Skip to content

Latest commit

 

History

History
716 lines (603 loc) · 34.9 KB

Changelog.md

File metadata and controls

716 lines (603 loc) · 34.9 KB

Changelog

6.1.0 - 2023-12-16

  • Released version 6.1.0, PageHelper provides direct dependency on jsqlparser as intermediate interfaces, allowing default implementation replacement through SPI.
  • Upgraded jsqlparser version to 4.7, re-implemented order by, pagination, and count queries.
  • Simplified pom.xml configuration, removed shade-embedded jsqlparser approach, and switched to selecting different jsqlparser versions through external dependencies, allowing self-SPI extension.
  • jsqlparser parsing no longer uses a thread pool, supporting SPI extension to override SqlParser implementation.
  • Changed SqlServer pagination to SqlServerSqlParser interface, added parameter sqlServerSqlParser to override the default value.
  • Extracted OrderByParser to OrderBySqlParser interface, added orderBySqlParser parameter to override the default implementation.
  • Changed static methods of OrderByParser to regular methods, preparing for future interface changes.
  • JSqlParser interface is no longer needed after JDK 8+, removed the interface, and marked the parameter in the documentation (This parameter was used in the early stages to support special configuration for SQL Server). Compatible with jsqlparser 4.7 version.
  • Fixed maven-compiler-plugin version to remove warnings and improve build stability. qxo
  • Added .vscode to .gitignore for vscode IDE. qxo
  • Fixed bug #779. chenyuehui

To ensure compatibility with jsqlparser 4.5, 4.7, and possible future versions, a new project called pagehelper-sqlparser has been created. Currently, it provides two implementations: 4.5 and 4.7. To use it, exclude jsqlparser from pagehelper and select one jsqlparser implementation. The current version defaults to using the code from version 4.7. If you want to switch to the 4.5 implementation, follow the configuration steps below:

<dependency>
    <groupId>com.github.pagehelper</groupId>
    <artifactId>pagehelper</artifactId>
    <version>6.1.0</version>
    <exclusions>
        <exclusion>
            <groupId>com.github.jsqlparser</groupId>
            <artifactId>jsqlparser</artifactId>
        </exclusion>
    </exclusions>
</dependency>
<dependency>
    <groupId>com.github.pagehelper</groupId>
    <artifactId>sqlparser4.5</artifactId>
    <version>6.1.0</version>
</dependency>

The priority of replacing default values with SPI is lower than the implementations specified by the sqlServerSqlParser, orderBySqlParser, and countSqlParser parameters. If no specific implementation is specified, the SPI implementation will take effect if available. You can refer to the code in the pagehelper-sqlsource module for SPI implementation examples.

By default, JSqlParser uses a temporarily created Executors.newSingleThreadExecutor() for parsing SQL. Here, the thread pool is bypassed through the API:

CCJSqlParser parser = CCJSqlParserUtil.newParser(statementReader);
parser.withSquareBracketQuotation(true);
return parser.Statement();

The purpose of using a thread pool in JSqlParser is to prevent parsing timeouts. Therefore, if you have encountered timeout situations, you can introduce the following dependency (which overrides the default implementation through SPI with a timeout of 10 seconds):

<dependency>
    <groupId>com.github.pagehelper</groupId>
    <artifactId>sqlparser-timeout</artifactId>
    <version>6.1.0</version>
</dependency>

6.0.0 - 2023-11-05

  • Based on JDK 8 adaptation, JDK 6 and 7 are not supported from 6.0 onwards, and 5.x versions can be used if necessary
  • Added asynchronous count support, configure asyncCount globally, default false,single activation by PageHelper.startPage(1, 10).enableAsyncCount(); Asynchronous queries are performed using independent connections (transactions), and it is not suitable to enable asynchronous queries when the query is affected by addition, deletion, and modification operations. closed #334
  • JSqlParser opens parser.withSquareBracketQuotation(true) by default and supports SqlServer [](6.1.0 remove this parameter)
  • feat: A new method for data object conversion has been added to the PageInfo, method: <E> PageInfo<E> convert(Page.Function<T, E> function) by codeke
  • CountSqlParser is changed to an interface, allowing the countSqlParser parameter to be replaced with your own implementation, which is supported #772
  • dialectAlias supports simplified configurations, e.g. dm=oracle;oracle=oracle9i, a direct reference to the current abbreviation, without writing the full name of the class
  • countColumnadd injection detection, fixed #686
  • Add the PageParam class, which does not embed objects (will affect the use), if you want to use, you can inherit the object,closed #562
  • All exception messages have been changed to English
  • open setLocalPage method, support #771
  • Solve the problem of handling order by error when sqlserver with union sql parsing,fixed #768
  • Optimized the total logic to solve the problem that the query is not pagination and the order by is invalid. fixed #641
  • Modify the dialect instantiation logic to ensure that the class is used after the configuration is completed. fixed #742
  • dialectAliasMap change to LinkedHashMap type, support matching in configuration order, fixed #758
  • fixed the pagination bug of xingyun database by maimaitiyaer_bonc

5.3.3 - 2023-06-03

  • Ignoring unnecessarily generated surefire-report by java-codehunger
  • Supports parsing of the dialect corresponding to the openGauss database from the URL by saxisuer
  • Fixed the issue that SQL injection verification was incorrect #716 by uyong
  • Support parsing of the dialect corresponding to Kingbase 8 from the URL by univ
  • Add support for cirrodata #705 by sxh0570

5.3.2 - 2022-09-18

  • Use document update, all parameters are included, the default home page document changed to Chinese.
  • Add support for kingbase. by HanHuimin001
  • Add 'debug' parameter, default 'false', turn on 'debug' mode when 'true', call stack will be recorded after 'debug' mode starts. by huyingqian
  • add supports for count sql hint syntax. by zhanliquan
  • Add a PageProperties interface, which can be used by the instantiated extension class inside the framework to obtain the paging plug-in configuration.
  • To add the CountMsIdGen interface, you can configure a custom implementation class by CountMsIdGen, which is used to generate the msId for the query corresponding to the COUNT query. The default implementation is still countSuffix , with extensions like 'selectByExample' mapped to the corresponding 'selectCountByExample' method.
  • Added keepOrderBy and keepSubSelectOrderBy configurations.
  • Add the sqlParser configuration and add the JSqlParser interface to solve the problem that JSqlParser and JDK compatibility cause no additional configuration.(6.1.0 remove this parameter)
  • The test uses the Logback logging framework and removes log4j.
  • Resolve that 'dialectKey' is empty resulting in NPE,fixed #656

For a detailed description of the above parameters, see How to use PageHelper.

5.3.1 - 2022-06-14

  • Resolve CVE-2022-28111 vulnerability, limit the order by parameter, avoid SQL injection
  • Add support for as400. by bluezealot
  • Optimize generic parameters of Page class **by Zhang Fulai * *
  • Standardize the order of PostgreSQL paging parameters by outian

5.3.0 - 2021-10-07

  • Add AutoDialect interface to automatically obtain the database type, which can be configured as its own implementation class through autoDialectClass. By default, DataSourceNegotiationAutoDialect is used, which is obtained according to the connection pool first. In the default implementation, special processing is added for hikari,druid,tomcat-jdbc,c3p0,dbcp type database connection pools, and jdbcUrl are obtained directly from the configuration. When other types of data sources are used, the connection is still obtained in the old way. You can configure autoDialectClass=old when you want to use exactly the same way as the old version. When the database connection pool type is very clear, it is recommended to configure it as a specific value. For example, when using hikari, configure autoDialectClass=hikari, and when using other connection pools, configure it as its own implementation class.
  • Enable dynamic designation of dialect implementation at runtime, such as PageHelper.startPage(1, 10).using("oracle"); Or PageHelper.startPage(2, 10).using("org.exmaple.CustomDialect");
  • PageInfo adds the empty instance constant attribute PageInfo.EMPTY and the content judgment boolean hasContent() .
  • Adding banner to startup requires log level debug, which can be closed by -Dpagehelper.banner=false or environment variable PAGEHELPER_BANNER=false.
     DEBUG [main] -
    
     ,------.                           ,--.  ,--.         ,--.
     |  .--. '  ,--,--.  ,---.   ,---.  |  '--'  |  ,---.  |  |  ,---.   ,---.  ,--.--.
     |  '--' | ' ,-.  | | .-. | | .-. : |  .--.  | | .-. : |  | | .-. | | .-. : |  .--'
     |  | --'  \ '-'  | ' '-' ' \   --. |  |  |  | \   --. |  | | '-' ' \   --. |  |
     `--'       `--`--' .`-  /   `----' `--'  `--'  `----' `--' |  |-'   `----' `--'
     `---'                                   `--'                        is intercepting.
    
    The purpose of adding banner is that if you configure paging plug-ins multiple times, you will see banner output multiple times, and you can see where it has been instantiated at the breakpoint of the PageInterceptor constructor.
  • Improve the Count query. When having exists, the query column is not optimized. The query column is not optimized when there are functions or operations with aliases in the column, so as to avoid that aliases used in order by or having do not exist.
  • It is judged that processing some data (such as TDEngine) returns null when there is no result in querying count.
  • Adding Firebird database support is the same as SqlServer2012 paging syntax.
  • Add impala database automatic recognition.
  • Upgrade JSqlParser to version 4.2.

5.2.1 - 2021-06-20

  • Upgrade dependency jsqlparser 4.0, mybatis 3.5.7
  • Automatically recognize the following databases:
    • 虚谷数据库 xugu #599
    • 神通数据库 oscar by ranqing
    • 瀚高数据库 highgo by ashaiqing
  • BoundSqlInterceptorChain interceptor index parameter bug, fixed #587
  • fixed #558
  • Add PostgreSQL dialect by liym@home
  • fixed #604, Solve the problem of total loss
  • Add code comments, fixed #547

5.2.0 - 2020-07-26

  • Upgrading jsqlparser to version 3.2 makes sql parsing better and supports sqlserver better.

  • Modify the substitution regularity in sqlserver mode, and now allow spaces in with( nolock) brackets.

  • Solving the bugs in reasonable, pageSizeZero and offset usage, the meaning and result are more consistent now.

  • In the process of splicing paging SQL, a new line character is added to avoid invalid paging part caused by comments in the original SQL.

  • ROW_ID alias in Oracle and Db2 is changed to PAGEHELPER_ROW_ID to avoid conflict with common names.

  • Solve the special problem when using other interceptors with a single parameter ProviderSql (support mybatis 3.4.0+) by Luo Zhenyu

  • Automatic identification of clickhouse is supported, and paging is performed by MySQL.

  • Change startRow, endRow type from int to long.

  • Page adds a public <T> PageInfo<T> toPageInfo(Function<E, T> function) method to convert the data in the query results.

  • Refer to Oracle9iDialect provided by pr#476, which is also a paging method used before. You can test and select the appropriate paging method by yourself.

    At present, there are two kinds of Oracle pagination as follows:

    -- OracleDialect outer control range
    WHERE ROW_ID <= ? AND ROW_ID > ?
    -- Oracle9iDialect's internal and external control scope respectively
    TMP_PAGE WHERE ROWNUM <= ? ) WHERE ROW_ID > ?
  • Adding BoundSqlInterceptor of PageHelper plug-in can process or simply read SQL in three stages, adding boundSqlInterceptors, and configuring multiple implementation class names that implement BoundSqlInterceptor interface, separated by English commas. PageHelper can also be set for this paging through a PageHelper.startPage(x,x).boundSqlInterceptor(BoundSqlInterceptor boundSqlInterceptor).

The biggest change of this update is the addition of BoundSqlInterceptor, which can intercept the SQL(BoundSQL object) of paging processing at runtime:

/**
 * BoundSql 处理器
 */
public interface BoundSqlInterceptor {
    /**
     * boundsql 处理
     *
     * @param type     类型
     * @param boundSql 当前类型的 boundSql
     * @param cacheKey 缓存 key
     * @param chain    处理器链,通过 chain.doBoundSql 方法继续执行后续方法,也可以直接返回 boundSql 终止后续方法的执行
     * @return 允许修改 boundSql 并返回修改后的
     */
    BoundSql boundSql(Type type, BoundSql boundSql, CacheKey cacheKey, Chain chain);

    enum Type {
        /**
         * 原始SQL,分页插件执行前,先执行这个类型
         */
        ORIGINAL,
        /**
         * count SQL,第二个执行这里
         */
        COUNT_SQL,
        /**
         * 分页 SQL,最后执行这里
         */
        PAGE_SQL
    }

    /**
     * 处理器链,可以控制是否继续执行
     */
    interface Chain {
        Chain DO_NOTHING = new Chain() {
            @Override
            public BoundSql doBoundSql(Type type, BoundSql boundSql, CacheKey cacheKey) {
                return boundSql;
            }
        };

        BoundSql doBoundSql(Type type, BoundSql boundSql, CacheKey cacheKey);
    }
}

The interface includes boundSql interface method, Type enumeration, and the definition of Chain interface, and you don't need to consider Chain when you implement it yourself.

The interceptor is configured by boundSqlInterceptors parameter, and there are three situations when executing:

  1. Regardless of whether the currently executed SQL will be paged or not, interceptor methods of Type.ORIGINAL will be executed.

  2. When the paging method is called, the interceptor will continue to execute the interceptor method of Type.COUNT_SQL, which will only be executed when paging is executed and count query is specified.

  3. When paging method is called, if count > 0, interceptor method of Type.PAGE_SQL will be executed, which will only be executed when paging is executed.

With the specified parameter PageHelper.startPage(1, Integer.MAX_VALUE, false).boundSqlInterceptor(BoundSqlInterceptor boundSqlInterceptor), it can also play the role of not paging and count query, but can execute interceptor method of Type.ORIGINAL.

If you want to get the page before SQL execution, you only need to pay attention to Type.ORIGINAL, and the other two are before count execution and before page execution (when count=0, the page method will not be executed and will not be executed here).

Take the test code as an example:

public class TestBoundSqlInterceptor implements BoundSqlInterceptor {
    public static final String COMMENT = "\n /* TestBoundSqlInterceptor */\n";

    @Override
    public BoundSql boundSql(Type type, BoundSql boundSql, CacheKey cacheKey, Chain chain) {
        if (type == Type.ORIGINAL) {
            String sql = boundSql.getSql();
            MetaObject metaObject = MetaObjectUtil.forObject(boundSql);
            metaObject.setValue("sql", sql + COMMENT);
        }
        return chain.doBoundSql(type, boundSql, cacheKey);
    }

}

The above code modifies the original sql before SQL execution, but only adds a comment at the end, which does not affect SQL execution. It is configured in the following way:

<plugin interceptor="com.github.pagehelper.PageInterceptor">
    <!-- 支持通过Mapper接口参数来传递分页参数 -->
    <property name="helperDialect" value="mysql"/>
    <property name="boundSqlInterceptors"
              value="com.github.pagehelper.test.basic.provider.TestBoundSqlInterceptor,com.github.pagehelper.test.basic.provider.TestBoundSqlInterceptor"/>
</plugin>

Here, in order to explain that the parameter value can be multiple, it is repeatedly configured once, that is, the above interceptor will execute it twice.

With this configuration, the above SQL will modify the SQL when the page is executed.

In addition to this configuration mode, temporary designation when PageHelper.startPage is also supported. This mode will put the interceptor at the chain head and execute it first, so you can control whether to execute it later or not, or you can do the final processing before returning after all subsequent executions.

Example:

PageHelper.startPage(1, 10).boundSqlInterceptor(new BoundSqlInterceptor() {
    @Override
    public BoundSql boundSql(Type type, BoundSql boundSql, CacheKey cacheKey, Chain chain) {
        System.out.println("before: " + boundSql.getSql());
        BoundSql doBoundSql = chain.doBoundSql(type, boundSql, cacheKey);
        System.out.println("after: " + doBoundSql.getSql());
        if (type == Type.ORIGINAL) {
            Assert.assertTrue(doBoundSql.getSql().contains(TestBoundSqlInterceptor.COMMENT));
        }
        return doBoundSql;
    }
});

5.1.11 - 2019-11-26

  • Added support for Shentong database wangss
  • Add support for HerdDB - support HerdDB, mostly like MySQL - auto detect HerdDB Enrico Olivelli
  • fix some typos and grammar issues LHearen

5.1.10 - 2019-06-05

In version 5.1.0 - 2017-08-28. Added ReplaceSql interface for handling sqlServer with (nolock) problem, add the replaceSql parameters, the optional value is simple and regex, or to achieve the ReplaceSql interface fully qualified class name. The default value is simple, still using the original way to deal with, the new regex will be convert with (nolock) to table_PAGEWITHNOLOCK.

This update only changes the default value from simple to regex, which can almost 100% solve the paging problem of sqlServer.

The following are examples from two issue.

issue #76

Original SQL:

SELECT *
FROM
forum_post_info a with(nolock)
LEFT JOIN forum_carcase_tags as b with(nolock) on a.id = b.carcase_id where b.tag_id = 127

Converted Count SQL:

SELECT COUNT(0)
FROM forum_post_info a WITH (NOLOCK)
	LEFT JOIN forum_carcase_tags b WITH (NOLOCK) ON a.id = b.carcase_id
WHERE b.tag_id = 127

Converted paging SQL:

SELECT TOP 10 *
FROM (
	SELECT ROW_NUMBER() OVER (ORDER BY RAND()) AS PAGE_ROW_NUMBER, *
	FROM (
		SELECT *
		FROM forum_post_info a WITH (NOLOCK)
			LEFT JOIN forum_carcase_tags b WITH (NOLOCK) ON a.id = b.carcase_id
		WHERE b.tag_id = 127
	) PAGE_TABLE_ALIAS
) PAGE_TABLE_ALIAS
WHERE PAGE_ROW_NUMBER > 1
ORDER BY PAGE_ROW_NUMBER

issue #398

Original SQL:

Select AUS.ScheduleID, AUS.SystemID, AUS.ClinicID, AUS.DoctorID, AUS.ScheduleDate,
	AUS.StartTime, AUS.EndTime, AUS.Status, AUS.BookBy, AUS.Note, AUS.Remark, AUS.SourceType, CM.CompanyName,
	AU.UserName As DoctorName, AU.UserNumber As DoctorNumber, CC.CodeDesc As ClinicName, CD.Lat, CD.Lng,
	CD.ContactTel, CD.Address, CR.ConsultationStatusID, CR.RegisterStatus,A1.CodeDesc as AreaLevel1, A2.CodeDesc as AreaLevel2
	From ACM_User_Schedule AUS with(nolock)
	Left Join Client_Register CR with(nolock) On AUS.BookBy=CR.ClientID And CR.SourceType='F' And AUS.ClientRegisterNum=CR.ClientRegisterNum
	Inner Join ACM_User AU with(nolock) On AU.UserID = AUS.DoctorID
	Inner Join Code_Clinic CC with(nolock) On AUS.ClinicID=CC.CodeID
	Inner Join Clinic_Detail CD with(nolock) On CC.CodeID = CD.ClinicID
	Inner Join Code_Area A1 with(nolock) On CD.AreaLevel1ID=A1.CodeID
	Inner Join Code_Area A2 with(nolock) On CD.AreaLevel2ID=A2.CodeID
	Inner Join Company_Master CM with(nolock) On CC.SystemID = CM.SystemID
	Where BookBy=1

Converted Count SQL:

SELECT COUNT(0)
FROM ACM_User_Schedule AUS WITH (NOLOCK)
	LEFT JOIN Client_Register CR WITH (NOLOCK)
	ON AUS.BookBy = CR.ClientID
		AND CR.SourceType = 'F'
		AND AUS.ClientRegisterNum = CR.ClientRegisterNum
	INNER JOIN ACM_User AU WITH (NOLOCK) ON AU.UserID = AUS.DoctorID
	INNER JOIN Code_Clinic CC WITH (NOLOCK) ON AUS.ClinicID = CC.CodeID
	INNER JOIN Clinic_Detail CD WITH (NOLOCK) ON CC.CodeID = CD.ClinicID
	INNER JOIN Code_Area A1 WITH (NOLOCK) ON CD.AreaLevel1ID = A1.CodeID
	INNER JOIN Code_Area A2 WITH (NOLOCK) ON CD.AreaLevel2ID = A2.CodeID
	INNER JOIN Company_Master CM WITH (NOLOCK) ON CC.SystemID = CM.SystemID
WHERE BookBy = 1

Converted paging SQL:

SELECT TOP 10 ScheduleID, SystemID, ClinicID, DoctorID, ScheduleDate
	, StartTime, EndTime, Status, BookBy, Note
	, Remark, SourceType, CompanyName, DoctorName, DoctorNumber
	, ClinicName, Lat, Lng, ContactTel, Address
	, ConsultationStatusID, RegisterStatus, AreaLevel1, AreaLevel2
FROM (
	SELECT ROW_NUMBER() OVER (ORDER BY RAND()) AS PAGE_ROW_NUMBER, ScheduleID, SystemID, ClinicID, DoctorID
		, ScheduleDate, StartTime, EndTime, Status, BookBy
		, Note, Remark, SourceType, CompanyName, DoctorName
		, DoctorNumber, ClinicName, Lat, Lng, ContactTel
		, Address, ConsultationStatusID, RegisterStatus, AreaLevel1, AreaLevel2
	FROM (
		SELECT AUS.ScheduleID, AUS.SystemID, AUS.ClinicID, AUS.DoctorID, AUS.ScheduleDate
			, AUS.StartTime, AUS.EndTime, AUS.Status, AUS.BookBy, AUS.Note
			, AUS.Remark, AUS.SourceType, CM.CompanyName, AU.UserName AS DoctorName, AU.UserNumber AS DoctorNumber
			, CC.CodeDesc AS ClinicName, CD.Lat, CD.Lng, CD.ContactTel, CD.Address
			, CR.ConsultationStatusID, CR.RegisterStatus, A1.CodeDesc AS AreaLevel1, A2.CodeDesc AS AreaLevel2
		FROM ACM_User_Schedule AUS WITH (NOLOCK)
			LEFT JOIN Client_Register CR WITH (NOLOCK)
			ON AUS.BookBy = CR.ClientID
				AND CR.SourceType = 'F'
				AND AUS.ClientRegisterNum = CR.ClientRegisterNum
			INNER JOIN ACM_User AU WITH (NOLOCK) ON AU.UserID = AUS.DoctorID
			INNER JOIN Code_Clinic CC WITH (NOLOCK) ON AUS.ClinicID = CC.CodeID
			INNER JOIN Clinic_Detail CD WITH (NOLOCK) ON CC.CodeID = CD.ClinicID
			INNER JOIN Code_Area A1 WITH (NOLOCK) ON CD.AreaLevel1ID = A1.CodeID
			INNER JOIN Code_Area A2 WITH (NOLOCK) ON CD.AreaLevel2ID = A2.CodeID
			INNER JOIN Company_Master CM WITH (NOLOCK) ON CC.SystemID = CM.SystemID
		WHERE BookBy = 1
	) PAGE_TABLE_ALIAS
) PAGE_TABLE_ALIAS
WHERE PAGE_ROW_NUMBER > 1
ORDER BY PAGE_ROW_NUMBER

SQL is formatted by https://tool.oschina.net/codeformat/sql

5.1.9 - 2019-05-29

  • Upgrade jsqlparser to 2.0, upgrade mybatis to 3.5.1. resolve compatibility issues.
  • Improve paging logic judgment. fixed #389
  • Solve MetaObject version compatibility issues. fixed #349
  • Processing order by output warning log when parsing fails, not throwing exception
  • Solve three problems that may cause countColumn to fail fixed #325
  • Add a comma with less BIT_ fixed #341
  • Handling invalid links in documents isea533
  • Document sample error. fixed #366
  • fixed #373 NPE problem

5.1.8 - 2018-11-11

  • Resolve the problem of with(nolock) in SQLServer (#pr10) by lvshuyan

5.1.7 - 2018-10-11

  • Support Aliyun PPAS database. Automatic identification of edb. fixed #281

5.1.6 - 2018-09-05

  • Add the parameter useSqlserver2012, set to true, and use sqlserver2012(Dialect) as the default paging method for SQL Server databases, which is useful in dynamic data sources.
  • Add an IPage interface. Currently, there is only one parameter to support the MyBatis query method, and when the parameter implements the IPage interface, paging query will be automatically performed if paging parameters exist. Thanks to moonfruit Issue two years ago.
  • fixed # 276 to resolve hashset concurrency issue
  • Optimize code structure and streamline interceptor code

5.1.5 - 2018-09-02

  • Optimize the code and remove unnecessary checks(by lenosp)
  • Solve the small problem of pageKey multi-processing once #268
  • Added javadoc documentation on gitee(https://apidoc.gitee.com/free/Mybatis_PageHelper)
  • Solve the problem of default reflection without cache fixed #275
  • Optimizing mysql ifnull function causes paging performance problems (by miaogr)(This change was eventually changed to the following aggregateFunctions
  • Jsqlparser has been upgraded to version 1.2, which is incompatible with 1.0 and has been resolved. fixed 273
  • Remove the g(s)etFirstPage and g(s)etLastPage methods that are ambiguous in PageInfo
  • Throws an exception that failed to parse when sorting fixed #257
  • Resolve the initialization problem when there is no properties property when configuring the spring use <bean>. fixed #26
  • Fix the problem that Oracle paging will leak data (by muyun12)
  • aggregateFunctions: The default is the aggregate function of all common databases, allowing you to manually add aggregate functions ( affecting the number of rows ). All functions that start with aggregate functions will be wrap as subquery. Other functions and columns will be replaced with count(0).

After adding the aggregateFunctions parameter, the biggest difference from the original is that if there is select ifnull(XXX,YY) from table ..., the original count query is select count(0) from (select ifnull(xxx,yy) from table ... ) temp_count now distinguishes aggregate functions, if not aggregate functions, it will become select count(0) from table ....

The aggregate function prefixes included by default are as follows:

/**
 * Aggregate functions, beginning with the following functions are considered aggregate functions
 */
private static final Set<String> AGGREGATE_FUNCTIONS = new HashSet<String>(Arrays.asList(
        ("APPROX_COUNT_DISTINCT," +
        "ARRAY_AGG," +
        "AVG," +
        "BIT_" +
        //"BIT_AND," +
        //"BIT_OR," +
        //"BIT_XOR," +
        "BOOL_," +
        //"BOOL_AND," +
        //"BOOL_OR," +
        "CHECKSUM_AGG," +
        "COLLECT," +
        "CORR," +
        //"CORR_," +
        //"CORRELATION," +
        "COUNT," +
        //"COUNT_BIG," +
        "COVAR," +
        //"COVAR_POP," +
        //"COVAR_SAMP," +
        //"COVARIANCE," +
        //"COVARIANCE_SAMP," +
        "CUME_DIST," +
        "DENSE_RANK," +
        "EVERY," +
        "FIRST," +
        "GROUP," +
        //"GROUP_CONCAT," +
        //"GROUP_ID," +
        //"GROUPING," +
        //"GROUPING," +
        //"GROUPING_ID," +
        "JSON_," +
        //"JSON_AGG," +
        //"JSON_ARRAYAGG," +
        //"JSON_OBJECT_AGG," +
        //"JSON_OBJECTAGG," +
        //"JSONB_AGG," +
        //"JSONB_OBJECT_AGG," +
        "LAST," +
        "LISTAGG," +
        "MAX," +
        "MEDIAN," +
        "MIN," +
        "PERCENT_," +
        //"PERCENT_RANK," +
        //"PERCENTILE_CONT," +
        //"PERCENTILE_DISC," +
        "RANK," +
        "REGR_," +
        "SELECTIVITY," +
        "STATS_," +
        //"STATS_BINOMIAL_TEST," +
        //"STATS_CROSSTAB," +
        //"STATS_F_TEST," +
        //"STATS_KS_TEST," +
        //"STATS_MODE," +
        //"STATS_MW_TEST," +
        //"STATS_ONE_WAY_ANOVA," +
        //"STATS_T_TEST_*," +
        //"STATS_WSR_TEST," +
        "STD," +
        //"STDDEV," +
        //"STDDEV_POP," +
        //"STDDEV_SAMP," +
        //"STDDEV_SAMP," +
        //"STDEV," +
        //"STDEVP," +
        "STRING_AGG," +
        "SUM," +
        "SYS_OP_ZONE_ID," +
        "SYS_XMLAGG," +
        "VAR," +
        //"VAR_POP," +
        //"VAR_SAMP," +
        //"VARIANCE," +
        //"VARIANCE_SAMP," +
        //"VARP," +
        "XMLAGG").split(",")));

5.1.4 - 2018-04-22

  • Add the DaMeng Database (dm) to page using Oracle. If you want to change SqlServer, you can refer to the dialectAlias parameter in the 5.1.3 update log.

5.1.3 - 2018-04-07

  • Page toString method adds super.toString(). The final output form is Page{Attribute}[Collection].
  • New defaultCount parameter is used to control whether to perform count query in the default method without count query. By default, true will execute count query. This is a globally valid parameter, and it is a uniform behavior when multiple data sources are used.
  • New dialogAlias parameter that allows you to configure aliases for custom implementations. it can be used to automatically obtain corresponding implementations based on JDBC URL. it allows you to overwrite existing implementations in this way. configuration examples are ( Separate multiple configurations with semicolons ):
    <property name="dialectAlias" value="oracle=com.github.pagehelper.dialect.helper.OracleDialect"/>
  • The new PageSerializable class, a simplified version of the PageInfo class, is recommended to use or refer to this class when it does not require much information.

5.1.2 - 2017-09-18

  • Solve the problem when using the PageHelper.orderBy method alone #110;

5.1.1 - 2017-08-30

  • The update to solve the problem and only SqlServer 2005,2008 related.
  • Resolve RegexWithNolockReplaceSql in the Wrong regular w?, it should be w+.
  • Resolved SqlServerDialect did not initialize the default SimpleWithNolockReplaceSql error.
  • SqlServerRowBoundsDialect support for the replaceSql parameter.

5.1.0 - 2017-08-28

  • Added the sorting functionality included in the previous version of 4.x, and the usage is consistent (PageHelper adds several sort-related methods).
  • Paging SQL is converted to PreparedStatement SQL.
  • Added ReplaceSql interface for handling sqlServer with (nolock) problem, add the replaceSql parameters, the optional value is simple and regex, or to achieve the ReplaceSql interface fully qualified class name. The default value is simple, still using the original way to deal with, the new regex will be convert with (nolock) to table_PAGEWITHNOLOCK.
  • PageRowBounds add count attribute, you can control whether execute the count query.

5.0.4 - 2017-08-01

  • Add a simple configuration support for the Phoenix database, You can configure helperDialect=phoenix. Can also automatically identify the Phoenix database jdbc url.
  • Simplified cache of msCountMap
  • Add countSuffix count query suffix configuration parameters, this parameter is configured for PageInterceptor, the default value is _COUNT.
  • Add custom count query support, see below for details.

Add custom count query support

Add countSuffix count query suffix configuration parameters, this parameter is configured for PageInterceptor, the default value is _COUNT.

The paging plugin will preferentially find the handwritten paging query by the current query msId + countSuffix.

If there is no custom query, the query is still automatically created using the previous way.

For example, if there are two queries:

<select id="selectLeftjoin" resultType="com.github.pagehelper.model.User">
    select a.id,b.name,a.py from user a
    left join user b on a.id = b.id
    order by a.id
</select>
<select id="selectLeftjoin_COUNT" resultType="Long">
    select count(distinct a.id) from user a
    left join user b on a.id = b.id
</select>

The above countSuffix uses the default value of _COUNT, and the paging plugin will automatically get the query to selectLeftjoin_COUNT. This query needs to ensure that the result is correct.

The value of the return value must be resultType =" Long ", and the same parameter used by selectLeftjoin 'is used, so it is used in SQL to follow the selection of selectLeftjoin'.

Because the selectLeftjoin_COUNT method is invoked automatically, there is no need to provide the appropriate method on the interface, or if it is required to be invoked separately.

The above method to perform the portion of the output log is as follows:

DEBUG [main] - ==>  Preparing: select count(distinct a.id) from user a left join user b on a.id = b.id
DEBUG [main] - ==> Parameters:
TRACE [main] - <==    Columns: C1
TRACE [main] - <==        Row: 183
DEBUG [main] - <==      Total: 1
DEBUG [main] - Cache Hit Ratio [com.github.pagehelper.mapper.CountryMapper]: 0.0
DEBUG [main] - ==>  Preparing: select a.id,b.name,a.py from user a left join user b on a.id = b.id order by a.id LIMIT 10
DEBUG [main] - ==> Parameters:
TRACE [main] - <==    Columns: ID, COUNTRYNAME, COUNTRYCODE
TRACE [main] - <==        Row: 1, Angola, AO
TRACE [main] - <==        Row: 2, Afghanistan, AF
TRACE [main] - <==        Row: 3, Albania, AL

5.0.3 -2017-06-20

  • Solve the supportMethodsArguments parameter problem. It is recommended to upgrade to the latest version.

5.0.2 - 2017-05-30

  • Page<E> implements Closeable interface, in JDK7+ which can use the in try () call, it will automatically call PageHelper.clearPage ();#58
  • fixed: DB2 paging must be specified sub-query alias, or an exception will occur #52
  • fixed:if page.size() == 0 then pageInfo.isIsLastPage() is false #50

5.0.1 - 2017-04-23

  • Add the new parameter countColumn used to configure the automatic count column, the default value 0, that is, `count(0).
  • The Page class is also added with the countColumn parameter, which can be configured for a specific query.
  • Modify the document display problem, by liumian* PR #30
  • Resolved sqlserver2012 paging error 42

5.0.0 - 2017-01-02

  • Use Use QueryInterceptor spec to handle paging logic
  • New pagination plugin interceptor com.github.pagehelper.PageInterceptor
  • New Dialect PageHelper is a special implementation class, the previous function is implemented in more user-friendly ways
  • New pagination plugin only a dialect parameter, the default dialect is PageHelper
  • PageHelper continue to support previously provided parameters, Among the latest to use the document has been fully updated
  • PageHelper has a helperDialect parameter which is the same functional as the previous dialect
  • Added paging implementation based on pure RowBounds and PageRowBounds, in com.github. pagehelper. dialect. rowbounds package, it is used as dialect Parameter sample implementation, more detailed documentation will be added later
  • Removed inappropriate orderby functions that appear in pagination plugin. It will provide a separate sort plug-ins in the future
  • Remove PageHelper are less commonly used methods
  • A new document, an important part of the update has been mentioned in the changelog, provides the English version of this document
  • fix bug #149
  • renamed Db2RowDialect to Db2RowBoundsDialect
  • All thrown exceptions being replaced by PageException

Older Changelogs are written in Chinese

You can view here