Merge pull request #434 from YuukiToriyama/release/v0.1.16

release/v0.1.16をmainブランチにマージ

Cargo.toml

@@ -8,7 +8,7 @@ members = [
 resolver = "2"
 
 [workspace.package]
-version = "0.1.15"
+version = "0.1.16"
 edition = "2021"
 description = "A Rust Library to parse japanese addresses."
 repository = "https://github.com/YuukiToriyama/japanese-address-parser"

core/Cargo.toml

@@ -24,7 +24,7 @@ eliminate-whitespaces = []
 [dependencies]
 itertools = "0.13.0"
 rapidfuzz = "0.5.0"
-regex = "1.10.2"
+regex = { version = "1.10.6", default-features = false, features = ["std", "unicode-perl"] }
 serde.workspace = true
 reqwest = { version = "0.12.5", default-features = false, features = ["json", "rustls-tls"] }
 js-sys = "0.3.67"

core/src/parser/adapter/orthographical_variant_adapter.rs

@@ -30,10 +30,11 @@ pub trait OrthographicalVariants {
     const 籠: Variant;
     const 濱: Variant;
     const 祗: Variant;
+    const 曾: Variant;
 }
 
 impl OrthographicalVariants for Variant {
-    const の: Variant = &["の", "ノ"];
+    const の: Variant = &["の", "ノ", "之"];
     const ツ: Variant = &["ツ", "ッ"];
     const ケ: Variant = &["ケ", "ヶ", "が", "ガ"];
     const 薮: Variant = &["薮", "藪", "籔"];
@@ -60,6 +61,7 @@ impl OrthographicalVariants for Variant {
     const 籠: Variant = &["籠", "篭"];
     const 濱: Variant = &["濱", "浜"];
     const 祗: Variant = &["祗", "祇"];
+    const 曾: Variant = &["曾", "曽"];
 }
 
 pub struct OrthographicalVariantAdapter {

core/src/tokenizer/read_town.rs

@@ -113,6 +113,7 @@ fn find_town(input: &str, candidates: &Vec<String>) -> Option<(String, String)>
                 Variant::籠,
                 Variant::濱,
                 Variant::祗,
+                Variant::曾,
             ],
         };
         if let Some(result) = adapter.apply(input, candidate) {

tests/integration_tests.rs

@@ -18,6 +18,11 @@ async fn 異字体旧字体への対応テスト() {
     run_data_driven_tests("./test_data/異字体旧字体への対応.csv").await
 }
 
+#[tokio::test]
+async fn 異字体ではない表記ゆれへの対応テスト() {
+    run_data_driven_tests("./test_data/異字体ではない表記ゆれへの対応.csv").await
+}
+
 #[tokio::test]
 async fn 丁目が算用数字の場合への対応テスト() {
     run_data_driven_tests("./test_data/丁目が算用数字の場合への対応.csv").await

tests/test_data/異字体ではない表記ゆれへの対応.csv

@@ -0,0 +1,5 @@
+address,prefecture,city,town,rest
+# 「ノ」「の」「之」の表記ゆれへの対応
+神奈川県鎌倉市山ノ内189,神奈川県,鎌倉市,山ノ内,189
+神奈川県鎌倉市山の内189,神奈川県,鎌倉市,山ノ内,189
+神奈川県鎌倉市山之内189,神奈川県,鎌倉市,山ノ内,189

tests/test_data/異字体旧字体への対応.csv

@@ -81,3 +81,6 @@ address,prefecture,city,town,rest
 # 「竃」と「竈」の表記ゆれへの対応
 静岡県御殿場市竈1032,静岡県,御殿場市,竈,1032
 静岡県御殿場市竃1032,静岡県,御殿場市,竈,1032
+# 「小曾根」と「小曽根」の表記ゆれへの対応
+埼玉県熊谷市小曽根1220,埼玉県,熊谷市,小曽根,1220
+埼玉県熊谷市小曾根1220,埼玉県,熊谷市,小曽根,1220