R92 A2 修复发票 OCR provider 与二进制写库

server/src/services/ocr/mock.js

@@ -1,20 +1,39 @@
 const crypto = require("node:crypto");
 
-function decodeMarker(imageBase64) {
+function decodeImageBase64(imageBase64) {
   try {
-    return Buffer.from(imageBase64, "base64").toString("utf8");
+    return Buffer.from(imageBase64, "base64");
   } catch {
-    return "";
+    return Buffer.alloc(0);
   }
 }
 
-function buildInvoiceKey(markerText) {
+function isTextLike(value) {
+  if (!value) {
+    return false;
+  }
+
+  const sample = value.slice(0, 256);
+  const controlChars = sample.match(/[\u0000-\u0008\u000B\u000C\u000E-\u001F]/g) ?? [];
+  return controlChars.length === 0;
+}
+
+function decodeMarker(buffer) {
+  const text = buffer.toString("utf8");
+  if (!isTextLike(text)) {
+    return "";
+  }
+
+  return text.replace(/[\u0000-\u001F\u007F]/g, " ").replace(/\s+/g, " ").trim();
+}
+
+function buildInvoiceKey(buffer, markerText) {
   const duplicateMatch = markerText.match(/MOCK_DUP:([A-Za-z0-9_-]+)/);
   if (duplicateMatch?.[1]) {
     return `dup:${duplicateMatch[1]}`;
   }
 
-  const hash = crypto.createHash("sha256").update(markerText || "mock-empty").digest("hex");
+  const hash = crypto.createHash("sha256").update(buffer.length ? buffer : "mock-empty").digest("hex");
   return `hash:${hash.slice(0, 20)}`;
 }
 
@@ -40,12 +59,13 @@ function toAmount(seedHex, fallback) {
 }
 
 async function runMockOcr({ ocrType, imageBase64 }) {
-  const markerText = decodeMarker(imageBase64);
+  const buffer = decodeImageBase64(imageBase64);
+  const markerText = decodeMarker(buffer);
   if (markerText.includes("MOCK_OCR_FAIL")) {
     throw new Error("mock_ocr_failed");
   }
 
-  const invoiceKey = buildInvoiceKey(markerText);
+  const invoiceKey = buildInvoiceKey(buffer, markerText);
   const hash = crypto.createHash("sha256").update(invoiceKey).digest("hex");
   const invoiceCode = `MOCK${hash.slice(0, 8).toUpperCase()}`;
   const invoiceNumber = hash.slice(8, 16).toUpperCase();
@@ -62,7 +82,8 @@ async function runMockOcr({ ocrType, imageBase64 }) {
     raw: {
       provider: "mock",
       invoiceKey,
-      marker: markerText.slice(0, 128),
+      marker: markerText ? markerText.slice(0, 128) : null,
+      sourceSha256: crypto.createHash("sha256").update(buffer).digest("hex"),
     },
   };