Do not report the Unicode data files licence name as a tracker key

actions/repo-hygiene/check-repo-hygiene.py

@@ -81,6 +81,13 @@
 # evidenced exceptions rather than a way to quiet a genuine hit.
 NOT_A_TRACKER_KEY = re.compile(r"\b(?:ISO-\d+|RFC-\d+|SHA-\d+|UTF-\d+|AES-\d+|PEP-\d+|CVE-\d+|SMETS-\d+|FLEX-\d+|GPL-\d+|LGPL-\d+|AGPL-\d+)\b")
 
+# Same shape at the end of a longer standard's name, matched against the line up to the key's
+# end, since the key alone cannot see what comes before it: Unicode-DFS-2016, the SPDX name of
+# Unicode's data files licence, which a Debian copyright file must spell exactly.  Only that name
+# is let through, so a project whose keys start DFS-, or a key that follows "Unicode-", is still
+# reported.
+PART_OF_A_STANDARD = re.compile(r"\bUnicode-DFS-\d+$")
+
 # Any link into the internal Atlassian site, not only a wiki one.  Review asked whether this
 # should require a /wiki/ path, since the name said "wiki link" while the pattern also matched a
 # Jira /browse/ URL.  Widened the name rather than narrowing the pattern: a Jira board URL
@@ -406,7 +413,9 @@ def check_no_internal_references(root, tracked, files):
         for number, line in enumerate(text.splitlines(), 1):
             for match in TRACKER_KEY.finditer(line):
                 span = match.group(0)
-                if not NOT_A_TRACKER_KEY.search(span):
+                if not NOT_A_TRACKER_KEY.search(span) and not PART_OF_A_STANDARD.search(
+                    line[: match.end()]
+                ):
                     findings.append(f"{relative}:{number}: tracker key {span}")
             for match in INTERNAL_LINK.finditer(line):
                 findings.append(f"{relative}:{number}: internal Atlassian link ({match.group(0)[:48]})")

tests/conftest.py

@@ -0,0 +1,15 @@
+"""Shared by the tests: the hygiene check, loaded from the action's script as it ships."""
+
+import importlib.util
+import pathlib
+
+ROOT = pathlib.Path(__file__).resolve().parent.parent
+
+
+def load_hygiene():
+    spec = importlib.util.spec_from_file_location(
+        "check_repo_hygiene", ROOT / "actions" / "repo-hygiene" / "check-repo-hygiene.py"
+    )
+    module = importlib.util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    return module

tests/test_internal_references.py

@@ -0,0 +1,36 @@
+"""The hygiene check's search for tracker keys: what it reports, and the standards it does not.
+
+The check is loaded from the action's script, so what is tested is the script that ships.  The
+keys it should report are assembled at run time, so that this file holds none for the check to
+find.
+"""
+
+import pytest
+from conftest import load_hygiene
+
+
+def findings(tmp_path, line):
+    path = tmp_path / "copyright"
+    path.write_text(line + "\n", encoding="utf-8")
+    return load_hygiene().check_no_internal_references(tmp_path, [path], [path])
+
+
+@pytest.mark.parametrize("line", [
+    "License: Unicode-DFS-2016",
+    "License: GPL-3+ and Expat and Unicode-DFS-2016 and Unicode-3.0",
+    "Encoded as UTF-8, per RFC-3629 and ISO-8859",
+])
+def test_an_external_standard_is_not_a_tracker_key(tmp_path, line):
+    assert findings(tmp_path, line) == []
+
+
+@pytest.mark.parametrize("before,key", [
+    ("Fixes ", "XY" + "-22"),
+    ("See ", "ABC" + "-1"),
+    # Only the Unicode licence's name is let through, not every key ending its shape.
+    ("Fixes ", "DFS" + "-2016"),
+    ("MyUnicode-", "DFS" + "-2016"),
+    ("Unicode-", "XY" + "-22"),
+])
+def test_a_tracker_key_is_reported(tmp_path, before, key):
+    assert findings(tmp_path, before + key) == [f"copyright:1: tracker key {key}"]

tests/test_slack_workflows.py

@@ -6,7 +6,6 @@
 environment, and Slack itself.
 """
 
-import importlib.util
 import json
 import os
 import pathlib
@@ -16,6 +15,7 @@
 
 import pytest
 import yaml
+from conftest import load_hygiene
 
 ROOT = pathlib.Path(__file__).resolve().parent.parent
 WORKFLOWS = ROOT / ".github" / "workflows"
@@ -258,15 +258,6 @@ def test_the_token_reaches_curl_only_in_a_header_file(stubs, workflow, environme
     assert json.loads((stubs / "curl.stdin").read_text())["channel"] == channel
 
 
-def hygiene():
-    spec = importlib.util.spec_from_file_location(
-        "check_repo_hygiene", ROOT / "actions" / "repo-hygiene" / "check-repo-hygiene.py"
-    )
-    module = importlib.util.module_from_spec(spec)
-    spec.loader.exec_module(module)
-    return module
-
-
 @pytest.mark.parametrize("uses", [
     "L337-org/github-workflows/.github/workflows/slack-on-pull-request.yaml@" + "0" * 40,
     "L337-org/github-workflows/.github/workflows/slack-on-failure.yaml@" + "0" * 40,
@@ -289,7 +280,7 @@ def test_hygiene_requires_v2_inputs_on_every_slack_call(uses, given, secrets, fo
         body["with"] = given
     if secrets is not None:
         body["secrets"] = secrets
-    findings = hygiene().slack_call_findings([(".github/workflows/x.yaml", {"jobs": {"slack": body}})])
+    findings = load_hygiene().slack_call_findings([(".github/workflows/x.yaml", {"jobs": {"slack": body}})])
     assert len(findings) == len(found)
     for finding, expected in zip(findings, found):
         assert expected in finding
@@ -299,4 +290,4 @@ def test_every_shipped_caller_here_names_a_channel_and_passes_the_token():
     documents = [
         (p.name, yaml.safe_load(p.read_text())) for p in sorted(WORKFLOWS.glob("*.yaml"))
     ]
-    assert hygiene().slack_call_findings(documents) == []
+    assert load_hygiene().slack_call_findings(documents) == []