CODEWITHSUNDEEP
gitshell
Vedant Agarwala
Vedant Agarwala

Reputation: 18819

How to aggregate code changes between commits

I want to see something like an aggregated git log --shortstat --oneline.

Instead of the following,

2fd8b62 quote sending successfully
 5 files changed, 26 insertions(+), 40 deletions(-)
5bc977e Hackedup old (redundant) code so that project compiles
 14 files changed, 90 insertions(+), 80 deletions(-)

I want something like

 19 files changed, 116 insertions, 120 deletions.

I understand this will contain a lot of redundant data (as files changed may be common, etc.) But I want to track the work done in a day (say) or any time period.

I cannot figure out a simple way to parse the output generated by

git log --shortstat --oneline commit1...commit2

To find the total change, I can do 

git diff --shortstat --oneline commit1...commit2

but I do not want this. I am considering each commit to be a valid change, even if it was partially undone in a later commit.

Upvotes: 2

Views: 1021

Answers (1)

VonC
VonC

Reputation: 1324547

This gist gives a possible solution:

git log --shortstat  | grep "files changed" | gawk '{files+=$1; inserted+=$4; deleted+=$6} END {print "files changed", files, "lines inserted:", inserted, "lines deleted:", deleted}'

On multiple line, for readability:

git log --shortstat  | grep "files changed" | \
gawk '{files+=$1; inserted+=$4; deleted+=$6} END \ 
{print "files changed", files, "lines inserted:", inserted, "lines deleted:", deleted}'

I just tried it on the git repo, for the 10 past days of commits (it works even on Windows):

c:\prgs\vonc\git\git>
git log --oneline --shortstat --since="10 days ago" | grep "files changed" | gawk '{files+=$1; inserted+=$4; deleted+=$6} END {print "files changed", files, "lines inserted:", inserted, "lines deleted:", deleted}'
files changed 47 lines inserted: 397 lines deleted: 30

That same technique (using gawk) lets you aggregate other data, like the number of commits per authors:

git log --pretty=format:%an | gawk '{ ++c[$0]; } END { for(cc in c) printf "%5d %s\n",c[cc],cc; }'| gsort -r

git log --pretty=format:%an | \
gawk '{ ++c[$0]; } END \
{ for(cc in c) printf "%5d %s\n",c[cc],cc; }'| gsort

Again, on the git repo, on Windows, it does work:

c:\prgs\vonc\git\git>
git log --pretty=format:%an --since="10 days ago" | gawk '{ ++c[$0]; } END { for(cc in c) printf "%5d %s\n",c[cc],cc; }'| gsort
   23 Junio C Hamano
     8 Jeff King
     3 Johannes Schindelin

Note that it is a bit overkill, in that it includes the merge commits.
Simpler would be

c:\prgs\vonc\git\git>
git shortlog -sn --no-merges --since="10 days ago"
     8  Jeff King
     6  Junio C Hamano
     3  Johannes Schindelin

Upvotes: 2

Related Questions