Merge pull request #542 from YuukiToriyama/feature/notation-distortio…

…n/matsuho-ichida 「松帆櫟田」の表記ゆれに対応
YuukiToriyama · Nov 16, 2024 · ecfe5c8 · ecfe5c8
2 parents bb18bf9 + c32547c
commit ecfe5c8
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
@@ -33,6 +33,7 @@ pub enum OrthographicalVariant {
     鷆,
     斑,
     櫻,
+    櫟,
 }
 
 impl OrthographicalVariant {
@@ -71,6 +72,7 @@ impl OrthographicalVariant {
             OrthographicalVariant::鷆 => &['鷆', '鷏'],
             OrthographicalVariant::斑 => &['斑', '班'],
             OrthographicalVariant::櫻 => &['櫻', '桜'],
+            OrthographicalVariant::櫟 => &['櫟', '擽'],
         }
     }
 

diff --git a/core/src/tokenizer/read_town.rs b/core/src/tokenizer/read_town.rs
@@ -95,6 +95,7 @@ fn find_town(input: &str, candidates: &Vec<String>) -> Option<(String, String)>
                 OrthographicalVariant::鷆,
                 OrthographicalVariant::斑,
                 OrthographicalVariant::櫻,
+                OrthographicalVariant::櫟,
             ],
         };
         if let Some(result) = adapter.apply(input, candidate) {

diff --git a/tests/test_data/異字体ではない表記ゆれへの対応.csv b/tests/test_data/異字体ではない表記ゆれへの対応.csv
@@ -9,3 +9,6 @@ address,prefecture,city,town,rest
 # 「斑目」と「班目」の表記揺れへの対応
 神奈川県南足柄市班目639,神奈川県,南足柄市,班目,639
 神奈川県南足柄市斑目639,神奈川県,南足柄市,班目,639
+# 「櫟」と「擽」の表記ゆれへの対応
+兵庫県南あわじ市松帆櫟田196,兵庫県,南あわじ市,松帆櫟田,196
+兵庫県南あわじ市松帆擽田196,兵庫県,南あわじ市,松帆櫟田,196