex. I may be something I'm doing, but it doesn't seems to work quite right and none of the \n characters are getting replaced with <BR> - even if there are no tags at all in the original input - they all remain as \n. Can you talk me through the structure of the expression?

s/(?<!table>)(?<!tr>)(?<!td>)\\n/<br>/g;

Cheers,

Nick

 

[duncdude]

Hi Nick

That is weird!?

I am in Oxford - quite close to the city centre

I assume you are prepending with $foo =~ ...

The 3 parenthesized sections are all negative lookbehind assertions

The ?<! does this - they just mean "only replace if none of the 3 parenthesized patterns exist - immediately before the newline"


Kind Regards
Duncan

 

[duncdude]

this works fine for me:-

[b]#!/usr/bin/perl[/b]

$foo = '<table>\n<tr>\n<td>line1\nline2\nline3</td>\n</tr>\n</table>';

print "$foo\n";

$foo =~ s/(?<!table>)(?<!tr>)(?<!td>)\\n/<br>/g;
$foo =~ s/\\n/{n}/g;

print "$foo\n";

Kind Regards
Duncan

 

[nickroper]

Yup, I had created a very similar script, except it includes a header for the browser. I just tried yours and it works fine, but mine doesn't. I think I'm not seeing wood for trees - can you spot where I'm going wrong ?

Cheers,

Nick

=============================================
#!/usr/bin/perl

#initialise a string
$foo = "\n here is\n some text\n <table>\n<tr>";

print "Content-type: text/html\n\n";

#display unmodified string
print "$foo\n";

#display a couple of blank lines in the browser
print "\n<br><br>\n";

#modify the string
#convert non-table linefeeds to <BR>
$foo =~ s/(?<!table>)(?<!tr>)(?<!td>)\\n/<br>/g;
#now convert remaining linefeeds to '{n}'
$foo =~ s/\\n/{n}/g;

#display modified string
print $foo;

==============================================

 

[TrojanWarBlade]

Nice use of shallow depth of field Duncs.
Nice bike too.
I guess you know what R1's and GSX-R1000's look like.
I've had two of each.
Without a bike at the mo though :-(


Trojan.