Merge pull request #695 from w3lld1/feat/github-workflow-timeout-hook

Add GitHub workflow timeout pre-commit hook

Author: sirosen

.pre-commit-hooks.yaml

@@ -352,3 +352,14 @@
       ^\.woodpecker/.+\.(yml|yaml)$
     )$
   types: [yaml]
+
+# this hook is autogenerated from a script
+# to modify this hook, update `src/check_jsonschema/catalog.py`
+# and run `just generate-hooks` or `tox run -e generate-hooks-config`
+- id: check-github-workflows-require-timeout
+  name: Require timeout-minutes in GitHub Workflows
+  description: 'Validate that all GitHub Workflow jobs set timeout-minutes'
+  entry: check-jsonschema --builtin-schema custom.github-workflows-require-timeout
+  language: python
+  files: ^\.github/workflows/[^/]+$
+  types: [yaml]

CHANGELOG.rst

@@ -10,6 +10,9 @@ Unreleased
 
 .. vendor-insert-here
 
+- Add a dedicated ``check-github-workflows-require-timeout`` pre-commit hook for
+  requiring ``timeout-minutes`` on all GitHub Workflow jobs. (:issue:`639`)
+
 0.37.4
 ------
 

docs/precommit_usage.rst

@@ -407,6 +407,20 @@ Validate Woodpecker Config against the schema provided by SchemaStore
         - id: check-woodpecker-ci
 
 
+``check-github-workflows-require-timeout``
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Validate that all GitHub Workflow jobs set timeout-minutes
+
+.. code-block:: yaml
+    :caption: example config
+
+    - repo: https://github.com/python-jsonschema/check-jsonschema
+      rev: 0.37.4
+      hooks:
+        - id: check-github-workflows-require-timeout
+
+
 .. generated-hook-list-end
 
 

scripts/generate-hooks-config.py

@@ -4,25 +4,31 @@
 import importlib.metadata
 import typing as t
 
-from check_jsonschema.catalog import SCHEMA_CATALOG
+from check_jsonschema.catalog import CUSTOM_SCHEMA_CATALOG, SCHEMA_CATALOG
 
 version = importlib.metadata.version("check_jsonschema")
 
 
 def iter_catalog_hooks():
-    for name in SCHEMA_CATALOG:
-        # copy config (new dict)
-        config = dict(SCHEMA_CATALOG[name]["hook_config"])
-        # set computed attributes
-        config["schema_name"] = name
-        config["id"] = f"check-{name}"
-        config["description"] = (
-            config.get("description")
-            or f"{config['name']} against the schema provided by SchemaStore"
-        )
-        if "types" in config and isinstance(config["types"], str):
-            config["types"] = [config["types"]]
-        yield config
+    catalogs = (
+        ("vendor", SCHEMA_CATALOG),
+        ("custom", CUSTOM_SCHEMA_CATALOG),
+    )
+    for schema_namespace, catalog in catalogs:
+        for name in catalog:
+            # copy config (new dict)
+            config = dict(catalog[name]["hook_config"])
+            # set computed attributes
+            config["schema_name"] = name
+            config["schema_namespace"] = schema_namespace
+            config["id"] = f"check-{name}"
+            config["description"] = (
+                config.get("description")
+                or f"{config['name']} against the schema provided by SchemaStore"
+            )
+            if "types" in config and isinstance(config["types"], str):
+                config["types"] = [config["types"]]
+            yield config
 
 
 def update_hook_config(new_config: str) -> None:
@@ -54,7 +60,7 @@ def generate_hook_lines(config) -> t.Iterator[str]:
         add_args = " " + add_args
     yield (
         "  entry: check-jsonschema --builtin-schema "
-        f"vendor.{config['schema_name']}{add_args}"
+        f"{config['schema_namespace']}.{config['schema_name']}{add_args}"
     )
 
     yield "  language: python"

src/check_jsonschema/catalog.py

@@ -11,10 +11,20 @@ def _githubusercontent_url(owner: str, repo: str, ref: str, path: str) -> str:
     return f"https://raw.githubusercontent.com/{owner}/{repo}/{ref}/{path}"
 
 
-# this lists custom schemas which are *not* part of the catalog
-CUSTOM_SCHEMA_NAMES = [
-    "github-workflows-require-timeout",
-]
+# this lists custom schemas which are *not* part of the vendored schema catalog
+CUSTOM_SCHEMA_CATALOG: dict[str, dict[str, t.Any]] = {
+    "github-workflows-require-timeout": {
+        "hook_config": {
+            "name": "Require timeout-minutes in GitHub Workflows",
+            "description": (
+                "Validate that all GitHub Workflow jobs set timeout-minutes"
+            ),
+            "files": r"^\.github/workflows/[^/]+$",
+            "types": "yaml",
+        },
+    },
+}
+CUSTOM_SCHEMA_NAMES = list(CUSTOM_SCHEMA_CATALOG)
 
 # Known configs. The SchemaCatalog lists known schema URLs with their names.
 # kept in alphabetical order by name

tests/unit/test_catalog.py

@@ -21,3 +21,19 @@ def test_hooks_cover_catalog():
     config_hook_ids = {x["id"] for x in config}
     catalog_hook_ids = {f"check-{name}" for name in SCHEMA_CATALOG}
     assert catalog_hook_ids <= config_hook_ids
+
+
+def test_github_workflows_require_timeout_hook_uses_custom_schema():
+    with open(CONFIG_FILE, "rb") as fp:
+        config = yaml.load(fp)
+
+    hook = next(
+        item
+        for item in config
+        if item["id"] == "check-github-workflows-require-timeout"
+    )
+    assert hook["entry"] == (
+        "check-jsonschema --builtin-schema custom.github-workflows-require-timeout"
+    )
+    assert hook["files"] == r"^\.github/workflows/[^/]+$"
+    assert hook["types"] == ["yaml"]