Merge pull request #539 from YuukiToriyama/feature/notation-distortio…

…n/tenwa 「鷏和」の表記ゆれに対応
YuukiToriyama · Nov 16, 2024 · fe0fa53 · fe0fa53
2 parents 87aff2f + 493fe89
commit fe0fa53
Show file tree

Hide file tree

Showing 3 changed files with 6 additions and 0 deletions.
diff --git a/core/src/adapter/orthographical_variant_adapter.rs b/core/src/adapter/orthographical_variant_adapter.rs
@@ -30,6 +30,7 @@ pub enum OrthographicalVariant {
     曾,
     國,
     鉋,
+    鷆,
 }
 
 impl OrthographicalVariant {
@@ -65,6 +66,7 @@ impl OrthographicalVariant {
             OrthographicalVariant::曾 => &['曾', '曽'],
             OrthographicalVariant::國 => &['國', '国'],
             OrthographicalVariant::鉋 => &['鉋', '飽'],
+            OrthographicalVariant::鷆 => &['鷆', '鷏'],
         }
     }
 

diff --git a/core/src/tokenizer/read_town.rs b/core/src/tokenizer/read_town.rs
@@ -92,6 +92,7 @@ fn find_town(input: &str, candidates: &Vec<String>) -> Option<(String, String)>
                 OrthographicalVariant::曾,
                 OrthographicalVariant::國,
                 OrthographicalVariant::鉋,
+                OrthographicalVariant::鷆,
             ],
         };
         if let Some(result) = adapter.apply(input, candidate) {

diff --git a/tests/test_data/異字体旧字体への対応.csv b/tests/test_data/異字体旧字体への対応.csv
@@ -87,3 +87,6 @@ address,prefecture,city,town,rest
 # 「神代國衙」と「神代国衙」の表記揺れへの対応
 兵庫県南あわじ市神代國衙1680,兵庫県,南あわじ市,神代國衙,1680
 兵庫県南あわじ市神代国衙1680,兵庫県,南あわじ市,神代國衙,1680
+# 「鷏和」と「鷆和」の表記揺れへの対応
+兵庫県赤穂市鷏和422,兵庫県,赤穂市,鷏和,422
+兵庫県赤穂市鷆和422,兵庫県,赤穂市,鷏和,422