R73-A2 收敛 VAT OCR 归一化与质量门槛

server/src/services/ocr/index.js

@@ -1,6 +1,10 @@
 const { runTencentVatOcr } = require("./tencent");
 const { runMockOcr } = require("./mock");
 
+const REQUIRED_OCR_FIELDS_BY_TYPE = {
+  vat: ["invoiceCode", "invoiceNumber", "invoiceDate", "totalAmount", "sellerName"],
+};
+
 function isOcrEnabledForType(config, ocrType) {
   if (config.ocrProvider === "mock") {
     return ocrType === "vat";
@@ -9,6 +13,27 @@ function isOcrEnabledForType(config, ocrType) {
   return config.ocrProvider === "tencent" && ocrType === "vat";
 }
 
+function getMissingRequiredOcrFields(ocrType, normalized) {
+  const requiredFields = REQUIRED_OCR_FIELDS_BY_TYPE[ocrType] ?? [];
+  if (!requiredFields.length) {
+    return [];
+  }
+
+  return requiredFields.filter((field) => {
+    const value = normalized?.[field];
+    return value == null || value === "";
+  });
+}
+
+function buildMissingRequiredOcrError(ocrType, normalized) {
+  const missingFields = getMissingRequiredOcrFields(ocrType, normalized);
+  if (!missingFields.length) {
+    return null;
+  }
+
+  return `missing_required_fields:${missingFields.join(",")}`;
+}
+
 async function runOcrForInvoice({ config, ocrType, imageBase64 }) {
   if (config.ocrProvider === "mock") {
     return runMockOcr({ ocrType, imageBase64 });
@@ -26,6 +51,8 @@ async function runOcrForInvoice({ config, ocrType, imageBase64 }) {
 }
 
 module.exports = {
+  buildMissingRequiredOcrError,
+  getMissingRequiredOcrFields,
   isOcrEnabledForType,
   runOcrForInvoice,
 };

server/src/services/ocr/mock.js

@@ -18,6 +18,18 @@ function buildInvoiceKey(markerText) {
   return `hash:${hash.slice(0, 20)}`;
 }
 
+function parseMissingFields(markerText) {
+  const match = markerText.match(/MOCK_OCR_MISSING:([A-Za-z0-9_,-]+)/);
+  if (!match?.[1]) {
+    return [];
+  }
+
+  return match[1]
+    .split(",")
+    .map((field) => field.trim())
+    .filter(Boolean);
+}
+
 function toAmount(seedHex, fallback) {
   const value = Number.parseInt(seedHex.slice(0, 8), 16);
   if (!Number.isFinite(value)) {
@@ -38,7 +50,7 @@ async function runMockOcr({ ocrType, imageBase64 }) {
   const invoiceCode = `MOCK${hash.slice(0, 8).toUpperCase()}`;
   const invoiceNumber = hash.slice(8, 16).toUpperCase();
 
-  return {
+  const result = {
     invoiceCode,
     invoiceNumber,
     invoiceDate: "2026-02-08",
@@ -53,6 +65,14 @@ async function runMockOcr({ ocrType, imageBase64 }) {
       marker: markerText.slice(0, 128),
     },
   };
+
+  for (const field of parseMissingFields(markerText)) {
+    if (Object.prototype.hasOwnProperty.call(result, field)) {
+      result[field] = null;
+    }
+  }
+
+  return result;
 }
 
 module.exports = {

server/src/services/ocr/tencent.js

@@ -46,12 +46,41 @@ function toNumberString(value) {
   return num.toFixed(2);
 }
 
+function addNumberStrings(...values) {
+  const normalized = values.map((value) => toNumberString(value));
+  if (normalized.some((value) => value == null)) {
+    return null;
+  }
+
+  const total = normalized.reduce((sum, value) => sum + Number(value), 0);
+  return total.toFixed(2);
+}
+
+function subtractNumberStrings(minuend, subtrahend) {
+  const normalizedMinuend = toNumberString(minuend);
+  const normalizedSubtrahend = toNumberString(subtrahend);
+  if (normalizedMinuend == null || normalizedSubtrahend == null) {
+    return null;
+  }
+
+  return (Number(normalizedMinuend) - Number(normalizedSubtrahend)).toFixed(2);
+}
+
+function normalizeIdentifier(value) {
+  if (typeof value !== "string") {
+    return null;
+  }
+
+  const normalized = value.replace(/\s+/g, "").trim();
+  return normalized || null;
+}
+
 function normalizeString(value) {
   if (typeof value !== "string") {
     return null;
   }
 
-  const normalized = value.trim();
+  const normalized = value.replace(/\s+/g, " ").trim();
   return normalized || null;
 }
 
@@ -76,13 +105,19 @@ function createTencentOcrClient(config) {
 }
 
 function normalizeVatInvoiceResult(result) {
+  const amount = toNumberString(result?.AmountWithoutTax);
+  const taxAmount = toNumberString(result?.TaxAmount);
+  const totalAmountFromSource = toNumberString(result?.TotalAmount ?? result?.AmountInFiguers);
+  const derivedAmount = amount ?? subtractNumberStrings(totalAmountFromSource, taxAmount);
+  const derivedTotalAmount = totalAmountFromSource ?? addNumberStrings(derivedAmount, taxAmount);
+
   return {
-    invoiceCode: normalizeString(result?.InvoiceCode),
-    invoiceNumber: normalizeString(result?.InvoiceNum),
+    invoiceCode: normalizeIdentifier(result?.InvoiceCode),
+    invoiceNumber: normalizeIdentifier(result?.InvoiceNum),
     invoiceDate: normalizeDate(result?.InvoiceDate),
-    amount: toNumberString(result?.AmountWithoutTax),
-    taxAmount: toNumberString(result?.TaxAmount),
-    totalAmount: toNumberString(result?.TotalAmount ?? result?.AmountInFiguers),
+    amount: derivedAmount,
+    taxAmount,
+    totalAmount: derivedTotalAmount,
     sellerName: normalizeString(result?.SellerName),
     buyerName: normalizeString(result?.BuyerName),
     raw: result ?? {},
@@ -99,5 +134,6 @@ async function runTencentVatOcr(config, imageBase64) {
 }
 
 module.exports = {
+  normalizeVatInvoiceResult,
   runTencentVatOcr,
 };