japanese_address_parser 2.2.0 → 2.2.1

data/js/node_modules/@geolonia/normalize-japanese-addresses/dist/main-node.js

@@ -27,6 +27,17 @@ OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
 PERFORMANCE OF THIS SOFTWARE.
 ***************************************************************************** */
 
+var __assign = function() {
+    __assign = Object.assign || function __assign(t) {
+        for (var s, i = 1, n = arguments.length; i < n; i++) {
+            s = arguments[i];
+            for (var p in s) if (Object.prototype.hasOwnProperty.call(s, p)) t[p] = s[p];
+        }
+        return t;
+    };
+    return __assign.apply(this, arguments);
+};
+
 function __awaiter(thisArg, _arguments, P, generator) {
     function adopt(value) { return value instanceof P ? value : new P(function (resolve) { resolve(value); }); }
     return new (P || (P = Promise))(function (resolve, reject) {
@@ -65,6 +76,11 @@ function __generator(thisArg, body) {
     }
 }
 
+var currentConfig = {
+    japaneseAddressesApi: 'https://geolonia.github.io/japanese-addresses/api/ja',
+    townCacheSize: 1000,
+};
+
 var kan2num = function (string) {
     var kanjiNumbers = japaneseNumeral.findKanjiNumbers(string);
     for (var i = 0; i < kanjiNumbers.length; i++) {
@@ -115,8 +131,8 @@ var patchAddr = function (pref, city, town, addr) {
 };
 
 // JIS 第2水準 => 第1水準 及び 旧字体 => 新字体
-var JIS_OLD_KANJI = '亞,圍,壹,榮,驛,應,櫻,假,會,懷,覺,樂,陷,歡,氣,戲,據,挾,區,徑,溪,輕,藝,儉,圈,權,嚴,恆,國,齋,雜,蠶,殘,兒,實,釋,從,縱,敍,燒,條,剩,壤,釀,眞,盡,醉,髓,聲,竊,淺,錢,禪,爭,插,騷,屬,對,滯,擇,單,斷,癡,鑄,敕,鐵,傳,黨,鬪,屆,腦,廢,發,蠻,拂,邊,瓣,寶,沒,滿,藥,餘,樣,亂,兩,禮,靈,爐,灣,惡,醫,飮,營,圓,歐,奧,價,繪,擴,學,罐,勸,觀,歸,犧,擧,狹,驅,莖,經,繼,缺,劍,檢,顯,廣,鑛,碎,劑,參,慘,絲,辭,舍,壽,澁,肅,將,證,乘,疊,孃,觸,寢,圖,穗,樞,齊,攝,戰,潛,雙,莊,裝,藏,續,體,臺,澤,膽,彈,蟲,廳,鎭,點,燈,盜,獨,貳,霸,賣,髮,祕,佛,變,辯,豐,飜,默,與,譽,謠,覽,獵,勵,齡,勞,壓,爲,隱,衞,鹽,毆,穩,畫,壞,殼,嶽,卷,關,顏,僞,舊,峽,曉,勳,惠,螢,鷄,縣,險,獻,驗,效,號,濟,册,棧,贊,齒,濕,寫,收,獸,處,稱,奬,淨,繩,讓,囑,愼,粹,隨,數,靜,專,踐,纖,壯,搜,總,臟,墮,帶,瀧,擔,團,遲,晝,聽,遞,轉,當,稻,讀,惱,拜,麥,拔,濱,竝,辨,舖,襃,萬,譯,豫,搖,來,龍,壘,隸,戀,樓,鰺,鶯,蠣,攪,竈,灌,諫,頸,礦,蘂,靱,賤,壺,礪,檮,濤,邇,蠅,檜,儘,藪,籠'.split(/,/);
-var JIS_NEW_KANJI = '亜,囲,壱,栄,駅,応,桜,仮,会,懐,覚,楽,陥,歓,気,戯,拠,挟,区,径,渓,軽,芸,倹,圏,権,厳,恒,国,斎,雑,蚕,残,児,実,釈,従,縦,叙,焼,条,剰,壌,醸,真,尽,酔,髄,声,窃,浅,銭,禅,争,挿,騒,属,対,滞,択,単,断,痴,鋳,勅,鉄,伝,党,闘,届,脳,廃,発,蛮,払,辺,弁,宝,没,満,薬,余,様,乱,両,礼,霊,炉,湾,悪,医,飲,営,円,欧,奥,価,絵,拡,学,缶,勧,観,帰,犠,挙,狭,駆,茎,経,継,欠,剣,検,顕,広,鉱,砕,剤,参,惨,糸,辞,舎,寿,渋,粛,将,証,乗,畳,嬢,触,寝,図,穂,枢,斉,摂,戦,潜,双,荘,装,蔵,続,体,台,沢,胆,弾,虫,庁,鎮,点,灯,盗,独,弐,覇,売,髪,秘,仏,変,弁,豊,翻,黙,与,誉,謡,覧,猟,励,齢,労,圧,為,隠,衛,塩,殴,穏,画,壊,殻,岳,巻,関,顔,偽,旧,峡,暁,勲,恵,蛍,鶏,県,険,献,験,効,号,済,冊,桟,賛,歯,湿,写,収,獣,処,称,奨,浄,縄,譲,嘱,慎,粋,随,数,静,専,践,繊,壮,捜,総,臓,堕,帯,滝,担,団,遅,昼,聴,逓,転,当,稲,読,悩,拝,麦,抜,浜,並,弁,舗,褒,万,訳,予,揺,来,竜,塁,隷,恋,楼,鯵,鴬,蛎,撹,竃,潅,諌,頚,砿,蕊,靭,賎,壷,砺,梼,涛,迩,蝿,桧,侭,薮,篭'.split(/,/);
+var JIS_OLD_KANJI = '亞,圍,壹,榮,驛,應,櫻,假,會,懷,覺,樂,陷,歡,氣,戲,據,挾,區,徑,溪,輕,藝,儉,圈,權,嚴,恆,國,齋,雜,蠶,殘,兒,實,釋,從,縱,敍,燒,條,剩,壤,釀,眞,盡,醉,髓,聲,竊,淺,錢,禪,爭,插,騷,屬,對,滯,擇,單,斷,癡,鑄,敕,鐵,傳,黨,鬪,屆,腦,廢,發,蠻,拂,邊,瓣,寶,沒,滿,藥,餘,樣,亂,兩,禮,靈,爐,灣,惡,醫,飮,營,圓,歐,奧,價,繪,擴,學,罐,勸,觀,歸,犧,擧,狹,驅,莖,經,繼,缺,劍,檢,顯,廣,鑛,碎,劑,參,慘,絲,辭,舍,壽,澁,肅,將,證,乘,疊,孃,觸,寢,圖,穗,樞,齊,攝,戰,潛,雙,莊,裝,藏,續,體,臺,澤,膽,彈,蟲,廳,鎭,點,燈,盜,獨,貳,霸,賣,髮,祕,佛,變,辯,豐,飜,默,與,譽,謠,覽,獵,勵,齡,勞,壓,爲,隱,衞,鹽,毆,穩,畫,壞,殼,嶽,卷,關,顏,僞,舊,峽,曉,勳,惠,螢,鷄,縣,險,獻,驗,效,號,濟,册,棧,贊,齒,濕,寫,收,獸,處,稱,奬,淨,繩,讓,囑,愼,粹,隨,數,靜,專,踐,纖,壯,搜,總,臟,墮,帶,瀧,擔,團,遲,晝,聽,遞,轉,當,稻,讀,惱,拜,麥,拔,濱,竝,辨,舖,襃,萬,譯,豫,搖,來,龍,壘,隸,戀,樓,鰺,鶯,蠣,攪,竈,灌,諫,頸,礦,蘂,靱,賤,壺,礪,檮,濤,邇,蠅,檜,儘,藪,籠,彌'.split(/,/);
+var JIS_NEW_KANJI = '亜,囲,壱,栄,駅,応,桜,仮,会,懐,覚,楽,陥,歓,気,戯,拠,挟,区,径,渓,軽,芸,倹,圏,権,厳,恒,国,斎,雑,蚕,残,児,実,釈,従,縦,叙,焼,条,剰,壌,醸,真,尽,酔,髄,声,窃,浅,銭,禅,争,挿,騒,属,対,滞,択,単,断,痴,鋳,勅,鉄,伝,党,闘,届,脳,廃,発,蛮,払,辺,弁,宝,没,満,薬,余,様,乱,両,礼,霊,炉,湾,悪,医,飲,営,円,欧,奥,価,絵,拡,学,缶,勧,観,帰,犠,挙,狭,駆,茎,経,継,欠,剣,検,顕,広,鉱,砕,剤,参,惨,糸,辞,舎,寿,渋,粛,将,証,乗,畳,嬢,触,寝,図,穂,枢,斉,摂,戦,潜,双,荘,装,蔵,続,体,台,沢,胆,弾,虫,庁,鎮,点,灯,盗,独,弐,覇,売,髪,秘,仏,変,弁,豊,翻,黙,与,誉,謡,覧,猟,励,齢,労,圧,為,隠,衛,塩,殴,穏,画,壊,殻,岳,巻,関,顔,偽,旧,峡,暁,勲,恵,蛍,鶏,県,険,献,験,効,号,済,冊,桟,賛,歯,湿,写,収,獣,処,称,奨,浄,縄,譲,嘱,慎,粋,随,数,静,専,践,繊,壮,捜,総,臓,堕,帯,滝,担,団,遅,昼,聴,逓,転,当,稲,読,悩,拝,麦,抜,浜,並,弁,舗,褒,万,訳,予,揺,来,竜,塁,隷,恋,楼,鯵,鴬,蛎,撹,竃,潅,諌,頚,砿,蕊,靭,賎,壷,砺,梼,涛,迩,蝿,桧,侭,薮,篭,弥'.split(/,/);
 var JIS_KANJI_REGEX_PATTERNS = JIS_OLD_KANJI.map(function (old, i) {
     var pattern = old + "|" + JIS_NEW_KANJI[i];
     return [pattern, old, JIS_NEW_KANJI[i]];
@@ -163,11 +179,6 @@ var toRegexPattern = function (string) {
     return _str;
 };
 
-var currentConfig = {
-    japaneseAddressesApi: 'https://geolonia.github.io/japanese-addresses/api/ja',
-    townCacheSize: 1000,
-};
-
 var cachedTownRegexes = new LRU__default['default']({
     max: currentConfig.townCacheSize,
     maxAge: 60 * 60 * 24 * 7 * 1000, // 7日間
@@ -249,8 +260,16 @@ var getTowns = function (pref, city) { return __awaiter(void 0, void 0, void 0,
         }
     });
 }); };
+// 十六町 のように漢数字と町が連結しているか
+var isKanjiNumberFollewedByCho = function (targetTownName) {
+    var xCho = targetTownName.match(/.町/g);
+    if (!xCho)
+        return false;
+    var kanjiNumbers = japaneseNumeral.findKanjiNumbers(xCho[0]);
+    return kanjiNumbers.length > 0;
+};
 var getTownRegexPatterns = function (pref, city) { return __awaiter(void 0, void 0, void 0, function () {
-    var cachedResult, towns, patterns;
+    var cachedResult, pre_towns, townSet, towns, isKyoto, _i, pre_towns_1, town, originalTown, townAbbr, patterns;
     return __generator(this, function (_a) {
         switch (_a.label) {
             case 0:
@@ -258,11 +277,32 @@ var getTownRegexPatterns = function (pref, city) { return __awaiter(void 0, void
                 if (typeof cachedResult !== 'undefined') {
                     return [2 /*return*/, cachedResult];
                 }
-                return [4 /*yield*/, getTowns(pref, city)
-                    // 少ない文字数の地名に対してミスマッチしないように文字の長さ順にソート
-                ];
+                return [4 /*yield*/, getTowns(pref, city)];
             case 1:
-                towns = _a.sent();
+                pre_towns = _a.sent();
+                townSet = new Set(pre_towns.map(function (town) { return town.town; }));
+                towns = [];
+                isKyoto = city.match(/^京都市/);
+                // 町丁目に「○○町」が含まれるケースへの対応
+                // 通常は「○○町」のうち「町」の省略を許容し同義語として扱うが、まれに自治体内に「○○町」と「○○」が共存しているケースがある。
+                // この場合は町の省略は許容せず、入力された住所は書き分けられているものとして正規化を行う。
+                // 更に、「愛知県名古屋市瑞穂区十六町1丁目」漢数字を含むケースだと丁目や番地・号の正規化が不可能になる。このようなケースも除外。
+                for (_i = 0, pre_towns_1 = pre_towns; _i < pre_towns_1.length; _i++) {
+                    town = pre_towns_1[_i];
+                    towns.push(town);
+                    originalTown = town.town;
+                    if (originalTown.indexOf('町') === -1)
+                        continue;
+                    townAbbr = originalTown.replace(/(?!^町)町/g, '') // NOTE: 冒頭の「町」は明らかに省略するべきではないので、除外
+                    ;
+                    if (!isKyoto && // 京都は通り名削除の処理があるため、意図しないマッチになるケースがある。これを除く
+                        !townSet.has(townAbbr) &&
+                        !townSet.has("\u5927\u5B57" + townAbbr) && // 大字は省略されるため、大字〇〇と〇〇町がコンフリクトする。このケースを除外
+                        !isKanjiNumberFollewedByCho(originalTown)) {
+                        // エイリアスとして町なしのパターンを登録
+                        towns.push(__assign(__assign({}, town), { originalTown: originalTown, town: townAbbr }));
+                    }
+                }
                 // 少ない文字数の地名に対してミスマッチしないように文字の長さ順にソート
                 towns.sort(function (a, b) {
                     var aLen = a.town.length;
@@ -303,12 +343,7 @@ var getTownRegexPatterns = function (pref, city) { return __awaiter(void 0, void
                         var _pattern = "(" + patterns.join('|') + ")((\u4E01|\u753A)\u76EE?|\u756A(\u753A|\u4E01)|\u6761|\u8ED2|\u7DDA|\u306E\u753A?|\u5730\u5272|\u53F7|[-\uFF0D\uFE63\u2212\u2010\u2043\u2011\u2012\u2013\u2014\uFE58\u2015\u23AF\u23E4\u30FC\uFF70\u2500\u2501])";
                         return _pattern; // デバッグのときにめんどくさいので変数に入れる。
                     }));
-                    if (city.match(/^京都市/)) {
-                        return [town, ".*" + pattern];
-                    }
-                    else {
-                        return [town, "^" + pattern];
-                    }
+                    return [town, pattern];
                 });
                 cachedTownRegexes.set(pref + "-" + city, patterns);
                 return [2 /*return*/, patterns];
@@ -340,35 +375,48 @@ var getSameNamedPrefectureCityRegexPatterns = function (prefs, prefList) {
     return cachedSameNamedPrefectureCityRegexPatterns;
 };
 
+var config$1 = currentConfig;
 var defaultOption = {
     level: 3,
 };
+/**
+ * @internal
+ */
 var __internals = {
     // default fetch
     fetch: function (input) {
-        var fileURL = new URL("" + currentConfig.japaneseAddressesApi + input);
+        var fileURL = new URL("" + config$1.japaneseAddressesApi + input);
         return unfetch__default['default'](fileURL.toString());
     },
 };
 var normalizeTownName = function (addr, pref, city) { return __awaiter(void 0, void 0, void 0, function () {
-    var townPatterns, i, _a, _town, pattern, match;
-    return __generator(this, function (_b) {
-        switch (_b.label) {
+    var townPatterns, regexPrefixes, _i, regexPrefixes_1, regexPrefix, _a, townPatterns_1, _b, town, pattern, regex, match;
+    return __generator(this, function (_c) {
+        switch (_c.label) {
             case 0:
                 addr = addr.trim().replace(/^大字/, '');
                 return [4 /*yield*/, getTownRegexPatterns(pref, city)];
             case 1:
-                townPatterns = _b.sent();
-                for (i = 0; i < townPatterns.length; i++) {
-                    _a = townPatterns[i], _town = _a[0], pattern = _a[1];
-                    match = addr.match(pattern);
-                    if (match) {
-                        return [2 /*return*/, {
-                                town: _town.town,
-                                addr: addr.substr(match[0].length),
-                                lat: _town.lat,
-                                lng: _town.lng,
-                            }];
+                townPatterns = _c.sent();
+                regexPrefixes = ['^'];
+                if (city.match(/^京都市/)) {
+                    // 京都は通り名削除のために後方一致を使う
+                    regexPrefixes.push('.*');
+                }
+                for (_i = 0, regexPrefixes_1 = regexPrefixes; _i < regexPrefixes_1.length; _i++) {
+                    regexPrefix = regexPrefixes_1[_i];
+                    for (_a = 0, townPatterns_1 = townPatterns; _a < townPatterns_1.length; _a++) {
+                        _b = townPatterns_1[_a], town = _b[0], pattern = _b[1];
+                        regex = new RegExp("" + regexPrefix + pattern);
+                        match = addr.match(regex);
+                        if (match) {
+                            return [2 /*return*/, {
+                                    town: town.originalTown || town.town,
+                                    addr: addr.substr(match[0].length),
+                                    lat: town.lat,
+                                    lng: town.lng,
+                                }];
+                        }
                     }
                 }
                 return [2 /*return*/];
@@ -383,6 +431,7 @@ var normalize$1 = function (address, option) {
             switch (_e.label) {
                 case 0:
                     addr = address
+                        .normalize('NFC')
                         .replace(/　/g, ' ')
                         .replace(/ +/g, ' ')
                         .replace(/([０-９Ａ-Ｚａ-ｚ]+)/g, function (match) {
@@ -490,6 +539,10 @@ var normalize$1 = function (address, option) {
                         addr = normalized.addr;
                         lat = parseFloat(normalized.lat);
                         lng = parseFloat(normalized.lng);
+                        if (Number.isNaN(lat) || Number.isNaN(lng)) {
+                            lat = null;
+                            lng = null;
+                        }
                     }
                     addr = addr
                         .replace(/^-/, '')
@@ -551,7 +604,7 @@ var normalize$1 = function (address, option) {
 var fetchOrReadFile = function (input) { return __awaiter(void 0, void 0, void 0, function () {
     var fileURL, filePath_1;
     return __generator(this, function (_a) {
-        fileURL = new URL("" + currentConfig.japaneseAddressesApi + input);
+        fileURL = new URL("" + config$1.japaneseAddressesApi + input);
         if (fileURL.protocol === 'http:' || fileURL.protocol === 'https:') {
             return [2 /*return*/, unfetch__default['default'](fileURL.toString())];
         }
@@ -577,7 +630,7 @@ var fetchOrReadFile = function (input) { return __awaiter(void 0, void 0, void 0
     });
 }); };
 __internals.fetch = fetchOrReadFile;
-var config = currentConfig;
+var config = config$1;
 var normalize = normalize$1;
 
 exports.config = config;

data/js/node_modules/@geolonia/normalize-japanese-addresses/dist/normalize.d.ts

@@ -1,3 +1,13 @@
+/**
+ * normalize {@link Normalizer} の動作オプション。
+ */
+export interface Config {
+    /** 住所データを URL 形式で指定。 file:// 形式で指定するとローカルファイルを参照できます。 */
+    japaneseAddressesApi: string;
+    /** 町丁目のデータを何件までキャッシュするか。デフォルト 1,000 */
+    townCacheSize: number;
+}
+export declare const config: Config;
 /**
  * 住所の正規化結果として戻されるオブジェクト
  */
@@ -48,6 +58,9 @@ export declare type Normalizer = (input: string, option?: Option) => Promise<Nor
 export declare type FetchLike = (input: string) => Promise<Response | {
     json: () => Promise<unknown>;
 }>;
+/**
+ * @internal
+ */
 export declare const __internals: {
     fetch: FetchLike;
 };

data/js/node_modules/@geolonia/normalize-japanese-addresses/package.json

@@ -1,17 +1,17 @@
 {
   "name": "@geolonia/normalize-japanese-addresses",
-  "version": "2.5.1",
+  "version": "2.5.8",
   "description": "",
   "browser": "./dist/main-browser.js",
   "main": "./dist/main-node.js",
-  "types": "./dist/main-node.d.ts",
+  "types": "./dist/normalize.d.ts",
   "scripts": {
     "test": "npm run test:main && npm run test:addresses && npm run test:node",
     "test:main": "jest test/main.test.ts",
     "test:addresses": "jest test/addresses.test.ts",
     "test:node": "curl -sL https://github.com/geolonia/japanese-addresses/archive/refs/heads/master.tar.gz | tar xvfz - -C ./test > /dev/null 2>&1 && jest test/fs.test.ts",
     "test:generate-test-data": "npx ts-node -O '{\"module\":\"commonjs\"}' test/build-test-data.ts > test/addresses.csv",
-    "lint": "eslint \"src/**/*.ts\" --fix",
+    "lint": "eslint \"src/**/*.ts\" \"test/**/*.test.ts\" --fix",
     "build": "rm -rf ./dist && rollup -c ./rollup.config.ts"
   },
   "author": "",
@@ -51,7 +51,7 @@
     }
   },
   "dependencies": {
-    "@geolonia/japanese-numeral": "^0.1.15",
+    "@geolonia/japanese-numeral": "^0.1.16",
     "isomorphic-unfetch": "^3.1.0",
     "lru-cache": "^6.0.0"
   }

data/js/node_modules/@geolonia/normalize-japanese-addresses/src/config.ts

@@ -1,9 +1,4 @@
-export interface Config {
-  japaneseAddressesApi: string
-
-  /** 町丁目のデータを何件までキャッシュするか。デフォルト 1,000 */
-  townCacheSize: number
-}
+import { Config } from './normalize'
 
 export const currentConfig: Config = {
   japaneseAddressesApi: 'https://geolonia.github.io/japanese-addresses/api/ja',

data/js/node_modules/@geolonia/normalize-japanese-addresses/src/lib/cacheRegexes.ts

@@ -1,12 +1,14 @@
 import { toRegexPattern } from './dict'
 import { kan2num } from './kan2num'
-import { currentConfig } from '../config'
 import LRU from 'lru-cache'
+import { currentConfig } from '../config'
 import { __internals } from '../normalize'
+import { findKanjiNumbers } from '@geolonia/japanese-numeral'
 
 type PrefectureList = { [key: string]: string[] }
 interface SingleTown {
   town: string
+  originalTown?: string
   koaza: string
   lat: string
   lng: string
@@ -22,8 +24,9 @@ let cachedPrefecturePatterns: [string, string][] | undefined = undefined
 const cachedCityPatterns: { [key: string]: [string, string][] } = {}
 let cachedPrefectures: PrefectureList | undefined = undefined
 const cachedTowns: { [key: string]: TownList } = {}
-let cachedSameNamedPrefectureCityRegexPatterns: [string, string][] | undefined =
-  undefined
+let cachedSameNamedPrefectureCityRegexPatterns:
+  | [string, string][]
+  | undefined = undefined
 
 export const getPrefectures = async () => {
   if (typeof cachedPrefectures !== 'undefined') {
@@ -90,13 +93,50 @@ export const getTowns = async (pref: string, city: string) => {
   return (cachedTowns[cacheKey] = towns)
 }
 
+// 十六町 のように漢数字と町が連結しているか
+const isKanjiNumberFollewedByCho = (targetTownName: string) => {
+  const xCho = targetTownName.match(/.町/g)
+  if (!xCho) return false
+  const kanjiNumbers = findKanjiNumbers(xCho[0])
+  return kanjiNumbers.length > 0
+}
+
 export const getTownRegexPatterns = async (pref: string, city: string) => {
   const cachedResult = cachedTownRegexes.get(`${pref}-${city}`)
   if (typeof cachedResult !== 'undefined') {
     return cachedResult
   }
 
-  const towns = await getTowns(pref, city)
+  const pre_towns = await getTowns(pref, city)
+  const townSet = new Set(pre_towns.map((town) => town.town))
+  const towns = []
+
+  const isKyoto = city.match(/^京都市/)
+
+  // 町丁目に「○○町」が含まれるケースへの対応
+  // 通常は「○○町」のうち「町」の省略を許容し同義語として扱うが、まれに自治体内に「○○町」と「○○」が共存しているケースがある。
+  // この場合は町の省略は許容せず、入力された住所は書き分けられているものとして正規化を行う。
+  // 更に、「愛知県名古屋市瑞穂区十六町1丁目」漢数字を含むケースだと丁目や番地・号の正規化が不可能になる。このようなケースも除外。
+  for (const town of pre_towns) {
+    towns.push(town)
+
+    const originalTown = town.town
+    if (originalTown.indexOf('町') === -1) continue
+    const townAbbr = originalTown.replace(/(?!^町)町/g, '') // NOTE: 冒頭の「町」は明らかに省略するべきではないので、除外
+    if (
+      !isKyoto && // 京都は通り名削除の処理があるため、意図しないマッチになるケースがある。これを除く
+      !townSet.has(townAbbr) &&
+      !townSet.has(`大字${townAbbr}`) && // 大字は省略されるため、大字〇〇と〇〇町がコンフリクトする。このケースを除外
+      !isKanjiNumberFollewedByCho(originalTown)
+    ) {
+      // エイリアスとして町なしのパターンを登録
+      towns.push({
+        ...town,
+        originalTown,
+        town: townAbbr,
+      })
+    }
+  }
 
   // 少ない文字数の地名に対してミスマッチしないように文字の長さ順にソート
   towns.sort((a, b) => {
@@ -153,11 +193,7 @@ export const getTownRegexPatterns = async (pref: string, city: string) => {
         ),
     )
 
-    if (city.match(/^京都市/)) {
-      return [town, `.*${pattern}`]
-    } else {
-      return [town, `^${pattern}`]
-    }
+    return [town, pattern]
   }) as [SingleTown, string][]
 
   cachedTownRegexes.set(`${pref}-${city}`, patterns)

data/js/node_modules/@geolonia/normalize-japanese-addresses/src/lib/dict.ts

@@ -1,8 +1,8 @@
 // JIS 第2水準 => 第1水準 及び 旧字体 => 新字体
-const JIS_OLD_KANJI = '亞,圍,壹,榮,驛,應,櫻,假,會,懷,覺,樂,陷,歡,氣,戲,據,挾,區,徑,溪,輕,藝,儉,圈,權,嚴,恆,國,齋,雜,蠶,殘,兒,實,釋,從,縱,敍,燒,條,剩,壤,釀,眞,盡,醉,髓,聲,竊,淺,錢,禪,爭,插,騷,屬,對,滯,擇,單,斷,癡,鑄,敕,鐵,傳,黨,鬪,屆,腦,廢,發,蠻,拂,邊,瓣,寶,沒,滿,藥,餘,樣,亂,兩,禮,靈,爐,灣,惡,醫,飮,營,圓,歐,奧,價,繪,擴,學,罐,勸,觀,歸,犧,擧,狹,驅,莖,經,繼,缺,劍,檢,顯,廣,鑛,碎,劑,參,慘,絲,辭,舍,壽,澁,肅,將,證,乘,疊,孃,觸,寢,圖,穗,樞,齊,攝,戰,潛,雙,莊,裝,藏,續,體,臺,澤,膽,彈,蟲,廳,鎭,點,燈,盜,獨,貳,霸,賣,髮,祕,佛,變,辯,豐,飜,默,與,譽,謠,覽,獵,勵,齡,勞,壓,爲,隱,衞,鹽,毆,穩,畫,壞,殼,嶽,卷,關,顏,僞,舊,峽,曉,勳,惠,螢,鷄,縣,險,獻,驗,效,號,濟,册,棧,贊,齒,濕,寫,收,獸,處,稱,奬,淨,繩,讓,囑,愼,粹,隨,數,靜,專,踐,纖,壯,搜,總,臟,墮,帶,瀧,擔,團,遲,晝,聽,遞,轉,當,稻,讀,惱,拜,麥,拔,濱,竝,辨,舖,襃,萬,譯,豫,搖,來,龍,壘,隸,戀,樓,鰺,鶯,蠣,攪,竈,灌,諫,頸,礦,蘂,靱,賤,壺,礪,檮,濤,邇,蠅,檜,儘,藪,籠'.split(
+const JIS_OLD_KANJI = '亞,圍,壹,榮,驛,應,櫻,假,會,懷,覺,樂,陷,歡,氣,戲,據,挾,區,徑,溪,輕,藝,儉,圈,權,嚴,恆,國,齋,雜,蠶,殘,兒,實,釋,從,縱,敍,燒,條,剩,壤,釀,眞,盡,醉,髓,聲,竊,淺,錢,禪,爭,插,騷,屬,對,滯,擇,單,斷,癡,鑄,敕,鐵,傳,黨,鬪,屆,腦,廢,發,蠻,拂,邊,瓣,寶,沒,滿,藥,餘,樣,亂,兩,禮,靈,爐,灣,惡,醫,飮,營,圓,歐,奧,價,繪,擴,學,罐,勸,觀,歸,犧,擧,狹,驅,莖,經,繼,缺,劍,檢,顯,廣,鑛,碎,劑,參,慘,絲,辭,舍,壽,澁,肅,將,證,乘,疊,孃,觸,寢,圖,穗,樞,齊,攝,戰,潛,雙,莊,裝,藏,續,體,臺,澤,膽,彈,蟲,廳,鎭,點,燈,盜,獨,貳,霸,賣,髮,祕,佛,變,辯,豐,飜,默,與,譽,謠,覽,獵,勵,齡,勞,壓,爲,隱,衞,鹽,毆,穩,畫,壞,殼,嶽,卷,關,顏,僞,舊,峽,曉,勳,惠,螢,鷄,縣,險,獻,驗,效,號,濟,册,棧,贊,齒,濕,寫,收,獸,處,稱,奬,淨,繩,讓,囑,愼,粹,隨,數,靜,專,踐,纖,壯,搜,總,臟,墮,帶,瀧,擔,團,遲,晝,聽,遞,轉,當,稻,讀,惱,拜,麥,拔,濱,竝,辨,舖,襃,萬,譯,豫,搖,來,龍,壘,隸,戀,樓,鰺,鶯,蠣,攪,竈,灌,諫,頸,礦,蘂,靱,賤,壺,礪,檮,濤,邇,蠅,檜,儘,藪,籠,彌'.split(
   /,/,
 )
-const JIS_NEW_KANJI = '亜,囲,壱,栄,駅,応,桜,仮,会,懐,覚,楽,陥,歓,気,戯,拠,挟,区,径,渓,軽,芸,倹,圏,権,厳,恒,国,斎,雑,蚕,残,児,実,釈,従,縦,叙,焼,条,剰,壌,醸,真,尽,酔,髄,声,窃,浅,銭,禅,争,挿,騒,属,対,滞,択,単,断,痴,鋳,勅,鉄,伝,党,闘,届,脳,廃,発,蛮,払,辺,弁,宝,没,満,薬,余,様,乱,両,礼,霊,炉,湾,悪,医,飲,営,円,欧,奥,価,絵,拡,学,缶,勧,観,帰,犠,挙,狭,駆,茎,経,継,欠,剣,検,顕,広,鉱,砕,剤,参,惨,糸,辞,舎,寿,渋,粛,将,証,乗,畳,嬢,触,寝,図,穂,枢,斉,摂,戦,潜,双,荘,装,蔵,続,体,台,沢,胆,弾,虫,庁,鎮,点,灯,盗,独,弐,覇,売,髪,秘,仏,変,弁,豊,翻,黙,与,誉,謡,覧,猟,励,齢,労,圧,為,隠,衛,塩,殴,穏,画,壊,殻,岳,巻,関,顔,偽,旧,峡,暁,勲,恵,蛍,鶏,県,険,献,験,効,号,済,冊,桟,賛,歯,湿,写,収,獣,処,称,奨,浄,縄,譲,嘱,慎,粋,随,数,静,専,践,繊,壮,捜,総,臓,堕,帯,滝,担,団,遅,昼,聴,逓,転,当,稲,読,悩,拝,麦,抜,浜,並,弁,舗,褒,万,訳,予,揺,来,竜,塁,隷,恋,楼,鯵,鴬,蛎,撹,竃,潅,諌,頚,砿,蕊,靭,賎,壷,砺,梼,涛,迩,蝿,桧,侭,薮,篭'.split(
+const JIS_NEW_KANJI = '亜,囲,壱,栄,駅,応,桜,仮,会,懐,覚,楽,陥,歓,気,戯,拠,挟,区,径,渓,軽,芸,倹,圏,権,厳,恒,国,斎,雑,蚕,残,児,実,釈,従,縦,叙,焼,条,剰,壌,醸,真,尽,酔,髄,声,窃,浅,銭,禅,争,挿,騒,属,対,滞,択,単,断,痴,鋳,勅,鉄,伝,党,闘,届,脳,廃,発,蛮,払,辺,弁,宝,没,満,薬,余,様,乱,両,礼,霊,炉,湾,悪,医,飲,営,円,欧,奥,価,絵,拡,学,缶,勧,観,帰,犠,挙,狭,駆,茎,経,継,欠,剣,検,顕,広,鉱,砕,剤,参,惨,糸,辞,舎,寿,渋,粛,将,証,乗,畳,嬢,触,寝,図,穂,枢,斉,摂,戦,潜,双,荘,装,蔵,続,体,台,沢,胆,弾,虫,庁,鎮,点,灯,盗,独,弐,覇,売,髪,秘,仏,変,弁,豊,翻,黙,与,誉,謡,覧,猟,励,齢,労,圧,為,隠,衛,塩,殴,穏,画,壊,殻,岳,巻,関,顔,偽,旧,峡,暁,勲,恵,蛍,鶏,県,険,献,験,効,号,済,冊,桟,賛,歯,湿,写,収,獣,処,称,奨,浄,縄,譲,嘱,慎,粋,随,数,静,専,践,繊,壮,捜,総,臓,堕,帯,滝,担,団,遅,昼,聴,逓,転,当,稲,読,悩,拝,麦,抜,浜,並,弁,舗,褒,万,訳,予,揺,来,竜,塁,隷,恋,楼,鯵,鴬,蛎,撹,竃,潅,諌,頚,砿,蕊,靭,賎,壷,砺,梼,涛,迩,蝿,桧,侭,薮,篭,弥'.split(
   /,/,
 )
 

data/js/node_modules/@geolonia/normalize-japanese-addresses/src/main-browser.ts

@@ -1,5 +1,4 @@
 import * as Normalize from './normalize'
-import { currentConfig } from './config'
 
-export const config = currentConfig
+export const config = Normalize.config
 export const normalize = Normalize.normalize

data/js/node_modules/@geolonia/normalize-japanese-addresses/src/main-node.ts

@@ -1,12 +1,11 @@
 import * as Normalize from './normalize'
-import { currentConfig } from './config'
 import { promises as fs } from 'fs'
 import unfetch from 'isomorphic-unfetch'
 
 const fetchOrReadFile = async (
   input: string,
 ): Promise<Response | { json: () => Promise<unknown> }> => {
-  const fileURL = new URL(`${currentConfig.japaneseAddressesApi}${input}`)
+  const fileURL = new URL(`${Normalize.config.japaneseAddressesApi}${input}`)
   if (fileURL.protocol === 'http:' || fileURL.protocol === 'https:') {
     return unfetch(fileURL.toString())
   } else if (fileURL.protocol === 'file:') {
@@ -23,5 +22,5 @@ const fetchOrReadFile = async (
 }
 
 Normalize.__internals.fetch = fetchOrReadFile
-export const config = currentConfig
+export const config = Normalize.config
 export const normalize = Normalize.normalize

data/js/node_modules/@geolonia/normalize-japanese-addresses/src/normalize.ts

@@ -1,5 +1,5 @@
 import { number2kanji } from '@geolonia/japanese-numeral'
-
+import { currentConfig } from './config'
 import { kan2num } from './lib/kan2num'
 import { zen2han } from './lib/zen2han'
 import { patchAddr } from './lib/patchAddr'
@@ -11,7 +11,18 @@ import {
   getSameNamedPrefectureCityRegexPatterns,
 } from './lib/cacheRegexes'
 import unfetch from 'isomorphic-unfetch'
-import { currentConfig } from './config'
+
+/**
+ * normalize {@link Normalizer} の動作オプション。
+ */
+export interface Config {
+  /** 住所データを URL 形式で指定。 file:// 形式で指定するとローカルファイルを参照できます。 */
+  japaneseAddressesApi: string
+
+  /** 町丁目のデータを何件までキャッシュするか。デフォルト 1,000 */
+  townCacheSize: number
+}
+export const config: Config = currentConfig
 
 /**
  * 住所の正規化結果として戻されるオブジェクト
@@ -74,10 +85,13 @@ const defaultOption: Option = {
   level: 3,
 }
 
+/**
+ * @internal
+ */
 export const __internals: { fetch: FetchLike } = {
   // default fetch
   fetch: (input: string) => {
-    const fileURL = new URL(`${currentConfig.japaneseAddressesApi}${input}`)
+    const fileURL = new URL(`${config.japaneseAddressesApi}${input}`)
     return unfetch(fileURL.toString())
   },
 }
@@ -86,16 +100,23 @@ const normalizeTownName = async (addr: string, pref: string, city: string) => {
   addr = addr.trim().replace(/^大字/, '')
   const townPatterns = await getTownRegexPatterns(pref, city)
 
-  for (let i = 0; i < townPatterns.length; i++) {
-    const [_town, pattern] = townPatterns[i]
-    const match = addr.match(pattern)
+  const regexPrefixes = ['^']
+  if (city.match(/^京都市/)) {
+    // 京都は通り名削除のために後方一致を使う
+    regexPrefixes.push('.*')
+  }
 
-    if (match) {
-      return {
-        town: _town.town,
-        addr: addr.substr(match[0].length),
-        lat: _town.lat,
-        lng: _town.lng,
+  for (const regexPrefix of regexPrefixes) {
+    for (const [town, pattern] of townPatterns) {
+      const regex = new RegExp(`${regexPrefix}${pattern}`)
+      const match = addr.match(regex)
+      if (match) {
+        return {
+          town: town.originalTown || town.town,
+          addr: addr.substr(match[0].length),
+          lat: town.lat,
+          lng: town.lng,
+        }
       }
     }
   }
@@ -113,6 +134,7 @@ export const normalize: Normalizer = async (
    * 3. 最初に出てくる `1-` や `五-` のような文字列を町丁目とみなして、それ以前のスペースをすべて削除する。
    */
   let addr = address
+    .normalize('NFC')
     .replace(/　/g, ' ')
     .replace(/ +/g, ' ')
     .replace(/([０-９Ａ-Ｚａ-ｚ]+)/g, (match) => {
@@ -145,8 +167,10 @@ export const normalize: Normalizer = async (
   const prefectures = await getPrefectures()
   const prefs = Object.keys(prefectures)
   const prefPatterns = getPrefectureRegexPatterns(prefs)
-  const sameNamedPrefectureCityRegexPatterns =
-    getSameNamedPrefectureCityRegexPatterns(prefs, prefectures)
+  const sameNamedPrefectureCityRegexPatterns = getSameNamedPrefectureCityRegexPatterns(
+    prefs,
+    prefectures,
+  )
 
   // 県名が省略されており、かつ市の名前がどこかの都道府県名と同じ場合(例.千葉県千葉市)、
   // あらかじめ県名を補完しておく。
@@ -223,7 +247,7 @@ export const normalize: Normalizer = async (
     }
   }
 
-  // 町丁目以降の正規化'
+  // 町丁目以降の正規化
   if (city && option.level >= 3) {
     const normalized = await normalizeTownName(addr, pref, city)
     if (normalized) {
@@ -231,6 +255,10 @@ export const normalize: Normalizer = async (
       addr = normalized.addr
       lat = parseFloat(normalized.lat)
       lng = parseFloat(normalized.lng)
+      if (Number.isNaN(lat) || Number.isNaN(lng)) {
+        lat = null
+        lng = null
+      }
     }
 
     addr = addr

data/js/package-lock.json

@@ -11,16 +11,16 @@
       "devDependencies": {}
     },
     "node_modules/@geolonia/japanese-numeral": {
-      "version": "0.1.15",
-      "resolved": "https://registry.npmjs.org/@geolonia/japanese-numeral/-/japanese-numeral-0.1.15.tgz",
-      "integrity": "sha512-g8xizIJH7sfd5JSJyN6pCbNNWVkasMpw84gpYQZJ09VDuTT5qzT5cKhr3+pFyS4SpfwQ9rrsJBmeD6z3vI2bjA=="
+      "version": "0.1.16",
+      "resolved": "https://registry.npmjs.org/@geolonia/japanese-numeral/-/japanese-numeral-0.1.16.tgz",
+      "integrity": "sha512-xzSsIHhyyjqNpW8qSh5bFMu3FJvyJGBbNZ/t8clPHkigjUdZ7Ck8wkzCkcwlVd00RkoHwGDLnvXx5W0WiGK0TQ=="
     },
     "node_modules/@geolonia/normalize-japanese-addresses": {
-      "version": "2.5.1",
-      "resolved": "https://registry.npmjs.org/@geolonia/normalize-japanese-addresses/-/normalize-japanese-addresses-2.5.1.tgz",
-      "integrity": "sha512-aIAkkyNnK7AVtP7ArTxNtx11pNpuQF3xiQ60PVG0N0MkpZom5qQ2JpgthHFdxh5ZEQA/ol5yFqPNdopKRIXytA==",
+      "version": "2.5.8",
+      "resolved": "https://registry.npmjs.org/@geolonia/normalize-japanese-addresses/-/normalize-japanese-addresses-2.5.8.tgz",
+      "integrity": "sha512-mv3+vHU2qXPjn2wbZH2p8kOKp7zTdcwCAWz8yryyyM5OGCRjHiqGBg+1t78Ut8b0wopyLJ7GjNanIoflEEKVeg==",
       "dependencies": {
-        "@geolonia/japanese-numeral": "^0.1.15",
+        "@geolonia/japanese-numeral": "^0.1.16",
         "isomorphic-unfetch": "^3.1.0",
         "lru-cache": "^6.0.0"
       }
@@ -96,16 +96,16 @@
   },
   "dependencies": {
     "@geolonia/japanese-numeral": {
-      "version": "0.1.15",
-      "resolved": "https://registry.npmjs.org/@geolonia/japanese-numeral/-/japanese-numeral-0.1.15.tgz",
-      "integrity": "sha512-g8xizIJH7sfd5JSJyN6pCbNNWVkasMpw84gpYQZJ09VDuTT5qzT5cKhr3+pFyS4SpfwQ9rrsJBmeD6z3vI2bjA=="
+      "version": "0.1.16",
+      "resolved": "https://registry.npmjs.org/@geolonia/japanese-numeral/-/japanese-numeral-0.1.16.tgz",
+      "integrity": "sha512-xzSsIHhyyjqNpW8qSh5bFMu3FJvyJGBbNZ/t8clPHkigjUdZ7Ck8wkzCkcwlVd00RkoHwGDLnvXx5W0WiGK0TQ=="
     },
     "@geolonia/normalize-japanese-addresses": {
-      "version": "2.5.1",
-      "resolved": "https://registry.npmjs.org/@geolonia/normalize-japanese-addresses/-/normalize-japanese-addresses-2.5.1.tgz",
-      "integrity": "sha512-aIAkkyNnK7AVtP7ArTxNtx11pNpuQF3xiQ60PVG0N0MkpZom5qQ2JpgthHFdxh5ZEQA/ol5yFqPNdopKRIXytA==",
+      "version": "2.5.8",
+      "resolved": "https://registry.npmjs.org/@geolonia/normalize-japanese-addresses/-/normalize-japanese-addresses-2.5.8.tgz",
+      "integrity": "sha512-mv3+vHU2qXPjn2wbZH2p8kOKp7zTdcwCAWz8yryyyM5OGCRjHiqGBg+1t78Ut8b0wopyLJ7GjNanIoflEEKVeg==",
       "requires": {
-        "@geolonia/japanese-numeral": "^0.1.15",
+        "@geolonia/japanese-numeral": "^0.1.16",
         "isomorphic-unfetch": "^3.1.0",
         "lru-cache": "^6.0.0"
       }

data/lib/japanese_address_parser/data/01-01204.csv

@@ -1998,7 +1998,7 @@ name,name_kana,name_romaji,nickname,latitude,longitude
 東旭川町桜岡,ヒガシアサヒカワチョウサクラオカ,HIGASHIASAHIKAWACHO SAKURAOKA,,43.794973,142.481861
 東旭川町下兵村,ヒガシアサヒカワチョウシモヘイソン,HIGASHIASAHIKAWACHO SHIMOHEISON,,43.759467,142.435123
 東旭川町忠別,ヒガシアサヒカワチョウチュウベツ,HIGASHIASAHIKAWACHO CHUBETSU,,43.721892,142.467918
-東旭川町豊田,ヒガシアサヒカワチョウトヨタ,HIGASHIASAHIKAWACHO TOYOTA(1-9-BANC,,43.767835,142.503332
+東旭川町豊田,ヒガシアサヒカワチョウトヨタ,HIGASHIASAHIKAWACHO TOYOTA,,43.767835,142.503332
 東旭川町東桜岡,ヒガシアサヒカワチョウヒガシサクラオカ,HIGASHIASAHIKAWACHO HIGASHISAKURAOK,,43.781441,142.514455
 東旭川町日ノ出,ヒガシアサヒカワチョウヒノデ,HIGASHIASAHIKAWACHO HINODE,,43.767631,142.464338
 東旭川町瑞穂,ヒガシアサヒカワチョウミズホ,HIGASHIASAHIKAWACHO MIZUHO,,43.72067,142.658772

data/lib/japanese_address_parser/data/01-01206.csv

@@ -630,7 +630,7 @@ name,name_kana,name_romaji,nickname,latitude,longitude
 阿寒町上舌辛二十線,,,,43.139904,144.116077
 音別町中音別,オンベツチョウナカオンベツ,OMBETSUCHO NAKAOMBETSU,,42.943491,143.933951
 音別町馬主来原野,,,,42.943918,143.970656
-音別町音別,オンベツチョウオンベツゲンヤキセン,OMBETSUCHO OMBETSUGENYAKISEN(FUTAMA,,42.937629,143.886981
+音別町音別,オンベツチョウオンベツゲンヤキセン,OMBETSUCHO OMBETSUGENYAKISEN,,42.937629,143.886981
 音別町音別原野西五線,,,,42.916643,143.910843
 音別町音別原野西四線,,,,42.916038,143.913518
 音別町尺別原野東二線,,,,42.874096,143.909832
@@ -663,7 +663,7 @@ name,name_kana,name_romaji,nickname,latitude,longitude
 音別町音別原野東二線,,,,43.052403,143.899845
 音別町音別原野西二線,,,,43.003435,143.86926
 音別町音別原野西一線,,,,43.022109,143.865227
-音別町音別原野基線,オンベツチョウオンベツゲンヤキセン,OMBETSUCHO OMBETSUGENYAKISEN(FUTAMA,,43.012674,143.880033
+音別町音別原野基線,オンベツチョウオンベツゲンヤキセン,OMBETSUCHO OMBETSUGENYAKISEN,,43.012674,143.880033
 音別町チャンベツ右岸,,,,43.002629,143.879439
 阿寒町シュリコマベツ,,,,43.423547,144.088424
 阿寒町ルベシベ,,,,43.329351,144.136311

data/lib/japanese_address_parser/data/01-01210.csv

@@ -517,7 +517,7 @@ name,name_kana,name_romaji,nickname,latitude,longitude
 栗沢町万字睦町,クリサワチョウマンジムツミチョウ,KURISAWACHO MANJIMUTSUMICHO,,43.141359,141.983121
 栗沢町南幸穂,クリサワチョウミナミサチホ,KURISAWACHO MINAMISACHIHO,,43.113302,141.743097
 栗沢町南本町,クリサワチョウミナミホンチョウ,KURISAWACHO MINAMIHONCHO,,43.115685,141.741278
-栗沢町宮村,クリサワチョウミヤムラ,KURISAWACHO MIYAMURA(248.339.726.78,,43.146497,141.858
+栗沢町宮村,クリサワチョウミヤムラ,KURISAWACHO MIYAMURA,,43.146497,141.858
 栗沢町美流渡楓町,クリサワチョウミルトカエデチョウ,KURISAWACHO MIRUTOKAEDECHO,,43.124886,141.941392
 栗沢町美流渡栄町,クリサワチョウミルトサカエチョウ,KURISAWACHO MIRUTOSAKAECHO,,43.158414,141.91956
 栗沢町美流渡桜町,クリサワチョウミルトサクラマチ,KURISAWACHO MIRUTOSAKURAMACHI,,43.150256,141.922398