Help with diff output

Shell Programming and Scripting
1

I am running diff between two directories dir1 and dir2.

diff --exclude --recursive --brief -b dir1 dir2

The output of the above command is

Files dir1/java/abc/bcd/abc9991.java and dir2/java/abc/bcd/abc9991.java differ
Files dir1/java/abc/bcd/abc9933.java and dir2/java/abc/bcd/abc9933.java differ
Only in dir2/ora/dpc: g025_abc.dpc
Only in dir2/ora/dpc: gc_cvb_jrp_1650_abc.dpc
Only in dir2/ora/dpc: gc_k_jrp_502901690_abc.dpc
Files dir1/ora/ps/abc_bcda640921_bcd.sp and dir2/ora/ps/abc_bcda640921_bcd.sp differ
Files dir1/ora/ps/cvg_fgh640931_drv_msd.sp and dir2/ora/sp/cvg_fgh640931_drv_msd.sp differ

Want to extract file names from the diff output based on file extensions(have over 6 extensions in the subdirectories .java,.dpc,.cvg, and so on)

and delete all the files in dir2 except these in the diff output.

example: from the above diff output I want to extract files

abc9991.java
gc_cvb_jrp_1650_abc.dpc
gc_k_jrp_502901690_abc.dpc
abc_bcda640921_bcd.sp
cvg_fgh640931_drv_msd.sp   from dir2 and delete rest of the files from dir2.

Tried using the cut command but too many sub directories inside and each time the position has to change.

2

Why are abc9933.java and g025_abc.dpc excluded?

3

example: from the above diff output I want to extract files

abc9991.java
abc9933.java
g025_abc.dpc
gc_cvb_jrp_1650_abc.dpc
gc_k_jrp_502901690_abc.dpc
abc_bcda640921_bcd.sp
cvg_fgh640931_drv_msd.sp

from dir2 and delete rest of the files.
.

---------- Post updated at 04:35 PM ---------- Previous update was at 04:34 PM ----------

@CarloM: My bad I missed them

4

The aim being to delete duplicate files from dir2 and sub-directories?

5

Correct. I only need newer and changed files in dir2.

6

If your diff supports -s (report identical files):

diff -s --exclude --recursive --brief -b dir1 dir2 | awk '/identical$/ {print $4}' | xargs echo rm

echo is just for error checking, remove it if the output looks right.

EDIT: If not then you could do something a bit more complicated:

diff <(diff --exclude --recursive --brief -b dir1 dir2 | awk '/^Only/ {printf "%s/%s\n", $3, $5} /differ$/ {print $4}' FS=" |:" | sort) <(find dir2 -type f|sort) | awk '{print $2}' | xargs echo rm
7

Tried this initially. Did not work. The output was rm.

diff -s --exclude --recursive --brief -b dir1 dir2 | awk '/identical$/ {print $4}' | xargs echo rm

Then tried this it showed everything in dir2 as different and when I put the rm command at the end it deleted all the files in dir2.

diff <(diff --exclude --recursive --brief -b dir1 dir2 | awk '/^Only/ {printf "%s/%s\n", $3, $5} /differ$/ {print $4}' FS=" |:" | sort) <(find dir2 -type f|sort) | awk '{print $2}' | xargs echo rm
8

What output do you get from the 'same files' diff?

diff -s --exclude --recursive --brief -b dir1 dir2 | head -10
9 
diff -s --exclude --recursive --brief -b dir1 dir2 | head -10

Common subdirectories: dir1/aix and dir2/aix
Common subdirectories: dir1/envsql and dir2/envsql
Common subdirectories: dir1/java and dir2/java
Common subdirectories: dir1/ora and dir2/ora
10

Apparently there's nothing to delete, so xargs is getting a blank input (which is something I didn't consider).

Change the xargs call to:

xargs -r echo rm

and it should get rid of the error for this case.

11

I see no files being deleted. There are many files that are similar in dir1 and dir2 and none of them got deleted.

Attached is the zip I created for your reference. It has two directories dir_1 and dir_2. If a compare is run between dir_1 and dir_2 , the only files that should be present in dir_2 after the compare should be the files inside dir_2/ora/seq/

12

What is the --exclude for? With no parameters that appears to be excluding everything, at least on my diff (GNU diffutils 3.3).

Without that it works as expected:

$ diff -s --recursive -b dir_1 dir_2
Files dir_1/java/abc/test/Constants.java and dir_2/java/abc/test/Constants.java are identical
Files dir_1/java/abc/test/constants_1.txt and dir_2/java/abc/test/constants_1.txt are identical
Files dir_1/java/abc/test/delete.java and dir_2/java/abc/test/delete.java are identical
Files dir_1/java/abc/test/test.java and dir_2/java/abc/test/test.java are identical
Only in dir_1/ora: java
Only in dir_2/ora: seq
$ diff -s --recursive -b dir_1 dir_2 | awk '/identical$/ {print $4}' | xargs echo rm
rm dir_2/java/abc/test/Constants.java dir_2/java/abc/test/constants_1.txt dir_2/java/abc/test/delete.java dir_2/java/abc/test/test.java