Add static workflow test for ci.yml

Closes #34

Task: MIL-001#7

tests/test_workflow.py

@@ -0,0 +1,167 @@
+"""Static checks of the CI workflow (QC-WF-001 criterion 24, MIL-001 task 7)."""
+
+import re
+from collections.abc import Hashable, Iterator
+from pathlib import Path
+from typing import Any
+
+import pytest
+import yaml
+
+ROOT = Path(__file__).parents[1]
+WORKFLOW_PATH = ROOT / ".gitea" / "workflows" / "ci.yml"
+GITHUB_WORKFLOW_PATH = ROOT / ".github" / "workflows" / "ci.yml"
+MAX_RUN_LINES = 5
+EXACT_TAG = re.compile(r"@v\d+\.\d+\.\d+$")
+FULL_SHA = re.compile(r"@[0-9a-f]{40}$")
+
+# Parsed YAML is untyped data, so Any is the honest type here.
+Workflow = dict[Any, Any]
+
+
+class UniqueKeyLoader(yaml.SafeLoader):
+    """Safe loader that rejects a mapping with a repeated key."""
+
+    def construct_mapping(
+        self, node: yaml.MappingNode, deep: bool = False
+    ) -> dict[Hashable, Any]:
+        """Build the mapping, raising ValueError on a duplicate key."""
+        seen: set[Hashable] = set()
+        for key_node, _ in node.value:
+            key = self.construct_object(key_node, deep=deep)
+            if key in seen:
+                raise ValueError(f"duplicate key {key!r} at {key_node.start_mark}")
+            seen.add(key)
+        return super().construct_mapping(node, deep=deep)
+
+
+def load_workflow() -> Workflow:
+    """Parse the workflow with the safe loader."""
+    loaded = yaml.safe_load(WORKFLOW_PATH.read_text(encoding="utf-8"))
+    assert isinstance(loaded, dict)
+    return loaded
+
+
+def triggers(workflow: Workflow) -> Any:
+    """Return the trigger section; YAML 1.1 reads the key `on` as True."""
+    return workflow["on"] if "on" in workflow else workflow[True]
+
+
+def trigger_names(workflow: Workflow) -> set[str]:
+    """Return the names of the events that trigger the workflow."""
+    section = triggers(workflow)
+    if isinstance(section, str):
+        return {section}
+    return {str(name) for name in section}
+
+
+def jobs(workflow: Workflow) -> dict[str, Workflow]:
+    """Return the jobs by id."""
+    found: dict[str, Workflow] = workflow["jobs"]
+    return found
+
+
+def steps(job: Workflow) -> list[Workflow]:
+    """Return the steps of a job."""
+    found: list[Workflow] = job.get("steps", [])
+    return found
+
+
+def all_steps(workflow: Workflow) -> Iterator[Workflow]:
+    """Yield every step of every job."""
+    for job in jobs(workflow).values():
+        yield from steps(job)
+
+
+def exceeds_line_limit(script: str, limit: int = MAX_RUN_LINES) -> bool:
+    """Tell whether a script has more lines than the limit, as written.
+
+    Blank lines and comments count.
+    """
+    return len(script.splitlines()) > limit
+
+
+def test_workflow_parses_with_required_keys() -> None:
+    """The file is valid YAML with a name, a trigger and jobs."""
+    workflow = load_workflow()
+    assert "name" in workflow
+    assert "on" in workflow or True in workflow
+    assert "jobs" in workflow
+
+
+def test_workflow_has_no_duplicate_keys() -> None:
+    """No mapping in the file repeats a key."""
+    text = WORKFLOW_PATH.read_text(encoding="utf-8")
+    assert yaml.load(text, Loader=UniqueKeyLoader) is not None
+
+
+def test_duplicate_key_loader_rejects_repeated_key() -> None:
+    """The duplicate-key loader bites on a repeated key."""
+    with pytest.raises(ValueError, match="duplicate key"):
+        yaml.load("a: 1\na: 2\n", Loader=UniqueKeyLoader)
+
+
+def test_permissions_are_declared() -> None:
+    """Permissions are set at workflow level or on every job."""
+    workflow = load_workflow()
+    if "permissions" in workflow:
+        return
+    assert all("permissions" in job for job in jobs(workflow).values())
+
+
+def test_every_job_has_runner_and_timeout() -> None:
+    """Each job names its runner and a timeout."""
+    for job_id, job in jobs(load_workflow()).items():
+        assert "runs-on" in job, job_id
+        assert "timeout-minutes" in job, job_id
+
+
+def test_every_step_has_a_name() -> None:
+    """Each step carries a name."""
+    for job_id, job in jobs(load_workflow()).items():
+        for step in steps(job):
+            assert "name" in step, f"{job_id}: {step}"
+
+
+def test_step_names_are_unique_within_a_job() -> None:
+    """No two steps of one job share a name."""
+    for job_id, job in jobs(load_workflow()).items():
+        names = [step["name"] for step in steps(job) if "name" in step]
+        assert len(names) == len(set(names)), job_id
+
+
+def test_uses_are_pinned_to_exact_release_or_sha() -> None:
+    """Each action is pinned to an exact tag or a full commit SHA."""
+    for step in all_steps(load_workflow()):
+        if "uses" in step:
+            ref = str(step["uses"])
+            assert EXACT_TAG.search(ref) or FULL_SHA.search(ref), ref
+
+
+def test_run_blocks_stay_within_line_limit() -> None:
+    """No inline script is longer than the limit."""
+    for step in all_steps(load_workflow()):
+        if "run" in step:
+            assert not exceeds_line_limit(str(step["run"])), step.get("name")
+
+
+def test_line_limit_check_rejects_six_lines() -> None:
+    """The line-limit helper counts lines as written, blank lines included."""
+    assert exceeds_line_limit("a\nb\nc\nd\ne\nf\n")
+    assert exceeds_line_limit("a\nb\nc\nd\ne\n\n")
+    assert not exceeds_line_limit("a\nb\n\nc\nd\n")
+
+
+def test_workflow_uses_no_secrets() -> None:
+    """The file refers to no secret."""
+    assert "secrets." not in WORKFLOW_PATH.read_text(encoding="utf-8")
+
+
+def test_workflow_is_not_triggered_by_pull_request_target() -> None:
+    """The privileged pull_request_target trigger is not used."""
+    assert "pull_request_target" not in trigger_names(load_workflow())
+
+
+def test_workflow_is_not_duplicated_for_github() -> None:
+    """The same workflow does not also live under .github/workflows."""
+    assert not GITHUB_WORKFLOW_PATH.exists()