Allowing Partially Trusted Callers
Sharing code libraries is a common scenario with common language runtime (CLR) integration, where an assembly containing a user-defined type, stored procedure, user-defined function, user-defined aggregate, trigger, or utility class is often accessed by another assembly or application. Code libraries that are to be shared by multiple applications must be signed with a strong name.
Only applications that are fully trusted by the runtime code access security system are allowed to access a shared managed code assembly that is not explicitly marked with the System.Security.AllowPartiallyTrustedCallers attribute. A partially trusted assembly (one that is registered in SQL Server with the SAFE or EXTERNAL_ACCESS permission set) that attempts to access a strong-name signed assembly without this attribute causes a System.Security.SecurityException to be thrown. The error message you see is similar to the following:
Msg 6522, Level 16, State 1, Procedure usp_RSTest, Line 0
A .NET Framework error occurred during execution of user defined
routine or aggregate 'usp_RSTest': System.Security.SecurityException: That assembly does not allow partially trusted callers.
System.Security.SecurityException: at
System.Security.CodeAccessSecurityEngine.ThrowSecurityException(
Assembly asm, PermissionSet granted,PermissionSet refused,
RuntimeMethodHandle rmh, SecurityAction action, Object demand,
IPermission permThatFailed) at
Microsoft.Samples.SqlServer.TestResultSet.Test()
We recommend that all assemblies registered in SQL Server, except those assemblies added to the global assembly cache, be marked with the AllowPartiallyTrustedCallers attribute so that assemblies loaded by SQL Server can access each other. Assemblies that are to be added to the global assembly cache should be thoroughly reviewed for safety before adding the AllowPartiallyTrustedCallers attribute, as the assembly would then be available to partially trusted callers from unexpected contexts. An assembly should not be made fully trusted (registered with the UNSAFE permission set in SQL Server).
For more information, see the "Using Libraries from Partially Trusted Code" section in the .NET Framework software development kit.
Example
Description
Suppose there is a utility class that would be useful for many server-side CLR integration applications. For example, it might be a class that represents the results of invoking a query. To enable sharing of this component, this utility class is placed in a separate assembly. Then that assembly is referenced from various other assemblies that contain CLR integration objects. Because this utility class is used in many different server applications, it is reviewed carefully and any security issues are resolved. The AllowPartiallyTrustedCallers attribute is then applied to the assembly that contains the utility class, so that CLR integration objects contained in assemblies marked with SAFE or EXTERNAL_ACCESS permission sets can use the utility class and methods, even though they are in a separate assembly.
Sometimes it is useful to be able to execute commands while reading through results of a query without opening a new connection and without reading all the results into memory. The Multiple Active Result Set (MARS) feature in ADO.NET 2.0 is a technology that can help you achieve that. Currently, MARS is not implemented for the in-process provider used for server-side programming. To work around this limitation, you can use server-side cursors. This sample demonstrates how to use server-side cursors to work around the lack of MARS support for server-side programming.
Using server-side cursors is expensive in terms of server resources and can sometimes prevent the query optimizer in SQL Server from enhancing the performance of your queries. Therefore, you might consider rewriting your code to use JOINs whenever possible.
The API for this class is similar to a data reader except that you can move forward or backward throughout the result set, and other commands can be issued on the connection while the result set is open.
This implementation is highly simplified to make the sample easy to understand. A more efficient implementation would fetch multiple rows to avoid a database turnaround per row fetched.
Using this class can have a significantly smaller memory footprint than filling a dataset with all the results of a query, which is very important for server-side programming.
This sample also demonstrates using the "Allow partially trusted callers" attribute to indicate that the Result Set assembly is a library which may be safely called from other assemblies. This approach is a little more complex but much more secure than registering the calling assembly using the unsafe permission. It is more secure because registering the calling assembly as safe the calling assembly limits affecting resources off the server and prevents damage to the integrity of the server.
The build instructions for this sample assume that the source code files are in a directory called c:\samples. If you use another directory, you will have to modify the Transact-SQL scripts. The Transact-SQL scripts also requires the AdventureWorks database. You can download the AdventureWorks sample database from the Microsoft SQL Server Samples and Community Projects home page.
To build and run the sample, paste the first code listing into a file called ResultSet.cs and compile with csc /target:library ResultSet.cs.
Paste the second code listing into a file called TestResultSet.cs and compile with csc /target:library /reference:ResultSet.dll TestResultSet.cs.
Paste the third code listing into a file called InstallCS.sql and compile with sqlcmd -E -I -i InstallCS.sql.
Paste the fourth code listing into a file called test.sql and compile with sqlcmd -E -I -i test.sql.
Paste the fifth code listing into a file called Cleanup.sql and compile with sqlcmd -E -I -i Cleanup.sql.
Code
// ResultSet.cs
using System;
using System.Collections;
using System.Collections.Generic;
using System.Text;
using System.Data;
using System.Data.Common;
using System.Data.Sql;
using System.Data.SqlTypes;
using System.Data.SqlClient;
using System.IO;
using System.Globalization;
using Microsoft.SqlServer.Server;
using System.Runtime.Serialization;
[assembly: System.Diagnostics.CodeAnalysis.SuppressMessage("Microsoft.Design", "CA1020:AvoidNamespacesWithFewTypes", Scope = "namespace", Target = "Microsoft.Samples.SqlServer")]
[assembly: System.Diagnostics.CodeAnalysis.SuppressMessage("Microsoft.Globalization", "CA1303:DoNotPassLiteralsAsLocalizedParameters", Scope = "member", Target = "Microsoft.Samples.SqlServer.ResultSetEnumerator.System.Collections.IEnumerator.Current", MessageId = "System.InvalidOperationException.#ctor(System.String)")]
namespace Microsoft.Samples.SqlServer {
// This is the base class for exceptions raised in the Adventure Works Cycles Storefront application.
[Serializable]
public class ResultSetException : System.Exception {
public ResultSetException() {}
public ResultSetException(string message) : base(message) {}
public ResultSetException(String message, Exception innerException) : base(message, innerException) {}
protected ResultSetException(SerializationInfo info, StreamingContext context) : base(info, context) {}
}
// This exception is raised when a user tries to register an email address which has already been taken.
[Serializable]
public class InvalidStateException : ResultSetException {
public InvalidStateException() {}
public InvalidStateException(string message) : base(message) {}
public InvalidStateException(String message, Exception innerException) : base(message, innerException) {}
protected InvalidStateException(SerializationInfo info, StreamingContext context) : base(info, context) {}
}
// This class is used to provide an API similar to a SQL data reader except that it is possible to navigate through any
// part of the result set. It is also possible to execute commands while this class is still "open" even though the
// CLR in-proc provider does not support MARS at this time. This implementation is highly simplified to make the
// sample easy to understand. A better implementation would fetch multiple rows to avoid a database turnaround per row
// fetched. Using this class can have a significantly smaller memory footprint than filling a dataset with
// all the results of a query which is very important for server side programming.
[System.Diagnostics.CodeAnalysis.SuppressMessage("Microsoft.Design", "CA1010:CollectionsShouldImplementGenericInterface"), System.Diagnostics.CodeAnalysis.SuppressMessage("Microsoft.Naming", "CA1710:IdentifiersShouldHaveCorrectSuffix")]
public class ResultSet : IDataReader, IDataRecord, IEnumerable, IDisposable {
// Which database to access
private SqlConnection connection;
// A cache or a single row's worth of data
private Microsoft.SqlServer.Server.SqlDataRecord results;
//A cache of whether the columns of the current cached row are set to the default value for the column
private int visibleFieldCount;
private string cursorName = string.Format(CultureInfo.InvariantCulture, "[{0}]", Guid.NewGuid().ToString());
internal Microsoft.SqlServer.Server.SqlDataRecord CurrentRecord {
get {
EnsureState(StateValues.read);
return results;
}
}
// Which database to access. The connection passed to the setter must be open.
[System.Diagnostics.CodeAnalysis.SuppressMessage("Microsoft.Globalization", "CA1303:DoNotPassLiteralsAsLocalizedParameters", MessageId = "Microsoft.Samples.SqlServer.InvalidStateException.#ctor(System.String)")/*, System.CLSCompliant(false)*/]
public SqlConnection Connection {
get {
return connection;
}
set {
if (value == null)
throw new ArgumentNullException("value");
if (state != StateValues.created && state != StateValues.initialized)
throw new InvalidStateException("Cannot alter select command after opening the result set.");
if (value.State != ConnectionState.Open)
throw new InvalidStateException("Connection must be opened prior to setting result set connection value.");
connection = value;
if (selectCommand != null && connection != null)
state = StateValues.initialized;
else
state = StateValues.created;
}
}
private string selectCommand; //What query to execute against the database
[System.Diagnostics.CodeAnalysis.SuppressMessage("Microsoft.Globalization", "CA1303:DoNotPassLiteralsAsLocalizedParameters", MessageId = "Microsoft.Samples.SqlServer.InvalidStateException.#ctor(System.String)")]
public string SelectCommand {
get {
return selectCommand;
}
set {
if (state != StateValues.created && state != StateValues.initialized)
throw new InvalidStateException("Cannot alter select command after opening the result set.");
selectCommand = value;
if (selectCommand != null && connection != null)
state = StateValues.initialized;
else
state = StateValues.created;
}
}
private bool isScrollable = true;
[System.Diagnostics.CodeAnalysis.SuppressMessage("Microsoft.Globalization", "CA1303:DoNotPassLiteralsAsLocalizedParameters", MessageId = "Microsoft.Samples.SqlServer.InvalidStateException.#ctor(System.String)")]
public bool IsScrollable {
get {
return isScrollable;
}
set {
if (state != StateValues.created && state != StateValues.initialized)
throw new InvalidStateException("Cannot alter IsScrollable property after opening the result set.");
isScrollable = value;
}
}
// The possible phases this object can go through from creation through termination.
// created: Result set is instantiated but connection and select command are not set
// initialized: Result set is instantiated and connection and select command are set
// opened: The cursor has been created for browsing data
// read: At least one row of data has been read and cached.
// closed: Cursor is closed and deallocated, and connection is closed. No further operations are possible.
private enum StateValues { created, initialized, opened, read, closed };
// What the current phase is
private StateValues state = StateValues.created;
// It is possible to create the result set with this constructor and then set the connection
// and select command using the public properties.
public ResultSet() {}
// Or you can create the result set all at once using this form of the constructor.
public ResultSet(SqlConnection connection, string selectCommand) {
this.Connection = connection;
this.SelectCommand = selectCommand;
state = StateValues.initialized;
}
// By default the result set is scrollable, but you can create a forward-only result set using this constructor
// and specifying false for isScrollable.
public ResultSet(SqlConnection connection, string selectCommand, bool isScrollable) {
this.Connection = connection;
this.SelectCommand = selectCommand;
this.IsScrollable = isScrollable;
state = StateValues.initialized;
}
// This method should be called just before the result set is to be used to pull data. The open doesn't happen as
// part of creation as server resources are being consumed while the result set is open. Be sure to close an open
// result set as soon as possible to release server side resources.
[System.Diagnostics.CodeAnalysis.SuppressMessage("Microsoft.Globalization", "CA1303:DoNotPassLiteralsAsLocalizedParameters", MessageId = "Microsoft.Samples.SqlServer.InvalidStateException.#ctor(System.String)")]
public void Open() {
switch (state) {
case StateValues.created:
throw new InvalidStateException("Cannot open result set until it is properly initialized");
case StateValues.opened:
case StateValues.read:
throw new InvalidStateException("Result set already open.");
case StateValues.closed:
throw new InvalidStateException("Cannot reopen closed result set.");
default:
break;
}
if (state == StateValues.initialized) {
// There are many possible cursor options. We only allow control of scrollability.
// To keep the implementation simple we only permit read operations.
string scrollMode = (isScrollable) ? "SCROLL" : "FORWARD_ONLY";
// Caller must ensure that any user input which is part of selectCommand is free from injection attacks.
SqlCommand cmd = connection.CreateCommand();
cmd.CommandText = string.Format(CultureInfo.InvariantCulture, "DECLARE {0} CURSOR GLOBAL {1} READ_ONLY FOR {2}; OPEN {0};",
cursorName, scrollMode, selectCommand);
cmd.ExecuteNonQuery();
state = StateValues.opened;
}
}
// Free all server and client side resources being consumed for this result set. Most operations
// are no longer valid after calling this method on the result set.
[System.Diagnostics.CodeAnalysis.SuppressMessage("Microsoft.Globalization", "CA1303:DoNotPassLiteralsAsLocalizedParameters", MessageId = "Microsoft.Samples.SqlServer.InvalidStateException.#ctor(System.String)")]
public void Close() {
switch (state) {
case StateValues.created:
case StateValues.initialized:
throw new InvalidStateException("Cannot close unopened result set.");
case StateValues.closed:
throw new InvalidStateException("Result set already closed.");
case StateValues.opened:
case StateValues.read:
SqlCommand cmd = connection.CreateCommand();
cmd.CommandText =
string.Format(CultureInfo.InvariantCulture, "CLOSE {0}; DEALLOCATE {0};", cursorName);
try {
cmd.ExecuteNonQuery();
}
finally {
results = null;
state = StateValues.closed;
}
break;
default:
throw new InvalidStateException(
string.Format(CultureInfo.InvariantCulture,
"Unknown result set state {0}", state.ToString()));
}
}
// This implementation does not support multi-dimensional data. Always zero.
public int Depth {
get {
EnsureState(StateValues.read);
return 0;
}
}
// Normally this would return a data table with schema information about the data being returned.
// Unfortunately this isn't implemented in the in-proc provider, so always throw a NYI exception.
[System.Diagnostics.CodeAnalysis.SuppressMessage("Microsoft.Globalization", "CA1303:DoNotPassLiteralsAsLocalizedParameters", MessageId = "System.NotImplementedException.#ctor(System.String)")]
public DataTable GetSchemaTable() {
throw new NotImplementedException("GetSchemaTable is not currently implemented.");
}
// Determines if there is any data to read. True if there is data to read, otherwise false.
[System.Diagnostics.CodeAnalysis.SuppressMessage("Microsoft.Globalization", "CA1303:DoNotPassLiteralsAsLocalizedParameters", MessageId = "System.NotImplementedException.#ctor(System.String)")]
public bool HasRows {
get {
throw new NotImplementedException("HasRows is not currently implemented.");
}
}
// Determines if the result set is available for use
// Returns false if the result set is available for use, otherwise true.
public bool IsClosed {
get { return state == StateValues.closed; }
}
// Multiple result sets are not supported. Always throws an invalid operation exception.
[System.Diagnostics.CodeAnalysis.SuppressMessage("Microsoft.Globalization", "CA1303:DoNotPassLiteralsAsLocalizedParameters", MessageId = "System.InvalidOperationException.#ctor(System.String)")]
public bool NextResult() {
throw new InvalidOperationException("Multiple result sets are not support for result set class");
}
// Fetches the next row from the result set on the server and makes that data available to the caller.
// Returns True if and only if there is data to read.
public bool Read() {
return fetch(string.Format(CultureInfo.InvariantCulture, "FETCH NEXT FROM {0};", cursorName));
}
// Returns the number of records affected by processing this result set. This may only be called when the result set
// is closed. Since this implementation supports only read-only access, there can never be any rows affected (always -1).
public int RecordsAffected {
get {
EnsureState(StateValues.closed);
return -1;
}
}
// Move to a particular record number in the results returned by the query. Requires a scrollable result set.
// Returns True if and only if there is data to read.
public bool ReadAbsolute(int position) {
EnsureScroll();
return fetch(string.Format(CultureInfo.InvariantCulture, "FETCH ABSOLUTE {0} FROM {1};",
position, cursorName));
}
// Move to the first result in the results returned by the query and read it. Requires a scrollable result set.
// Returns True if and only if there is data to read.
public bool ReadFirst() {
EnsureScroll();
return fetch(string.Format(CultureInfo.InvariantCulture, "FETCH FIRST FROM {0};", cursorName));
}
// Move to the last result in the results returned by the query and read it. Requires a scrollable result set.
// Returns True if and only if there is data to read.
public bool ReadLast() {
EnsureScroll();
return fetch(string.Format(CultureInfo.InvariantCulture, "FETCH LAST FROM {0};", cursorName));
}
// Move to the immediate prior result in the results returned by the query, and read it. Requires a scrollable
// result set. Returns True if and only if there is data to read.
public bool ReadPrevious() {
EnsureScroll();
return fetch(string.Format(CultureInfo.InvariantCulture, "FETCH PREVIOUS FROM {0};", cursorName));
}
// Move the specified number of records forward or backward from the current position
// in the results returned by the query and read it. Requires a scrollable result set.
/// position is how many records forward or backward. Returns True if and only if there is data to read.
public bool ReadRelative(int position) {
EnsureScroll();
return fetch(string.Format(CultureInfo.InvariantCulture, "FETCH RELATIVE {0} FROM {1};", position, cursorName));
}
// Dispose is idential to closing the result set.
public void Dispose() {
Dispose(true);
}
protected virtual void Dispose(bool shouldClearManagedResources) {
Close();
if (shouldClearManagedResources)
connection = null;
}
// The number of columns available in the results just read
public int FieldCount {
get {
EnsureState(StateValues.read);
return results.FieldCount;
}
}
public int VisibleFieldCount {
get {
EnsureState(StateValues.read);
return visibleFieldCount;
}
}
[System.Diagnostics.CodeAnalysis.SuppressMessage("Microsoft.Globalization", "CA1303:DoNotPassLiteralsAsLocalizedParameters", MessageId = "System.InvalidOperationException.#ctor(System.String)")]
public IDataReader GetData(int i) {
throw new InvalidOperationException("GetData is no longer implemented");
}
// What kind of data is acceptable for the specified column
public string GetDataTypeName(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetDataTypeName(i);
}
// Returns the object which represents the kind of data which is acceptable for the specified column.
public Type GetFieldType(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetFieldType(i);
}
// Returns the symbolic id of the specified column
public string GetName(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetName(i);
}
// Given a column name, returns the integer id the column.
public int GetOrdinal(string name) {
EnsureState(StateValues.read);
//Note: this implementation does not deal with all collation and internationalization issues.
return results.GetOrdinal(name);
}
// Places all the values contained in the current row into the supplied object array. The elements
// of the array are based on the SQL type system.
public int GetSqlValues(object[] values) {
EnsureState(StateValues.read);
return results.GetSqlValues(values);
}
// Places all the values contained in the current row into the supplied object array. The
// elements of the array are based on the CLR type system.
public int GetValues(object[] values) {
EnsureState(StateValues.read);
return results.GetValues(values);
}
// Accesses the data in the specified column by name
public object this[string name] {
get {
EnsureState(StateValues.read);
return results[name];
}
}
// Accesses the data in the specified column by numerical id
public object this[int i] {
get {
EnsureState(StateValues.read);
return results[i];
}
}
// The following methods provide access to column data for the current row as objects using the Sql type system.
// The advantage of using this type system is that you can deal very precisely with the database's concept of null.
// Also, some types (such as SqlXml) provide more efficient methods of data access than the CLR type system.
// Access the data in the specified column in the current row as a SqlBinary object
public SqlBinary GetSqlBinary(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlBinary(i);
}
// Access the data in the specified column in the current row as a SqlBoolean object
public SqlBoolean GetSqlBoolean(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlBoolean(i);
}
// Access the data in the specified column in the current row as a SqlByte object
public SqlByte GetSqlByte(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlByte(i);
}
// Access the data in the specified column in the current row as a SqlBytes object
public SqlBytes GetSqlBytes(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlBytes(i);
}
// Access the data in the specified column in the current row as a SqlChars object
public SqlChars GetSqlChars(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlChars(i);
}
// Access the data in the specified column in the current row as a SqlDateTime object
public SqlDateTime GetSqlDateTime(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlDateTime(i);
}
// Access the data in the specified column in the current row as a SqlDecimal object
public SqlDecimal GetSqlDecimal(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlDecimal(i);
}
// Access the data in the specified column in the current row as a SqlDouble object
public SqlDouble GetSqlDouble(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlDouble(i);
}
// Access the data in the specified column in the current row as a SqlGuid object
public SqlGuid GetSqlGuid(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlGuid(i);
}
// Access the data in the specified column in the current row as a SqlInt16 object
public SqlInt16 GetSqlInt16(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlInt16(i);
}
// Access the data in the specified column in the current row as a SqlInt32 object
public SqlInt32 GetSqlInt32(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlInt32(i);
}
// Access the data in the specified column in the current row as a SqlInt64 object
public SqlInt64 GetSqlInt64(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlInt64(i);
}
// Return schema information for the specified column
public Microsoft.SqlServer.Server.SqlMetaData GetSqlMetaData(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlMetaData(i);
}
// Access the data in the specified column in the current row as a SqlMoney object
public SqlMoney GetSqlMoney(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlMoney(i);
}
// Access the data in the specified column in the current row as a SqlSingle object
public SqlSingle GetSqlSingle(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlSingle(i);
}
// Access the data in the specified column in the current row as a SqlString object
public SqlString GetSqlString(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlString(i);
}
// Access the data in the specified column in the current row as a Sql object
public object GetSqlValue(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlValue(i);
}
// Access the data in the specified column in the current row as a SqlXml object
public SqlXml GetSqlXml(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetSqlXml(i);
}
// The following methods provide access to column data for the current row using the CLR type system. The advantage
// of using these methods is seamless integration with other CLR components.
// Access the data in the specified column in the current row as a Boolean
public bool GetBoolean(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetBoolean(i);
}
// Access the data in the specified column in the current row as a byte
public byte GetByte(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetByte(i);
}
// Copies bytes from the specified column of the current row into a buffer supplied by the caller.
// i is the columm, fieldOffset is where to start in the column's data, buffer is where to place the bytes,
// bufferOffset is where to start in the buffer, and length is how many bytes to copy.
// Returns how many bytes were actually copied.
public long GetBytes(int i, long fieldOffset, byte[] buffer, int bufferoffset, int length) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetBytes(i, fieldOffset, buffer, bufferoffset, length);
}
// Access the data in the specified column in the current row as a char.
public char GetChar(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetChar(i);
}
// Copies characters from the specified column of the current row into a buffer supplied by the caller.
// i is the column, fieldoffset is where to start in the column's data, buffer is where to place the characters,
// bufferoffset is where to start in the buffer, length is how many characters to copy.
// Returns how many characters were actually copied.
public long GetChars(int i, long fieldoffset, char[] buffer, int bufferoffset, int length) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetChars(i, fieldoffset, buffer, bufferoffset, length);
}
// Access the data in the specified column in the current row as a DateTime
public DateTime GetDateTime(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetDateTime(i);
}
// Access the data in the specified column in the current row as a decimal.
public decimal GetDecimal(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetDecimal(i);
}
// Access the data in the specified column in the current row as a double
public double GetDouble(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetDouble(i);
}
// Access the data in the specified column in the current row as a float
public float GetFloat(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetFloat(i);
}
// Access the data in the specified column in the current row as a Guid
public Guid GetGuid(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetGuid(i);
}
// Access the data in the specified column in the current row as a short
public short GetInt16(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetInt16(i);
}
// Access the data in the specified column in the current row as an int
public int GetInt32(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetInt32(i);
}
// Access the data in the specified column in the current row as a long
public long GetInt64(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetInt64(i);
}
// Access the data in the specified column in the current row as a string
public string GetString(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetString(i);
}
// Access the data in the specified column in the current row as an object
public object GetValue(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.GetValue(i);
}
// Returns true if and only if the data in the specified column of the current row is null as defined by the database.
// Note this is different than the CLR concept of null. The parameter is the column.
public bool IsDBNull(int i) {
EnsureState(StateValues.read);
EnsureOrdinal(i);
return results.IsDBNull(i);
}
// Utility method which executes a server side cursor motion command and reads the row at the current location of
// the cursor after the motion is complete. Reading one row at a time can be inefficient, see comments at the start
// of this class. The parameter is the SQL cursor motion command to execute. Returns True if and only if there is data to read.
[System.Diagnostics.CodeAnalysis.SuppressMessage("Microsoft.Globalization", "CA1303:DoNotPassLiteralsAsLocalizedParameters", MessageId = "Microsoft.Samples.SqlServer.ResultSet.EnsureState(Microsoft.Samples.SqlServer.ResultSet+StateValues,Microsoft.Samples.SqlServer.ResultSet+StateValues,System.String)")]
private bool fetch(string commandText) {
EnsureState(StateValues.read, StateValues.opened, "The result set must be open to perform the read operation.");
SqlCommand cmd = connection.CreateCommand();
cmd.CommandText = commandText;
SqlDataReader reader = null;
try {
reader = cmd.ExecuteReader();
// If there is no data, return false;
if (!reader.Read()) {
state = StateValues.opened;
return false;
}
else {
// Otherwise, advance the state to 'read'
state = StateValues.read;
// If we haven't cached schema information get it now
if (results == null) FetchMetadata(reader);
// Cache data values
object[] values = new object[FieldCount];
reader.GetSqlValues(values);
results.SetValues(values);
// Indicate there is data read by returning true
return true;
}
}
finally {
if (reader != null) {
reader.Close();
reader = null;
}
}
}
// Helper method for the fetch method. Initializes the data structures which cache the schema information for the
// current query. The parameter is the reader from where schema information may be read.
private void FetchMetadata(SqlDataReader reader) {
int columnCount = reader.FieldCount;
visibleFieldCount = columnCount;
// Initialize the data cache
SqlMetaData[] metadata = new SqlMetaData[visibleFieldCount];
// Fill the schema cache for each column
DataTable schemaTable = reader.GetSchemaTable();
SqlString collationString = new SqlString(String.Empty);
for (int i = 0 ; i < metadata.Length ; i++) {
DataRow columnSchema = schemaTable.Rows[i];
metadata[i] = new SqlMetaData(
reader.GetName(i),
(SqlDbType)columnSchema["ProviderType"],
(int)columnSchema["ColumnSize"],
(byte)((short)columnSchema["NumericPrecision"]),
(byte)((short)columnSchema["NumericScale"]),
collationString.LCID,
collationString.SqlCompareOptions,
(Type)columnSchema["DataType"]
);
}
results = new Microsoft.SqlServer.Server.SqlDataRecord(metadata);
}
// Helper method which validates the state of the result set before processing a
// requested operation. If the result set is in an invalid state an exception is thrown.
private void EnsureState(StateValues desiredState) {
if (state != desiredState) {
// Compute the default message
string message = string.Format(CultureInfo.InvariantCulture, "Expected result set state to be {0} but it was {1}",
desiredState.ToString(), state.ToString());
// If we have more specific information, provide a different message.
if (desiredState == StateValues.read)
message = "You must read a valid row of the result set before performing the requested operation.";
throw new InvalidStateException(message);
}
}
// Similar to the single argument EnsureState method except that the two specified states
// are both legal and a custom error message must be provided.
private void EnsureState(StateValues desiredState1, StateValues desiredState2, string message) {
if (state != desiredState1 && state != desiredState2)
throw new InvalidStateException(message);
}
// Makes certain that the requested column is valid, and throws an exception with an error message if it isn't.
// This would be more useful than the error message given when accessing beyond the valid boundary of an array,
// for example. The parameter is the column trying to be accessed
private void EnsureOrdinal(int i) {
if (i >= results.FieldCount)
throw new ArgumentException(
string.Format(CultureInfo.InvariantCulture, "Attempt to access column {0} when only {1} columns exist",
i, results.FieldCount));
}
// Makes certain that the result set is in a proper state to read data, and that
// scrolling is allowed. This method is called from methods which require scrolling capabilities.
[System.Diagnostics.CodeAnalysis.SuppressMessage("Microsoft.Globalization", "CA1303:DoNotPassLiteralsAsLocalizedParameters", MessageId = "System.InvalidOperationException.#ctor(System.String)")]
private void EnsureScroll() {
if (!isScrollable)
throw new InvalidOperationException("Attempting to execute a scroll operation on a non-scrollable result set.");
}
public IEnumerator GetEnumerator() {
return new ResultSetEnumerator(this);
}
}
public class ResultSetEnumerator : IEnumerator {
private bool isRead;
private ResultSet resultSetToEnumerate;
public ResultSetEnumerator(ResultSet resultSetToEnumerate) {
this.resultSetToEnumerate = resultSetToEnumerate;
}
// Returns the current element of a result set which is represented by a SqlDataRecord
object IEnumerator.Current {
get {
if (!isRead)
throw new InvalidOperationException(
"The enumerator is positioned before the first element "
+ "of the collection or after the last element");
return resultSetToEnumerate.CurrentRecord;
}
}
[System.Diagnostics.CodeAnalysis.SuppressMessage("Microsoft.Globalization", "CA1303:DoNotPassLiteralsAsLocalizedParameters", MessageId = "System.InvalidOperationException.#ctor(System.String)")]
public SqlDataRecord Current {
get {
if (!isRead)
throw new InvalidOperationException(
"The enumerator is positioned before the first element "
+ "of the collection or after the last element");
return resultSetToEnumerate.CurrentRecord;
}
}
// Advance to the next valid element of a result set; returns True if this position has a valid element
public bool MoveNext() {
if (!isRead) {
if (resultSetToEnumerate.ReadFirst()) {
isRead = true;
return true;
}
else
return false;
}
else {
return resultSetToEnumerate.Read();
}
}
// Reset the current position of the enumerator to just prior to the current element
public void Reset() {
isRead = false;
}
}
}
Code
// TestResultSet.cs
using System;
using System.Collections.Generic;
using System.Text;
using System.Configuration;
using System.Data;
using System.Data.SqlClient;
using System.Data.SqlTypes;
using Microsoft.SqlServer.Server;
// Main application that increases the prices for the most popular bikes in order to demonstrate using server side cursors
// to browser results and update in parallel.
namespace Microsoft.Samples.SqlServer {
public sealed class TestResultSet {
// The factor to mulitply by the current price to get the new price for each color
private const String priceIncreases = "Black, 1.02, Silver, 1.01, Red, 1.05, Yellow, 1.02, Green, 1.03";
// Given the name of a color, return the price increase factor as a SqlMoney object.
private static readonly Dictionary<String, SqlMoney> increaseDictionary
= new Dictionary<String, SqlMoney>(priceIncreases.Length);
// Don't allow construction of instances since the public portion of this class is only used to contain static methods.
private TestResultSet() {}
// The Adventure Works Cycles corporation needs to increase the standard costs and list price for its most popular
// bikes due to price increases for the paint used on those bikes. This program demonstrates browsing popular bikes
// and updating prices in parallel using the result set sample to implement that scenario. Note that programmers
// should always consider whether using a JOIN in a server side query or update would be more efficient that using this
// style of programming.
public static void Test() {
// Use the priceIncreases constant to initialize entries in the increaseDictionary collection.
InitializeIncreaseDictionary();
SqlConnection myConnection = null;
bool isRSOpened = false;
// Initialize the command to get most popular bikes
myConnection = new SqlConnection("context connection=true");
ResultSet rs = null;
try {
myConnection.Open();
rs = new ResultSet(myConnection,
"SELECT TOP 10 P.ProductID, P.Name, P.Color, P.StandardCost, P.ListPrice, "
+ "sum(SOD.OrderQty) FROM Production.Product AS P "
+ "JOIN Sales.SalesOrderDetail AS SOD ON P.ProductID = SOD.ProductID "
+ "JOIN Production.ProductSubcategory AS PSC "
+ "ON P.ProductSubcategoryID = PSC.ProductSubcategoryID "
+ "JOIN Production.ProductCategory AS PC "
+ "ON PSC.ProductCategoryID = PC.ProductCategoryID "
+ "WHERE PC.Name = 'Bikes' "
+ "GROUP BY P.ProductID, P.Name, P.Color, P.StandardCost, P.ListPrice "
+ "ORDER BY sum(SOD.OrderQty) desc;");
// Initialize the command to update the price of a product
SqlCommand updateCommand = myConnection.CreateCommand();
updateCommand.CommandText = "usp_UpdateProductPrice";
updateCommand.CommandType = CommandType.StoredProcedure;
SqlParameter productIDParameter = new SqlParameter("@ProductID", SqlDbType.Int);
updateCommand.Parameters.Add(productIDParameter);
SqlParameter standardCostParameter = new SqlParameter("@StandardCost", SqlDbType.Money);
updateCommand.Parameters.Add(standardCostParameter);
SqlParameter listPriceParameter = new SqlParameter("@ListPrice", SqlDbType.Money);
updateCommand.Parameters.Add(listPriceParameter);
rs.Open();
isRSOpened = true;
while (rs.Read()) {
// Column two in the result set contains the color of the bike
SqlMoney rateOfIncrease = increaseDictionary[rs.GetString(2)];
// Column zero in the result set contains the product ID
productIDParameter.Value = rs.GetInt32(0);
// Column three in the result set contains the current standard cost
standardCostParameter.Value = rs.GetSqlMoney(3) * rateOfIncrease;
// Column four in the result set contains the current list price
listPriceParameter.Value = rs.GetSqlMoney(4) * rateOfIncrease;
// Update the prices of one of the popular bikes based on its color and current prices.
updateCommand.ExecuteNonQuery();
}
}
finally {
if (myConnection != null) {
if (isRSOpened)
rs.Close();
myConnection.Close();
}
}
}
// Helper function which fills the dictonary containing the price increases for each color
static void InitializeIncreaseDictionary() {
// Initialize a dictionary which contains the price increase factors for each color
String[] splitIncreases = priceIncreases.Split(new Char[] { ',' });
for ( int i = 0 ; i < splitIncreases.Length - 1 ; i += 2 ) {
increaseDictionary[splitIncreases[i].Trim()] = SqlMoney.Parse(splitIncreases[i + 1].Trim());
}
}
}
}
Code
-- InstallCS.sql
USE AdventureWorks
GO
IF EXISTS (SELECT [name] from sys.procedures WHERE [name] = N'usp_RSTest')
DROP PROCEDURE usp_RSTest;
GO
IF EXISTS (SELECT [name] from sys.procedures WHERE [name] = N'usp_UpdateProductPrice')
DROP PROCEDURE usp_UpdateProductPrice;
GO
IF EXISTS (SELECT [name] FROM sys.assemblies WHERE [name] = N'TestResultSet')
DROP ASSEMBLY TestResultSet;
GO
IF EXISTS (SELECT [name] FROM sys.assemblies WHERE [name] = N'ResultSet')
DROP ASSEMBLY ResultSet;
GO
-- Add the assembly which contains the CLR methods we want to invoke on the server.
DECLARE @SamplesPath nvarchar(1024);
CREATE ASSEMBLY [ResultSet]
FROM 'C:\ResultSet\ResultSet.dll'
WITH permission_set = safe;
CREATE ASSEMBLY [TestResultSet]
FROM 'C:\ResultSet\TestResultSet.dll'
WITH permission_set = safe;
GO
CREATE PROCEDURE [usp_RSTest]
AS EXTERNAL NAME [TestResultSet].[Microsoft.Samples.SqlServer.TestResultSet].Test
GO
CREATE PROCEDURE usp_UpdateProductPrice(
@ProductID int,
@StandardCost money,
@ListPrice money
)
AS
SET NOCOUNT ON;
UPDATE Production.Product
SET StandardCost = @StandardCost,
ListPrice = @ListPrice
WHERE ProductID = @ProductID;
GO
Code
-- test.sql
USE AdventureWorks
GO
SELECT TOP 10 P.ProductID, P.Name, P.Color, P.StandardCost, P.ListPrice, sum(SOD.OrderQty) FROM Production.Product AS P
JOIN Sales.SalesOrderDetail AS SOD ON P.ProductID = SOD.ProductID
JOIN Production.ProductSubcategory AS PSC ON P.ProductSubcategoryID = PSC.ProductSubcategoryID
JOIN Production.ProductCategory AS PC ON PSC.ProductCategoryID = PC.ProductCategoryID
WHERE PC.Name = 'Bikes'
GROUP BY P.ProductID, P.Name, P.Color, P.StandardCost, P.ListPrice
ORDER BY sum(SOD.OrderQty) desc;
GO
EXEC usp_RSTest
GO
SELECT TOP 10 P.ProductID, P.Name, P.Color, P.StandardCost, P.ListPrice, sum(SOD.OrderQty) FROM Production.Product AS P
JOIN Sales.SalesOrderDetail AS SOD ON P.ProductID = SOD.ProductID
JOIN Production.ProductSubcategory AS PSC ON P.ProductSubcategoryID = PSC.ProductSubcategoryID
JOIN Production.ProductCategory AS PC ON PSC.ProductCategoryID = PC.ProductCategoryID
WHERE PC.Name = 'Bikes'
GROUP BY P.ProductID, P.Name, P.Color, P.StandardCost, P.ListPrice
ORDER BY sum(SOD.OrderQty) desc;
GO
Code
-- Cleanup.sql
USE AdventureWorks
GO
IF EXISTS (SELECT [name] from sys.procedures WHERE [name] = N'usp_RSTest')
DROP PROCEDURE usp_RSTest;
GO
IF EXISTS (SELECT [name] from sys.procedures WHERE [name] = N'usp_UpdateProductPrice')
DROP PROCEDURE usp_UpdateProductPrice;
GO
IF EXISTS (SELECT [name] FROM sys.assemblies WHERE [name] = N'TestResultSet')
DROP ASSEMBLY TestResultSet;
GO
IF EXISTS (SELECT [name] FROM sys.assemblies WHERE [name] = N'ResultSet')
DROP ASSEMBLY ResultSet;
GO