Shell Programming and Scripting

This might fail on large files...

Code:

awk 'gsub (/{[^}]*} /,"")' RS= file

It will also remove empty lines and paragraphs that do not contain curly brackets..

This one intrigued me because I'm more of an awk person than a perl person...

Code:

awk '{print $2} NR==1' RS={ FS="} " ORS= file

I wasn't familiar with RS, FS and ORS as built in variables and now realize that's some very powerful stuff there. (by the way, I don't know how I made it so many years never knowing about FNR... argh!!).

To explain your example, RS is the record separator, which is how a line is defined. Normally I think the default RS is "\n" i.e. the newline. FS is the field separator and normally that's whitespace. ORS is output record separator.

In other words, define a new line by finding {. Break the "line" into two fields separated by "} " (note space after brackets... So "my big}date yesterday" is not a match). The first field is what's inside the {}, the second is what comes after. Print the second field (printing the first field would only show what was inside the brackets). Make the output record separator nothing (otherwise it would insert line breaks any time there's a {}. One downside of this though is that if the line break is within the {}, that break disappears. So

Yesterday we ate {some
giant} cake and
drank tea

becomes

Yesterday we ate cake and
drank tea

which is not the end of the world I suppose.

Great answer, Scrutinizer!

Thanks climatron . See if this works better:

Code:

awk 'NR>1{gsub(/[^\n]/,x,$1)}1' RS={ FS="} ?" OFS= ORS= file

Code:

#!/usr/bin/env perl

open( $fh, "<", "yourfile") or die "Cannot open file: $!\n";

my $content = do { local $/; <$fh> };		# slurp in whole file

$content =~ s,{.*?},,smg;			        # remove from { to } multiline match

print $content."\n";

And just to be a completist on this issue, building on scrutinizer's suggestion, here's the code that would do everything I asked for in the original message

awk 'NR>1{gsub(/[^\n]/,x,$1)}1' RS={ FS="} ?" OFS= ORS= filename | awk 'NF > 0 { gsub(/^[ \t]+|[ \t]+$/, ""); print }'

It gets rid of all text within brackets AND trims whitespace and removes empty lines.