优化

integration-service/src/main/java/com/baiyue/integration/dto/ocr/OcrCandidate.java

@@ -1,16 +1,22 @@
 package com.baiyue.integration.dto.ocr;
 
+import java.math.BigDecimal;
+
 /**
  * 百度 OCR 返回的文字候选及其识别框。
  */
 public record OcrCandidate(
         String text,
-        Double confidence,
+        BigDecimal confidence,
         Integer left,
         Integer top,
         Integer width,
         Integer height
 ) {
+    public OcrCandidate {
+        confidence = OcrConfidence.normalize(confidence);
+    }
+
     public long area() {
         if (width == null || height == null || width <= 0 || height <= 0) {
             return 0;

integration-service/src/main/java/com/baiyue/integration/dto/ocr/OcrConfidence.java

@@ -0,0 +1,27 @@
+package com.baiyue.integration.dto.ocr;
+
+import java.math.BigDecimal;
+import java.math.RoundingMode;
+
+/** Canonical confidence representation shared by OCR response models. */
+public final class OcrConfidence {
+    private static final int SCALE = 5;
+    private static final BigDecimal ZERO = BigDecimal.ZERO.setScale(SCALE);
+    private static final BigDecimal ONE = BigDecimal.ONE.setScale(SCALE);
+
+    private OcrConfidence() {
+    }
+
+    public static BigDecimal normalize(BigDecimal value) {
+        if (value == null) {
+            return null;
+        }
+        if (value.compareTo(BigDecimal.ZERO) < 0) {
+            return ZERO;
+        }
+        if (value.compareTo(BigDecimal.ONE) > 0) {
+            return ONE;
+        }
+        return value.setScale(SCALE, RoundingMode.HALF_UP);
+    }
+}

integration-service/src/main/java/com/baiyue/integration/dto/ocr/WeightOcrImageResult.java

@@ -12,10 +12,15 @@ public record WeightOcrImageResult(
         String rawResponseFile,
         Status status,
         BigDecimal weight,
-        Double confidence,
+        BigDecimal confidence,
         List<OcrCandidate> candidates,
+        String ocrRawText,
         String error
 ) {
+    public WeightOcrImageResult {
+        confidence = OcrConfidence.normalize(confidence);
+    }
+
     public enum Status {
         RECOGNIZED,
         NO_WEIGHT_FOUND,

integration-service/src/main/java/com/baiyue/integration/provider/ocr/BaiduMeterClient.java

@@ -12,6 +12,7 @@ import com.fasterxml.jackson.databind.ObjectMapper;
 import org.springframework.stereotype.Component;
 
 import java.io.IOException;
+import java.math.BigDecimal;
 import java.net.URI;
 import java.net.URLEncoder;
 import java.net.http.HttpClient;
@@ -163,7 +164,7 @@ public class BaiduMeterClient {
             JsonNode location = item.path("location");
             candidates.add(new OcrCandidate(
                     item.path("words").asText(),
-                    nullableDouble(probability.get("average")),
+                    nullableDecimal(probability.get("average")),
                     nullableInt(location.get("left")),
                     nullableInt(location.get("top")),
                     nullableInt(location.get("width")),
@@ -195,8 +196,8 @@ public class BaiduMeterClient {
         return URLEncoder.encode(value, StandardCharsets.UTF_8);
     }
 
-    private static Double nullableDouble(JsonNode node) {
-        return node == null || !node.isNumber() ? null : node.doubleValue();
+    private static BigDecimal nullableDecimal(JsonNode node) {
+        return node == null || !node.isNumber() ? null : node.decimalValue();
     }
 
     private static Integer nullableInt(JsonNode node) {

integration-service/src/main/java/com/baiyue/integration/service/ocr/WeightCandidateSelector.java

@@ -25,7 +25,8 @@ public class WeightCandidateSelector {
                 .flatMap(Optional::stream)
                 .max(Comparator
                         .comparingLong((SelectedWeight selected) -> selected.candidate().area())
-                        .thenComparing(selected -> selected.confidence() == null ? -1 : selected.confidence()));
+                        .thenComparing(selected -> selected.confidence() == null
+                                ? BigDecimal.valueOf(-1) : selected.confidence()));
     }
 
     private Optional<SelectedWeight> toSelectedWeight(OcrCandidate candidate) {
@@ -55,6 +56,6 @@ public class WeightCandidateSelector {
                 .replace('－', '-');
     }
 
-    public record SelectedWeight(BigDecimal weight, Double confidence, OcrCandidate candidate) {
+    public record SelectedWeight(BigDecimal weight, BigDecimal confidence, OcrCandidate candidate) {
     }
 }

integration-service/src/main/java/com/baiyue/integration/service/ocr/WeightOcrBatchService.java

@@ -34,6 +34,7 @@ import java.util.stream.Stream;
 @Service
 public class WeightOcrBatchService {
     private static final Set<String> SUPPORTED_EXTENSIONS = Set.of("jpg", "jpeg", "png", "bmp");
+    private static final int RAW_TEXT_MAX_LENGTH = 1000;
 
     @Resource private IntegrationProperties properties;
     @Resource private WeightOcrStoragePolicy storagePolicy;
@@ -113,7 +114,7 @@ public class WeightOcrBatchService {
         return new WeightOcrImageResult(
                 source.toString(), outputRelative(paths, cropped), outputRelative(paths, rawResponse),
                 Status.RECOGNIZED, selected.weight(), selected.confidence(),
-                response.candidates(), null);
+                response.candidates(), summarizeCandidates(response.candidates()), null);
     }
 
     private WeightOcrImageResult noWeight(Path source, BatchPaths paths, Path cropped,
@@ -121,6 +122,7 @@ public class WeightOcrBatchService {
         return new WeightOcrImageResult(
                 source.toString(), outputRelative(paths, cropped), outputRelative(paths, rawResponse),
                 Status.NO_WEIGHT_FOUND, null, null, response.candidates(),
+                summarizeCandidates(response.candidates()),
                 "百度响应中没有符合重量格式的候选值");
     }
 
@@ -129,7 +131,33 @@ public class WeightOcrBatchService {
         return new WeightOcrImageResult(
                 source.toString(), Files.exists(cropped) ? outputRelative(paths, cropped) : null,
                 Files.exists(rawResponse) ? outputRelative(paths, rawResponse) : null,
-                Status.FAILED, null, null, List.of(), error);
+                Status.FAILED, null, null, List.of(), null, error);
+    }
+
+    private String summarizeCandidates(List<com.baiyue.integration.dto.ocr.OcrCandidate> candidates) {
+        if (candidates == null || candidates.isEmpty()) {
+            return null;
+        }
+        StringBuilder summary = new StringBuilder();
+        for (var candidate : candidates) {
+            if (candidate == null || candidate.text() == null) {
+                continue;
+            }
+            String text = candidate.text().strip().replaceAll("[\\p{Cc}\\p{Cf}]+", " ");
+            if (text.isBlank()) {
+                continue;
+            }
+            int separatorLength = summary.isEmpty() ? 0 : 1;
+            int remaining = RAW_TEXT_MAX_LENGTH - summary.length() - separatorLength;
+            if (remaining <= 0) {
+                break;
+            }
+            if (separatorLength > 0) {
+                summary.append('\n');
+            }
+            summary.append(text, 0, Math.min(text.length(), remaining));
+        }
+        return summary.isEmpty() ? null : summary.toString();
     }
 
     private WeightOcrBatchResult buildResult(String batchId, Instant startedAt, BatchPaths paths,

integration-service/src/main/java/com/baiyue/integration/service/ocr/WeightOcrSingleImageService.java

@@ -56,7 +56,9 @@ public class WeightOcrSingleImageService {
             if (batch.results() == null || batch.results().size() != 1) {
                 throw new BizException(ResultCode.INTERNAL_ERROR, "OCR单图识别结果数量异常");
             }
-            return batch.results().get(0);
+            WeightOcrImageResult result = batch.results().get(0);
+            return new WeightOcrImageResult(null, null, null, result.status(), result.weight(),
+                result.confidence(), java.util.List.of(), result.ocrRawText(), result.error());
         } finally {
             deleteRequestDirectory(requestDirectory);
         }