[improvement] remove number condition (#375)

This commit is contained in:
mainmain
2023-11-13 19:52:18 +08:00
committed by GitHub
parent 5feac0c14e
commit 47be92d5f6
4 changed files with 12 additions and 6 deletions

View File

@@ -5,6 +5,7 @@ import com.tencent.supersonic.chat.api.pojo.SemanticSchema;
import com.tencent.supersonic.chat.api.pojo.request.QueryReq; import com.tencent.supersonic.chat.api.pojo.request.QueryReq;
import com.tencent.supersonic.common.util.ContextUtils; import com.tencent.supersonic.common.util.ContextUtils;
import com.tencent.supersonic.common.util.jsqlparser.SqlParserAddHelper; import com.tencent.supersonic.common.util.jsqlparser.SqlParserAddHelper;
import com.tencent.supersonic.common.util.jsqlparser.SqlParserRemoveHelper;
import com.tencent.supersonic.common.util.jsqlparser.SqlParserSelectFunctionHelper; import com.tencent.supersonic.common.util.jsqlparser.SqlParserSelectFunctionHelper;
import com.tencent.supersonic.common.util.jsqlparser.SqlParserSelectHelper; import com.tencent.supersonic.common.util.jsqlparser.SqlParserSelectHelper;
import com.tencent.supersonic.knowledge.service.SchemaService; import com.tencent.supersonic.knowledge.service.SchemaService;
@@ -26,6 +27,11 @@ public class HavingCorrector extends BaseSemanticCorrector {
//add having expression filed to select //add having expression filed to select
addHavingToSelect(semanticParseInfo); addHavingToSelect(semanticParseInfo);
//remove number condition
String correctorSql = semanticParseInfo.getSqlInfo().getCorrectS2SQL();
correctorSql = SqlParserRemoveHelper.removeNumberCondition(correctorSql);
semanticParseInfo.getSqlInfo().setCorrectS2SQL(correctorSql);
} }
private void addHaving(SemanticParseInfo semanticParseInfo) { private void addHaving(SemanticParseInfo semanticParseInfo) {

View File

@@ -317,8 +317,8 @@ public class QueryServiceImpl implements QueryService {
correctorSql = SqlParserAddHelper.addWhere(correctorSql, addWhereConditions); correctorSql = SqlParserAddHelper.addWhere(correctorSql, addWhereConditions);
correctorSql = SqlParserAddHelper.addHaving(correctorSql, addHavingConditions); correctorSql = SqlParserAddHelper.addHaving(correctorSql, addHavingConditions);
log.info("correctorSql after replacing:{}", correctorSql); log.info("correctorSql after replacing:{}", correctorSql);
correctorSql = SqlParserRemoveHelper.removeNumberCondition(correctorSql);
parseInfo.getSqlInfo().setCorrectS2SQL(correctorSql); parseInfo.getSqlInfo().setCorrectS2SQL(correctorSql);
semanticQuery.setParseInfo(parseInfo); semanticQuery.setParseInfo(parseInfo);
String explainSql = semanticQuery.explain(user); String explainSql = semanticQuery.explain(user);

View File

@@ -54,7 +54,7 @@ public class SqlParserRemoveHelper {
} }
removeWhereExpression(whereExpression, removeFieldNames); removeWhereExpression(whereExpression, removeFieldNames);
} }
public static String removeWhereCondition(String sql) { public static String removeNumberCondition(String sql) {
Select selectStatement = SqlParserSelectHelper.getSelect(sql); Select selectStatement = SqlParserSelectHelper.getSelect(sql);
SelectBody selectBody = selectStatement.getSelectBody(); SelectBody selectBody = selectStatement.getSelectBody();

View File

@@ -14,7 +14,7 @@ class SqlParserRemoveHelperTest {
void removeWhereHavingCondition() { void removeWhereHavingCondition() {
String sql = "select 歌曲名 from 歌曲库 where sum(粉丝数) > 20000 and 2>1 and " String sql = "select 歌曲名 from 歌曲库 where sum(粉丝数) > 20000 and 2>1 and "
+ "sum(播放量) > 20000 and 1=1 HAVING sum(播放量) > 20000 and 3>1"; + "sum(播放量) > 20000 and 1=1 HAVING sum(播放量) > 20000 and 3>1";
sql = SqlParserRemoveHelper.removeWhereCondition(sql); sql = SqlParserRemoveHelper.removeNumberCondition(sql);
System.out.println(sql); System.out.println(sql);
Assert.assertEquals( Assert.assertEquals(
"SELECT 歌曲名 FROM 歌曲库 WHERE sum(粉丝数) > 20000 AND sum(播放量) > 20000 HAVING sum(播放量) > 20000", "SELECT 歌曲名 FROM 歌曲库 WHERE sum(粉丝数) > 20000 AND sum(播放量) > 20000 HAVING sum(播放量) > 20000",
@@ -22,7 +22,7 @@ class SqlParserRemoveHelperTest {
sql = "SELECT 歌曲,sum(播放量) FROM 歌曲库\n" sql = "SELECT 歌曲,sum(播放量) FROM 歌曲库\n"
+ "WHERE (歌手名 = '张三' AND 2 > 1) AND 数据日期 = '2023-11-07'\n" + "WHERE (歌手名 = '张三' AND 2 > 1) AND 数据日期 = '2023-11-07'\n"
+ "GROUP BY 歌曲名 HAVING sum(播放量) > 100000"; + "GROUP BY 歌曲名 HAVING sum(播放量) > 100000";
sql = SqlParserRemoveHelper.removeWhereCondition(sql); sql = SqlParserRemoveHelper.removeNumberCondition(sql);
System.out.println(sql); System.out.println(sql);
Assert.assertEquals( Assert.assertEquals(
"SELECT 歌曲, sum(播放量) FROM 歌曲库 WHERE (歌手名 = '张三') " "SELECT 歌曲, sum(播放量) FROM 歌曲库 WHERE (歌手名 = '张三') "
@@ -31,7 +31,7 @@ class SqlParserRemoveHelperTest {
sql = "SELECT 歌曲名,sum(播放量) FROM 歌曲库 WHERE (1 = 1 AND 1 = 1 AND 2 > 1 )" sql = "SELECT 歌曲名,sum(播放量) FROM 歌曲库 WHERE (1 = 1 AND 1 = 1 AND 2 > 1 )"
+ "AND 1 = 1 AND 歌曲类型 IN ('类型一', '类型二') AND 歌手名 IN ('林俊杰', '周杰伦')" + "AND 1 = 1 AND 歌曲类型 IN ('类型一', '类型二') AND 歌手名 IN ('林俊杰', '周杰伦')"
+ "AND 数据日期 = '2023-11-07' GROUP BY 歌曲名 HAVING 2 > 1 AND SUM(播放量) >= 1000"; + "AND 数据日期 = '2023-11-07' GROUP BY 歌曲名 HAVING 2 > 1 AND SUM(播放量) >= 1000";
sql = SqlParserRemoveHelper.removeWhereCondition(sql); sql = SqlParserRemoveHelper.removeNumberCondition(sql);
System.out.println(sql); System.out.println(sql);
Assert.assertEquals( Assert.assertEquals(
"SELECT 歌曲名, sum(播放量) FROM 歌曲库 WHERE 歌曲类型 IN ('类型一', '类型二') " "SELECT 歌曲名, sum(播放量) FROM 歌曲库 WHERE 歌曲类型 IN ('类型一', '类型二') "
@@ -41,7 +41,7 @@ class SqlParserRemoveHelperTest {
sql = "SELECT 品牌名称,法人 FROM 互联网企业 WHERE (2 > 1 AND 1 = 1) AND 数据日期 = '2023-10-31'" sql = "SELECT 品牌名称,法人 FROM 互联网企业 WHERE (2 > 1 AND 1 = 1) AND 数据日期 = '2023-10-31'"
+ "GROUP BY 品牌名称, 法人 HAVING 2 > 1 AND sum(注册资本) > 100000000 AND sum(营收占比) = 0.5 and 1 = 1"; + "GROUP BY 品牌名称, 法人 HAVING 2 > 1 AND sum(注册资本) > 100000000 AND sum(营收占比) = 0.5 and 1 = 1";
sql = SqlParserRemoveHelper.removeWhereCondition(sql); sql = SqlParserRemoveHelper.removeNumberCondition(sql);
System.out.println(sql); System.out.println(sql);
Assert.assertEquals( Assert.assertEquals(
"SELECT 品牌名称, 法人 FROM 互联网企业 WHERE 数据日期 = '2023-10-31' GROUP BY " "SELECT 品牌名称, 法人 FROM 互联网企业 WHERE 数据日期 = '2023-10-31' GROUP BY "