发布 v1.10 #5
@@ -22,7 +22,6 @@ import com.easyagents.search.engine.service.DocumentSearcher;
|
||||
import com.easyagents.search.engine.service.KeywordSearchMetadataKeys;
|
||||
import com.mybatisflex.core.keygen.impl.FlexIDKeyGenerator;
|
||||
import com.mybatisflex.core.paginate.Page;
|
||||
import com.mybatisflex.core.query.QueryMethods;
|
||||
import com.mybatisflex.core.query.QueryWrapper;
|
||||
import com.mybatisflex.spring.service.impl.ServiceImpl;
|
||||
import org.slf4j.Logger;
|
||||
@@ -110,7 +109,7 @@ public class DocumentServiceImpl extends ServiceImpl<DocumentMapper, Document> i
|
||||
|
||||
@Override
|
||||
public Page<Document> getDocumentList(String knowledgeId, int pageSize, int pageNum, String fileName) {
|
||||
return queryDocumentList(knowledgeId, pageSize, pageNum, fileName, null);
|
||||
return queryDocumentList(knowledgeId, pageSize, pageNum, fileName, null, false);
|
||||
}
|
||||
|
||||
/**
|
||||
@@ -129,7 +128,7 @@ public class DocumentServiceImpl extends ServiceImpl<DocumentMapper, Document> i
|
||||
int pageNum,
|
||||
BigInteger documentId
|
||||
) {
|
||||
return queryDocumentList(knowledgeId, pageSize, pageNum, null, documentId);
|
||||
return queryDocumentList(knowledgeId, pageSize, pageNum, null, documentId, true);
|
||||
}
|
||||
|
||||
/**
|
||||
@@ -140,6 +139,7 @@ public class DocumentServiceImpl extends ServiceImpl<DocumentMapper, Document> i
|
||||
* @param pageNum 页码
|
||||
* @param fileName 可选的文件标题筛选
|
||||
* @param documentId 可选的文档 ID
|
||||
* @param includeDetailFields 是否返回正文、存储路径和扩展配置等详情字段
|
||||
* @return 文档分页
|
||||
*/
|
||||
private Page<Document> queryDocumentList(
|
||||
@@ -147,27 +147,45 @@ public class DocumentServiceImpl extends ServiceImpl<DocumentMapper, Document> i
|
||||
int pageSize,
|
||||
int pageNum,
|
||||
String fileName,
|
||||
BigInteger documentId
|
||||
BigInteger documentId,
|
||||
boolean includeDetailFields
|
||||
) {
|
||||
QueryWrapper queryWrapper=QueryWrapper.create()
|
||||
.select(
|
||||
DOCUMENT.ALL_COLUMNS,
|
||||
QueryMethods.count(DOCUMENT_CHUNK.DOCUMENT_ID).as("chunk_count")
|
||||
|
||||
)
|
||||
.from(Document.class)
|
||||
.leftJoin(DocumentChunk.class).on(DOCUMENT.ID.eq(DOCUMENT_CHUNK.DOCUMENT_ID))
|
||||
.where(DOCUMENT.COLLECTION_ID.eq(knowledgeId))
|
||||
.orderBy(DOCUMENT.ID, false)
|
||||
;
|
||||
QueryWrapper queryWrapper = QueryWrapper.create().from(Document.class);
|
||||
if (includeDetailFields) {
|
||||
// 公开文档分页继续保留历史字段契约,但分块数直接使用文档表中的持久化统计。
|
||||
queryWrapper.select(
|
||||
DOCUMENT.ALL_COLUMNS,
|
||||
DOCUMENT.TOTAL_CHUNKS.as("chunk_count")
|
||||
);
|
||||
} else {
|
||||
// 列表页只读取展示字段,避免批量导入期间反复传输 LONGTEXT 正文和扩展配置。
|
||||
queryWrapper.select(
|
||||
DOCUMENT.ID,
|
||||
DOCUMENT.COLLECTION_ID,
|
||||
DOCUMENT.DOCUMENT_TYPE,
|
||||
DOCUMENT.TITLE,
|
||||
DOCUMENT.CONTENT_TYPE,
|
||||
DOCUMENT.PROCESS_STATUS,
|
||||
DOCUMENT.TOTAL_CHUNKS,
|
||||
DOCUMENT.COMPLETED_CHUNKS,
|
||||
DOCUMENT.FAILED_CHUNKS,
|
||||
DOCUMENT.PROGRESS_PERCENT,
|
||||
DOCUMENT.LAST_TASK_ERROR,
|
||||
DOCUMENT.TASK_MODIFIED_AT,
|
||||
DOCUMENT.CREATED,
|
||||
DOCUMENT.MODIFIED,
|
||||
DOCUMENT.TOTAL_CHUNKS.as("chunk_count")
|
||||
);
|
||||
}
|
||||
queryWrapper
|
||||
.where(DOCUMENT.COLLECTION_ID.eq(knowledgeId))
|
||||
.orderBy(DOCUMENT.ID, false);
|
||||
if (fileName != null && !fileName.trim().isEmpty()) {
|
||||
queryWrapper.and(DOCUMENT.TITLE.like(fileName));
|
||||
}
|
||||
if (documentId != null) {
|
||||
queryWrapper.and(DOCUMENT.ID.eq(documentId));
|
||||
}
|
||||
// 分组
|
||||
queryWrapper.groupBy(DOCUMENT.ID);
|
||||
return documentMapper.paginateAs(pageNum, pageSize, queryWrapper, Document.class);
|
||||
}
|
||||
|
||||
|
||||
@@ -5,6 +5,7 @@ import com.easyagents.core.store.StoreResult;
|
||||
import com.mybatisflex.core.query.QueryWrapper;
|
||||
import org.junit.Assert;
|
||||
import org.junit.Test;
|
||||
import org.mockito.ArgumentCaptor;
|
||||
import org.mockito.Mockito;
|
||||
import tech.easyflow.ai.config.SearcherFactory;
|
||||
import tech.easyflow.ai.entity.Document;
|
||||
@@ -21,6 +22,8 @@ import tech.easyflow.common.web.exceptions.BusinessException;
|
||||
import java.lang.reflect.Field;
|
||||
import java.math.BigInteger;
|
||||
import java.util.List;
|
||||
import java.util.Locale;
|
||||
import java.util.regex.Pattern;
|
||||
|
||||
/**
|
||||
* {@link DocumentServiceImpl} 文档维护回归测试。
|
||||
@@ -30,6 +33,70 @@ import java.util.List;
|
||||
*/
|
||||
public class DocumentServiceImplTest {
|
||||
|
||||
/**
|
||||
* 验证管理端列表只查询展示字段,并直接使用文档表中的分块统计。
|
||||
*
|
||||
* @throws Exception 反射注入异常
|
||||
*/
|
||||
@Test
|
||||
public void getDocumentListShouldAvoidChunkJoinAndLargeFields() throws Exception {
|
||||
DocumentMapper documentMapper = Mockito.mock(DocumentMapper.class);
|
||||
DocumentServiceImpl service = new DocumentServiceImpl();
|
||||
setField(service, "documentMapper", documentMapper);
|
||||
|
||||
service.getDocumentList("1001", 10, 1, null);
|
||||
|
||||
ArgumentCaptor<QueryWrapper> queryCaptor =
|
||||
ArgumentCaptor.forClass(QueryWrapper.class);
|
||||
Mockito.verify(documentMapper).paginateAs(
|
||||
Mockito.eq(1),
|
||||
Mockito.eq(10),
|
||||
queryCaptor.capture(),
|
||||
Mockito.eq(Document.class)
|
||||
);
|
||||
String sql = normalizeSql(queryCaptor.getValue().toSQL());
|
||||
Assert.assertFalse(sql.contains("tb_document_chunk"));
|
||||
Assert.assertFalse(containsSqlIdentifier(sql, "content"));
|
||||
Assert.assertFalse(containsSqlIdentifier(sql, "options"));
|
||||
Assert.assertFalse(sql.contains("group by"));
|
||||
Assert.assertTrue(containsSqlIdentifier(sql, "total_chunks"));
|
||||
Assert.assertTrue(containsSqlIdentifier(sql, "chunk_count"));
|
||||
}
|
||||
|
||||
/**
|
||||
* 验证公开文档分页继续返回历史详情字段,同时不再关联分块表聚合。
|
||||
*
|
||||
* @throws Exception 反射注入异常
|
||||
*/
|
||||
@Test
|
||||
public void getDocumentListByIdShouldKeepDetailFieldsWithoutChunkJoin()
|
||||
throws Exception {
|
||||
DocumentMapper documentMapper = Mockito.mock(DocumentMapper.class);
|
||||
DocumentServiceImpl service = new DocumentServiceImpl();
|
||||
setField(service, "documentMapper", documentMapper);
|
||||
|
||||
service.getDocumentListById(
|
||||
"1001",
|
||||
10,
|
||||
1,
|
||||
BigInteger.valueOf(2002)
|
||||
);
|
||||
|
||||
ArgumentCaptor<QueryWrapper> queryCaptor =
|
||||
ArgumentCaptor.forClass(QueryWrapper.class);
|
||||
Mockito.verify(documentMapper).paginateAs(
|
||||
Mockito.eq(1),
|
||||
Mockito.eq(10),
|
||||
queryCaptor.capture(),
|
||||
Mockito.eq(Document.class)
|
||||
);
|
||||
String sql = normalizeSql(queryCaptor.getValue().toSQL());
|
||||
Assert.assertFalse(sql.contains("tb_document_chunk"));
|
||||
Assert.assertFalse(sql.contains("group by"));
|
||||
Assert.assertTrue(sql.contains("*"));
|
||||
Assert.assertTrue(containsSqlIdentifier(sql, "chunk_count"));
|
||||
}
|
||||
|
||||
/**
|
||||
* 验证删除链路在外部索引和分块清理后删除文档主记录。
|
||||
*
|
||||
@@ -252,4 +319,27 @@ public class DocumentServiceImplTest {
|
||||
field.setAccessible(true);
|
||||
field.set(target, value);
|
||||
}
|
||||
|
||||
/**
|
||||
* 统一 SQL 文本格式,便于断言查询结构。
|
||||
*
|
||||
* @param sql 原始 SQL
|
||||
* @return 去除标识符引号并转为小写的 SQL
|
||||
*/
|
||||
private static String normalizeSql(String sql) {
|
||||
return sql.replace("`", "").toLowerCase(Locale.ROOT);
|
||||
}
|
||||
|
||||
/**
|
||||
* 判断 SQL 是否包含完整列标识符,避免与同前缀列名混淆。
|
||||
*
|
||||
* @param sql 已标准化的 SQL
|
||||
* @param identifier 列标识符
|
||||
* @return 包含完整标识符时返回 true
|
||||
*/
|
||||
private static boolean containsSqlIdentifier(String sql, String identifier) {
|
||||
return Pattern.compile("\\b" + Pattern.quote(identifier) + "\\b")
|
||||
.matcher(sql)
|
||||
.find();
|
||||
}
|
||||
}
|
||||
|
||||
@@ -0,0 +1,4 @@
|
||||
ALTER TABLE `tb_document_chunk`
|
||||
ADD INDEX `idx_document_chunk_document_id` (`document_id`) USING BTREE,
|
||||
ALGORITHM=INPLACE,
|
||||
LOCK=NONE;
|
||||
Reference in New Issue
Block a user