R92 A2 修复发票 OCR provider 与二进制写库

server/prisma/migrations/20260505211000_r92_ocr_provider_mock/migration.sql

@@ -0,0 +1 @@
+ALTER TYPE "OcrProvider" ADD VALUE IF NOT EXISTS 'mock';

server/prisma/schema.prisma

@@ -160,6 +160,7 @@ enum InvoiceArchivedReason {
 
 enum OcrProvider {
   tencent
+  mock
 }
 
 enum OcrType {

server/src/jobs/ocrJob.js

@@ -4,6 +4,16 @@ const { buildMissingRequiredOcrError, runOcrForInvoice } = require("../services/
 const { readFileBuffer } = require("../services/storage");
 
 const OCR_DUPLICATE_ERROR = "duplicate_invoice_code_number";
+const SUPPORTED_OCR_PROVIDERS = new Set(["tencent", "mock"]);
+
+function getEffectiveOcrProvider(config) {
+  const provider = config?.ocrProvider;
+  if (!SUPPORTED_OCR_PROVIDERS.has(provider)) {
+    throw new Error(`Unsupported OCR provider: ${provider}`);
+  }
+
+  return provider;
+}
 
 function trimErrorMessage(error) {
   const code = typeof error?.code === "string" ? error.code : null;
@@ -59,7 +69,7 @@ async function claimPendingInvoices(prisma, config, workerId) {
         ocr_lock_owner = ${workerId}
     FROM candidates
     WHERE i.id = candidates.id
-    RETURNING i.id, i.file_sha256, i.ocr_type, i.storage_path, i.mime_type, i.ocr_attempts
+    RETURNING i.id, i.file_sha256, i.ocr_provider, i.ocr_type, i.storage_path, i.mime_type, i.ocr_attempts
   `;
 }
 
@@ -89,10 +99,11 @@ async function findDuplicateInvoice(prisma, invoiceId, invoiceCode, invoiceNumbe
 }
 
 async function getOrCreateOcrResult(prisma, config, claimedInvoice, imageBase64) {
+  const ocrProvider = getEffectiveOcrProvider(config);
   const where = {
     file_sha256_ocr_provider_ocr_type: {
       file_sha256: claimedInvoice.file_sha256,
-      ocr_provider: "tencent",
+      ocr_provider: ocrProvider,
       ocr_type: claimedInvoice.ocr_type,
     },
   };
@@ -123,7 +134,7 @@ async function getOrCreateOcrResult(prisma, config, claimedInvoice, imageBase64)
     const created = await prisma.ocrResult.create({
       data: {
         file_sha256: claimedInvoice.file_sha256,
-        ocr_provider: "tencent",
+        ocr_provider: ocrProvider,
         ocr_type: claimedInvoice.ocr_type,
         result_json: normalized,
       },
@@ -160,12 +171,13 @@ async function getOrCreateOcrResult(prisma, config, claimedInvoice, imageBase64)
   }
 }
 
-async function markInvoiceAsDuplicate(prisma, invoiceId, normalized, ocrResultId) {
+async function markInvoiceAsDuplicate(prisma, config, invoiceId, normalized, ocrResultId) {
   const now = new Date();
   await prisma.invoicePool.update({
     where: { id: invoiceId },
     data: {
       status: "ocr_failed",
+      ocr_provider: getEffectiveOcrProvider(config),
       ocr_result_id: ocrResultId,
       invoice_code: normalized.invoiceCode ?? null,
       invoice_number: normalized.invoiceNumber ?? null,
@@ -184,12 +196,13 @@ async function markInvoiceAsDuplicate(prisma, invoiceId, normalized, ocrResultId
   });
 }
 
-async function markInvoiceAsAvailable(prisma, invoiceId, normalized, ocrResultId) {
+async function markInvoiceAsAvailable(prisma, config, invoiceId, normalized, ocrResultId) {
   const now = new Date();
   await prisma.invoicePool.update({
     where: { id: invoiceId },
     data: {
       status: "available",
+      ocr_provider: getEffectiveOcrProvider(config),
       ocr_result_id: ocrResultId,
       invoice_code: normalized.invoiceCode ?? null,
       invoice_number: normalized.invoiceNumber ?? null,
@@ -208,12 +221,13 @@ async function markInvoiceAsAvailable(prisma, invoiceId, normalized, ocrResultId
   });
 }
 
-async function markInvoiceAsQualityFailed(prisma, invoiceId, normalized, ocrResultId, errorMessage) {
+async function markInvoiceAsQualityFailed(prisma, config, invoiceId, normalized, ocrResultId, errorMessage) {
   const now = new Date();
   await prisma.invoicePool.update({
     where: { id: invoiceId },
     data: {
       status: "ocr_failed",
+      ocr_provider: getEffectiveOcrProvider(config),
       ocr_result_id: ocrResultId,
       invoice_code: normalized.invoiceCode ?? null,
       invoice_number: normalized.invoiceNumber ?? null,
@@ -241,6 +255,7 @@ async function markInvoiceAsFailed(prisma, claimedInvoice, errorMessage, config)
     where: { id: claimedInvoice.id },
     data: {
       status: isTerminal ? "ocr_failed" : "pending_ocr",
+      ocr_provider: getEffectiveOcrProvider(config),
       ocr_attempts: nextAttempts,
       ocr_last_error: errorMessage,
       ocr_completed_at: isTerminal ? now : null,
@@ -269,6 +284,7 @@ async function processClaimedInvoice(prisma, config, claimedInvoice, workerId, l
     if (qualityError) {
       await markInvoiceAsQualityFailed(
         prisma,
+        config,
         claimedInvoice.id,
         normalized,
         ocrResultId,
@@ -295,7 +311,7 @@ async function processClaimedInvoice(prisma, config, claimedInvoice, workerId, l
     );
 
     if (duplicate) {
-      await markInvoiceAsDuplicate(prisma, claimedInvoice.id, normalized, ocrResultId);
+      await markInvoiceAsDuplicate(prisma, config, claimedInvoice.id, normalized, ocrResultId);
       logger.warn(
         {
           invoiceId: claimedInvoice.id,
@@ -308,7 +324,7 @@ async function processClaimedInvoice(prisma, config, claimedInvoice, workerId, l
       return { outcome: "duplicate" };
     }
 
-    await markInvoiceAsAvailable(prisma, claimedInvoice.id, normalized, ocrResultId);
+    await markInvoiceAsAvailable(prisma, config, claimedInvoice.id, normalized, ocrResultId);
     return { outcome: "available" };
   } catch (error) {
     const message = trimErrorMessage(error);
@@ -427,6 +443,7 @@ function startOcrJob(fastify) {
 module.exports = {
   OCR_DUPLICATE_ERROR,
   claimPendingInvoices,
+  getEffectiveOcrProvider,
   markInvoiceAsQualityFailed,
   runOcrOnce,
   startOcrJob,

server/src/routes/invoicePool.js

@@ -356,7 +356,8 @@ async function invoicePoolRoutes(fastify) {
         contentType: fileMimeType,
       });
 
-      const status = isOcrEnabledForType(fastify.config, ocrType) ? "pending_ocr" : "available";
+      const ocrEnabled = isOcrEnabledForType(fastify.config, ocrType);
+      const status = ocrEnabled ? "pending_ocr" : "available";
 
       const invoice = await fastify.prisma.$transaction(async (tx) => {
         const now = new Date();
@@ -369,6 +370,7 @@ async function invoicePoolRoutes(fastify) {
             mime_type: fileMimeType,
             file_size: fileSize,
             file_sha256: fileSha256,
+            ocr_provider: ocrEnabled ? fastify.config.ocrProvider : undefined,
             ocr_type: ocrType,
             status,
             ocr_last_requested_at: status === "pending_ocr" ? now : null,
@@ -1036,6 +1038,7 @@ async function invoicePoolRoutes(fastify) {
           where: { id: invoiceId },
           data: {
             status: "pending_ocr",
+            ocr_provider: fastify.config.ocrProvider,
             ocr_last_requested_at: now,
             ocr_last_error: null,
             ocr_completed_at: null,

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"),
     },
   };
 

server/tests/unit/jobs/ocrJob.test.js

@@ -8,8 +8,10 @@ function buildClaimedInvoice(overrides = {}) {
   return {
     id: "11111111-1111-4111-8111-111111111111",
     file_sha256: "sha256-1",
+    ocr_provider: "mock",
     ocr_type: "vat",
     storage_path: "20260428/invoice.txt",
+    mime_type: "text/plain",
     ocr_attempts: 0,
     ...overrides,
   };
@@ -74,10 +76,18 @@ describe("ocrJob quality gate", () => {
         where: { id: "11111111-1111-4111-8111-111111111111" },
         data: expect.objectContaining({
           status: "available",
+          ocr_provider: "mock",
           ocr_last_error: null,
         }),
       }),
     );
+    expect(prisma.ocrResult.create).toHaveBeenCalledWith(
+      expect.objectContaining({
+        data: expect.objectContaining({
+          ocr_provider: "mock",
+        }),
+      }),
+    );
   });
 
   it("marks invoice ocr_failed when required VAT fields are missing", async () => {
@@ -112,6 +122,7 @@ describe("ocrJob quality gate", () => {
         where: { id: "11111111-1111-4111-8111-111111111111" },
         data: expect.objectContaining({
           status: "ocr_failed",
+          ocr_provider: "mock",
           ocr_last_error: "missing_required_fields:invoiceCode,sellerName",
         }),
       }),
@@ -124,4 +135,45 @@ describe("ocrJob quality gate", () => {
       "OCR invoice marked as failed for missing required fields",
     );
   });
+
+  it("stores binary PDF mock OCR metadata without unsafe null marker text", async () => {
+    const pdfPayload = Buffer.from([0x25, 0x50, 0x44, 0x46, 0x00, 0xff, 0x10, 0x20, 0x41]);
+    await fs.writeFile(path.join(tempDir, "20260428/invoice.pdf"), pdfPayload);
+
+    const prisma = buildPrisma([
+      buildClaimedInvoice({
+        storage_path: "20260428/invoice.pdf",
+        mime_type: "application/pdf",
+        file_sha256: "sha256-pdf-binary",
+      }),
+    ]);
+    const logger = {
+      warn: vi.fn(),
+      error: vi.fn(),
+    };
+
+    const result = await runOcrOnce(
+      prisma,
+      {
+        storageDir: tempDir,
+        ocrBatchSize: 1,
+        ocrLockTtlSeconds: 300,
+        ocrMaxAttempts: 3,
+        ocrProvider: "mock",
+      },
+      logger,
+      "ocr-test-worker",
+    );
+
+    expect(result).toEqual({ claimed: 1, available: 1, duplicate: 0, failed: 0 });
+    const createdPayload = prisma.ocrResult.create.mock.calls[0][0].data.result_json;
+    expect(createdPayload.raw).toEqual(
+      expect.objectContaining({
+        provider: "mock",
+        marker: null,
+        sourceSha256: expect.any(String),
+      }),
+    );
+    expect(JSON.stringify(createdPayload)).not.toContain("\\u0000");
+  });
 });

server/tests/unit/routes/invoicePool.upload.test.js

@@ -197,6 +197,7 @@ describe("invoicePool upload route", () => {
         expect.objectContaining({
           data: expect.objectContaining({
             ocr_type: "vat",
+            ocr_provider: "mock",
             status: "pending_ocr",
           }),
         }),