From b6cd79fa03014c94e366cf112a6d9224d49134ec Mon Sep 17 00:00:00 2001
From: David Mandelberg <david@mandelberg.org>
Date: Wed, 8 Nov 2023 15:36:58 -0500
Subject: [PATCH] Add functions to get labels and descriptions

---
 rock_paper_sand/wikidata.py      | 24 +++++++++++++++-
 rock_paper_sand/wikidata_test.py | 47 +++++++++++++++++++++++++++++++-
 2 files changed, 69 insertions(+), 2 deletions(-)

diff --git a/rock_paper_sand/wikidata.py b/rock_paper_sand/wikidata.py
index cf10a12..905b1a0 100644
--- a/rock_paper_sand/wikidata.py
+++ b/rock_paper_sand/wikidata.py
@@ -14,7 +14,7 @@
 """Code that uses Wikidata's APIs."""
 
 import collections
-from collections.abc import Generator, Iterable, Sequence, Set
+from collections.abc import Generator, Iterable, Mapping, Sequence, Set
 import contextlib
 import dataclasses
 import datetime
@@ -71,6 +71,28 @@ def requests_session() -> Generator[requests.Session, None, None]:
         yield session
 
 
+def _language_keyed_string(
+    mapping: Mapping[str, Any],
+    languages: Sequence[str],
+) -> str | None:
+    # https://doc.wikimedia.org/Wikibase/master/php/docs_topics_json.html#json_fingerprint
+    for language in languages:
+        if language in mapping:
+            return mapping[language]["value"]
+        for other_language, record in mapping.items():
+            if other_language.startswith(f"{language}-"):
+                return record["value"]
+    return None
+
+
+def _label(item: Any, languages: Sequence[str]) -> str | None:
+    return _language_keyed_string(item["labels"], languages)
+
+
+def _description(item: Any, languages: Sequence[str]) -> str | None:
+    return _language_keyed_string(item["descriptions"], languages)
+
+
 def _truthy_statements(
     item: Any, prop: wikidata_value.Property
 ) -> Sequence[Any]:
diff --git a/rock_paper_sand/wikidata_test.py b/rock_paper_sand/wikidata_test.py
index 302d333..77580fb 100644
--- a/rock_paper_sand/wikidata_test.py
+++ b/rock_paper_sand/wikidata_test.py
@@ -14,7 +14,7 @@
 
 # pylint: disable=missing-module-docstring
 
-from collections.abc import Collection, Mapping, Sequence, Set
+from collections.abc import Callable, Collection, Mapping, Sequence, Set
 import datetime
 from typing import Any
 from unittest import mock
@@ -318,6 +318,51 @@ def test_related_media_error(self) -> None:
 
 class WikidataUtilsTest(parameterized.TestCase):
     # pylint: disable=protected-access
+    @parameterized.product(
+        (
+            dict(function=wikidata._label, section="labels"),
+            dict(function=wikidata._description, section="descriptions"),
+        ),
+        (
+            dict(
+                mapping={},
+                languages=("en",),
+                expected_value=None,
+            ),
+            dict(
+                mapping={"en": {"value": "foo"}},
+                languages=(),
+                expected_value=None,
+            ),
+            dict(
+                mapping={
+                    "en": {"value": "foo"},
+                    "en-us": {"value": "bar"},
+                },
+                languages=("qa", "en"),
+                expected_value="foo",
+            ),
+            dict(
+                mapping={"en-us": {"value": "foo"}},
+                languages=("en",),
+                expected_value="foo",
+            ),
+        ),
+    )
+    def test_language_keyed_string(
+        self,
+        *,
+        function: Callable[[Any, Sequence[str]], str | None],
+        section: str,
+        mapping: Mapping[str, Any],
+        languages: Sequence[str],
+        expected_value: str | None,
+    ) -> None:
+        self.assertEqual(
+            expected_value,
+            function({section: mapping}, languages),
+        )
+
     @parameterized.named_parameters(
         dict(
             testcase_name="preferred",