[已完成]: getArea (#5)

* feat(utility): add get-area utility function

* fix(@xstools/utility): small fix

* fix(utility): improve getArea utility function and addressNormalizer

* feat(utility): improve getArea utility function and addressNormalizer

* refactor(utility): optimize address cleaning functions

* refactor: rename

* test: update getDistrict test

* docs: update changeset

---------

Co-authored-by: bingtsingw <10382462+bingtsingw@users.noreply.github.com>

.changeset/eight-moose-push.md

@@ -0,0 +1,5 @@
+---
+'@xstools/utility': minor
+---
+
+增加 getDistrict: 根据title address 获取城市区名

packages/utility/src/misc/get-district.test.ts

@@ -0,0 +1,76 @@
+import { describe, expect, test } from 'bun:test';
+import { addressTrimEnd, addressTrimParenthesis, districtStartWith, getDistrict } from './get-district';
+
+describe('misc', () => {
+  test('getDistrict', () => {
+    // 标准地址
+    [
+      {
+        address: '北京市北京市大兴区行远街道101号',
+        title: '星巴克',
+        expected: '大兴区',
+      },
+
+      {
+        address: '广东省深圳市罗湖区科技园路0号',
+        title: '',
+        expected: '罗湖区',
+      },
+      {
+        address: '北京市北京市大兴区行远街道101号',
+        title: '大兴星巴克',
+        expected: '',
+      },
+    ].forEach(({ address, title, expected }) => expect(getDistrict({ address, title })).toEqual(expected));
+
+    // 特殊地址
+    [
+      {
+        address: '',
+        title: '',
+        expected: '',
+      },
+      {
+        address: '',
+        title: '大兴星巴克',
+        expected: '',
+      },
+      {
+        address: '上海市上海市黄浦区南京西路325号上海市历史博物馆内',
+        title: 'COFFEE满坡栗咖啡(上海历史博物馆臻选店）',
+        expected: '黄浦区',
+      },
+      {
+        address: '上海市静安区乌鲁木齐北路4号(近市西中学后门)',
+        title: 'SilverFlow银流咖啡(元善里店)',
+        expected: '静安区',
+      },
+      {
+        address: '吉林省长春市朝阳区吉大南校北门致远街408号',
+        title: 'Im Fine Cafe Bar 吉林省长春市朝阳区吉大南校',
+        expected: '朝阳区',
+      },
+    ].forEach(({ address, title, expected }) => expect(getDistrict({ address, title })).toEqual(expected));
+  });
+
+  test('addressTrimParenthesis', () => {
+    expect(addressTrimParenthesis('hello')).toEqual('hello');
+    expect(addressTrimParenthesis('北京市海淀区知春路[五道口]10号')).toEqual('北京市海淀区知春路10号');
+    expect(addressTrimParenthesis('上海市黄浦区南京东路100号(步行街)')).toEqual('上海市黄浦区南京东路100号');
+    expect(addressTrimParenthesis('深圳市南山区高新南一道(软件园)[南门]300号')).toEqual('深圳市南山区高新南一道300号');
+  });
+
+  test('addressTrimEnd', () => {
+    expect(addressTrimEnd('hello')).toEqual('hello');
+    expect(addressTrimEnd('北京市海淀区知春路10号xxxx')).toEqual('北京市海淀区知春路10号');
+  });
+
+  test('districtStartWith', () => {
+    expect(districtStartWith({ title: '朝阳区xx酒吧', district: '朝阳区' })).toBeFalse();
+    expect(districtStartWith({ title: 'xx朝阳区xx酒吧', district: '朝阳区' })).toBeFalse();
+    expect(districtStartWith({ title: 'xxx朝阳区xx酒吧', district: '朝阳区' })).toBeFalse();
+
+    expect(districtStartWith({ title: 'xxxx朝阳区xx酒吧', district: '朝阳区' })).toBeTrue();
+    expect(districtStartWith({ title: 'xx酒吧', district: '朝阳区' })).toBeTrue();
+  });
+});

packages/utility/src/misc/get-district.ts

@@ -0,0 +1,57 @@
+export const addressTrimParenthesis = (address: string) => {
+  const brackets = ['()', '[]', '{}', '<>', '（）', '【】', '｛｝', '《》', '〈〉', '〔〕', '〖〗', '〘〙'];
+  const left: string[] = [];
+  const right: string[] = [];
+
+  brackets.forEach((bracket) => {
+    const [open, close] = bracket.split('').filter((bracket) => bracket.trim());
+    left.push(open!);
+    right.push(close!);
+  });
+
+  const regexParts = left.flatMap((open) => right.map((close) => `\\${open}[^\\${open}\\${close}]*\\${close}`));
+
+  const regexPattern = regexParts.join('|');
+  return address.replace(new RegExp(regexPattern, 'g'), '');
+};
+
+export const addressTrimEnd = (address: string) => {
+  return address.replace(/((?:街|道|路)\d+号).*/, '$1');
+};
+
+/**
+ * 判断是否为有效的区域
+ * - 如果 title 中从第 0 到第 3 个字符开始的子串与 district 或去掉“区”后的 district 相匹配，则认为无效。
+ * - 后续可以添加更多的规则
+ */
+export const districtStartWith = ({ title, district }: { title: string; district: string }): boolean => {
+  const trimmedDistrict = district.replace('区', '');
+  for (let i = 0; i <= 3; i++) {
+    if (title.slice(i).startsWith(district) || title.slice(i).startsWith(trimmedDistrict)) {
+      return false;
+    }
+  }
+
+  return true;
+};
+
+export const getDistrict = ({ title = '', address }: { title?: string; address?: string }): string => {
+  if (!address) {
+    return '';
+  }
+
+  const cleanedAddress = addressTrimEnd(addressTrimParenthesis(address));
+
+  const regex =
+    /(?<_>[^省]+省|.+自治区|[^澳门]+澳门|[^香港]+香港|[^市]+市)?(?<__>[^自治州]+自治州|[^特别行政区]+特别行政区|[^市]+市|.*?地区|.*?行政单位|.+盟|市辖区|[^县]+县)(?<district>[^县]+县|[^市]+市|[^镇]+镇|[^区]+区|[^乡]+乡|.+场|.+旗|.+海域|.+岛)?(?<___>.*)/;
+  const match = cleanedAddress.match(regex);
+
+  const district = match?.groups?.['district'];
+  if (district) {
+    if (districtStartWith({ title, district })) {
+      return district;
+    }
+  }
+
+  return '';
+};

packages/utility/src/misc/index.ts

@@ -1,2 +1,3 @@
 export * from './get-distance';
+export { getDistrict } from './get-district';
 export * from './stringify';