Update federatedcode pipes (#1831)

* Separate commiting and pushing functionality in federatedcode pipes

Signed-off-by: Jono Yang <[email protected]>

* Add write_data_as_yaml function to federatedcode pipes

Signed-off-by: Jono Yang <[email protected]>

* Update CHANGELOG.rst

Signed-off-by: Jono Yang <[email protected]>

---------

Signed-off-by: Jono Yang <[email protected]>
Co-authored-by: Ayan Sinha Mahapatra <[email protected]>

Author: JonoYang

CHANGELOG.rst

@@ -29,6 +29,13 @@ v35.4.0 (unreleased)
   * Add support when the "components" entry is missing.
   https://github.com/aboutcode-org/scancode.io/issues/1727
 
+- Split the functionality of
+  ``scanpipe.pipes.federatedcode.commit_and_push_changes`` into
+  ``scanpipe.pipes.federatedcode.commit_changes`` and
+  ``scanpipe.pipes.federatedcode.push_changes``. Add
+  ``scanpipe.pipes.federatedcode.write_data_as_yaml``.
+
+
 v35.3.0 (2025-08-20)
 --------------------
 

scanpipe/pipes/federatedcode.py

@@ -31,6 +31,7 @@
 from django.conf import settings
 
 import requests
+import saneyaml
 from git import Repo
 from packageurl import PackageURL
 
@@ -161,6 +162,19 @@ def add_scan_result(project, repo, package_scan_file, logger=None):
     return relative_scan_file_path
 
 
+def commit_changes(repo, files_to_commit, commit_message):
+    """Commit changes to remote repository."""
+    repo.index.add(files_to_commit)
+    repo.index.commit(textwrap.dedent(commit_message))
+
+
+def push_changes(repo, remote_name="origin", branch_name=""):
+    """Push changes to remote repository."""
+    if not branch_name:
+        branch_name = repo.active_branch.name
+    repo.git.push(remote_name, branch_name, "--no-verify")
+
+
 def commit_and_push_changes(
     repo, file_to_commit, purl, remote_name="origin", logger=None
 ):
@@ -177,14 +191,27 @@ def commit_and_push_changes(
 
     Signed-off-by: {author_name} <{author_email}>
     """
-
-    default_branch = repo.active_branch.name
-
-    repo.index.add([file_to_commit])
-    repo.index.commit(textwrap.dedent(commit_message))
-    repo.git.push(remote_name, default_branch, "--no-verify")
+    files_to_commit = [file_to_commit]
+    commit_changes(
+        repo=repo, files_to_commit=files_to_commit, commit_message=commit_message
+    )
+    push_changes(
+        repo=repo,
+        remote_name=remote_name,
+    )
 
 
 def delete_local_clone(repo):
     """Remove local clone."""
     shutil.rmtree(repo.working_dir)
+
+
+def write_data_as_yaml(base_path, file_path, data):
+    """
+    Write the ``data`` as YAML to the ``file_path`` in the ``base_path`` root directory.
+    Create directories in the path as needed.
+    """
+    write_to = Path(base_path) / Path(file_path)
+    write_to.parent.mkdir(parents=True, exist_ok=True)
+    with open(write_to, encoding="utf-8", mode="w") as f:
+        f.write(saneyaml.dump(data))

scanpipe/tests/pipes/test_federatedcode.py

@@ -28,6 +28,7 @@
 from django.test import TestCase
 
 import git
+import saneyaml
 
 from scanpipe import models
 from scanpipe.pipes import federatedcode
@@ -77,3 +78,28 @@ def test_scanpipe_pipes_federatedcode_delete_local_clone(self):
         federatedcode.delete_local_clone(repo)
 
         self.assertEqual(False, Path(local_dir).exists())
+
+    def test_scanpipe_pipes_federatedcode_write_data_as_yaml(self):
+        # create local repo
+        local_dir = tempfile.mkdtemp()
+        repo = git.Repo.init(local_dir)
+
+        # write data
+        data = ["123", "abc", 3]
+        federatedcode.write_data_as_yaml(
+            base_path=repo.working_dir,
+            file_path="test.yml",
+            data=data,
+        )
+
+        # Check if file was written
+        test_file_path = Path(repo.working_dir) / "test.yml"
+        self.assertEqual(True, test_file_path.exists())
+        with open(test_file_path) as f:
+            contents = f.read()
+        yml = saneyaml.load(contents)
+        expected_results = ["123", "abc", "3"]
+        self.assertEqual(expected_results, yml)
+
+        # clean up
+        shutil.rmtree(repo.working_dir)