Fix get-authors script to not extract email addresses, canonicalize names, add .mailmap

1 files changed, 9 insertions, 2 deletions

diff --git a/contrib/get-authors b/contrib/get-authors
index 118ca4466..0fe71dca3 100755
--- a/contrib/get-authors
+++ b/contrib/get-authors
@@ -11,6 +11,13 @@
 
 gituser="`git config user.name` <`git config user.email`>"
 
+if [ "$1" = "--email" ]; then
+  strip_email=cat
+  shift
+else
+  strip_email="sed 's, *<[^>]*@[^>]*>,,g'"
+fi
+
 while [ $# -gt 0 ]; do
   f=$1
   shift
@@ -18,7 +25,7 @@ while [ $# -gt 0 ]; do
   major_contributors=
   minor_contributors=
   total_lines=`wc -l "$f" | awk '{print$1}'`
-  original_author=`git log --pretty="format:%aN <%aE>" "$f" | tail -1`
+  original_author=`git log --use-mailmap --pretty="format:%aN <%aE>" "$f" | tail -1`
   git blame --incremental "$f" | gawk '/^[0-9a-f]+ [0-9]+ [0-9]+ [0-9]+$/ {nl=$4;} /^author / {$1=""; author=$0;} /^author-mail / {mail=$2} /^filename / {while(nl--) {print author,mail}}' | sed "s,Not Committed Yet <not.committed.yet>,$gituser," | sort | uniq -c | sort -n |
     ( while read lines author; do
         pct=$((100*$lines/$total_lines))
@@ -32,5 +39,5 @@ while [ $# -gt 0 ]; do
       done; \
       echo "$original_author"
       echo "${major_contributors:-none}"
-      echo "${minor_contributors:-none}" )
+      echo "${minor_contributors:-none}" ) | eval "$strip_email"
 done