Skip to content

Latest commit

 

History

History
468 lines (349 loc) · 19.1 KB

README.textile

File metadata and controls

468 lines (349 loc) · 19.1 KB

Persist

Persist is a Java-based ORM/DAO tool. It provides only the minimal amount of functionalities necessary to map
objects or maps from database queries and to statement parameters.

Persist works around a java.sql.Connection object. This means that it does not care about customer query
languages (it uses plain SQL with placeholders, as PreparedStatement objects use), connection pool handling,
transaction handling (for the most part), and so on. This also means it is very flexible, and can be integrated with any
code that depends on JDBC (including code that already use another ORM/DAO tool).

Persist does not require explicit mappings from POJOs to database tables. As long as there is some sort of naming
conventions that relate database names with POJO names, Persist will require virtually no mappings. It can, however, be
instructed to map Java classes and fields to database tables and columns using annotations.

Persist requires no singletons, no ThreadLocal’s, no global objects and no configuration files. It has
no external dependencies (although it will use Log4j, if available). It has a very small, robust and straightforward
codebase, which can be easily debugged in case you face an unforeseen problem. It is actively tested (with a high code
coverage) for MySQL, PostgreSQL, H2/HSQLDB, Derby, Oracle, DB2 and MS SQL Server. It imposes a very small overhead to
comparable plain JDBC operations, making it very attractive for high performance and/or low footprint applications.

Persist is distributed under a BSD license.

Quickstart

A Persist engine only requires a java.sql.Connection to be created:

 Persist persist = new Persist(connection);

Persist supports several different mapping strategies:

POJOs mapped to tables

By default, if no annotations specify a given class should not be mapped to a table, Persist will try to find a
table that matches that class and create a mapping between fields and columns.

 // inserts a new customer (the class _Customer_ is mapped to the table _customer_ automatically)
 persist.insert(customer);

 // reads a customer by its primary key
 Customer c = persist.readByPrimaryKey(Customer.class, 42);

 // retrieves customers using a custom query (note the usage of varargs)
 List<Customer> list = persist.readList(Customer.class, "select * from customer where id > ?", 10);

 // fetch all customers and assign the ResultSet to an Iterator
 Iterator allCustomersIterator = persist.readIterator(Customer.class, "select * from customer");

POJOs not mapped to tables

If a class is annotated with @NoTable, Persist will not try to map it to a table, and the class will
only be able to hold data produced by queries.

 @NoTable
 class QueryData {
    private int count;
    private String concatName;
   
    public long getCount() { return count; }
    public void setCount(long count) { this.count = count; }
   
    public String getConcatName() { return concatName; }
    public void setConcatName(String concatName) { this.concatName = concatName; }
 }

 QueryData qd1 = persist.read(QueryData.class, "select 1 as count, 'hello' as concat_name from dual");
 QueryData qd2 = persist.read(QueryData.class, "select 2 as counts, null as concatnames from dual");

java.util.Map’s

Map instances can be used to hold data from queries. Persist will convert values returned from the query to Java
types. Keys in the table are the names of the columns returned in lower case.

 // fetch a customer using a custom query and return the result as a map
 Map<String,Object> customerMap = persist.readMap("select * from customer where id=?", 10);

 // fetch all customers and result the results as Map instances in a List 
 List<Map<String,Object>> customerMapList = persist.readMapList("select * from customer");

 // fetch all customers and assign the ResultSet to an Iterator which maps rows to Map instances
 Iterator allCustomersIterator = persist.readMapIterator("select * from customer");

Java primitive types

If a query returns a single column, Persist can map data directly into primitive types (either single values or lists):

 // return customer name as String
 String name = persist.read(String.class, "select name from customer where id=?", 55);

 // fetch all customer id's as a list of integers
 List<Integer> ids = persist.readList(int.class, "select id from customer");

Custom queries with no returning data

Arbitrary queries that return no data can be easily executed.

 // execute arbitrary SQL with parameters
 persist.executeUpdate("delete from customer where id in (?,?)", 10, 20);

For the POJO mapping strategies, persist can map names either using explicit annotations or through a name
guesser
which translates class and field names to table and column names. A name guesser is a class that implements an
interface with a single method, and provide a generic, uniform way of translating names from the database schema to the
Java name conventions.

That’s almost everthing you need to know before using Persist! To get a better understanding on its internals,
please consult the following sections.

Creating Persist instances

Persist only requires a java.sql.Connection object to be created:

 Persist persist = new Persist(connection);

Caches

Internally, Persist will maintain a cache for all mapped objects it interacts with. Caches exist on a classloader
basis. If an application has to deal with connections with different databases, different mapping caches must be used.

To specify different caches, use a cache name in the constructor:

 // Create a persist instance for MySQL using the default cache
 Persist persistMysql = new Persist(connectionMysql);

 // Create a persist instance for Oracle using the "oracle" cache name
 Persist persistOracle = new Persist("oracle", connectionOracle);

Logging

If Persist can find Log4J in the classpath, it will use it. The following channels are used, all in debug
mode:

  • persist.engine
  • persist.parameters
  • persist.results
  • persist.profiling

Annotations

There are a few annotations that control the mapping behavior from classes to database tables:

  • @Table can be associated with a class and specify the name of the table that class is mapped to
  • @Column can be associated with a setter or getter of a field and specify the name of the column to be
    associated with that field, and/or specify if the field is auto-incremented by the database upon insertion.
  • @NoTable can be associated with a class to specify the class should not be mapped to a table in the
    database. Classes annotated with @NoTable can only be used to hold data from queries.
  • @NoColumn can be associated with a setter or getter of a field and specify it should not be mapped to
    a column in the database (by default, Persist attempts to map all fields of a given class to columns in the table
    associated with their class).

Mapping POJOs to tables

By default, if Persist is given a class, it will try to map it to a database table. To do so, it relies either on
explicit annotations (such as @Table and @Column) or name guessers which are configurable and
determine a global translation mechanism between class and field names to table and column names.

Consider the following table definition and its associated bean

 create table customer (
   id int auto_increment,
   name varchar(255),
   primary key (id)
 )
 class Customer {
    private int id;
    private String name;
   
    @Column(autoIncrement=true)
    public long getId() { return id; }
    public void setId(long id) { this.id = id; }
   
    public String getName() { return name; }
    public void setName(String name) { this.name = name; }
 }

class Customer and its fields don’t specify annotations to explicitly define which tables and columns
should be used. Therefore, when Persist tries to map this class, it will use a name guesser. Name guessers are
responsible for programatically converting class and field names to table and column names, using whichever convention
is in place.

Since no name guesser was specified, the DefaultNameGuesser will be used. The DefaultNameGuesser
converts class and field names in the form CompoundName to this list of guessed names: [compound_name, compound_names, compoundname, compoundnames]

While performing the automatic mapping for the Customer class above, Persist would try to find any of
those tables in the database: [customer, customers]. Since the table customer exists, it picks it for
the mapping and start mapping fields from Customer to columns in customer using the same approach.

During the process of mapping columns to fields, Persist stores information about which columns are primary keys
and which ones are auto-incremented upon insertion (this must be specified using @Column(autoIncrement=true)
since there’s no deterministic way of doing this automatically).

This mapping process only happens once (per class per classloader). Persist stores mappings for each class it
interacts with in an internal cache

After a mapping is created, Persist can perform CRUD operations directly on instances of the Customer
POJO. To illustrate the whole process, consider the following code

 // create a new customer instance
 Customer customer = new Customer();
 customer.setName("a new customer");

 // fetch a customer using its primary key
 Customer c = persist.readByPrimaryKey(Customer.class, 10);

The moment persist has contact with Customer, it will build a mapping automatically and cache it. With
the mapping, it will know that the Customer class is mapped to the customer table, and that id
is the primary key in that table. With that information, Persist can issue a select SQL statement querying for
all columns in customer having the specified primary key (id).

Other CRUD operations can be used directly as well

 // insert
 persist.insert(customer);

 // update
 persist.update(customer);

 // delete
 persist.delete(customer);

Important note: Persist can only perform readByPrimaryKey, update and delete
operations for classes mapped to tables that have primary keys. insert and all the read
operations can work on any POJO mapped on any table.

Persist supports several different ways of reading data from mapped tables:

 // fetch a single customer using a custom query
 Customer customer = persist.read(Customer.class, "select * from customer where id = 10");

 // fetch all customers
 List<Customer> allCustomersList = persist.readList(Customer.class);

 // fetch a set of customers using a custom query
 List<Customer> customersList = persist.readList(Customer.class, "select * from customer where id < ?", 100);

 // fetch all customers using a custom query and assign the ResultSet to an Iterator which maps rows to Customer instances
 Iterator allCustomersIterator = persist.readIterator(Customer.class, "select * from customer where id in (?,?)", 10, 20);

Types and conversions

Persist will respect the Java types of the fields on a given POJO as much as it can while retrieving data from
ResultSet’s. Furthermore, Persist can perform type conversions to/from query parameters and ResultSet columns.

The following tables depict the ResultSet.get and PreparedStatement.set methods used for each Java type:

Java type ResultSet.get method PreparedStatement.set method
Boolean/boolean getBoolean setBoolean
Byte/byte getByte setByte
Short/short getShort setShort
Integer/int getInt setInt
Long/long getLong setLong
Float/float getFloat setFloat
Double/double getDouble setDouble
Character/char getString setString
Character[]/char[] getString setString
Byte[]/byte[] getBytes setBytes
String setString setString
java.math.BigDecimal getBigDecimal setBigDecimal
java.io.Reader getCharacterStream setCharacterStream
java.io.InputStream getBinaryStream setBinaryStream
java.util.Date getTimestamp setTimestamp
java.sql.Date getDate setDate
java.sql.Time getTime setTime
java.sql.Timestamp getTimestamp setTimestamp
java.sql.Clob getClob setClob
java.sql.Blob getBlob setBlob

POJOs not mapped to tables (@NoTable)

POJOs that are annotated with NoTable can only be used to hold data from queries. Mapping for classes
annotated with NoTable is performed using these rules:

  • Class names won’t affect the mapping
  • If a field contains a @Column(name="...") annotation, then only the specified column name will be
    used for that field
  • Otherwise, all column names returned by the name guesser will be associated to the field
  • If more than one field have conflicting table names (either from @Column annotations or from guessed
    names), Persist will throw an exception while trying to use the class

To illustrate how this works, consider the following class:

 class QueryData {
    private int count;
    private String concatName;
   
    public long getCount() { return count; }
    public void setCount(long count) { this.count = count; }
   
    public String getConcatName() { return concatName; }
    public void setConcatName(String concatName) { this.concatName = concatName; }
 }

Persist would create the following mapping for this class:

Column names Field name
count, counts count
concat_name, concat_names, concatname, concatnames concatName

Some examples of how this would work using dummy queries:

 QueryData qd1 = persist.read(QueryData.class, "select 1 as count, 'hello' as concat_name from dual");
 QueryData qd2 = persist.read(QueryData.class, "select 2 as counts, null as concatnames from dual");

Type conversions are performed using the same conversion table as POJOs mapped to tables use.

java.util.Map’s

Query results can be mapped directly to java.util.Map instances, using the readMap-prefixed methods.
Keys in the map are the names of the columns in lower case, and values are fetched from the ResultSet.

 // fetch a customer using a custom query and return the result as a map
 Map<String,Object> customerMap = persist.readMap("select * from customer where id=?", 10);

 // fetch all customers and result the results as Map instances in a List 
 List<Map<String,Object>> customerMapList = persist.readMapList("select * from customer");

 // fetch all customers and assign the ResultSet to an Iterator which maps rows to Map instances
 Iterator allCustomersIterator = persist.readMapIterator("select * from customer");

Values are retrieved from the ResultSet according with their SQL types (as defined in java.sql.Types), so that
for each SQL type Persist will request the value according with a pre-defined Java type, as specified in the following
table:

SQL type ResultSet.get method
ARRAY getArray
BIGINT getLong
BIT getBoolean
BLOB getBytes
BOOLEAN getBoolean
CHAR getString
CLOB getString
DATALINK getBinaryStream
DATE getDate
DECIMAL getBigDecimal
DOUBLE getDouble
FLOAT getFloat
INTEGER getInt
JAVA_OBJECT getObject
LONGVARBINARY getBytes
LONGVARCHAR getString
NULL getNull
NCHAR getString
NUMERIC getBigDecimal
OTHER getObject
REAL getDouble
REF getRef
SMALLINT getInt
TIME getTime
TIMESTAMP getTimestamp
TINYINT getInt
VARBINARY getBytes
VARCHAR getString
100 (Oracle specific) getFloat
101 (Oracle specific) getDouble

Primitive types

Persist can map query results having a single column directly to primitive Java types (such as int, Double,
String, etc.), either as single values or lists.

 // return customer name as String
 String name = persist.read(String.class, "select name from customer where id=?", 55);

 // fetch all customer id's as a list of integers
 List<Integer> ids = persist.readList(int.class, "select id from customer");

Type conversions are performed using the same conversion table as POJOs mapped to tables use.

Developing for the project

In order to run the tests for all the supported databases you’ll need the following databases installed:

  • MySQL
  • PostgreSQL
  • H2/HSQLDB
  • Derby
  • Oracle
  • DB2
  • MS SQL Server

One easy way to test against all databases supported in Linux (all the ones listed above except MS SQL Server) is to build a VM (using VMWare, VirtualBox, Parallels, KVM, Xen, etc.).

In order to build a Ubuntu VM with the databases ready for testing:

  1. Install Ubuntu Server using your preferred virtualization server having the user ‘ubuntu’ with sudo access
  2. Upgrade the VM packages: apt-get dist-upgrade
  3. Install openssh in the VM: apt-get install openssh-server
  4. Copy your ssh public key to the VM’s ~ubuntu/.ssh/authorized_keys
  5. In the VM, run sudo visudo and add the following line to the end of the file: ubuntu ALL=(ALL) NOPASSWD: ALL
  6. Install Python and Fabric (easy_install fabric)
  7. Run the command: cd build && fab install
  8. Configure your /etc/hosts file to map 127.0.0.1 to dbvm

To access the databases in the VM, use:

Database Admin command line Persist db command line
MySQL mysql -u root mysql -u persist persist
PostgreSQL sudo -u postgres psql psql -d persist -U persist
Oracle sqlplus SYS/root as SYSDBA sqlplus persist/persist
Derby /usr/local/derby/bin/ij then: connect ‘jdbc:derby://127.0.0.1/persist;create=true;user=persist;password=persist’;
H2 java -cp /usr/local/h2/bin/h2-1.2.147.jar org.h2.tools.Shell -url jdbc:h2:tcp://127.0.0.1/persist -user persist -password persist

License

Copyright 2011, persist committers. All rights reserved.

Redistribution and use in source and binary forms, with or without modification, are
permitted provided that the following conditions are met:

   1. Redistributions of source code must retain the above copyright notice, this list of
      conditions and the following disclaimer.

   2. Redistributions in binary form must reproduce the above copyright notice, this list
      of conditions and the following disclaimer in the documentation and/or other materials
      provided with the distribution.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY EXPRESS OR IMPLIED
WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS OR
CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

The views and conclusions contained in the software and documentation are those of the
authors and should not be interpreted as representing official policies, either expressed
or implied by the copyright holders.