From ff0a73f5ad300110defffdc2593af129911c0644 Mon Sep 17 00:00:00 2001
From: Tim Hockin <thockin@google.com>
Date: Mon, 15 Aug 2022 10:19:29 -0700
Subject: [PATCH] Support shallow sync when the rev is not in-range

If I ask for `--depth 1` and a branch, it's fine.

If I ask for `--depth 1` and a tag, and that tag is not within 1 commit
of the branch that was cloned, it will give an error.  Oddly, if the
initial `clone` was OK, and subsequent syncs drift, it is OK, because of
how we `fetch`.  But if it is too far away at the beginning, kaboom.

This betrays that the current model of `--branch` and `--rev` is really
broken, and should be revamped.  For now, I did something simple - if
the rev can't be found, try a fetch.  A "real" fix is more involved.

Also add tests.
---
 cmd/git-sync/main.go | 42 ++++++++++++++++++++++++++++++++++-
 test_e2e.sh          | 53 ++++++++++++++++++++++++++++++++++++++++++--
 2 files changed, 92 insertions(+), 3 deletions(-)

diff --git a/cmd/git-sync/main.go b/cmd/git-sync/main.go
index ea5d376..b1e74c0 100644
--- a/cmd/git-sync/main.go
+++ b/cmd/git-sync/main.go
@@ -1345,7 +1345,7 @@ func (git *repoSync) SyncRepo(ctx context.Context, refreshCreds func(context.Con
 		if err != nil {
 			return false, "", err
 		}
-		hash, err = git.LocalHashForRev(ctx, git.rev)
+		hash, err = git.ensureLocalHashForRev(ctx, git.rev)
 		if err != nil {
 			return false, "", err
 		}
@@ -1368,6 +1368,46 @@ func (git *repoSync) SyncRepo(ctx context.Context, refreshCreds func(context.Con
 	return true, hash, git.AddWorktreeAndSwap(ctx, hash)
 }
 
+func (git *repoSync) ensureLocalHashForRev(ctx context.Context, rev string) (string, error) {
+	hash, err := git.LocalHashForRev(ctx, rev)
+	if err == nil {
+		return hash, nil
+	}
+
+	// git might return either error, based on flags used internal to
+	// LocalHashForRev, so we will consider either one to be a "rev not found".
+	if es := err.Error(); !stringContainsOneOf(es, "unknown revision", "bad revision") {
+		return "", err
+	}
+
+	// The rev was not found, try to fetch it.
+	git.log.V(1).Info("rev was not found, trying fetch", "rev", git.rev)
+	args := []string{"fetch", "-f", "--tags"}
+	if git.depth != 0 {
+		args = append(args, "--depth", strconv.Itoa(git.depth))
+	}
+	args = append(args, git.repo, "--")
+	if _, err := git.run.Run(ctx, git.root, nil, git.cmd, args...); err != nil {
+		return "", err
+	}
+
+	// Try again.
+	hash, err = git.LocalHashForRev(ctx, git.rev)
+	if err != nil {
+		return "", err
+	}
+	return hash, nil
+}
+
+func stringContainsOneOf(s string, matches ...string) bool {
+	for _, m := range matches {
+		if strings.Contains(s, m) {
+			return true
+		}
+	}
+	return false
+}
+
 // GetRevs returns the local and upstream hashes for rev.
 func (git *repoSync) GetRevs(ctx context.Context) (string, string, error) {
 	// Ask git what the exact hash is for rev.
diff --git a/test_e2e.sh b/test_e2e.sh
index e24f105..9dcd8d8 100755
--- a/test_e2e.sh
+++ b/test_e2e.sh
@@ -934,10 +934,10 @@ function e2e::sync_slow_git_long_timeout() {
 ##############################################
 # Test depth syncing
 ##############################################
-function e2e::sync_depth_shallow() {
+function e2e::sync_branch_depth_shallow() {
     # First sync
-    echo "$FUNCNAME 1" > "$REPO"/file
     expected_depth="1"
+    echo "$FUNCNAME 1" > "$REPO"/file
     git -C "$REPO" commit -qam "$FUNCNAME 1"
 
     GIT_SYNC \
@@ -981,6 +981,55 @@ function e2e::sync_depth_shallow() {
     fi
 }
 
+##############################################
+# Test depth syncing with a tag not within depth
+##############################################
+function e2e::sync_tag_depth_shallow_out_of_range() {
+    TAG="e2e-tag"
+    expected_depth="1"
+
+    # First commits, tag is not within --depth
+    echo "$FUNCNAME 1" > "$REPO"/file
+    git -C "$REPO" commit -qam "$FUNCNAME 1"
+    git -C "$REPO" tag -af "$TAG" -m "$FUNCNAME 1" >/dev/null
+    echo "$FUNCNAME 2" > "$REPO"/file
+    git -C "$REPO" commit -qam "$FUNCNAME 2"
+
+    GIT_SYNC \
+        --period=100ms \
+        --repo="file://$REPO" \
+        --branch="$MAIN_BRANCH" \
+        --rev="$TAG" \
+        --depth="$expected_depth" \
+        --root="$ROOT" \
+        --link="link" \
+        >> "$1" 2>&1 &
+    sleep 3
+    assert_link_exists "$ROOT"/link
+    assert_file_exists "$ROOT"/link/file
+    assert_file_eq "$ROOT"/link/file "$FUNCNAME 1"
+    depth=$(GIT_DIR="$ROOT"/link/.git git log | grep commit | wc -l)
+    if [[ $expected_depth != $depth ]]; then
+        fail "initial depth mismatch expected=$expected_depth actual=$depth"
+    fi
+
+    # Make 2 new commits, and tag the older one, so it's not within --depth
+    echo "$FUNCNAME 3" > "$REPO"/file
+    git -C "$REPO" commit -qam "$FUNCNAME 3"
+    SHA=$(git -C "$REPO" rev-parse HEAD)
+    echo "$FUNCNAME 4" > "$REPO"/file
+    git -C "$REPO" commit -qam "$FUNCNAME 4"
+    git -C "$REPO" tag -af "$TAG" -m "$FUNCNAME 3" "$SHA" >/dev/null
+    sleep 3
+    assert_link_exists "$ROOT"/link
+    assert_file_exists "$ROOT"/link/file
+    assert_file_eq "$ROOT"/link/file "$FUNCNAME 3"
+    depth=$(GIT_DIR="$ROOT"/link/.git git log | grep commit | wc -l)
+    if [[ $expected_depth != $depth ]]; then
+        fail "forward depth mismatch expected=$expected_depth actual=$depth"
+    fi
+}
+
 ##############################################
 # Test fetch skipping commit
 ##############################################