Merge pull request #541 from YuukiToriyama/feature/notation-distortio…

…n/minami-zakura 「南櫻」の表記ゆれに対応
YuukiToriyama · Nov 16, 2024 · bb18bf9 · bb18bf9
2 parents 8f5c0e4 + d47240e
commit bb18bf9
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
@@ -32,6 +32,7 @@ pub enum OrthographicalVariant {
     鉋,
     鷆,
     斑,
+    櫻,
 }
 
 impl OrthographicalVariant {
@@ -69,6 +70,7 @@ impl OrthographicalVariant {
             OrthographicalVariant::鉋 => &['鉋', '飽'],
             OrthographicalVariant::鷆 => &['鷆', '鷏'],
             OrthographicalVariant::斑 => &['斑', '班'],
+            OrthographicalVariant::櫻 => &['櫻', '桜'],
         }
     }
 

diff --git a/core/src/tokenizer/read_town.rs b/core/src/tokenizer/read_town.rs
@@ -94,6 +94,7 @@ fn find_town(input: &str, candidates: &Vec<String>) -> Option<(String, String)>
                 OrthographicalVariant::鉋,
                 OrthographicalVariant::鷆,
                 OrthographicalVariant::斑,
+                OrthographicalVariant::櫻,
             ],
         };
         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
@@ -90,3 +90,6 @@ address,prefecture,city,town,rest
 # 「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
+# 「南桜」と「南櫻」の表記揺れへの対応
+滋賀県野洲市南桜1792,滋賀県,野洲市,南櫻,1792
+滋賀県野洲市南櫻1792,滋賀県,野洲市,南櫻,1792