chore: release updated anxinyan version

admin-web/src/utils/address-recognition.ts

@@ -0,0 +1,213 @@
+import regionSource from "../static/regions/pca.json";
+
+type RegionNode = {
+  code: string;
+  name: string;
+  children?: RegionNode[];
+};
+
+export type RecognizedReturnAddress = {
+  consignee: string;
+  mobile: string;
+  province: string;
+  city: string;
+  district: string;
+  detail_address: string;
+};
+
+export type RecognizeReturnAddressResult = {
+  ok: boolean;
+  address?: RecognizedReturnAddress;
+  message?: string;
+};
+
+const regionTree = regionSource as RegionNode[];
+const nameLabels = ["收货人", "收件人", "姓名", "联系人", "取件人"];
+const mobileLabels = ["收货电话", "联系电话", "手机号", "手机号码", "手机", "电话"];
+const addressLabels = ["收货地址", "收件地址", "寄回地址", "地址"];
+const allLabels = [...nameLabels, ...mobileLabels, ...addressLabels];
+
+function normalizeLines(raw: string) {
+  return raw
+    .replace(/\r/g, "\n")
+    .split("\n")
+    .map((line) => line.trim())
+    .filter(Boolean);
+}
+
+function labelMatch(line: string, labels: string[]) {
+  for (const label of labels) {
+    const pattern = new RegExp(`^\\s*${label}\\s*[:：]?\\s*(.*)$`);
+    const match = line.match(pattern);
+    if (match) {
+      return { label, value: String(match[1] || "").trim() };
+    }
+  }
+  return null;
+}
+
+function isKnownLabel(line: string) {
+  const normalized = line.replace(/\s+/g, "");
+  return allLabels.some((label) => normalized === label || normalized === `${label}:` || normalized === `${label}：`);
+}
+
+function extractLabeledValue(lines: string[], labels: string[], block = false) {
+  for (let index = 0; index < lines.length; index += 1) {
+    const match = labelMatch(lines[index], labels);
+    if (!match) continue;
+    if (match.value) return match.value;
+
+    const values: string[] = [];
+    for (let nextIndex = index + 1; nextIndex < lines.length; nextIndex += 1) {
+      const nextLine = lines[nextIndex];
+      if (labelMatch(nextLine, allLabels) && isKnownLabel(nextLine)) {
+        break;
+      }
+      if (labelMatch(nextLine, allLabels)?.value) {
+        break;
+      }
+      values.push(nextLine);
+      if (!block) break;
+    }
+    return values.join(block ? "" : " ").trim();
+  }
+  return "";
+}
+
+function normalizeMobile(value: string) {
+  const directMatch = value.match(/1[3-9]\d{9}/);
+  if (directMatch) return directMatch[0];
+
+  const digits = value.replace(/\D+/g, "");
+  return digits.match(/1[3-9]\d{9}/)?.[0] || "";
+}
+
+function stripKnownAddressPrefixes(value: string) {
+  return value
+    .replace(/\s+/g, "")
+    .replace(/^(中国大陆|中华人民共和国|中国|大陆)+/, "")
+    .replace(/^(收货地址|收件地址|寄回地址|地址)[:：]?/, "");
+}
+
+function aliases(name: string) {
+  const suffixes = ["特别行政区", "壮族自治区", "回族自治区", "维吾尔自治区", "自治区", "自治州", "自治县", "地区", "省", "市", "区", "县", "旗", "盟"];
+  const values = [name];
+  for (const suffix of suffixes) {
+    if (name.endsWith(suffix) && name.length > suffix.length) {
+      values.push(name.slice(0, -suffix.length));
+    }
+  }
+  return Array.from(new Set(values)).sort((a, b) => b.length - a.length);
+}
+
+function consumePrefix(text: string, names: string[]) {
+  for (const name of names) {
+    if (name && text.startsWith(name)) {
+      return { consumed: name.length, rest: text.slice(name.length) };
+    }
+  }
+  return null;
+}
+
+function isDirectCity(province: RegionNode, city: RegionNode) {
+  return province.name === city.name || aliases(province.name).some((name) => aliases(city.name).includes(name));
+}
+
+function matchDistrict(city: RegionNode, text: string) {
+  for (const district of city.children || []) {
+    const match = consumePrefix(text, aliases(district.name));
+    if (match) {
+      return { district, detail: match.rest };
+    }
+  }
+  return null;
+}
+
+function matchRegion(addressText: string) {
+  const address = stripKnownAddressPrefixes(addressText);
+  if (!address) return null;
+
+  for (const province of regionTree) {
+    const provinceMatch = consumePrefix(address, aliases(province.name));
+    if (!provinceMatch) continue;
+
+    for (const city of province.children || []) {
+      const cityMatch = consumePrefix(provinceMatch.rest, aliases(city.name));
+      const districtSource = cityMatch ? cityMatch.rest : (isDirectCity(province, city) ? provinceMatch.rest : "");
+      if (!districtSource) continue;
+
+      const districtMatch = matchDistrict(city, districtSource);
+      if (districtMatch) {
+        return {
+          province: province.name,
+          city: city.name,
+          district: districtMatch.district.name,
+          detail_address: districtMatch.detail,
+        };
+      }
+    }
+  }
+
+  for (const province of regionTree) {
+    for (const city of province.children || []) {
+      const cityMatch = consumePrefix(address, aliases(city.name));
+      if (!cityMatch) continue;
+
+      const districtMatch = matchDistrict(city, cityMatch.rest);
+      if (districtMatch) {
+        return {
+          province: province.name,
+          city: city.name,
+          district: districtMatch.district.name,
+          detail_address: districtMatch.detail,
+        };
+      }
+    }
+  }
+
+  return null;
+}
+
+function fallbackAddressLine(lines: string[], consignee: string, mobile: string) {
+  return lines
+    .map((line) => labelMatch(line, allLabels)?.value || line)
+    .filter((line) => line && line !== consignee && !line.includes(mobile) && !normalizeMobile(line))
+    .sort((a, b) => b.length - a.length)[0] || "";
+}
+
+export function recognizeReturnAddress(raw: string): RecognizeReturnAddressResult {
+  const lines = normalizeLines(raw);
+  if (!lines.length) {
+    return { ok: false, message: "请先粘贴寄回地址信息" };
+  }
+
+  const consignee = extractLabeledValue(lines, nameLabels).trim();
+  const mobile = normalizeMobile(extractLabeledValue(lines, mobileLabels) || raw);
+  const addressText = extractLabeledValue(lines, addressLabels, true) || fallbackAddressLine(lines, consignee, mobile);
+  const region = matchRegion(addressText);
+
+  if (!consignee) {
+    return { ok: false, message: "未识别到收件人，请检查文本中是否包含收货人或收件人" };
+  }
+  if (!mobile) {
+    return { ok: false, message: "未识别到有效手机号，请检查文本中的收货电话" };
+  }
+  if (!region) {
+    return { ok: false, message: "未识别到省市区，请检查地址是否包含城市和区县" };
+  }
+  if (!region.detail_address.trim()) {
+    return { ok: false, message: "未识别到详细地址，请检查区县后的街道门牌信息" };
+  }
+
+  return {
+    ok: true,
+    address: {
+      consignee,
+      mobile,
+      province: region.province,
+      city: region.city,
+      district: region.district,
+      detail_address: region.detail_address.trim(),
+    },
+  };
+}