在Linux下打印两个文件之间的差异

Question

I'm generating snapshot files from my system which contains user rights to directories (short example below):
first_snapshot

RO_sad;user_a  
RW_sad;user_b  
RO_5ka;user_c  
RO_wts;user_c  

After few moments I get another snapshot which contains as follows:
second_snapshot

RW_sad;user_a  
RW_sad;user_b  
RO_5ka;user_c  

Note that rights has changed for user a (from RO to RW) and for user c (rights RO_wts were removed). How can i print differences between those two files? When using

diff first_snapshot second_snapshot

i get something like this:

1,4c1,3
< RO_sad;user_a
< RW_sad;user_b
< RO_5ka;user_c
< RO_wts;user_c
\ No newline at end of file
---
> RW_sad;user_a
> RW_sad;user_b
> RO_5ka;user_c
\ No newline at end of file

Desired file that would help me should look like:

remove RO_sad;user_a
add RW_sad;user_a
remove RO_wts;user_c

Any ideas how could i achieve that? I'd prefer using diff but I need to remove entries that doesnt matter for me, like RW_sad;user_b (exists in both snapshots) or all addons from diff.

Answer 1

You want a "unified diff" like this:

diff -U0 f1.txt f2.txt

It will give you something like this:

--- f1.txt  2014-10-09 19:47:33.000000000 +0800
+++ f2.txt  2014-10-09 19:47:44.000000000 +0800
@@ -1 +1 @@
-RO_sad;user_a
+RW_sad;user_a
@@ -4 +3,0 @@
-RO_wts;user_c

You can then replace the first characters with sed:

diff -U0 f1.txt f2.txt | sed -n -e 's/^+R/add /p' -e 's/^-R/remove /p'

This gives you:

remove O_sad;user_a
add W_sad;user_a
remove O_wts;user_c

Answer 2

Try diff -u

- RO_sad;user_a
+ RW_sad;user_a
- RO_wts;user_c

Answer 3

@John
Not quite. When using diff -U0 i get:

--- first_snapshot  2014-10-09 14:02:27.000000000 +0200
+++ second_snapshot 2014-10-09 14:02:27.000000000 +0200
@@ -1,4 +1,3 @@
-RO_sad;user_a
-RW_sad;user_b
-RO_5ka;user_c
-RO_wts;user_c
\ No newline at end of file
+RW_sad;user_a 
+RW_sad;user_b 
+RO_5ka;user_c
\ No newline at end of file

So i still get this line

+RW_sad;user_b

plus entries

\ No newline at end of file

I've managed to remove duplicated values from both files using AWK

awk 'FNR==NR{a[$1];next};!($1 in a)' $first $second > $third
awk 'FNR==NR{a[$1];next};!($1 in a)' $second $first > $fourth

and then when using diff

diff -uU0 $fourth $third > diff.txt

I get something I could work with:

--- fourth  2014-10-09 14:21:07.000000000 +0200
+++ third   2014-10-09 14:21:07.000000000 +0200
@@ -1,2 +1 @@
-RO_sad;user_a
-RO_wts;user_b
+RW_sad;user_a

Using sed on diff.txt gives me all I need :)