Add function to parse string snaks

dseomn · Nov 10, 2023 · d9186ae · d9186ae
1 parent 5ce0c7c
commit d9186ae
Show file tree

Hide file tree

Showing 2 changed files with 59 additions and 0 deletions.
diff --git a/rock_paper_sand/wikidata_value.py b/rock_paper_sand/wikidata_value.py
@@ -230,6 +230,17 @@ def parse_snak_item(snak: Snak) -> ItemRef:
     return ItemRef(snak["datavalue"]["value"]["id"])
 
 
+def parse_snak_string(snak: Snak) -> str:
+    """Returns a string value from a snak."""
+    if snak["snaktype"] != "value":
+        raise NotImplementedError(
+            f"Cannot parse non-value snak as a string: {snak}"
+        )
+    if snak["datatype"] != "string" or snak["datavalue"]["type"] != "string":
+        raise ValueError(f"Cannot parse non-string snak as a string: {snak}")
+    return snak["datavalue"]["value"]
+
+
 def parse_snak_time(snak: Snak) -> tuple[datetime.datetime, datetime.datetime]:
     """Returns (earliest possible time, latest possible time) of a time snak."""
     # https://doc.wikimedia.org/Wikibase/master/php/docs_topics_json.html#json_datavalues_time

diff --git a/rock_paper_sand/wikidata_value_test.py b/rock_paper_sand/wikidata_value_test.py
@@ -186,6 +186,54 @@ def test_parse_snak_item(self) -> None:
             ),
         )
 
+    @parameterized.named_parameters(
+        dict(
+            testcase_name="not_value",
+            snak={"snaktype": "somevalue"},
+            error_class=NotImplementedError,
+            error_regex=r"non-value",
+        ),
+        dict(
+            testcase_name="datatype_not_string",
+            snak={"snaktype": "value", "datatype": "wikibase-item"},
+            error_class=ValueError,
+            error_regex=r"non-string",
+        ),
+        dict(
+            testcase_name="type_not_string",
+            snak={
+                "snaktype": "value",
+                "datatype": "string",
+                "datavalue": {"type": "wikibase-entityid"},
+            },
+            error_class=ValueError,
+            error_regex=r"non-string",
+        ),
+    )
+    def test_parse_snak_string_error(
+        self,
+        *,
+        snak: Any,
+        error_class: type[Exception],
+        error_regex: str,
+    ) -> None:
+        with self.assertRaisesRegex(error_class, error_regex):
+            wikidata_value.parse_snak_string(wikidata_value.Snak(snak))
+
+    def test_parse_snak_string(self) -> None:
+        self.assertEqual(
+            "foo",
+            wikidata_value.parse_snak_string(
+                wikidata_value.Snak(
+                    {
+                        "snaktype": "value",
+                        "datatype": "string",
+                        "datavalue": {"type": "string", "value": "foo"},
+                    }
+                )
+            ),
+        )
+
     @parameterized.named_parameters(
         dict(
             testcase_name="not_value",