[PATCH] shortlog: pass the mailmap into the revision walker

From: CB Bailey <redacted>

shortlog always respects the mailmap in its output. Pass the mailmap
into the revision walker to allow the mailmap to be used with revision
limiting options such as '--author'.

This removes some apparently inconsistent behaviors when using
'--author', such as not finding some or all commits for a given author
which do appear under that author in an unrestricted invocation of
shortlog or commits being summarized under a different author than the
specified author.
---
 builtin/shortlog.c |  2 ++
 t/t4203-mailmap.sh | 28 ++++++++++++++++++++++++++++
 2 files changed, 30 insertions(+)

diff --git a/builtin/shortlog.c b/builtin/shortlog.c
index 88f88e97b2..a6fb00ade8 100644
--- a/builtin/shortlog.c
+++ b/builtin/shortlog.c

@@ -188,6 +188,8 @@ static void get_from_rev(struct rev_info *rev, struct shortlog *log)
 {
 	struct commit *commit;
 
+	rev->mailmap = &log->mailmap;
+
 	if (prepare_revision_walk(rev))
 		die(_("revision walk setup failed"));
 	while ((commit = get_revision(rev)) != NULL)

diff --git a/t/t4203-mailmap.sh b/t/t4203-mailmap.sh
index 43b1522ea2..9bee35b06c 100755
--- a/t/t4203-mailmap.sh
+++ b/t/t4203-mailmap.sh

@@ -383,6 +383,34 @@ test_expect_success 'Shortlog output (complex mapping)' '
 
 '
 
+test_expect_success 'Shortlog output (complex mapping, filtered)' '
+
+	printf "     1\tA U Thor <author@example.com>\n" >expect &&
+
+	git shortlog -es --author="A U Thor" HEAD >actual &&
+	test_cmp expect actual &&
+
+	printf "     1\tCTO <cto@company.xx>\n" >expect &&
+
+	git shortlog -es --author=CTO HEAD >actual &&
+	test_cmp expect actual &&
+
+	printf "     2\tOther Author <other@author.xx>\n" >expect &&
+
+	git shortlog -es --author="Other Author" HEAD >actual &&
+	test_cmp expect actual &&
+
+	printf "     2\tSanta Claus <santa.claus@northpole.xx>\n" >expect &&
+
+	git shortlog -es --author="Santa Claus" HEAD >actual &&
+	test_cmp expect actual &&
+
+	printf "     1\tSome Dude <some@dude.xx>\n" >expect &&
+
+	git shortlog -es --author="Some Dude" HEAD >actual &&
+	test_cmp expect actual
+'
+
 # git log with --pretty format which uses the name and email mailmap placemarkers
 cat >expect <<\EOF
 Author CTO <cto@coompany.xx> maps to CTO <cto@company.xx>

-- 
2.17.0.rc0

From: CB Bailey <redacted>

shortlog always respects the mailmap in its output. Pass the mailmap
into the revision walker to allow the mailmap to be used with revision
limiting options such as '--author'.

This removes some apparently inconsistent behaviors when using
'--author', such as not finding some or all commits for a given author
which do appear under that author in an unrestricted invocation of
shortlog or commits being summarized under a different author than the
specified author.

Signed-off-by: CB Bailey <redacted>
---

Resending with omitted s-o-b.

 builtin/shortlog.c |  2 ++
 t/t4203-mailmap.sh | 28 ++++++++++++++++++++++++++++
 2 files changed, 30 insertions(+)

diff --git a/builtin/shortlog.c b/builtin/shortlog.c
index 88f88e97b2..a6fb00ade8 100644
--- a/builtin/shortlog.c
+++ b/builtin/shortlog.c

@@ -188,6 +188,8 @@ static void get_from_rev(struct rev_info *rev, struct shortlog *log)
 {
 	struct commit *commit;
 
+	rev->mailmap = &log->mailmap;
+
 	if (prepare_revision_walk(rev))
 		die(_("revision walk setup failed"));
 	while ((commit = get_revision(rev)) != NULL)

diff --git a/t/t4203-mailmap.sh b/t/t4203-mailmap.sh
index 43b1522ea2..9bee35b06c 100755
--- a/t/t4203-mailmap.sh
+++ b/t/t4203-mailmap.sh

@@ -383,6 +383,34 @@ test_expect_success 'Shortlog output (complex mapping)' '
 
 '
 
+test_expect_success 'Shortlog output (complex mapping, filtered)' '
+
+	printf "     1\tA U Thor <author@example.com>\n" >expect &&
+
+	git shortlog -es --author="A U Thor" HEAD >actual &&
+	test_cmp expect actual &&
+
+	printf "     1\tCTO <cto@company.xx>\n" >expect &&
+
+	git shortlog -es --author=CTO HEAD >actual &&
+	test_cmp expect actual &&
+
+	printf "     2\tOther Author <other@author.xx>\n" >expect &&
+
+	git shortlog -es --author="Other Author" HEAD >actual &&
+	test_cmp expect actual &&
+
+	printf "     2\tSanta Claus <santa.claus@northpole.xx>\n" >expect &&
+
+	git shortlog -es --author="Santa Claus" HEAD >actual &&
+	test_cmp expect actual &&
+
+	printf "     1\tSome Dude <some@dude.xx>\n" >expect &&
+
+	git shortlog -es --author="Some Dude" HEAD >actual &&
+	test_cmp expect actual
+'
+
 # git log with --pretty format which uses the name and email mailmap placemarkers
 cat >expect <<\EOF
 Author CTO <cto@coompany.xx> maps to CTO <cto@company.xx>

-- 
2.17.0.rc0

On Wed, Dec 12 2018, CB Bailey wrote:

Makes sense. Not saying this is how it should be, just an equivalently
working patch that I came up with on top while poing at this:

    diff --git a/builtin/shortlog.c b/builtin/shortlog.c
    index a6fb00ade8..ad84d70d07 100644
    --- a/builtin/shortlog.c
    +++ b/builtin/shortlog.c
    @@ -188,10 +188,9 @@ static void get_from_rev(struct rev_info *rev, struct shortlog *log)
     {
     	struct commit *commit;

    -	rev->mailmap = &log->mailmap;
    -
     	if (prepare_revision_walk(rev))
     		die(_("revision walk setup failed"));
    +	rev->mailmap = &log->mailmap;
     	while ((commit = get_revision(rev)) != NULL)
     		shortlog_add_commit(log, commit);
     }
    diff --git a/t/t4203-mailmap.sh b/t/t4203-mailmap.sh
    index 9bee35b06c..74a269052d 100755
    --- a/t/t4203-mailmap.sh
    +++ b/t/t4203-mailmap.sh
    @@ -384,17 +384,6 @@ test_expect_success 'Shortlog output (complex mapping)' '
     '

     test_expect_success 'Shortlog output (complex mapping, filtered)' '
    -
    -	printf "     1\tA U Thor [off-list ref]\n" >expect &&
    -
    -	git shortlog -es --author="A U Thor" HEAD >actual &&
    -	test_cmp expect actual &&
    -
    -	printf "     1\tCTO [off-list ref]\n" >expect &&
    -
    -	git shortlog -es --author=CTO HEAD >actual &&
    -	test_cmp expect actual &&
    -
     	printf "     2\tOther Author [off-list ref]\n" >expect &&

     	git shortlog -es --author="Other Author" HEAD >actual &&

I.e. we just need the assignment after prepare_revision_walk() and the
first 2x tests were things that passed before this change.

So that's not a "let's squash that on top" but "I was poking at this and
here's stuff that I fiddled with or surprised me slightly".