How can I show the number of positions lines have moved in diff output?

Question

top1

1,facebook.com
2,google.com
3,youtube.com
4,yahoo.com
5,wikipedia.org
6,baidu.com
7,live.com
8,amazon.com
9,qq.com
10,twitter.com

top2

1,facebook.com
2,google.com
3,youtube.com
4,yahoo.com
5,baidu.com
6,wikipedia.org
7,live.com
8,qq.com
9,amazon.com
10,twitter.com

diff -u  /Users/stofke/Desktop/top2a.csv /Users/stofke/Desktop/top1a.csv |
  sed -n '1,2d;/^[-+]/p' |
  sort --field-separator=',' --key=2 

+9,amazon.com
-8,amazon.com
+5,baidu.com
-6,baidu.com
+8,qq.com
-9,qq.com
+6,wikipedia.org
-5,wikipedia.org

But I actually need this:

 +1, amazon.com
 -1  baidu.com
 -1  qq.com
 +1  wikipedia.org

I don't know how to do that last step

(I would also need to calculate the % wise decline or raise)

Adding | paste -d, - -

gets me this

+9,amazon.com,-8,amazon.com
+5,baidu.com,-6,baidu.com
+8,qq.com,-9,qq.com
+6,wikipedia.org,-5,wikipedia.org

Answer 1

You could also consider an approach using join since you want sorted output in the end. Like:

join -t, -j2 <(sort -t, -k2 file1) <(sort -t, -k2 file2) |
   awk -F, '{n=$3-$2; if (n) print n, $1}'

Answer 2

Using awk

awk -F, '
    NR == FNR {n[$2] = $1; next}
    {n[$2] -= $1}
    END {for (k in n) if (n[k] != 0) printf " %2d  %s\n",n[k],k}
' top1 top2

outputs

 -1  amazon.com
  1  qq.com
 -1  wikipedia.org
  1  baidu.com

order not guaranteed.