git-summary: add --dedup-by-email to remove duplicate users

bin/git-summary

@@ -3,8 +3,32 @@
 
 cd "$(git root)" || { echo "Can't cd to top level directory";exit 1; }
 
+SUMMARY_BY_LINE=
+DEDUP_BY_EMAIL=
+for arg in "$@"; do
+    case "$arg" in
+        --line)
+            SUMMARY_BY_LINE=1
+            ;;
+        --dedup-by-email)
+            DEDUP_BY_EMAIL=1
+            ;;
+        *)
+            # set the argument back
+            set -- "$@" "$arg"
+            ;;
+    esac
+
+    shift
+done
+
+if [ -n "$DEDUP_BY_EMAIL" ] && [ -n "$SUMMARY_BY_LINE" ]; then
+    >&2 echo "--dedup-by-email used with --line is not supported"
+    exit 1
+fi
+
 commit=""
-test $# -ne 0 && commit=$@
+test $# -ne 0 && commit=$*
 project=${PWD##*/}
 
 #
@@ -31,6 +55,7 @@ active_days() {
 #
 
 commit_count() {
+  # shellcheck disable=SC2086
   git log --oneline $commit | wc -l | tr -d ' '
 }
 
@@ -42,6 +67,46 @@ file_count() {
   git ls-files | wc -l | tr -d ' '
 }
 
+#
+# remove duplicate authors who belong to the same email address
+#
+
+dedup_by_email() {
+    # in:
+    # 27  luo zexuan <[email protected]>
+    #  7  罗泽轩 <[email protected]>
+    # out:
+    # 34 luo zexuan
+    LC_ALL=C awk '
+    {
+        sum += $1
+        if ($NF in emails) {
+            emails[$NF] += $1
+        } else {
+            email = $NF
+            emails[email] = $1
+            # set commits/email to empty
+            $1=$NF=""
+            sub(/^[[:space:]]+/, "", $0)
+            sub(/[[:space:]]+$/, "", $0)
+            name = $0
+            if (name in names) {
+                # when the same name is associated with existed email,
+                # merge the previous email into the later one.
+                emails[email] += emails[names[name]]
+                emails[names[name]] = 0
+            }
+            names[name] = email
+        }
+    }
+    END {
+        for (name in names) {
+            email = names[name]
+            printf "%6d\t%s\n", emails[email], name
+        }
+    }' | sort -rn -k 1
+}
+
 #
 # list authors
 #
@@ -110,7 +175,7 @@ result() {
 echo
 echo " project  : $project"
 
-if test "$1" = "--line"; then
+if [ -n "$SUMMARY_BY_LINE" ]; then
   echo " lines    : $(line_count)"
   echo " authors  :"
   lines | sort | uniq -c | sort -rn | format_authors
@@ -120,8 +185,15 @@ else
   echo " active   :"  $(active_days) days
   echo " commits  :" $(commit_count)
   if test "$commit" = ""; then
-    echo " files    :" $(file_count)
+    echo " files    :" "$(file_count)"
   fi
   echo " authors  : "
-  git shortlog -n -s $commit | format_authors
+  if [ -n "$DEDUP_BY_EMAIL" ]; then
+    # the $commit can be empty
+    # shellcheck disable=SC2086
+    git shortlog -n -s -e $commit | dedup_by_email | format_authors
+  else
+    # shellcheck disable=SC2086
+    git shortlog -n -s $commit | format_authors
+  fi
 fi

etc/git-extras-completion.zsh

@@ -423,6 +423,7 @@ _git-standup() {
 
 _git-summary() {
     _arguments '--line[summarize with lines rather than commits]'
+    _arguments '--dedup-by-email[remove duplicate users by the email address]'
     __gitex_commits
 }
 

man/git-summary.1

@@ -7,7 +7,7 @@
 \fBgit\-summary\fR \- Show repository summary
 .
 .SH "SYNOPSIS"
-\fBgit\-summary\fR [\-\-line] [<commitish>]
+\fBgit\-summary\fR [\-\-line] [\-\-dedup\-by\-email] [<commitish>]
 .
 .SH "DESCRIPTION"
 Shows a summary of the repository\.
@@ -19,6 +19,32 @@ Shows a summary of the repository\.
 Summarize only the range of commits included in the <commitish>\.
 .
 .P
+\-\-dedup\-by\-email
+.
+.P
+Remove duplicate authors who belong to the same email address\. For example,
+.
+.IP "" 4
+.
+.nf
+
+$ git summary
+\.\.\.
+133  TJ Holowaychuk            9\.9%
+115  Tj Holowaychuk            8\.5%
+
+$ git summary \-\-dedup\-by\-email
+\.\.\.
+248  TJ Holowaychuk            18\.4%
+.
+.fi
+.
+.IP "" 0
+.
+.P
+This option can not be used together with \fB\-\-line\fR\.
+.
+.P
 \-\-line
 .
 .P

man/git-summary.md

@@ -3,7 +3,7 @@ git-summary(1) -- Show repository summary
 
 ## SYNOPSIS
 
-`git-summary` [--line] [<commitish>]
+`git-summary` [--line] [--dedup-by-email] [<commitish>]
 
 ## DESCRIPTION
 
@@ -15,6 +15,22 @@ Shows a summary of the repository.
 
   Summarize only the range of commits included in the <commitish>.
 
+  --dedup-by-email
+
+  Remove duplicate authors who belong to the same email address.
+  For example,
+
+    $ git summary
+    ...
+    133  TJ Holowaychuk            9.9%
+    115  Tj Holowaychuk            8.5%
+
+    $ git summary --dedup-by-email
+    ...
+    248  TJ Holowaychuk            18.4%
+
+  This option can not be used together with `--line`.
+
   --line
 
   Summarize with lines other than commits.