feat(fetch): expose fetch-mode on git fetch tasks

Lets a kind opt into fetch-content's init_and_fetch mode for repos whose git
server cannot serve a full clone. The flag and its digest entry are only added
when the mode is not the default, so existing fetch tasks keep their command
and their cached artifacts.

Author: JohanLorenzo

src/taskgraph/transforms/fetch.py

@@ -281,6 +281,11 @@ class GitFetchSchema(Schema, forbid_unknown_fields=False, kw_only=True):
     include_dot_git: Optional[bool] = None
     artifact_name: Optional[str] = None
     path_prefix: Optional[str] = None
+    # How to populate the checkout. "clone" (the default) clones the repo,
+    # "fetch" fetches only `revision` and leaves no branches or tags,
+    # for git servers that cannot serve a full clone. See
+    # https://bugzilla.mozilla.org/show_bug.cgi?id=2047876
+    fetch_mode: Optional[Literal["clone", "fetch"]] = None
     # ssh-key is a taskcluster secret path (e.g. project/civet/github-deploy-key)
     # In the secret dictionary, the key should be specified as
     #  "ssh_privkey": "-----BEGIN OPENSSH PRIVATE KEY-----\nkfksnb3jc..."
@@ -320,6 +325,11 @@ def create_git_fetch_task(config, name, fetch):
         args.append("--include-dot-git")
         digest_data.append(".git")
 
+    fetch_mode = fetch.get("fetch-mode")
+    if fetch_mode and fetch_mode != "clone":
+        args.extend(["--fetch-mode", fetch_mode])
+        digest_data.append(f"fetch-mode={fetch_mode}")
+
     return {
         "command": args,
         "artifact_name": artifact_name,

test/test_transforms_fetch.py

@@ -31,6 +31,34 @@ def assert_static_url(task):
     assert task["attributes"]["fetch-artifact"] == "public/resource"
 
 
+GIT_REVISION = "0123456789abcdef0123456789abcdef01234567"
+
+
+def expected_git_command(*extra_args):
+    return [
+        "fetch-content",
+        "git-checkout-archive",
+        "--path-prefix",
+        "repo",
+        "https://example.com/repo",
+        GIT_REVISION,
+        "/builds/worker/artifacts/repo.tar.zst",
+        *extra_args,
+    ]
+
+
+def assert_git(task):
+    assert task["run"]["command"] == expected_git_command()
+
+
+def assert_git_fetch_mode_clone(task):
+    assert task["run"]["command"] == expected_git_command()
+
+
+def assert_git_fetch_mode_fetch(task):
+    assert task["run"]["command"] == expected_git_command("--fetch-mode", "fetch")
+
+
 @pytest.mark.parametrize(
     "task_input",
     (
@@ -48,6 +76,38 @@ def assert_static_url(task):
             },
             id="static-url",
         ),
+        pytest.param(
+            {
+                "fetch": {
+                    "type": "git",
+                    "repo": "https://example.com/repo",
+                    "revision": GIT_REVISION,
+                },
+            },
+            id="git",
+        ),
+        pytest.param(
+            {
+                "fetch": {
+                    "type": "git",
+                    "repo": "https://example.com/repo",
+                    "revision": GIT_REVISION,
+                    "fetch-mode": "clone",
+                },
+            },
+            id="git-fetch-mode-clone",
+        ),
+        pytest.param(
+            {
+                "fetch": {
+                    "type": "git",
+                    "repo": "https://example.com/repo",
+                    "revision": GIT_REVISION,
+                    "fetch-mode": "fetch",
+                },
+            },
+            id="git-fetch-mode-fetch",
+        ),
     ),
 )
 def test_transforms(request, run_transform, task_input):
@@ -62,3 +122,30 @@ def test_transforms(request, run_transform, task_input):
     param_id = request.node.callspec.id
     assertion_func = globals()[f"assert_{param_id.replace('-', '_')}"]
     assertion_func(task)
+
+
+@pytest.mark.parametrize(
+    "fetch_mode,expected_extra_digest",
+    (
+        pytest.param(None, [], id="unset"),
+        pytest.param("clone", [], id="clone"),
+        pytest.param("fetch", ["fetch-mode=fetch"], id="fetch"),
+    ),
+)
+def test_git_fetch_mode_digest_data(fetch_mode, expected_extra_digest):
+    fetch_config = {
+        "type": "git",
+        "repo": "https://example.com/repo",
+        "revision": GIT_REVISION,
+    }
+    if fetch_mode:
+        fetch_config["fetch-mode"] = fetch_mode
+
+    result = fetch.create_git_fetch_task(None, "fake-task-name", fetch_config)
+
+    assert result["digest_data"] == [
+        GIT_REVISION,
+        "repo",
+        "repo.tar.zst",
+        *expected_extra_digest,
+    ]