react-native-address-parse 1.1.0

package/harmony/address_parse/src/main/ets/AddressParseTurboModule.ets

@@ -0,0 +1,600 @@
+/**
+ * Copyright (c) 2026 react-native-address-parse contributors.
+ *
+ * Ported from zh-address-parse (https://github.com/zhaocjcc/zh-address-parse, MIT).
+ */
+import { AnyThreadTurboModule, AnyThreadTurboModuleContext } from '@rnoh/react-native-openharmony';
+import { PROVINCES } from './data/Provinces';
+import { CITIES } from './data/Cities';
+import { AREAS } from './data/Areas';
+import { ZH_CN_NAMES } from './data/Names';
+import { GovItem } from './data/GovData';
+
+export interface ExtraGovData {
+  province?: GovItem[];
+  city?: GovItem[];
+  area?: GovItem[];
+}
+
+export interface ParseOptions {
+  type?: number;
+  textFilter?: string[];
+  nameMaxLength?: number;
+  extraGovData?: ExtraGovData;
+}
+
+export interface ParseResult {
+  province: string;
+  city: string;
+  area: string;
+  detail: string;
+  name: string;
+  phone: string;
+  postalCode: string;
+  provinceCode: string;
+  cityCode: string;
+  areaCode: string;
+}
+
+interface PhoneResult {
+  address: string;
+  phone: string;
+}
+
+interface PostalCodeResult {
+  address: string;
+  postalCode: string;
+}
+
+interface RegionParseResult {
+  province: GovItem[];
+  city: GovItem[];
+  area: GovItem[];
+  detail: string[];
+}
+
+interface ParseState {
+  phone: string;
+  postalCode: string;
+  name: string;
+  provinceCode: string;
+  cityCode: string;
+  areaCode: string;
+  province: GovItem[];
+  city: GovItem[];
+  area: GovItem[];
+  detail: string[];
+}
+
+const EMPTY_PROVINCES: GovItem[] = PROVINCES.slice();
+const EMPTY_CITIES: GovItem[] = CITIES.slice();
+const EMPTY_AREAS: GovItem[] = AREAS.slice();
+
+let provinces: GovItem[] = EMPTY_PROVINCES.slice();
+let cities: GovItem[] = EMPTY_CITIES.slice();
+let areas: GovItem[] = EMPTY_AREAS.slice();
+
+function setExtraGovData(extraGovData?: ExtraGovData): void {
+  if (extraGovData !== undefined && extraGovData.province !== undefined && extraGovData.province.length > 0) {
+    provinces = provinces.concat(extraGovData.province);
+  }
+  if (extraGovData !== undefined && extraGovData.city !== undefined && extraGovData.city.length > 0) {
+    cities = cities.concat(extraGovData.city);
+  }
+  if (extraGovData !== undefined && extraGovData.area !== undefined && extraGovData.area.length > 0) {
+    areas = areas.concat(extraGovData.area);
+  }
+}
+
+function sortAddress(splitAddress: string[]): string[] {
+  const result: string[] = [];
+  const suffixes: string[] = ['省', '市', '区', '县', '镇'];
+  suffixes.forEach((suffix: string) => {
+    const index = splitAddress.findIndex((item: string) => item.indexOf(suffix) >= 0);
+    if (index >= 0) {
+      result.push(splitAddress.splice(index, 1)[0]);
+    }
+  });
+  return result.concat(splitAddress);
+}
+
+function emptyRegionResult(): RegionParseResult {
+  const r: RegionParseResult = { province: [], city: [], area: [], detail: [] };
+  return r;
+}
+
+function parseRegionWithRegexp(fragment: string, hasParseResult: RegionParseResult): RegionParseResult {
+  let province = hasParseResult.province;
+  let city = hasParseResult.city;
+  let area = hasParseResult.area;
+  const detail: string[] = [];
+  const CH = '[\\u4E00-\\u9FA5]';
+
+  let matchStr = '';
+  if (province.length === 0) {
+    for (let i = 1; i < fragment.length; i++) {
+      const str = fragment.substring(0, i + 1);
+      const regexProvince = new RegExp(`\\{"code":"[0-9]{1,6}","name":"${str}${CH}*?"\\}`, 'g');
+      const matchProvince = JSON.stringify(provinces).match(regexProvince);
+      if (matchProvince) {
+        if (matchProvince.length === 1) {
+          province = [];
+          matchStr = str;
+          province.push(JSON.parse(matchProvince[0]) as GovItem);
+        }
+      } else {
+        break;
+      }
+    }
+    if (province.length > 0) {
+      fragment = fragment.replace(new RegExp(matchStr, 'g'), '');
+    }
+  }
+
+  if (city.length === 0) {
+    for (let i = 1; i < fragment.length; i++) {
+      const str = fragment.substring(0, i + 1);
+      const provinceCode = province.length > 0 ? province[0].code : '[0-9]{1,6}';
+      const regexCity = new RegExp(`\\{"code":"[0-9]{1,6}","name":"${str}${CH}*?","provinceCode":"${provinceCode}"\\}`, 'g');
+      const matchCity = JSON.stringify(cities).match(regexCity);
+      if (matchCity) {
+        if (matchCity.length === 1) {
+          city = [];
+          matchStr = str;
+          city.push(JSON.parse(matchCity[0]) as GovItem);
+        }
+      } else {
+        break;
+      }
+    }
+    if (city.length > 0) {
+      const provinceCode: string = city[0].provinceCode ?? '';
+      fragment = fragment.replace(new RegExp(matchStr, 'g'), '');
+      if (province.length === 0) {
+        const regexProvince = new RegExp(`\\{"code":"${provinceCode}","name":"${CH}+?"\\}`, 'g');
+        const matchProvince = JSON.stringify(provinces).match(regexProvince);
+        if (matchProvince) {
+          province.push(JSON.parse(matchProvince[0]) as GovItem);
+        }
+      }
+    }
+  }
+
+  if (area.length === 0) {
+    for (let i = 1; i < fragment.length; i++) {
+      const str = fragment.substring(0, i + 1);
+      const cityCode = city.length > 0 ? city[0].code : '[0-9]{1,6}';
+      const provinceCode = province.length > 0 ? province[0].code : '[0-9]{1,6}';
+      const regexArea = new RegExp(`\\{"code":"[0-9]{1,9}","name":"${str}${CH}*?","cityCode":"${cityCode}","provinceCode":"${provinceCode}"\\}`, 'g');
+      const matchArea = JSON.stringify(areas).match(regexArea);
+      if (matchArea) {
+        if (matchArea.length === 1) {
+          area = [];
+          matchStr = str;
+          area.push(JSON.parse(matchArea[0]) as GovItem);
+        }
+      } else {
+        break;
+      }
+    }
+    if (area.length > 0) {
+      const provinceCode: string = area[0].provinceCode ?? '';
+      const cityCode: string = area[0].cityCode ?? '';
+      fragment = fragment.replace(matchStr, '');
+      if (province.length === 0) {
+        const regexProvince = new RegExp(`\\{"code":"${provinceCode}","name":"${CH}+?"\\}`, 'g');
+        const matchProvince = JSON.stringify(provinces).match(regexProvince);
+        if (matchProvince) {
+          province.push(JSON.parse(matchProvince[0]) as GovItem);
+        }
+      }
+      if (city.length === 0) {
+        const regexCity = new RegExp(`\\{"code":"${cityCode}","name":"${CH}+?","provinceCode":"${provinceCode}"\\}`, 'g');
+        const matchCity = JSON.stringify(cities).match(regexCity);
+        if (matchCity) {
+          city.push(JSON.parse(matchCity[0]) as GovItem);
+        }
+      }
+    }
+  }
+
+  if (fragment.length > 0) {
+    detail.push(fragment);
+  }
+
+  const result: RegionParseResult = { province, city, area, detail };
+  return result;
+}
+
+function parseRegion(fragment: string, hasParseResult: RegionParseResult): RegionParseResult {
+  const province: GovItem[] = [];
+  const city: GovItem[] = [];
+  const area: GovItem[] = [];
+  const detail: string[] = [];
+
+  if (hasParseResult.province.length > 0) {
+    province.push(...hasParseResult.province);
+  } else {
+    for (const tempProvince of provinces) {
+      const name: string = tempProvince.name;
+      let replaceName = '';
+      for (let i = name.length; i > 1; i--) {
+        const temp = name.substring(0, i);
+        if (fragment.indexOf(temp) === 0) {
+          replaceName = temp;
+          break;
+        }
+      }
+      if (replaceName) {
+        province.push(tempProvince);
+        fragment = fragment.replace(new RegExp(replaceName, 'g'), '');
+        break;
+      }
+    }
+  }
+
+  if (hasParseResult.city.length > 0) {
+    city.push(...hasParseResult.city);
+  } else {
+    for (const tempCity of cities) {
+      const name: string = tempCity.name;
+      const provinceCode: string = tempCity.provinceCode ?? '';
+      const currentProvince = province.length > 0 ? province[0] : undefined;
+      if (currentProvince) {
+        if (currentProvince.code === provinceCode) {
+          let replaceName = '';
+          for (let i = name.length; i > 1; i--) {
+            const temp = name.substring(0, i);
+            if (fragment.indexOf(temp) === 0) {
+              replaceName = temp;
+              break;
+            }
+          }
+          if (replaceName) {
+            city.push(tempCity);
+            fragment = fragment.replace(new RegExp(replaceName, 'g'), '');
+            break;
+          }
+        }
+      } else {
+        for (let i = name.length; i > 1; i--) {
+          const replaceName = name.substring(0, i);
+          if (fragment.indexOf(replaceName) === 0) {
+            city.push(tempCity);
+            const p = provinces.find((item: GovItem) => item.code === provinceCode);
+            if (p) {
+              province.push(p);
+            }
+            fragment = fragment.replace(replaceName, '');
+            break;
+          }
+        }
+        if (city.length > 0) {
+          break;
+        }
+      }
+    }
+  }
+
+  for (const tempArea of areas) {
+    const name: string = tempArea.name;
+    const provinceCode: string = tempArea.provinceCode ?? '';
+    const cityCode: string = tempArea.cityCode ?? '';
+    const currentProvince = province.length > 0 ? province[0] : undefined;
+    const currentCity = city.length > 0 ? city[0] : undefined;
+
+    if (currentProvince || currentCity) {
+      if ((currentProvince !== undefined && currentProvince.code === provinceCode)
+        || (currentCity !== undefined && currentCity.code === cityCode)) {
+        let replaceName = '';
+        for (let i = name.length; i > 1; i--) {
+          const temp = name.substring(0, i);
+          if (fragment.indexOf(temp) === 0) {
+            replaceName = temp;
+            break;
+          }
+        }
+        if (replaceName) {
+          area.push(tempArea);
+          if (!currentCity) {
+            const c = cities.find((item: GovItem) => item.code === cityCode);
+            if (c) {
+              city.push(c);
+            }
+          }
+          if (!currentProvince) {
+            const p = provinces.find((item: GovItem) => item.code === provinceCode);
+            if (p) {
+              province.push(p);
+            }
+          }
+          fragment = fragment.replace(replaceName, '');
+          break;
+        }
+      }
+    } else {
+      for (let i = name.length; i > 1; i--) {
+        const replaceName = name.substring(0, i);
+        if (fragment.indexOf(replaceName) === 0) {
+          area.push(tempArea);
+          const c = cities.find((item: GovItem) => item.code === cityCode);
+          if (c) {
+            city.push(c);
+          }
+          const p = provinces.find((item: GovItem) => item.code === provinceCode);
+          if (p) {
+            province.push(p);
+          }
+          fragment = fragment.replace(replaceName, '');
+          break;
+        }
+      }
+      if (area.length > 0) {
+        break;
+      }
+    }
+  }
+
+  if (fragment.length > 0) {
+    detail.push(fragment);
+  }
+
+  const result: RegionParseResult = { province, city, area, detail };
+  return result;
+}
+
+function judgeFragmentIsName(fragment: string, nameMaxLength: number): string {
+  if (!fragment || !/[\u4E00-\u9FA5]/.test(fragment)) {
+    return '';
+  }
+
+  const nameCall: string[] = ['先生', '小姐', '同志', '哥哥', '姐姐', '妹妹', '弟弟', '妈妈', '爸爸', '爷爷', '奶奶', '姑姑', '舅舅'];
+  const hasCall = nameCall.some((item: string) => fragment.indexOf(item) >= 0);
+  if (hasCall) {
+    return fragment;
+  }
+
+  const filters: string[] = ['街道', '乡镇', '镇', '乡'];
+  const hasFilter = filters.some((item: string) => fragment.indexOf(item) >= 0);
+  if (hasFilter) {
+    return '';
+  }
+
+  const nameFirst = fragment.substring(0, 1);
+  if (fragment.length <= nameMaxLength && fragment.length > 1 && ZH_CN_NAMES.indexOf(nameFirst) >= 0) {
+    return fragment;
+  }
+
+  return '';
+}
+
+function filterPhone(address: string): PhoneResult {
+  let phone = '';
+  address = address.replace(/(\d{3})-(\d{4})-(\d{4})/g, '$1$2$3');
+  address = address.replace(/(\d{3}) (\d{4}) (\d{4})/g, '$1$2$3');
+  address = address.replace(/(\d{4}) (\d{4}) (\d{4})/g, '$1$2$3');
+
+  const mobileReg = /(0|\+?86-?|17951|)?1[3456789]\d{9}(-\d{4})?/g;
+  const mobile = mobileReg.exec(address);
+  if (mobile) {
+    phone = mobile[0];
+    address = address.replace(mobile[0], ' ');
+  }
+
+  if (!phone) {
+    const landlineReg = /(?:\+?86[- ]?)?(?:\(?0\d{2,3}\)?[- ]?)\d{7,8}(?:-\d{1,6})?/g;
+    const landline = landlineReg.exec(address);
+    if (landline) {
+      phone = landline[0];
+      address = address.replace(landline[0], ' ');
+    }
+  }
+
+  const cleaned = phone.replace(/^\+?86[- ]?/g, '').replace(/[（）]/g, '').trim();
+  const result: PhoneResult = { address, phone: cleaned };
+  return result;
+}
+
+function filterPostalCode(address: string): PostalCodeResult {
+  let postalCode = '';
+  const postalCodeReg = /[1-9]\d{5}(?!\d)/g;
+  const code = postalCodeReg.exec(address);
+  if (code) {
+    postalCode = code[0];
+    address = address.replace(code[0], ' ');
+  }
+  const result: PostalCodeResult = { address, postalCode };
+  return result;
+}
+
+function cleanAddress(address: string, textFilter: string[]): string {
+  address = address
+    .replace(/\r\n/g, ' ')
+    .replace(/\n/g, ' ')
+    .replace(/\t/g, ' ');
+
+  const search: string[] = [
+    '详细地址', '收货地址', '收件地址', '地址', '所在地区', '姓名', '收货人', '收件人',
+    '联系人', '收', '邮编', '联系电话', '电话', '联系人手机号码', '手机号码', '手机号',
+    '自治区直辖县级行政区划', '省直辖县级行政区划',
+  ].concat(textFilter);
+  search.forEach((str: string) => {
+    address = address.replace(new RegExp(str, 'g'), ' ');
+  });
+
+  const pattern = new RegExp("[`~!@$^&*()=|{}':;',\\[\\]\.<>/?~！@￥……&*（）——|{}【】‘；：”“’。，、？]", 'g');
+  address = address.replace(pattern, ' ');
+
+  address = address.replace(/ {2,}/g, ' ');
+  const municipalities: string[] = ['北京', '上海', '天津', '重庆'];
+  municipalities.forEach((str: string) => {
+    address = address.replace(str + str, str);
+  });
+  return address;
+}
+
+export function addressParse(address: string, options: ParseOptions): ParseResult {
+  const type = options.type !== undefined ? options.type : 0;
+  const textFilter = options.textFilter !== undefined ? options.textFilter : [];
+  const nameMaxLength = options.nameMaxLength !== undefined ? options.nameMaxLength : 4;
+  const extraGovData = options.extraGovData;
+
+  const emptyResult: ParseResult = {
+    province: '', city: '', area: '', detail: '', name: '',
+    phone: '', postalCode: '', provinceCode: '', cityCode: '', areaCode: '',
+  };
+  if (!address) {
+    return emptyResult;
+  }
+
+  setExtraGovData(extraGovData);
+
+  const parseResult: ParseState = {
+    phone: '', postalCode: '', name: '',
+    provinceCode: '', cityCode: '', areaCode: '',
+    province: [], city: [], area: [], detail: [],
+  };
+
+  address = cleanAddress(address, textFilter);
+
+  const resultPhone = filterPhone(address);
+  address = resultPhone.address;
+  parseResult.phone = resultPhone.phone;
+
+  const resultCode = filterPostalCode(address);
+  address = resultCode.address;
+  parseResult.postalCode = resultCode.postalCode;
+
+  let splitAddress = address.split(' ').filter((item: string) => item.length > 0 && !/^\d{6,}$/.test(item)).map((item: string) => item.trim());
+  splitAddress = sortAddress(splitAddress);
+
+  splitAddress.forEach((item: string) => {
+    if (!parseResult.province[0] || !parseResult.city[0] || !parseResult.area[0]) {
+      const seed: RegionParseResult = {
+        province: parseResult.province,
+        city: parseResult.city,
+        area: parseResult.area,
+        detail: [],
+      };
+      let parse: RegionParseResult;
+      if (type === 1) {
+        parse = parseRegion(item, seed);
+      } else {
+        parse = parseRegionWithRegexp(item, seed);
+      }
+      parseResult.province = parse.province;
+      parseResult.area = parse.area;
+      parseResult.city = parse.city;
+      parseResult.detail = parseResult.detail.concat(parse.detail);
+      if (parseResult.area[0] !== undefined) {
+        parseResult.areaCode = parseResult.area[0].code;
+      } else {
+        parseResult.areaCode = '';
+      }
+      if (parseResult.province[0] !== undefined) {
+        parseResult.provinceCode = parseResult.province[0].code;
+      } else {
+        parseResult.provinceCode = '';
+      }
+      if (parseResult.city[0] !== undefined) {
+        parseResult.cityCode = parseResult.city[0].code;
+      } else {
+        parseResult.cityCode = '';
+      }
+    } else {
+      parseResult.detail.push(item);
+    }
+  });
+
+  const province = parseResult.province.length > 0 ? parseResult.province[0] : undefined;
+  const city = parseResult.city.length > 0 ? parseResult.city[0] : undefined;
+  const area = parseResult.area.length > 0 ? parseResult.area[0] : undefined;
+  let detail = parseResult.detail;
+
+  const names: string[] = [];
+  if (province !== undefined && province.name) {
+    names.push(province.name);
+  }
+  if (city !== undefined && city.name) {
+    names.push(city.name);
+  }
+  if (area !== undefined && area.name) {
+    names.push(area.name);
+  }
+  if (names.length > 0) {
+    detail = detail.map((item: string) => item.replace(new RegExp(names.join('|'), 'g'), ''));
+  }
+  detail = Array.from(new Set(detail));
+
+  if (detail.length > 0) {
+    const copyDetail: string[] = detail.filter((item: string) => item.length > 0);
+    copyDetail.sort((a: string, b: string) => a.length - b.length);
+    const index = copyDetail.findIndex((item: string) => judgeFragmentIsName(item, nameMaxLength).length > 0);
+    let name = '';
+    if (index >= 0) {
+      name = copyDetail[index];
+    } else if (copyDetail.length > 0 && copyDetail[0].length <= nameMaxLength && /[\u4E00-\u9FA5]/.test(copyDetail[0])) {
+      name = copyDetail[0];
+    }
+
+    if (name.length > 0) {
+      parseResult.name = name;
+      const idx = detail.findIndex((item: string) => item === name);
+      if (idx >= 0) {
+        detail.splice(idx, 1);
+      }
+    }
+  }
+
+  let provinceName = '';
+  if (province !== undefined) {
+    provinceName = province.name;
+  }
+  let cityName = '';
+  if (city !== undefined) {
+    cityName = city.name;
+  }
+  if (cityName === '市辖区' || cityName === '区' || cityName === '县' || cityName === '镇') {
+    cityName = provinceName;
+  }
+  let areaName = '';
+  if (area !== undefined) {
+    areaName = area.name;
+  }
+
+  const finalResult: ParseResult = {
+    province: provinceName,
+    city: cityName,
+    area: areaName,
+    detail: detail.length > 0 ? detail.join('') : '',
+    name: parseResult.name,
+    phone: parseResult.phone,
+    postalCode: parseResult.postalCode,
+    provinceCode: parseResult.provinceCode,
+    cityCode: parseResult.cityCode,
+    areaCode: parseResult.areaCode,
+  };
+  return finalResult;
+}
+
+export class AddressParseTurboModule extends AnyThreadTurboModule {
+  public static readonly NAME = 'AddressParse';
+
+  constructor(ctx: AnyThreadTurboModuleContext) {
+    super(ctx);
+  }
+
+  parse(
+    address: string,
+    type: number,
+    textFilter: string[],
+    nameMaxLength: number,
+    extraGovData?: ExtraGovData,
+  ): ParseResult {
+    return addressParse(address, { type, textFilter, nameMaxLength, extraGovData });
+  }
+
+  __onDestroy__(): void {
+    super.__onDestroy__();
+  }
+}