device-voice-h5/utils/shippingAddress.js

import { regionData } from 'element-china-area-data';

const GENERIC_CITY_LABELS = new Set([
	'市辖区',
	'县',
	'自治区直辖县级行政区划',
	'省直辖县级行政区划'
]);

const NAME_LABEL_REG = /(收货人|收件人|联系人|姓名)/g;
const PHONE_LABEL_REG = /(手机号码|手机号|联系电话|联系方式|电话|手机)/g;
const ADDRESS_LABEL_REG = /(收货地址|详细地址|地址|所在地区|地区)/g;
const ADDRESS_SIGNAL_REG = /(省|市|区|县|旗|乡|镇|街道|大道|路|街|巷|弄|村|社区|小区|号|栋|单元|室|楼)/;
const PHONE_REG = /(?<!\d)(?:\+?86[-\s]?)?(1[3-9](?:[\s-]?\d){9})(?!\d)/;

const provinceOptions = regionData.map(({ label, value }) => ({ label, value }));

const districtRecords = regionData.flatMap((province) =>
	(province.children || []).flatMap((city) =>
		(city.children || []).map((district) => ({
			province,
			city,
			district,
			provinceSearchLabel: normalizeForSearch(province.label),
			citySearchLabel: normalizeForSearch(getDisplayCityLabel(city.label)),
			districtSearchLabel: normalizeForSearch(district.label),
			fullSearchLabel: normalizeForSearch(
				`${province.label}${getDisplayCityLabel(city.label)}${district.label}`
			)
		}))
	)
);

function normalizeText(value = '') {
	return String(value)
		.replace(/\r/g, '\n')
		.replace(/\u00a0/g, ' ')
		.replace(/\t/g, ' ')
		.trim();
}

function normalizeForSearch(value = '') {
	return normalizeText(value).replace(/[\s,，.。;；:：、\-_/]/g, '');
}

function getDisplayCityLabel(label = '') {
	return GENERIC_CITY_LABELS.has(label) ? '' : label;
}

function getProvinceByCode(provinceCode) {
	return regionData.find((item) => item.value === provinceCode) || null;
}

function getCityByCode(provinceCode, cityCode) {
	const province = getProvinceByCode(provinceCode);
	return province?.children?.find((item) => item.value === cityCode) || null;
}

function getDistrictByCode(provinceCode, cityCode, districtCode) {
	const city = getCityByCode(provinceCode, cityCode);
	return city?.children?.find((item) => item.value === districtCode) || null;
}

function getCityOptions(provinceCode) {
	const province = getProvinceByCode(provinceCode) || regionData[0];
	return (province?.children || []).map(({ label, value }) => ({ label, value }));
}

function getDistrictOptions(provinceCode, cityCode) {
	const province = getProvinceByCode(provinceCode) || regionData[0];
	const city =
		province?.children?.find((item) => item.value === cityCode) ||
		province?.children?.[0] ||
		null;

	return (city?.children || []).map(({ label, value }) => ({ label, value }));
}

function getLabelParts(regionCodes = []) {
	const [provinceCode, cityCode, districtCode] = regionCodes;
	const province = getProvinceByCode(provinceCode);
	const city = getCityByCode(provinceCode, cityCode);
	const district = getDistrictByCode(provinceCode, cityCode, districtCode);

	if (!province || !city || !district) {
		return [];
	}

	const displayCityLabel = getDisplayCityLabel(city.label);
	return [province.label, displayCityLabel, district.label].filter(Boolean);
}

function cleanupAddressText(value = '') {
	return normalizeText(value)
		.replace(NAME_LABEL_REG, ' ')
		.replace(PHONE_LABEL_REG, ' ')
		.replace(ADDRESS_LABEL_REG, ' ')
		.replace(/[：:]/g, ' ')
		.replace(/[，,;；]/g, ' ')
		.replace(/\s+/g, ' ')
		.trim();
}

function extractPhone(rawText = '') {
	const text = normalizeText(rawText);
	const match = text.match(PHONE_REG);
	if (!match) {
		return {
			phone: '',
			textWithoutPhone: text
		};
	}

	const digits = match[1].replace(/\D/g, '');
	const phone = digits.length > 11 ? digits.slice(-11) : digits;

	return {
		phone,
		textWithoutPhone: text.replace(match[0], ' ').replace(/\s+/g, ' ').trim()
	};
}

function pickBestMatch(records, matcher) {
	return records
		.filter(matcher)
		.sort((left, right) => {
			return (
				right.fullSearchLabel.length - left.fullSearchLabel.length ||
				right.districtSearchLabel.length - left.districtSearchLabel.length
			);
		})[0] || null;
}

function findRegionMatch(rawText = '') {
	const searchText = normalizeForSearch(rawText);
	if (!searchText) {
		return null;
	}

	let match = pickBestMatch(districtRecords, (record) => {
		return record.fullSearchLabel && searchText.includes(record.fullSearchLabel);
	});

	if (!match) {
		match = pickBestMatch(districtRecords, (record) => {
			return (
				record.provinceSearchLabel &&
				record.districtSearchLabel &&
				searchText.includes(record.provinceSearchLabel) &&
				searchText.includes(record.districtSearchLabel)
			);
		});
	}

	return match;
}

function getRegionStartIndex(text, match) {
	const cityLabel = getDisplayCityLabel(match.city.label);
	const candidates = [match.province.label, cityLabel, match.district.label]
		.filter(Boolean)
		.map((label) => text.indexOf(label))
		.filter((index) => index >= 0);

	return candidates.length ? Math.min(...candidates) : -1;
}

function stripMatchedRegion(addressText, match) {
	const cityLabel = getDisplayCityLabel(match.city.label);
	let detailAddress = normalizeText(addressText);

	[match.province.label, cityLabel, match.district.label]
		.filter(Boolean)
		.forEach((label) => {
			detailAddress = detailAddress.replace(label, '');
		});

	return detailAddress.replace(/^[,，;；:：\s]+/, '').trim();
}

function extractNameFromText(rawText = '', textWithoutPhone = '', regionStartIndex = -1) {
	const sourceText = regionStartIndex > 0
		? textWithoutPhone.slice(0, regionStartIndex)
		: textWithoutPhone || rawText;

	const directNameMatch = normalizeText(rawText).match(
		/(?:收货人|收件人|联系人|姓名)\s*[：:]?\s*([A-Za-z\u4e00-\u9fa5·]{2,20})/
	);
	if (directNameMatch?.[1]) {
		return directNameMatch[1].trim();
	}

	const cleaned = cleanupAddressText(sourceText)
		.replace(PHONE_REG, ' ')
		.replace(/\d+/g, ' ')
		.trim();
	const tokens = cleaned.split(/\s+/).filter(Boolean);
	const nameToken = tokens.find((token) => {
		return !ADDRESS_SIGNAL_REG.test(token) && /^[A-Za-z\u4e00-\u9fa5·]{2,20}$/.test(token);
	});

	return nameToken || '';
}

function extractAddressSource(textWithoutPhone = '', name = '', match = null) {
	let addressSource = normalizeText(textWithoutPhone);

	if (name) {
		addressSource = addressSource.replace(name, ' ').replace(/\s+/g, ' ').trim();
	}

	if (match) {
		const regionStartIndex = getRegionStartIndex(addressSource, match);
		if (regionStartIndex >= 0) {
			return addressSource.slice(regionStartIndex).trim();
		}
	}

	return cleanupAddressText(addressSource);
}

function formatRegionCodes(match) {
	return [match.province.value, match.city.value, match.district.value];
}

export function buildRegionColumns(regionCodes = []) {
	const provinceCode =
		getProvinceByCode(regionCodes[0])?.value || provinceOptions[0]?.value || '';
	const cityOptions = getCityOptions(provinceCode);
	const cityCode =
		cityOptions.find((item) => item.value === regionCodes[1])?.value ||
		cityOptions[0]?.value ||
		'';
	const districtOptions = getDistrictOptions(provinceCode, cityCode);
	const districtCode =
		districtOptions.find((item) => item.value === regionCodes[2])?.value ||
		districtOptions[0]?.value ||
		'';

	return {
		columns: [provinceOptions, cityOptions, districtOptions],
		defaultIndex: [
			provinceOptions.findIndex((item) => item.value === provinceCode),
			cityOptions.findIndex((item) => item.value === cityCode),
			districtOptions.findIndex((item) => item.value === districtCode)
		]
	};
}

export function getRegionDisplayText(regionCodes = []) {
	return getLabelParts(regionCodes).join(' / ');
}

export function composeRecipientAddress(regionCodes = [], detailAddress = '') {
	const regionText = getLabelParts(regionCodes).join('');
	return `${regionText}${normalizeText(detailAddress)}`.trim();
}

export function parseSavedRecipientAddress(rawAddress = '') {
	const addressText = normalizeText(rawAddress);
	if (!addressText) {
		return {
			regionCodes: [],
			detailAddress: ''
		};
	}

	const match = findRegionMatch(addressText);
	if (!match) {
		return {
			regionCodes: [],
			detailAddress: addressText
		};
	}

	const detailAddress = stripMatchedRegion(addressText, match);
	return {
		regionCodes: formatRegionCodes(match),
		detailAddress: detailAddress || addressText
	};
}

export function parseShippingText(rawText = '') {
	const text = normalizeText(rawText);
	if (!text) {
		return {
			recipient_name: '',
			recipient_phone: '',
			regionCodes: [],
			detailAddress: ''
		};
	}

	const { phone, textWithoutPhone } = extractPhone(text);
	const regionMatch = findRegionMatch(textWithoutPhone);
	const regionStartIndex = regionMatch ? getRegionStartIndex(textWithoutPhone, regionMatch) : -1;
	const recipientName = extractNameFromText(text, textWithoutPhone, regionStartIndex);
	const addressSource = extractAddressSource(textWithoutPhone, recipientName, regionMatch);
	const detailAddress = regionMatch
		? stripMatchedRegion(addressSource, regionMatch)
		: cleanupAddressText(addressSource);

	return {
		recipient_name: recipientName,
		recipient_phone: phone,
		regionCodes: regionMatch ? formatRegionCodes(regionMatch) : [],
		detailAddress
	};
}