Filter test collection based on files changed in upstream PRs (#14931)

conf/github.yaml.template

@@ -0,0 +1,13 @@
+GITHUB:
+  FOREMAN:
+    ORG: theforeman
+    REPO: foreman
+    RULES:
+     - PATH:
+       MARKER:
+  KATELLO:
+    ORG: Katello
+    REPO: katello
+    RULES:
+      - PATH:
+        MARKER:

conftest.py

@@ -23,6 +23,7 @@
     'pytest_plugins.sanity_plugin',
     'pytest_plugins.video_cleanup',
     'pytest_plugins.capsule_n-minus',
+    'pytest_plugins.upstream_pr',
     # Fixtures
     'pytest_fixtures.core.broker',
     'pytest_fixtures.core.sat_cap_factory',

pytest_plugins/upstream_pr.py

@@ -0,0 +1,68 @@
+from github import Github
+
+from robottelo.config import settings
+from robottelo.logging import collection_logger as logger
+
+
+def pytest_addoption(parser):
+    """Add CLI option to specify upstream GitHub PRs.
+
+    Add --upstream-pr option for filtering tests based on the files modified by upstream
+    PRs.
+    """
+    parser.addoption(
+        "--upstream-pr",
+        help=(
+            "Filter test collection based on files modified by upstream PR."
+            " Example: `--upstream-pr REPO/PR_ID`"
+        ),
+    )
+
+
+def pytest_collection_modifyitems(session, items, config):
+    """Filter tests based on upstream PRs.
+    1. Get the list of modified files in the upstream PRs.
+    2. Map each file to at most one component.
+    3. Filter the collected tests to include only those with matching components.
+    """
+    upstream_prs = [
+        pr_info for pr_info in (config.getoption('upstream_pr') or '').split(',') if pr_info != ''
+    ]
+    if not upstream_prs:
+        return
+
+    markers = set()
+    for pr_info in upstream_prs:
+        # Get all filenames modified by this PR
+        repo_key, pr_id = pr_info.split('/')
+        if not (repo_config := settings.github.get(repo_key)):
+            raise Exception(f"Key {repo_key} not found in settings file.")
+        pr = Github().get_repo(f"{repo_config.org}/{repo_config.repo}").get_pull(int(pr_id))
+        pr_filenames = {file.filename for commit in pr.get_commits() for file in commit.files}
+
+        # Get a list of markers from all of the matching rules
+        unprocessed_filenames = pr_filenames.copy()
+        for rule in repo_config.rules:
+            if matched_filenames := {
+                filename for filename in unprocessed_filenames if filename.startswith(rule.path)
+            }:
+                markers.add(rule.marker)
+                unprocessed_filenames.difference_update(matched_filenames)
+
+    # If no markers were found above, deselect all tests.
+    # Any unprocessed filenames are ignored.
+    selected = []
+    deselected = []
+    for item in items:
+        if markers:
+            if any(item_marker.name in markers for item_marker in item.iter_markers()):
+                selected.append(item)
+            else:
+                logger.debug(f'Deselected test {item.nodeid} due to PR filter {upstream_prs}')
+                deselected.append(item)
+        else:
+            logger.debug(f'Deselected test {item.nodeid} due to PR filter {upstream_prs}')
+            deselected.append(item)
+
+    config.hook.pytest_deselected(items=deselected)
+    items[:] = selected

requirements.txt

@@ -11,6 +11,7 @@ jinja2==3.1.3
 manifester==0.0.14
 navmazing==1.2.2
 productmd==1.38
+PyGithub==2.3.0
 pyotp==2.9.0
 python-box==7.1.1
 pytest==8.2.0

robottelo/config/validators.py

@@ -161,6 +161,14 @@
             must_exist=True,
         ),
     ],
+    github=[
+        Validator('github.foreman.org', default='theforeman'),
+        Validator('github.foreman.repo', default='foreman'),
+        Validator('github.foreman.rules', default=[], is_type_of=list),
+        Validator('github.katello.org', default='Katello'),
+        Validator('github.katello.repo', default='katello'),
+        Validator('github.katello.rules', default=[], is_type_of=list),
+    ],
     http_proxy=[
         Validator(
             'http_proxy.un_auth_proxy_url',