Framework Design Guidelines: Data Source Architectural Patterns
This chapter is from the book
This chapter is from the book
This chapter is from the book
Table Data Gateway
An object that acts as a Gateway (466) to a database table. One instance handles all the rows in the table.
Mixing SQL in application logic can cause several problems. Many developers aren't comfortable with SQL, and many who are comfortable may not write it well. Database administrators need to be able to find SQL easily so they can figure out how to tune and evolve the database.
ATable Data Gatewayholds all the SQL for accessing a single table or view: selects, inserts, updates, and deletes. Other code calls its methods for all interaction with the database.
How It Works
ATable Data Gatewayhas a simple interface, usually consisting of several find methods to get data from the database and update, insert, and delete methods. Each method maps the input parameters into a SQL call and executes the SQL against a database connection.TheTable Data Gatewayis usually stateless, as its role is to push data back and forth.
The trickiest thing about aTable Data Gatewayis how it returns information from a query. Even a simple find-by-ID query will return multiple data items. In environments where you can return multiple items you can use that for a single row, but many languages give you only a single return value and many queries return multiple rows.
一个替代方法是返回一些简单的数据结构cture, such as a map. A map works, but it forces data to be copied out of the record set that comes from the database into the map. I think that using maps to pass data around is bad form because it defeats compile time checking and isn't a very explicit interface, leading to bugs as people misspell what's in the map. A better alternative is to use aData Transfer Object (401).It's another object to create but one that may well be used elsewhere.
To save all this you can return theRecord Set (508)that comes from the SQL query. This is conceptually messy, as ideally the in-memory object doesn't have to know anything about the SQL interface. It may also make it difficult to substitute the database for a file if you can't easily create record sets in your own code. Nevertheless, in many environments that useRecord Set (508)widely, such as .NET, it's a very effective approach. ATable Data Gatewaythus goes very well withTable Module (125).If all of your updates are done through theTable Data Gateway,the returned data can be based on views rather than on the actual tables, which reduces the coupling between your code and the database.
If you're using aDomain Model (116),you can have theTable Data Gateway返回合适的域对象。这个问题with this is that you then have bidirectional dependencies the domain objects and the gateway. The two are closely connected, so that isn't necessarily a terrible thing, but it's something I'm always reluctant to do.
Most times when you useTable Data Gateway,you'll have one for each table in the database. For very simple cases, however, you can have a singleTable Data Gatewaythat handles all methods for all tables. You can also have one for views or even for interesting queries that aren't kept in the database as views. Obviously, view-basedTable Data Gatewaysoften can't update and so won't have update behavior. However, if you can make updates to the underlying tables, then encapsulating the updates behind update operations on theTable Data Gatewayis a very good technique.
When to Use It
As withRow Data Gateway (152)the decision regardingTable Data Gatewayis first whether to use aGateway (466)approach at all and then which one.
I find thatTable Data Gatewayis probably the simplest database interface pattern to use, as it maps so nicely onto a database table or record type. It also makes a natural point to encapsulate the precise access logic of the data source. I use it least withDomain Model (116)because I find thatData Mapper (165)gives a better isolation between theDomain Model (116)and the database.
Table Data Gatewayworks particularly well withTable Module (125),where it produces a record set data structure for theTable Module (125)to work on. Indeed, I can't really imagine any other database-mapping approach forTable Module (125).
Just likeRow Data Gateway (152),Table Data Gatewayis very suitable forTransaction Scripts (110).The choice between the two really boils down to how they deal with multiple rows of data. Many people like using aData Transfer Object (401),but that seems to me like more work than is worthwhile, unless the sameData Transfer Object (401)is used elsewhere. I preferTable Data Gateway当结果集表示方便for theTransaction Script (110)to work with.
Interestingly, it often makes sense to have theData Mappers (165)talk to the database viaTable Data Gateways.Although this isn't useful when everything is handcoded, it can be very effective if you want to use metadata for theTable Data Gatewaysbut prefer handcoding for the actual mapping to the domain objects.
One of the benefits of using aTable Data Gatewayto encapsulate database access is that the same interface can work both for using SQL to manipulate the database and for using stored procedures. Indeed, stored procedures themselves are often organized asTable Data Gateways.That way the insert and update stored procedures encapsulate the actual table structure. The find procedures in this case can return views, which helps to hide the underlying table structure.
Further Reading
[Alur et al.] discusses theData Access Objectpattern, which is aTable Data Gateway.They show returning a collection ofData Transfer Objects (401)on the query methods. It's not clear whether they see this pattern as always being table based; the intent and discussion seems to imply eitherTable Data GatewayorRow Data Gateway (152).
I've used a different name, partly because I see this pattern as a particular usage of the more generalGateway (466)concept and I want the pattern name to reflect that. Also, the termData Access Objectand its abbreviationDAOhas its own particular meaning within the Microsoft world.
Example: Person Gateway (C#)
Table Data Gateway数据库访问的形式通常是窗户吗s world, so it makes sense to illustrate one with C#. I have to stress, however, that this classic form ofTable Data Gatewaydoesn't quite fit in the .NET environment since it doesn't take advantage of the ADO.NET data set; instead, it uses the data reader, which is a cursor-like interface to database records. The data reader is the right choice for manipulating larger amounts of information when you don't want to bring everything into memory in one go.
For the example I'm using a Person Gateway class that connects to a person table in a database. The Person Gateway contains the finder code, returning ADO.NET's data reader to access the returned data.
class PersonGateway... public IDataReader FindAll() { String sql = "select * from person"; return new OleDbCommand(sql, DB.Connection).ExecuteReader(); } public IDataReader FindWithLastName(String lastName) { String sql = "SELECT * FROM person WHERE lastname = ?"; IDbCommand comm = new OleDbCommand(sql, DB.Connection); comm.Parameters.Add(new OleDbParameter("lastname", lastName)); return comm.ExecuteReader(); } public IDataReader FindWhere(String whereClause) { String sql = String.Format("select * from person where {0}", whereClause); return new OleDbCommand(sql, DB.Connection).ExecuteReader(); }
Almost always you'll want to pull back a bunch of rows with a reader. On a rare occasion you might want to get hold of an individual row of data with a method along these lines:
class PersonGateway... public Object[] FindRow (long key) { String sql = "SELECT * FROM person WHERE id = ?"; IDbCommand comm = new OleDbCommand(sql, DB.Connection); comm.Parameters.Add(new OleDbParameter("key",key)); IDataReader reader = comm.ExecuteReader(); reader.Read(); Object [] result = new Object[reader.FieldCount]; reader.GetValues(result); reader.Close(); return result; }
The update and insert methods receive the necessary data in arguments and invoke the appropriate SQL routines.
class PersonGateway... public void Update (long key, String lastname, String firstname, long numberOfDependents){ String sql = @" UPDATE person SET lastname = ?, firstname = ?, numberOfDependents = ? WHERE id = ?"; IDbCommand comm = new OleDbCommand(sql, DB.Connection); comm.Parameters.Add(new OleDbParameter ("last", lastname)); comm.Parameters.Add(new OleDbParameter ("first", firstname)); comm.Parameters.Add(new OleDbParameter ("numDep", numberOfDependents)); comm.Parameters.Add(new OleDbParameter ("key", key)); comm.ExecuteNonQuery(); } class PersonGateway... public long Insert(String lastName, String firstName, long numberOfDependents) { String sql = "INSERT INTO person VALUES (?,?,?,?)"; long key = GetNextID(); IDbCommand comm = new OleDbCommand(sql, DB.Connection); comm.Parameters.Add(new OleDbParameter ("key", key)); comm.Parameters.Add(new OleDbParameter ("last", lastName)); comm.Parameters.Add(new OleDbParameter ("first", firstName)); comm.Parameters.Add(new OleDbParameter ("numDep", numberOfDependents)); comm.ExecuteNonQuery(); return key; }
The deletion method just needs a key.
class PersonGateway... public void Delete (long key) { String sql = "DELETE FROM person WHERE id = ?"; IDbCommand comm = new OleDbCommand(sql, DB.Connection); comm.Parameters.Add(new OleDbParameter ("key", key)); comm.ExecuteNonQuery(); }
Example: Using ADO.NET Data Sets (C#)
The genericTable Data Gatewayworks with pretty much any kind of platform since it's nothing but a wrapper for SQL statements. With .NET you use data sets more often, butTable Data Gatewayis still useful although it comes in a different form.
A data set needs data adapters to load the data into it and update the data. In find it useful to define a holder for the data set and the adapters. A gateway then uses the holder to store them. Much of this behavior is generic and can be done in a superclass.
The holder indexes the data sets and adapters by the name of the table.
class DataSetHolder... public DataSet Data = new DataSet(); private Hashtable DataAdapters = new Hashtable();
)
){kind=link}
The gateway stores the holder and exposes the data set for its clients.
class DataGateway... public DataSetHolder Holder; public DataSet Data { get {return Holder.Data;} }
The gateway can act on an existing holder, or it can create a new one.
class DataGateway... protected DataSetGateway() { Holder = new DataSetHolder(); } protected DataSetGateway(DataSetHolder holder) { this.Holder = holder; }
The find behavior can work a bit differently here. A data set is a container for table-oriented data and can hold data from several tables. For that reason it's better to load data into a data set.
class DataGateway... public void LoadAll() { String commandString = String.Format("select * from {0}", TableName); Holder.FillData(commandString, TableName); } public void LoadWhere(String whereClause) { String commandString = String.Format("select * from {0}where {1}", TableName,whereClause); Holder.FillData(commandString, TableName); } abstract public String TableName {get;} class PersonGateway... public override String TableName { get {return "Person";} } class DataSetHolder... public void FillData(String query, String tableName) { if (DataAdapters.Contains(tableName)) throw new MutlipleLoadException(); OleDbDataAdapter da = new OleDbDataAdapter(query, DB.Connection); OleDbCommandBuilder builder = new OleDbCommandBuilder(da); da.Fill(Data, tableName); DataAdapters.Add(tableName, da); }
To update data you manipulate the data set directly in some client code.
person.LoadAll(); person[key]["lastname"] = "Odell"; person.Holder.Update();
The gateway can have an indexer to make it easier to get to specific rows.
class DataGateway... public DataRow this[long key] { get { String filter = String.Format("id = {0}", key); return Table.Select(filter)[0]; } } public override DataTable Table { get {return Data.Tables[TableName];} }
The update triggers update behavior on the holder.
class DataSetHolder... public void Update() { foreach (String table in DataAdapters.Keys) ((OleDbDataAdapter)DataAdapters[table]).Update(Data, table); } public DataTable this[String tableName] { get {return Data.Tables[tableName];} }
Insertion can be done much the same way: Get a data set, insert a new row in the data table, and fill in each column. However, an update method can do the insertion in one call.
class DataGateway... public long Insert(String lastName, String firstname, int numberOfDependents) { long key = new PersonGatewayDS().GetNextID(); DataRow newRow = Table.NewRow(); newRow["id"] = key; newRow["lastName"] = lastName; newRow["firstName"] = firstname; newRow["numberOfDependents"] = numberOfDependents; Table.Rows.Add(newRow); return key; }