project: avoid unnecessary sync revision enumeration

Skip the detached-history walk when its count will not be printed, ask
rev-list --count for upstream gain, and cap the published-but-unmerged
existence probe at one commit. These preserve the decisions while
reducing traversal and output work.

Bug: 553599402
Change-Id: I30f769c9aa5269428ece15ffcb45c50993d7c4b8
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/634004
Commit-Queue: Gavin Mak <gavinmak@google.com>
Tested-by: Gavin Mak <gavinmak@google.com>
Reviewed-by: Brian Gan <brgan@google.com>

project.py

@@ -2053,13 +2053,13 @@ class Project:
 
     def Sync_LocalHalf(
         self,
-        syncbuf,
-        force_sync=False,
-        force_checkout=False,
-        force_rebase=False,
-        submodules=False,
-        verbose=False,
-    ):
+        syncbuf: Any,
+        force_sync: bool = False,
+        force_checkout: bool = False,
+        force_rebase: bool = False,
+        submodules: bool = False,
+        verbose: bool = False,
+    ) -> None:
         """Perform only the local IO portion of the sync process.
 
         Network access is not required.
@@ -2203,9 +2203,11 @@ class Project:
                     self._CopyAndLinkFiles()
                     return
             else:
-                lost = self._revlist(not_rev(revid), HEAD)
-                if lost and verbose:
-                    syncbuf.info(self, "discarding %d commits", len(lost))
+                if verbose:
+                    lost_output = self._revlist("--count", not_rev(revid), HEAD)
+                    lost = int(lost_output[0]) if lost_output else 0
+                    if lost:
+                        syncbuf.info(self, "discarding %d commits", lost)
 
             try:
                 _checkout()
@@ -2244,7 +2246,8 @@ class Project:
             self._CopyAndLinkFiles()
             return
 
-        upstream_gain = self._revlist(not_rev(HEAD), revid)
+        gain_output = self._revlist("--count", not_rev(HEAD), revid)
+        upstream_gain = int(gain_output[0]) if gain_output else 0
 
         # See if we can perform a fast forward merge.  This can happen if our
         # branch isn't in the exact same state as we last published.
@@ -2258,7 +2261,7 @@ class Project:
             pub = self.WasPublished(branch.name, all_refs)
 
         if pub:
-            not_merged = self._revlist(not_rev(revid), pub)
+            not_merged = self._revlist("-1", not_rev(revid), pub)
             if not_merged:
                 if upstream_gain:
                     if force_rebase:
@@ -2274,7 +2277,7 @@ class Project:
                                 "branch %s is published (but not merged) and "
                                 "is now %d commits behind. Fix this manually "
                                 "or rerun with the --rebase option to force a "
-                                "rebase." % (branch.name, len(upstream_gain)),
+                                "rebase." % (branch.name, upstream_gain),
                                 project=self.name,
                             )
                         )

tests/test_project.py

@@ -2248,6 +2248,97 @@ class StatelessSyncTests(unittest.TestCase):
             )
             proj._CopyAndLinkFiles.assert_called_once_with()
 
+    def test_sync_local_half_skips_unused_detached_history_walk(self) -> None:
+        """Non-verbose detached sync does not enumerate discarded commits."""
+        with utils_for_test.TempGitTree() as tempdir:
+            proj = self._get_project(tempdir)
+            proj._InitWorkTree = mock.MagicMock()
+            proj.CleanPublishedCache = mock.MagicMock()
+            proj.GetRevisionId = mock.MagicMock(return_value="new")
+            proj._Checkout = mock.MagicMock()
+            proj._CopyAndLinkFiles = mock.MagicMock()
+            proj.IsRebaseInProgress = mock.MagicMock(return_value=False)
+            proj.IsCherryPickInProgress = mock.MagicMock(return_value=False)
+            proj._revlist = mock.MagicMock()
+            proj.bare_ref = mock.MagicMock()
+            proj.bare_ref.all = {"HEAD": "old"}
+            proj.bare_ref.head = "old"
+            proj.work_git = mock.MagicMock()
+            proj.work_git.GetHead.return_value = "old"
+
+            syncbuf = project.SyncBuffer(proj.config)
+            proj.Sync_LocalHalf(syncbuf, verbose=False)
+
+            proj._revlist.assert_not_called()
+            proj._Checkout.assert_called_once_with(
+                "new", force_checkout=False, quiet=True
+            )
+
+    def test_sync_local_half_verbose_detached_uses_count(self) -> None:
+        """Verbose detached sync queries commit count with --count."""
+        with utils_for_test.TempGitTree() as tempdir:
+            proj = self._get_project(tempdir)
+            proj._InitWorkTree = mock.MagicMock()
+            proj.CleanPublishedCache = mock.MagicMock()
+            proj.GetRevisionId = mock.MagicMock(return_value="new")
+            proj._Checkout = mock.MagicMock()
+            proj._CopyAndLinkFiles = mock.MagicMock()
+            proj.IsRebaseInProgress = mock.MagicMock(return_value=False)
+            proj.IsCherryPickInProgress = mock.MagicMock(return_value=False)
+            proj._revlist = mock.MagicMock(return_value=["3"])
+            proj.bare_ref = mock.MagicMock()
+            proj.bare_ref.all = {"HEAD": "old"}
+            proj.bare_ref.head = "old"
+            proj.work_git = mock.MagicMock()
+            proj.work_git.GetHead.return_value = "old"
+
+            syncbuf = mock.MagicMock()
+            syncbuf.detach_head = False
+            proj.Sync_LocalHalf(syncbuf, verbose=True)
+
+            proj._revlist.assert_called_once_with("--count", "^new", "HEAD")
+            syncbuf.info.assert_called_once_with(
+                proj, "discarding %d commits", 3
+            )
+
+    def test_sync_local_half_uses_count_and_limit_for_published_branch(
+        self,
+    ) -> None:
+        """Published sync uses --count for gain and -1 for probe."""
+        with utils_for_test.TempGitTree() as tempdir:
+            proj = self._get_project(tempdir)
+            proj._InitWorkTree = mock.MagicMock()
+            proj.CleanPublishedCache = mock.MagicMock()
+            proj.GetRevisionId = mock.MagicMock(return_value="new")
+            proj._CopyAndLinkFiles = mock.MagicMock()
+            proj.IsRebaseInProgress = mock.MagicMock(return_value=False)
+            proj.IsCherryPickInProgress = mock.MagicMock(return_value=False)
+            proj._revlist = mock.MagicMock(side_effect=[["2"], ["pub-sha"]])
+            proj.bare_ref = mock.MagicMock()
+            proj.bare_ref.all = {"refs/heads/topic": "old"}
+            proj.bare_ref.head = "refs/heads/topic"
+            proj.work_git = mock.MagicMock()
+            proj.work_git.GetHead.return_value = "refs/heads/topic"
+            proj.work_git.merge_base.side_effect = project.GitError("diverged")
+            branch = mock.MagicMock()
+            branch.name = "topic"
+            branch.LocalMerge = "refs/remotes/origin/main"
+            proj.GetBranch = mock.MagicMock(return_value=branch)
+            proj.WasPublished = mock.MagicMock(return_value="pub-sha")
+
+            syncbuf = mock.MagicMock()
+            syncbuf.detach_head = False
+            proj.Sync_LocalHalf(syncbuf, force_rebase=False)
+
+            self.assertEqual(
+                proj._revlist.call_args_list,
+                [
+                    mock.call("--count", "^HEAD", "new"),
+                    mock.call("-1", "^new", "pub-sha"),
+                ],
+            )
+            syncbuf.fail.assert_called_once()
+
     def test_sync_network_half_stateless_skips_if_stash(self):
         """Test stateless sync skips if stash exists."""
         with utils_for_test.TempGitTree() as tempdir:
@@ -3351,9 +3442,11 @@ class ReprojectCmdTests(unittest.TestCase):
         def _revlist(*args: Any, **kwargs: Any) -> List[str]:
             if kwargs.get("format"):
                 return list(local_changes)
-            if args[0] == project.not_rev(project.HEAD):
+            if project.not_rev(project.HEAD) in args:
+                if "--count" in args:
+                    return [str(len(upstream_gain))]
                 return list(upstream_gain)
-            if args[1] == self.PUB_ID:
+            if self.PUB_ID in args:
                 return [self.PUB_ID]
             return []