Merge pull request #540 from YuukiToriyama/feature/notation-distortio…

…n/madarame 「班目」の表記ゆれに対応
YuukiToriyama · Nov 16, 2024 · 8f5c0e4 · 8f5c0e4
2 parents fe0fa53 + 5b1fafb
commit 8f5c0e4
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
@@ -31,6 +31,7 @@ pub enum OrthographicalVariant {
     國,
     鉋,
     鷆,
+    斑,
 }
 
 impl OrthographicalVariant {
@@ -67,6 +68,7 @@ impl OrthographicalVariant {
             OrthographicalVariant::國 => &['國', '国'],
             OrthographicalVariant::鉋 => &['鉋', '飽'],
             OrthographicalVariant::鷆 => &['鷆', '鷏'],
+            OrthographicalVariant::斑 => &['斑', '班'],
         }
     }
 

diff --git a/core/src/tokenizer/read_town.rs b/core/src/tokenizer/read_town.rs
@@ -93,6 +93,7 @@ fn find_town(input: &str, candidates: &Vec<String>) -> Option<(String, String)>
                 OrthographicalVariant::國,
                 OrthographicalVariant::鉋,
                 OrthographicalVariant::鷆,
+                OrthographicalVariant::斑,
             ],
         };
         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
@@ -6,3 +6,6 @@ address,prefecture,city,town,rest
 # 「上氷鉋」と「上氷飽」の表記揺れへの対応
 長野県長野市川中島町上氷鉋1368,長野県,長野市,川中島町上氷鉋,1368
 長野県長野市川中島町上氷飽1368,長野県,長野市,川中島町上氷鉋,1368
+# 「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