Merge pull request #375 from YuukiToriyama/feature/optout-county-name…

…-completion/master

郡名補完機能のオプトアウトをrelease/v0.1.10にマージ

core/Cargo.toml

@@ -15,8 +15,9 @@ rust-version = "1.73.0"
 crate-type = ["rlib", "cdylib"]
 
 [features]
-default = []
+default = ["city-name-correction"]
 blocking = ["reqwest/blocking"]
+city-name-correction = []
 
 [dependencies]
 itertools = "0.13.0"

core/src/lib.rs

@@ -1,3 +1,9 @@
+//! A Rust library to parse japanese addresses.
+//!
+//! ## Feature flags
+//! - `blocking`: Provide method that works synchronously
+//! - `city-name-correction`*(enabled by default)*: Enable autocorrection if ambiguous city name was typed
+
 #[cfg(all(target_family = "wasm", feature = "blocking"))]
 compile_error! {
     "The `blocking` feature is not supported with wasm target."

core/src/parser.rs

@@ -98,11 +98,21 @@ pub async fn parse(api: Arc<AsyncApi>, input: &str) -> ParseResult {
         Ok(result) => result,
     };
     // 市町村名を特定
-    let Ok(tokenizer) = tokenizer.read_city(prefecture.cities) else {
-        return ParseResult {
-            address: Address::from(tokenizer),
-            error: Some(Error::new_parse_error(ParseErrorKind::City)),
-        };
+    let tokenizer = match tokenizer.read_city(&prefecture.cities) {
+        Ok(found) => found,
+        Err(not_found) => {
+            // 市区町村が特定できない場合かつフィーチャフラグが有効な場合、郡名が抜けている可能性を検討
+            match not_found.read_city_with_county_name_completion(&prefecture.cities) {
+                Ok(found) if cfg!(feature = "city-name-correction") => found,
+                _ => {
+                    // それでも見つからない場合は終了
+                    return ParseResult {
+                        address: Address::from(tokenizer),
+                        error: Some(Error::new_parse_error(ParseErrorKind::City)),
+                    };
+                }
+            }
+        }
     };
     // その市町村の町名リストを取得
     let city = match api
@@ -254,11 +264,19 @@ pub fn parse_blocking(api: Arc<BlockingApi>, input: &str) -> ParseResult {
         }
         Ok(result) => result,
     };
-    let Ok(tokenizer) = tokenizer.read_city(prefecture.cities) else {
-        return ParseResult {
-            address: Address::from(tokenizer),
-            error: Some(Error::new_parse_error(ParseErrorKind::City)),
-        };
+    let tokenizer = match tokenizer.read_city(&prefecture.cities) {
+        Ok(found) => found,
+        Err(not_found) => {
+            match not_found.read_city_with_county_name_completion(&prefecture.cities) {
+                Ok(found) if cfg!(feature = "city-name-correction") => found,
+                _ => {
+                    return ParseResult {
+                        address: Address::from(tokenizer),
+                        error: Some(Error::new_parse_error(ParseErrorKind::City)),
+                    };
+                }
+            }
+        }
     };
     let city = match api.get_city_master(
         tokenizer.prefecture_name.as_ref().unwrap(),

core/src/parser/adapter.rs

@@ -1,2 +1 @@
 pub mod orthographical_variant_adapter;
-pub mod vague_expression_adapter;

core/src/tokenizer.rs

@@ -1,4 +1,5 @@
 pub(crate) mod read_city;
+pub(crate) mod read_city_with_county_name_completion;
 pub(crate) mod read_prefecture;
 pub(crate) mod read_town;
 
@@ -11,6 +12,8 @@ pub(crate) struct PrefectureNameFound;
 #[derive(Debug)]
 pub(crate) struct CityNameFound;
 #[derive(Debug)]
+pub(crate) struct CityNameNotFound;
+#[derive(Debug)]
 pub(crate) struct TownNameFound;
 #[derive(Debug)]
 pub(crate) struct End;

core/src/tokenizer/read_city.rs

@@ -3,15 +3,14 @@ use std::marker::PhantomData;
 use crate::parser::adapter::orthographical_variant_adapter::{
     OrthographicalVariantAdapter, OrthographicalVariants, Variant,
 };
-use crate::parser::adapter::vague_expression_adapter::VagueExpressionAdapter;
-use crate::tokenizer::{CityNameFound, End, PrefectureNameFound, Tokenizer};
+use crate::tokenizer::{CityNameFound, CityNameNotFound, PrefectureNameFound, Tokenizer};
 
 impl Tokenizer<PrefectureNameFound> {
     pub(crate) fn read_city(
         &self,
-        candidates: Vec<String>,
-    ) -> Result<Tokenizer<CityNameFound>, Tokenizer<End>> {
-        for candidate in &candidates {
+        candidates: &Vec<String>,
+    ) -> Result<Tokenizer<CityNameFound>, Tokenizer<CityNameNotFound>> {
+        for candidate in candidates {
             if self.rest.starts_with(candidate) {
                 return Ok(Tokenizer {
                     input: self.input.clone(),
@@ -69,26 +68,13 @@ impl Tokenizer<PrefectureNameFound> {
             }
         }
 
-        // ここまでで市町村名の特定ができない場合はVagueExpressionAdapterを使用して市町村名を推測する
-        let vague_expression_adapter = VagueExpressionAdapter {};
-        if let Some(result) = vague_expression_adapter.apply(self.rest.as_str(), &candidates) {
-            return Ok(Tokenizer {
-                input: self.input.clone(),
-                prefecture_name: self.prefecture_name.clone(),
-                city_name: Some(result.0),
-                town_name: None,
-                rest: result.1,
-                _state: PhantomData::<CityNameFound>,
-            });
-        }
-
         Err(Tokenizer {
             input: self.input.clone(),
             prefecture_name: self.prefecture_name.clone(),
             city_name: None,
             town_name: None,
             rest: self.rest.clone(),
-            _state: PhantomData::<End>,
+            _state: PhantomData::<CityNameNotFound>,
         })
     }
 }
@@ -108,7 +94,7 @@ mod tests {
             rest: "横浜市保土ケ谷区川辺町2番地9".to_string(),
             _state: PhantomData::<PrefectureNameFound>,
         };
-        let result = tokenizer.read_city(vec![
+        let result = tokenizer.read_city(&vec![
             "横浜市保土ケ谷区".to_string(),
             "横浜市鶴見区".to_string(),
             "横浜市西区".to_string(),
@@ -132,7 +118,7 @@ mod tests {
             rest: "横浜市保土ヶ谷区川辺町2番地9".to_string(),
             _state: PhantomData::<PrefectureNameFound>,
         };
-        let result = tokenizer.read_city(vec![
+        let result = tokenizer.read_city(&vec![
             "横浜市保土ケ谷区".to_string(),
             "横浜市鶴見区".to_string(),
             "横浜市西区".to_string(),
@@ -146,31 +132,6 @@ mod tests {
         assert_eq!(tokenizer.rest, "川辺町2番地9");
     }
 
-    #[test]
-    fn read_city_vague_expression_adapterで成功() {
-        let tokenizer = Tokenizer {
-            input: "埼玉県東秩父村大字御堂634番地".to_string(), // 「秩父郡」が省略されている
-            prefecture_name: Some("埼玉県".to_string()),
-            city_name: None,
-            town_name: None,
-            rest: "東秩父村大字御堂634番地".to_string(),
-            _state: PhantomData::<PrefectureNameFound>,
-        };
-        let result = tokenizer.read_city(vec![
-            "秩父郡皆野町".to_string(),
-            "秩父郡長瀞町".to_string(),
-            "秩父郡小鹿野町".to_string(),
-            "秩父郡東秩父村".to_string(),
-        ]);
-        assert!(result.is_ok());
-        let tokenizer = result.unwrap();
-        assert_eq!(tokenizer.input, "埼玉県東秩父村大字御堂634番地");
-        assert_eq!(tokenizer.prefecture_name, Some("埼玉県".to_string()));
-        assert_eq!(tokenizer.city_name, Some("秩父郡東秩父村".to_string()));
-        assert_eq!(tokenizer.town_name, None);
-        assert_eq!(tokenizer.rest, "大字御堂634番地");
-    }
-
     #[test]
     fn read_city_失敗() {
         let tokenizer = Tokenizer {
@@ -181,7 +142,7 @@ mod tests {
             rest: "京都市上京区川辺町2番地9".to_string(),
             _state: PhantomData::<PrefectureNameFound>,
         };
-        let result = tokenizer.read_city(vec![
+        let result = tokenizer.read_city(&vec![
             "横浜市保土ケ谷区".to_string(),
             "横浜市鶴見区".to_string(),
             "横浜市西区".to_string(),