fix

core/src/main/java/com/tzld/videoVector/dao/mapper/pgVector/ext/VideoVectorMapperExt.java

@@ -75,4 +75,22 @@ public interface VideoVectorMapperExt {
     List<VideoVectorSearchResult> searchTopNByCosine(@Param("configCode") String configCode,
                                         @Param("queryVector") String queryVector,
                                         @Param("topN") int topN);
+
+    /**
+     * 分页查询指定 configCode 下的记录（不过滤 text 是否为空，只要有数据就返回）
+     * @param configCode 配置编码
+     * @param offset     偏移量
+     * @param limit      每页数量
+     * @return VideoVector 列表（包含 videoId、pointIndex）
+     */
+    List<VideoVector> selectByConfigCodePaged(@Param("configCode") String configCode,
+                                              @Param("offset") int offset,
+                                              @Param("limit") int limit);
+
+    /**
+     * 批量更新 text 字段
+     * @param id   主键ID
+     * @param text 文本内容
+     */
+    int updateTextById(@Param("id") Long id, @Param("text") String text);
 }

core/src/main/java/com/tzld/videoVector/job/DataMigrationJob.java

@@ -49,16 +49,16 @@ public class DataMigrationJob {
 
     // ==================== PG 数据源 Mapper ====================
     @Resource
-    private DeconstructContentMapper pgContentMapper;
+    private DeconstructContentMapper deconstructContentMapper;
 
     @Resource
-    private DeconstructVectorConfigMapper pgConfigMapper;
+    private DeconstructVectorConfigMapper deconstructVectorConfigMapper;
 
     @Resource
-    private ContentVectorMapperExt pgContentVectorMapper;
+    private ContentVectorMapperExt contentVectorMapperExt;
 
     @Resource
-    private VideoVectorMapperExt pgVideoVectorMapper;
+    private VideoVectorMapperExt videoVectorMapperExt;
 
     // ==================== Redis ====================
     @Resource
@@ -102,7 +102,7 @@ public class DataMigrationJob {
                     // 转换为 PG 实体
                     DeconstructContent pg = convertContent(mysql);
                     // 插入 PG（使用 insertSelective 避免主键冲突时可手动处理）
-                    pgContentMapper.insertSelective(pg);
+                    deconstructContentMapper.insertSelective(pg);
                     successCount++;
                 } catch (Exception e) {
                     if (e.getMessage() != null && e.getMessage().contains("duplicate key")) {
@@ -151,7 +151,7 @@ public class DataMigrationJob {
             for (MysqlDeconstructVectorConfig mysql : allRecords) {
                 try {
                     DeconstructVectorConfig pg = convertVectorConfig(mysql);
-                    pgConfigMapper.insertSelective(pg);
+                    deconstructVectorConfigMapper.insertSelective(pg);
                     successCount++;
                 } catch (Exception e) {
                     if (e.getMessage() != null && e.getMessage().contains("duplicate key")) {
@@ -206,7 +206,7 @@ public class DataMigrationJob {
                     }
 
                     // vectorData 是 JSON 数组格式 "[0.1,0.2,...]"，pgvector 接受此格式
-                    pgContentVectorMapper.upsertWithEmbedding(
+                    contentVectorMapperExt.upsertWithEmbedding(
                             mysql.getContentId(),
                             mysql.getTaskId(),
                             mysql.getConfigCode(),
@@ -301,7 +301,7 @@ public class DataMigrationJob {
 
                             // Redis 中存储的是归一化后的 JSON 数组 "[0.1,0.2,...]"
                             // pgvector 接受此格式
-                            pgVideoVectorMapper.upsertVector(videoId, configCode, 0, vectorJson, null);
+                            videoVectorMapperExt.upsertVector(videoId, configCode, 0, vectorJson, null);
                             totalSuccess++;
                         } catch (NumberFormatException e) {
                             log.warn("非法 videoId: {}", idStr);
@@ -343,7 +343,7 @@ public class DataMigrationJob {
         try {
             DeconstructVectorConfigExample example = new DeconstructVectorConfigExample();
             example.createCriteria().andEnabledEqualTo((short) 1);
-            List<DeconstructVectorConfig> configs = pgConfigMapper.selectByExample(example);
+            List<DeconstructVectorConfig> configs = deconstructVectorConfigMapper.selectByExample(example);
             if (configs != null) {
                 for (DeconstructVectorConfig config : configs) {
                     if (StringUtils.hasText(config.getConfigCode())

core/src/main/java/com/tzld/videoVector/job/VideoVectorTextBackfillJob.java

@@ -0,0 +1,487 @@
+package com.tzld.videoVector.job;
+
+import com.alibaba.fastjson.JSON;
+import com.alibaba.fastjson.JSONArray;
+import com.alibaba.fastjson.JSONObject;
+import com.aliyun.odps.data.Record;
+import com.google.common.collect.Lists;
+import com.tzld.videoVector.api.AigcApiService;
+import com.tzld.videoVector.dao.mapper.pgVector.DeconstructVectorConfigMapper;
+import com.tzld.videoVector.dao.mapper.pgVector.ext.VideoVectorMapperExt;
+import com.tzld.videoVector.model.po.pgVector.DeconstructVectorConfig;
+import com.tzld.videoVector.model.po.pgVector.DeconstructVectorConfigExample;
+import com.tzld.videoVector.model.po.pgVector.VideoVector;
+import com.tzld.videoVector.util.OdpsUtil;
+import com.tzld.videoVector.util.VectorUtils;
+import com.xxl.job.core.biz.model.ReturnT;
+import com.xxl.job.core.handler.annotation.XxlJob;
+import lombok.extern.slf4j.Slf4j;
+import org.springframework.stereotype.Component;
+import org.springframework.util.CollectionUtils;
+import org.springframework.util.StringUtils;
+
+import javax.annotation.Resource;
+import java.util.*;
+import java.util.concurrent.atomic.AtomicInteger;
+import java.util.stream.Collectors;
+
+/**
+ * 历史数据 text 字段补充任务
+ * <p>
+ * 扫描 video_vectors 中的现有数据，根据其 config_code 对应的配置，
+ * 从原始数据源（content_profile / result_log / aigc_deconstruct）中提取文本，
+ * 回填到 video_vectors.text 字段。
+ * <p>
+ * 不检查数据状态，只要有数据就根据 config 提取对应数据。
+ */
+@Slf4j
+@Component
+public class VideoVectorTextBackfillJob {
+
+    @Resource
+    private VideoVectorMapperExt videoVectorMapperExt;
+
+    @Resource
+    private DeconstructVectorConfigMapper vectorConfigMapper;
+
+    @Resource
+    private AigcApiService aigcApiService;
+
+    /** 每批处理数量 */
+    private static final int BATCH_SIZE = 500;
+
+    /**
+     * 历史数据 text 字段补充
+     * 遍历所有启用的配置，对 video_vectors 中已有记录补充 text 字段
+     *
+     * @param param 参数（可选：指定 configCode 只处理特定配置）
+     * @return 执行结果
+     */
+    @XxlJob("videoVectorTextBackfillJob")
+    public ReturnT<String> videoVectorTextBackfillJob(String param) {
+        log.info("开始执行历史数据 text 字段补充任务, param: {}", param);
+
+        try {
+            // 获取所有启用的配置
+            List<DeconstructVectorConfig> configs = getAllEnabledConfigs();
+            if (CollectionUtils.isEmpty(configs)) {
+                log.warn("未找到启用的向量化配置");
+                return ReturnT.SUCCESS;
+            }
+
+            // 如果指定了 configCode，只处理特定配置
+            if (StringUtils.hasText(param)) {
+                configs = configs.stream()
+                        .filter(c -> param.equals(c.getConfigCode()))
+                        .collect(Collectors.toList());
+                if (configs.isEmpty()) {
+                    log.warn("未找到 configCode={} 的启用配置", param);
+                    return ReturnT.SUCCESS;
+                }
+            }
+
+            log.info("共加载 {} 个向量化配置", configs.size());
+
+            int totalUpdateCount = 0;
+            int totalSkipCount = 0;
+            int totalFailCount = 0;
+
+            // 逐个配置处理
+            for (DeconstructVectorConfig config : configs) {
+                String configCode = config.getConfigCode();
+                String sourceField = config.getSourceField();
+                log.info("开始处理配置: configCode={}, sourceField={}", configCode, sourceField);
+
+                int configUpdateCount = 0;
+                int configSkipCount = 0;
+                int configFailCount = 0;
+                int offset = 0;
+
+                while (true) {
+                    // 分页查询该配置下的记录
+                    List<VideoVector> vectors = videoVectorMapperExt.selectByConfigCodePaged(configCode, offset, BATCH_SIZE);
+                    if (CollectionUtils.isEmpty(vectors)) {
+                        log.info("配置 {} 第 {} 偏移无数据，处理结束", configCode, offset);
+                        break;
+                    }
+
+                    // 按 sourceField 类型分发处理
+                    int[] counts = processVectorBatch(vectors, config);
+                    configUpdateCount += counts[0];
+                    configSkipCount += counts[1];
+                    configFailCount += counts[2];
+
+                    if (vectors.size() < BATCH_SIZE) {
+                        break;
+                    }
+                    offset += BATCH_SIZE;
+                }
+
+                log.info("配置 {} 处理完成，更新: {}, 跳过: {}, 失败: {}",
+                        configCode, configUpdateCount, configSkipCount, configFailCount);
+                totalUpdateCount += configUpdateCount;
+                totalSkipCount += configSkipCount;
+                totalFailCount += configFailCount;
+            }
+
+            log.info("历史数据 text 字段补充任务完成，总更新: {}, 总跳过: {}, 总失败: {}",
+                    totalUpdateCount, totalSkipCount, totalFailCount);
+            return ReturnT.SUCCESS;
+        } catch (Exception e) {
+            log.error("历史数据 text 字段补充任务执行失败: {}", e.getMessage(), e);
+            return new ReturnT<>(ReturnT.FAIL_CODE, "任务执行失败: " + e.getMessage());
+        }
+    }
+
+    /**
+     * 处理一批 VideoVector 记录
+     * <p>
+     * 多点模式处理策略：
+     * 原始向量化时 texts[i] 对应 point_index=i，回填时保持相同映射。
+     * 由于置信度过滤可能导致重新提取的文本列表与原始不同，
+     * 当 point_index 超出当前提取的 texts 范围时直接跳过，避免错误匹配。
+     * <p>
+     * 单点模式（point_index=0）：直接取第一个有效文本。
+     *
+     * @return int[]{更新数, 跳过数, 失败数}
+     */
+    private int[] processVectorBatch(List<VideoVector> vectors, DeconstructVectorConfig config) {
+        String sourceField = config.getSourceField();
+        boolean multiPoint = VectorUtils.isMultiPointConfig(config);
+        AtomicInteger updateCount = new AtomicInteger(0);
+        AtomicInteger skipCount = new AtomicInteger(0);
+        AtomicInteger failCount = new AtomicInteger(0);
+
+        // 收集需要处理的 videoId（去重）
+        List<Long> videoIds = vectors.stream()
+                .map(VideoVector::getVideoId)
+                .distinct()
+                .collect(Collectors.toList());
+
+        // 根据 sourceField 获取原始数据
+        Map<Long, String> rawDataMap;
+        switch (sourceField) {
+            case "result_json":
+                rawDataMap = batchQueryVideoRawResults(videoIds);
+                break;
+            case "result_log":
+                rawDataMap = batchQueryResultLogData(videoIds);
+                break;
+            case "aigc_deconstruct":
+                rawDataMap = batchQueryAigcData(videoIds);
+                break;
+            default:
+                log.warn("不支持的 sourceField: {}", sourceField);
+                return new int[]{0, vectors.size(), 0};
+        }
+
+        // 缓存每个 videoId 提取的文本列表，避免对同一 videoId 重复解析
+        Map<Long, List<String>> textsCache = new HashMap<>();
+
+        // 遍历每条记录，提取文本并更新
+        for (VideoVector vector : vectors) {
+            try {
+                Long videoId = vector.getVideoId();
+                String rawData = rawDataMap.get(videoId);
+                if (!StringUtils.hasText(rawData)) {
+                    skipCount.incrementAndGet();
+                    continue;
+                }
+
+                // 根据配置提取文本（相同 videoId 复用缓存）
+                List<String> texts = textsCache.computeIfAbsent(videoId,
+                        id -> extractTexts(rawData, config));
+                if (CollectionUtils.isEmpty(texts)) {
+                    skipCount.incrementAndGet();
+                    continue;
+                }
+
+                // 根据 pointIndex 取对应的文本
+                int pointIndex = vector.getPointIndex() != null ? vector.getPointIndex() : 0;
+                String text;
+
+                if (multiPoint) {
+                    // 多点模式：严格按 point_index 映射，超出范围则跳过
+                    if (pointIndex < texts.size()) {
+                        text = texts.get(pointIndex);
+                    } else {
+                        log.debug("videoId={} point_index={} 超出提取文本范围(size={})，跳过",
+                                videoId, pointIndex, texts.size());
+                        skipCount.incrementAndGet();
+                        continue;
+                    }
+                } else {
+                    // 单点模式：取第一个有效文本
+                    text = texts.stream()
+                            .filter(StringUtils::hasText)
+                            .findFirst()
+                            .orElse(null);
+                }
+
+                if (!StringUtils.hasText(text)) {
+                    skipCount.incrementAndGet();
+                    continue;
+                }
+
+                // 截断文本
+                Integer maxLength = config.getMaxLength();
+                if (maxLength != null && maxLength > 0 && text.length() > maxLength) {
+                    text = text.substring(0, maxLength);
+                }
+
+                // 更新 text 字段
+                videoVectorMapperExt.updateTextById(vector.getId(), text);
+                updateCount.incrementAndGet();
+            } catch (Exception e) {
+                log.error("处理 videoVector id={} videoId={} 时异常: {}",
+                        vector.getId(), vector.getVideoId(), e.getMessage());
+                failCount.incrementAndGet();
+            }
+        }
+
+        return new int[]{updateCount.get(), skipCount.get(), failCount.get()};
+    }
+
+    /**
+     * 根据配置从原始数据中提取文本
+     */
+    private List<String> extractTexts(String rawData, DeconstructVectorConfig config) {
+        List<String> texts = new ArrayList<>();
+        try {
+            JSONObject json = JSON.parseObject(rawData);
+            if (json == null) {
+                return texts;
+            }
+
+            String sourcePath = config.getSourcePath();
+            if (!StringUtils.hasText(sourcePath)) {
+                return texts;
+            }
+
+            String extractRule = config.getExtractRule();
+            if (StringUtils.hasText(extractRule)) {
+                // 多点模式：带置信度过滤
+                texts.addAll(extractTextsWithConfidence(json, sourcePath, extractRule));
+            } else {
+                // 单点模式：直接提取
+                texts.addAll(VectorUtils.extractFromJson(json, sourcePath));
+            }
+        } catch (Exception e) {
+            log.error("解析原始数据失败: {}", e.getMessage());
+        }
+        return texts;
+    }
+
+    /**
+     * 带置信度过滤的文本提取
+     */
+    private List<String> extractTextsWithConfidence(JSONObject json, String sourcePath, String extractRule) {
+        List<String> texts = new ArrayList<>();
+        try {
+            JSONObject rule = JSON.parseObject(extractRule);
+            String textField = rule.getString("text_field");
+            String confidenceField = rule.getString("confidence_field");
+            double confidenceThreshold = rule.getDoubleValue("confidence_threshold");
+
+            if (!StringUtils.hasText(textField) || !StringUtils.hasText(confidenceField)) {
+                return texts;
+            }
+
+            if (sourcePath.endsWith("[*]")) {
+                // 数组路径模式
+                String arrayPath = sourcePath.substring(0, sourcePath.length() - 3);
+                Object arrayObj = navigateToValue(json, arrayPath);
+                if (arrayObj instanceof JSONArray) {
+                    JSONArray array = (JSONArray) arrayObj;
+                    for (int i = 0; i < array.size(); i++) {
+                        JSONObject item = array.getJSONObject(i);
+                        if (item != null && isConfidenceQualified(item, confidenceField, confidenceThreshold)) {
+                            String text = item.getString(textField);
+                            if (StringUtils.hasText(text)) {
+                                texts.add(text);
+                            }
+                        }
+                    }
+                }
+            } else {
+                // 单对象路径模式
+                Object obj = navigateToValue(json, sourcePath);
+                if (obj instanceof JSONObject) {
+                    JSONObject item = (JSONObject) obj;
+                    if (isConfidenceQualified(item, confidenceField, confidenceThreshold)) {
+                        String text = item.getString(textField);
+                        if (StringUtils.hasText(text)) {
+                            texts.add(text);
+                        }
+                    }
+                }
+            }
+        } catch (Exception e) {
+            log.error("置信度提取失败: {}", e.getMessage());
+        }
+        return texts;
+    }
+
+    /**
+     * 根据 JSON 路径导航到目标值
+     */
+    private Object navigateToValue(JSONObject json, String path) {
+        if (json == null || !StringUtils.hasText(path) || !path.startsWith("$.")) {
+            return null;
+        }
+        try {
+            String pathContent = path.substring(2);
+            String[] parts = pathContent.split("\\.");
+            Object current = json;
+            for (String part : parts) {
+                if (current instanceof JSONObject) {
+                    current = ((JSONObject) current).get(part);
+                } else {
+                    return null;
+                }
+            }
+            return current;
+        } catch (Exception e) {
+            return null;
+        }
+    }
+
+    /**
+     * 判断置信度是否满足条件
+     */
+    private boolean isConfidenceQualified(JSONObject item, String confidenceField, double threshold) {
+        Object value = item.get(confidenceField);
+        if (value == null) {
+            return false;
+        }
+        if (value instanceof String) {
+            return "high".equalsIgnoreCase((String) value);
+        }
+        if (value instanceof Number) {
+            return ((Number) value).doubleValue() >= threshold;
+        }
+        return false;
+    }
+
+    // ========================== 数据源查询方法 ==========================
+
+    /**
+     * 批量查询 content_profile 的 raw_result
+     */
+    private Map<Long, String> batchQueryVideoRawResults(List<Long> videoIds) {
+        Map<Long, String> result = new HashMap<>();
+        for (List<Long> partition : Lists.partition(videoIds, 200)) {
+            String idsStr = partition.stream()
+                    .map(String::valueOf)
+                    .collect(Collectors.joining(","));
+            String sql = String.format(
+                    "SELECT content_id, raw_result " +
+                            "FROM videoods.content_profile " +
+                            "WHERE content_id IN (%s);",
+                    idsStr);
+            try {
+                List<Record> records = OdpsUtil.getOdpsData(sql);
+                if (records != null) {
+                    for (Record record : records) {
+                        Long videoId = Long.valueOf(record.getString(0));
+                        String rawResult = record.getString(1);
+                        if (videoId != null && rawResult != null) {
+                            result.put(videoId, rawResult);
+                        }
+                    }
+                }
+            } catch (Exception e) {
+                log.error("查询 content_profile raw_result 失败: {}", e.getMessage());
+            }
+        }
+        return result;
+    }
+
+    /**
+     * 批量查询 result_log 的 data 字段
+     */
+    private Map<Long, String> batchQueryResultLogData(List<Long> videoIds) {
+        Map<Long, String> result = new HashMap<>();
+        for (List<Long> partition : Lists.partition(videoIds, 200)) {
+            String idsStr = partition.stream()
+                    .map(String::valueOf)
+                    .collect(Collectors.joining(","));
+            String sql = String.format(
+                    "SELECT video_id, data " +
+                            "FROM loghubods.result_log " +
+                            "WHERE video_id IN (%s) AND dt > 20240001;",
+                    idsStr);
+            try {
+                List<Record> records = OdpsUtil.getOdpsData(sql);
+                if (records != null) {
+                    for (Record record : records) {
+                        Long videoId = Long.valueOf(record.getString(0));
+                        String data = record.getString(1);
+                        if (videoId != null && data != null) {
+                            result.put(videoId, data);
+                        }
+                    }
+                }
+            } catch (Exception e) {
+                log.error("查询 result_log data 失败: {}", e.getMessage());
+            }
+        }
+        return result;
+    }
+
+    /**
+     * 批量查询 AIGC 解构数据
+     * 通过 AIGC API 获取 dataContent
+     */
+    private Map<Long, String> batchQueryAigcData(List<Long> videoIds) {
+        Map<Long, String> result = new HashMap<>();
+        try {
+            // 获取 AIGC 任务输入列表
+            List<AigcApiService.AigcTaskInput> taskInputList = aigcApiService.getTaskInputList(46);
+            if (CollectionUtils.isEmpty(taskInputList)) {
+                return result;
+            }
+
+            // 构建 videoId -> taskInstanceId 映射
+            Map<Long, Long> videoIdToTaskInstanceId = new HashMap<>();
+            for (AigcApiService.AigcTaskInput input : taskInputList) {
+                try {
+                    Long videoId = Long.parseLong(input.getBizUniqueId());
+                    videoIdToTaskInstanceId.put(videoId, input.getTaskInstanceId());
+                } catch (NumberFormatException e) {
+                    // 忽略格式非法的
+                }
+            }
+
+            // 对需要的 videoId 查询 dataContent
+            Set<Long> targetIds = new HashSet<>(videoIds);
+            for (Map.Entry<Long, Long> entry : videoIdToTaskInstanceId.entrySet()) {
+                Long videoId = entry.getKey();
+                if (!targetIds.contains(videoId)) {
+                    continue;
+                }
+                try {
+                    JSONObject dataContent = aigcApiService.getTaskCallbackDetail(entry.getValue());
+                    if (dataContent != null) {
+                        result.put(videoId, dataContent.toJSONString());
+                    }
+                } catch (Exception e) {
+                    log.error("查询 AIGC dataContent 失败, videoId={}: {}", videoId, e.getMessage());
+                }
+            }
+        } catch (Exception e) {
+            log.error("批量查询 AIGC 数据失败: {}", e.getMessage());
+        }
+        return result;
+    }
+
+    /**
+     * 获取所有启用的向量化配置
+     */
+    private List<DeconstructVectorConfig> getAllEnabledConfigs() {
+        DeconstructVectorConfigExample example = new DeconstructVectorConfigExample();
+        example.createCriteria().andEnabledEqualTo((short) 1);
+        example.setOrderByClause("priority ASC");
+        return vectorConfigMapper.selectByExample(example);
+    }
+}

core/src/main/java/com/tzld/videoVector/service/impl/PgVectorStoreServiceImpl.java

@@ -31,7 +31,7 @@ import java.util.stream.Collectors;
 public class PgVectorStoreServiceImpl implements VectorStoreService {
 
     @Autowired
-    private VideoVectorMapperExt videoVectorMapper;
+    private VideoVectorMapperExt videoVectorMapperExt;
 
     // ---------------------------------------------------------------- CRUD
 
@@ -56,7 +56,7 @@ public class PgVectorStoreServiceImpl implements VectorStoreService {
         }
 
         String embedding = vectorToString(vector);
-        videoVectorMapper.upsertVector(videoId, configCode, pointIndex, embedding, text);
+        videoVectorMapperExt.upsertVector(videoId, configCode, pointIndex, embedding, text);
         log.debug("保存向量成功，configCode={}, videoId={}, pointIndex={}, 维度={}", configCode, videoId, pointIndex, vector.size());
     }
 
@@ -71,7 +71,7 @@ public class PgVectorStoreServiceImpl implements VectorStoreService {
         if (configCode == null || configCode.isEmpty()) {
             configCode = DEFAULT_CONFIG_CODE;
         }
-        return videoVectorMapper.existsByVideoIdAndConfigCode(videoId, configCode) > 0;
+        return videoVectorMapperExt.existsByVideoIdAndConfigCode(videoId, configCode) > 0;
     }
 
     @Override
@@ -93,7 +93,7 @@ public class PgVectorStoreServiceImpl implements VectorStoreService {
         for (int i = 0; i < idList.size(); i += batchSize) {
             int end = Math.min(i + batchSize, idList.size());
             List<Long> batch = idList.subList(i, end);
-            List<Long> found = videoVectorMapper.selectExistingVideoIds(batch, configCode);
+            List<Long> found = videoVectorMapperExt.selectExistingVideoIds(batch, configCode);
             if (found != null) {
                 existing.addAll(found);
             }
@@ -113,7 +113,7 @@ public class PgVectorStoreServiceImpl implements VectorStoreService {
             configCode = DEFAULT_CONFIG_CODE;
         }
 
-        String embedding = videoVectorMapper.selectEmbeddingByVideoIdAndConfigCode(videoId, configCode);
+        String embedding = videoVectorMapperExt.selectEmbeddingByVideoIdAndConfigCode(videoId, configCode);
         return VectorUtils.parseVectorString(embedding);
     }
 
@@ -136,7 +136,7 @@ public class PgVectorStoreServiceImpl implements VectorStoreService {
         for (int i = 0; i < idList.size(); i += batchSize) {
             int end = Math.min(i + batchSize, idList.size());
             List<Long> batch = idList.subList(i, end);
-            List<VideoVector> vectors = videoVectorMapper.selectVectorsByVideoIds(batch, configCode);
+            List<VideoVector> vectors = videoVectorMapperExt.selectVectorsByVideoIds(batch, configCode);
             if (vectors != null) {
                 for (VideoVector vv : vectors) {
                     List<Float> parsed = VectorUtils.parseVectorString(vv.getEmbedding());
@@ -159,7 +159,7 @@ public class PgVectorStoreServiceImpl implements VectorStoreService {
         if (configCode == null || configCode.isEmpty()) {
             configCode = DEFAULT_CONFIG_CODE;
         }
-        List<Long> ids = videoVectorMapper.selectAllVideoIdsByConfigCode(configCode);
+        List<Long> ids = videoVectorMapperExt.selectAllVideoIdsByConfigCode(configCode);
         if (ids == null) return Collections.emptySet();
         return new HashSet<>(ids);
     }
@@ -175,7 +175,7 @@ public class PgVectorStoreServiceImpl implements VectorStoreService {
         if (configCode == null || configCode.isEmpty()) {
             configCode = DEFAULT_CONFIG_CODE;
         }
-        videoVectorMapper.deleteByVideoIdAndConfigCode(videoId, configCode);
+        videoVectorMapperExt.deleteByVideoIdAndConfigCode(videoId, configCode);
         log.debug("删除向量成功，configCode={}, videoId={}", configCode, videoId);
     }
 
@@ -196,7 +196,7 @@ public class PgVectorStoreServiceImpl implements VectorStoreService {
         for (int i = 0; i < idList.size(); i += batchSize) {
             int end = Math.min(i + batchSize, idList.size());
             List<Long> batch = idList.subList(i, end);
-            videoVectorMapper.deleteBatchByVideoIdsAndConfigCode(batch, configCode);
+            videoVectorMapperExt.deleteBatchByVideoIdsAndConfigCode(batch, configCode);
         }
         log.info("批量删除向量成功，configCode={}, 数量={}", configCode, videoIds.size());
     }
@@ -220,7 +220,7 @@ public class PgVectorStoreServiceImpl implements VectorStoreService {
         String queryVectorStr = vectorToString(queryVector);
         log.info("开始pgvector搜索，configCode={}，topN={}", configCode, topN);
 
-        List<VideoVectorSearchResult> results = videoVectorMapper.searchTopNByCosine(configCode, queryVectorStr, topN);
+        List<VideoVectorSearchResult> results = videoVectorMapperExt.searchTopNByCosine(configCode, queryVectorStr, topN);
         if (results == null || results.isEmpty()) {
             log.info("向量库为空或无匹配结果，configCode={}", configCode);
             return Collections.emptyList();

core/src/main/java/com/tzld/videoVector/service/impl/VectorizeServiceImpl.java

@@ -28,10 +28,10 @@ import java.util.List;
 public class VectorizeServiceImpl implements VectorizeService {
 
     @Resource
-    private DeconstructVectorConfigMapperExt vectorConfigMapper;
+    private DeconstructVectorConfigMapperExt vectorConfigMapperExt;
 
     @Resource
-    private ContentVectorMapperExt pgContentVectorMapper;
+    private ContentVectorMapperExt contentVectorMapperExt;
 
     @Resource
     private EmbeddingService embeddingService;
@@ -39,7 +39,7 @@ public class VectorizeServiceImpl implements VectorizeService {
     @Override
     public List<DeconstructVectorConfig> getVectorConfigs(Integer bizType, Integer contentType) {
         // 使用 null-aware 查询：字段为 null（通用配置）和字段等于指定値的配置均可匹配
-        return vectorConfigMapper.selectMatchingConfigs(
+        return vectorConfigMapperExt.selectMatchingConfigs(
                 bizType != null ? bizType.shortValue() : null,
                 contentType != null ? contentType.shortValue() : null
         );
@@ -98,7 +98,7 @@ public class VectorizeServiceImpl implements VectorizeService {
 
     @Override
     public List<ContentVector> vectorizeByConfig(DeconstructContent content,
-                                                                 DeconstructVectorConfig config) {
+                                                 DeconstructVectorConfig config) {
         List<ContentVector> vectors = new ArrayList<>();
 
         // 提取文本内容
@@ -221,7 +221,7 @@ public class VectorizeServiceImpl implements VectorizeService {
         }
         try {
             // 从 pgvector 查询
-            ContentVector cached = pgContentVectorMapper.selectByTextHashAndConfigCode(textHash, configCode);
+            ContentVector cached = contentVectorMapperExt.selectByTextHashAndConfigCode(textHash, configCode);
             if (cached != null && StringUtils.hasText(cached.getEmbedding())) {
                 return VectorUtils.parseVectorString(cached.getEmbedding());
             }
@@ -243,7 +243,7 @@ public class VectorizeServiceImpl implements VectorizeService {
             try {
                 // 存储到 pgvector，embedding 是 JSON 数组格式 "[0.1,0.2,...]"
                 String embedding = vector.getEmbedding();
-                pgContentVectorMapper.upsertWithEmbedding(
+                contentVectorMapperExt.upsertWithEmbedding(
                         vector.getContentId(),
                         vector.getTaskId(),
                         vector.getConfigCode(),
@@ -268,17 +268,17 @@ public class VectorizeServiceImpl implements VectorizeService {
 
     @Override
     public List<ContentVector> getVectorsByContentId(Long contentId) {
-        return pgContentVectorMapper.selectByContentId(contentId);
+        return contentVectorMapperExt.selectByContentId(contentId);
     }
 
     @Override
     public List<ContentVector> getVectorsByContentId(Long contentId, String configCode) {
-        return pgContentVectorMapper.selectByContentIdAndConfigCode(contentId, configCode);
+        return contentVectorMapperExt.selectByContentIdAndConfigCode(contentId, configCode);
     }
 
     @Override
     public List<ContentVector> getVectorsByField(Long contentId, String sourceField) {
-        return pgContentVectorMapper.selectByContentIdAndField(contentId, sourceField);
+        return contentVectorMapperExt.selectByContentIdAndField(contentId, sourceField);
     }
 
 }

core/src/main/resources/mapper/pgVector/ext/VideoVectorMapperExt.xml

@@ -94,4 +94,19 @@
     LIMIT #{topN}
   </select>
 
+  <!-- 分页查询指定 configCode 下的记录 -->
+  <select id="selectByConfigCodePaged" resultMap="VectorWithEmbeddingResultMap">
+    SELECT id, video_id, config_code, point_index, created_at, updated_at
+    FROM video_vectors
+    WHERE config_code = #{configCode}
+    ORDER BY id
+    LIMIT #{limit} OFFSET #{offset}
+  </select>
+
+  <!-- 更新 text 字段 -->
+  <update id="updateTextById">
+    UPDATE video_vectors SET "text" = #{text}, updated_at = NOW()
+    WHERE id = #{id}
+  </update>
+
 </mapper>

server/src/main/java/com/tzld/videoVector/controller/XxlJobController.java

@@ -20,6 +20,9 @@ public class XxlJobController {
     @Autowired
     private MaterialVectorJob materialVectorJob;
 
+    @Autowired
+    private VideoVectorTextBackfillJob videoVectorTextBackfillJob;
+
     // ==================== 视频向量化任务 ====================
 
     @GetMapping("/vectorVideoJob")
@@ -66,5 +69,11 @@ public class XxlJobController {
         return CommonResponse.success();
     }
 
+    @GetMapping("/videoVectorTextBackfillJob")
+    public CommonResponse<Void> videoVectorTextBackfillJob() {
+        videoVectorTextBackfillJob.videoVectorTextBackfillJob(null);
+        return CommonResponse.success();
+    }
+
 
 }

server/src/main/resources/application.yml

@@ -20,7 +20,7 @@ pagehelper:
 
 mybatis:
   configuration:
-    log-impl: org.apache.ibatis.logging.stdout.StdOutImpl  # 直接输出到控制台（简单调试用）
+    log-impl: org.apache.ibatis.logging.slf4j.Slf4jImpl  # 通过 SLF4J 输出，受 logback 级别控制
 
 app:
   id: video-vector-server  # 应用标识

server/src/main/resources/logback-spring.xml

@@ -151,6 +151,8 @@
     <logger name="org.apache.http.impl.conn.Wire" level="WARN"/>
     <!-- 屏蔽 MyBatis SqlSession 创建/关闭的 DEBUG 日志 -->
     <logger name="org.mybatis.spring.SqlSessionUtils" level="WARN"/>
+    <!-- 屏蔽 JDBC Connection will not be managed by Spring 日志 -->
+    <logger name="org.mybatis.spring.transaction.SpringManagedTransaction" level="WARN"/>
     <!--aliyun loghub 为了防止进程退出时，内存中的数据丢失，请加上此选项-->
     <shutdownHook class="ch.qos.logback.core.hook.DelayingShutdownHook"/>