From 08b7f8028533c257c38e3033df1594b88bfd479d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tina=20M=C3=BCller?= <tina.mueller@suse.com>
Date: Wed, 31 Jul 2024 14:00:47 +0200
Subject: [PATCH] Support git_auto_clone for empty CASDIR/NEEDLES_DIR

Issue: https://progress.opensuse.org/issues/162632
---
 lib/OpenQA/Task/Git/Clone.pm | 68 ++++++++++++++++++++++++++----------
 lib/OpenQA/Utils.pm          |  4 +++
 2 files changed, 54 insertions(+), 18 deletions(-)

diff --git a/lib/OpenQA/Task/Git/Clone.pm b/lib/OpenQA/Task/Git/Clone.pm
index 9fc27698859..2b77844bb82 100644
--- a/lib/OpenQA/Task/Git/Clone.pm
+++ b/lib/OpenQA/Task/Git/Clone.pm
@@ -83,35 +83,67 @@ sub _git_fetch ($path, $branch_arg) {
     die "Failed to fetch from '$branch_arg': $r->{stderr}" unless $r->{status};
 }
 
+sub _git_gc ($path) {
+    my $r = run_cmd_with_log_return_error(_ssh_git_cmd(['-C', $path, 'gc', '--auto', '--quiet']));
+    die "Failed to gc '$path': $r->{stderr}" unless $r->{status};
+}
+
+sub _git_diff_index ($path) {
+    my $r = run_cmd_with_log_return_error(_ssh_git_cmd(['-C', $path, 'diff-index', 'HEAD', '--exit-code']));
+    unless ($r->{status}) {
+        # dirty git status
+        return "diff-index: $r->{stdout}" if $r->{return_code} >> 8 == 1;
+        die "Failed to diff-index '$path': $r->{stderr}";
+    }
+    return 0;
+}
+
 sub _git_reset_hard ($path, $branch) {
     my $r = run_cmd_with_log_return_error(['git', '-C', $path, 'reset', '--hard', "origin/$branch"]);
     die "Failed to reset to 'origin/$branch': $r->{stderr}" unless $r->{status};
 }
 
 sub _git_clone ($job, $ctx, $path, $url) {
-    $ctx->debug(qq{Updating $path to $url});
-    $url = Mojo::URL->new($url);
-    my $requested_branch = $url->fragment;
-    $url->fragment(undef);
-    my $remote_default_branch = _get_remote_default_branch($url);
-    $requested_branch ||= $remote_default_branch;
-    $ctx->debug(qq{Remote default branch $remote_default_branch});
-    die "Unable to detect remote default branch for '$url'" unless $remote_default_branch;
-
-    if (!-d $path) {
-        _git_clone_url_to_path($url, $path);
-        # update local branch to latest remote branch version
-        _git_fetch($path, "$requested_branch:$requested_branch")
-          if ($requested_branch ne $remote_default_branch);
+    my $requested_branch = '';
+    if ($url) {
+        # given a git url, do initial clone
+        $ctx->debug(qq{Updating $path to $url});
+        $url = Mojo::URL->new($url);
+        $requested_branch = $url->fragment;
+        $url->fragment(undef);
+        my $remote_default_branch = _get_remote_default_branch($url);
+        $requested_branch ||= $remote_default_branch;
+        $ctx->debug(qq{Remote default branch $remote_default_branch});
+        die "Unable to detect remote default branch for '$url'" unless $remote_default_branch;
+
+        if (!-d $path) {
+            _git_clone_url_to_path($url, $path);
+            # update local branch to latest remote branch version
+            _git_fetch($path, "$requested_branch:$requested_branch")
+              if ($requested_branch ne $remote_default_branch);
+        }
+
+        my $origin_url = _git_get_origin_url($path);
+        if ($url ne $origin_url) {
+            $ctx->warn("Local checkout at $path has origin $origin_url but requesting to clone from $url");
+            return;
+        }
     }
-
-    my $origin_url = _git_get_origin_url($path);
-    if ($url ne $origin_url) {
-        $ctx->warn("Local checkout at $path has origin $origin_url but requesting to clone from $url");
+    else {
+        # look for existing clone in the given path
+        my $origin_url = _git_get_origin_url($path);
+        $ctx->debug(qq{Updating $path to $origin_url});
+        $requested_branch = _get_remote_default_branch($origin_url);
+    }
+    if (my $is_dirty = _git_diff_index($path)) {
+        $ctx->warn("Clone at '$path' has dirty status, not updating: $is_dirty");
         return;
     }
 
+    _git_gc($path);
     my $current_branch = _get_current_branch($path);
+    # TODO deal with conflicts because of save_needle
+    # see fetchneedles
     if ($requested_branch eq $current_branch) {
         # updating default branch (including checkout)
         _git_fetch($path, $requested_branch);
diff --git a/lib/OpenQA/Utils.pm b/lib/OpenQA/Utils.pm
index e79a5727877..f6dc2aea601 100644
--- a/lib/OpenQA/Utils.pm
+++ b/lib/OpenQA/Utils.pm
@@ -603,6 +603,10 @@ sub create_downloads_list ($job_settings) {
 
 sub create_git_clone_list ($job_settings, $clones = {}) {
     my $distri = $job_settings->{DISTRI};
+    # Potential existing git clones to update
+    not $job_settings->{CASEDIR} and $clones->{testcasedir($distri)} = undef;
+    not $job_settings->{NEEDLES_DIR} and $clones->{needledir($distri)} = undef;
+
     my $case_url = Mojo::URL->new($job_settings->{CASEDIR} // '');
     my $needles_url = Mojo::URL->new($job_settings->{NEEDLES_DIR} // '');
     if ($case_url->scheme) {