R92 A2 修复发票 OCR provider 与二进制写库

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,