info: Parallelize project data gathering for JSON output

https://gerrit-review.googlesource.com/c/git-repo/+/581921 parallelized
`repo info` for text output. This commit does the same for JSON output
format.

Benchmarked `repo info --format=json` on an Android workspace with ~3k
projects (N=3):
- Before (sequential): 1m 30s average
- After (parallelized): 46s average (~2x speedup)

Verified that the JSON output is identical before and after.

Bug: 526685287
Change-Id: If573223aba584f8b932f87d29e34ed565c5c930a
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/601861
Commit-Queue: Gavin Mak <gavinmak@google.com>
Reviewed-by: Brian Gan <brgan@google.com>
Tested-by: Gavin Mak <gavinmak@google.com>

Author: gavinmak

subcmds/info.py

@@ -191,7 +191,8 @@ def _getSummaryData(self) -> Dict[str, Any]:
             "superproject_revision": srev,
         }
 
-    def _getProjectData(self, project) -> Dict[str, Any]:
+    @classmethod
+    def _getProjectData(cls, project) -> Dict[str, Any]:
         """Gather project data as a dict."""
         data = {
             "name": project.name,
@@ -206,6 +207,12 @@ def _getProjectData(self, project) -> Dict[str, Any]:
             data["current_branch"] = currentBranch
         return data
 
+    @classmethod
+    def _ProjectDataHelper(cls, project_idx: int) -> Dict[str, Any]:
+        """Helper to get project data in parallel."""
+        project = cls.get_parallel_context()["projects"][project_idx]
+        return cls._getProjectData(project)
+
     def _ExecuteJson(self, opt, args) -> None:
         """Output info as JSON."""
         result = {}
@@ -215,7 +222,22 @@ def _ExecuteJson(self, opt, args) -> None:
             projs = self.GetProjects(
                 args, all_manifests=not opt.this_manifest_only
             )
-            result["projects"] = [self._getProjectData(p) for p in projs]
+            project_data = []
+
+            def _ProcessResults(_pool, _output, results):
+                project_data.extend(results)
+
+            with self.ParallelContext():
+                self.get_parallel_context()["projects"] = projs
+                self.ExecuteInParallel(
+                    opt.jobs,
+                    self._ProjectDataHelper,
+                    range(len(projs)),
+                    callback=_ProcessResults,
+                    ordered=True,
+                    chunksize=1,
+                )
+            result["projects"] = project_data
 
         json_settings = {
             # JSON style guide says Unicode characters are fully allowed.

tests/test_subcmds_info.py

@@ -223,3 +223,33 @@ def test_get_project_data_uses_head_revision() -> None:
     project.GetHeadRevisionId.return_value = None
     data = cmd._getProjectData(project)
     assert data["current_revision"] == "manifest_sha_54321"
+
+
+def test_json_with_projects(capsys) -> None:
+    """--format=json should emit project data."""
+    cmd = _get_cmd()
+    opts, args = cmd.OptionParser.parse_args(["--format=json"])
+    opts.jobs = 1  # To avoid multiprocessing pickle issues with mocks
+
+    project = mock.MagicMock()
+    project.name = "foo"
+    project.worktree = "/path/to/foo"
+    project.revisionExpr = "refs/heads/main"
+    project.GetBranches.return_value = {"branch1": mock.MagicMock()}
+    project.GetHeadRevisionId.return_value = "head_sha_12345"
+    project.CurrentBranch = "branch1"
+
+    cmd.GetProjects = mock.MagicMock(return_value=[project])
+
+    cmd.Execute(opts, args)
+
+    data = json.loads(capsys.readouterr().out)
+    assert "projects" in data
+    assert len(data["projects"]) == 1
+    project_data = data["projects"][0]
+    assert project_data["name"] == "foo"
+    assert project_data["mount_path"] == "/path/to/foo"
+    assert project_data["current_revision"] == "head_sha_12345"
+    assert project_data["manifest_revision"] == "refs/heads/main"
+    assert project_data["local_branches"] == ["branch1"]
+    assert project_data["current_branch"] == "branch1"