hi,

i have a script which pipes the output of a diff -rq command into a separate file/ it would read something like this:


Files mod1/lala/xml/test1.txt and mod2/lala/xml/test1.txt differ
Only in mod2/lala/xml: test2.txt

What i need to do is to parse this file so i end up with just a list of directories in which the differences occur:

i.e.

mod2/lala/xml/

any ideas?

You can try something like that :

awk '
   /^Files/ { sub(/\/[^\/]*$/, "", $(NF-1)); print $(NF-1) }
   /^Only/  { sub(/\/[^\/]*$/, "", $3     ); print $3      }
    ' diff_file | sort -u

thanks jean-pierre, that works great!

oops, spoke to soon.

the problem i'm having is that when files are only in one directory, i'm losing part of the path on output.

i.e. if test1.txt is in new, but not in old, i'm getting

mod2/lala

instead of:

mod2/lala/xml

awk '
   /^Files/ { sub(/\/[^\/]*$/, "", $(NF-1)); print $(NF-1) }
   /^Only/  { sub(/:$/,        "", $3     ); print $3      }
    ' diff_file | sort -u

you're a life saver...thanks.