I want to use SQL's Like keyword in dynamic LINQ.
The query that I want to make is like this
select * from table_a where column_a like '%search%'
Where the column_a can be dynamically changed to other column etc
In this dynamic LINQ
var result = db.table_a.Where( a=> (a.column_a.Contains("search")) );
But the column can't be dynamically changed , only the search key can
How do we create a dynamic LINQ like
var result = db.table_a.Where("column_a == \"search\"");
That we can change the column and the search key dynamically
This should work for you:
.Where("AColumnName.Contains(#0)", "Criteria")
Create an ExtensionMethods class with this function
public static IQueryable<T> Like<T>(this IQueryable<T> source, string propertyName, string keyword)
{
var type = typeof(T);
var property = type.GetProperty(propertyName);
string number = "Int";
if (property.PropertyType.Name.StartsWith(number))
return source;
var parameter = Expression.Parameter(type, "p");
var propertyAccess = Expression.MakeMemberAccess(parameter, property);
var constant = Expression.Constant("%" + keyword + "%");
MethodCallExpression methodExp = Expression.Call(null, typeof(SqlMethods).GetMethod("Like", new Type[] { typeof(string), typeof(string) }), propertyAccess, constant);
Expression<Func<T, bool>> lambda = Expression.Lambda<Func<T, bool>>(methodExp, parameter);
return source.Where(lambda);
}
And then call it like this:
var result = db.table_a.Like("column_a", "%search%");
http://weblogs.asp.net/rajbk/archive/2007/09/18/dynamic-string-based-queries-in-linq.aspx
Addition:
Use an expression tree
How do I create an expression tree to represent 'String.Contains("term")' in C#?
That is what the dynamic linq library does internally.
I do not believe there is a direct translation to SQL for the LIKE keyword in LINQ. You could build one if you used expression trees, but I haven't gotten that good yet.
What I do is something like this:
using System.Data.Linq.SqlClient;
if (!string.IsNullOrEmpty(data.MailerName))
search = search.Where(a => SqlMethods.Like(a.Mailer.Name, string.Format("%{0}%", data.MailerName)));
where search is the query I'm building and data is the object containing the properties that hold the search criteria. I build the query dynamically by listing all of the possible search criteria in this way, which adds the appropriate Where methods to search.
Maybe a bit late but Another approach is add Extention method that use Contains to simulate Like keyword as :
public static class DbHelpers
{
public static IQueryable<T> Like<T>(this IQueryable<T> source, string propertyName, string propertyValue)
{
var prop = typeof(T).GetProperty(propertyName);
if (prop == null || prop.PropertyType.Name.StartsWith("Int"))
return source;
ParameterExpression parameter = Expression.Parameter(typeof(T), "row");
Expression property = Expression.Property(parameter, propertyName);
Expression value = Expression.Constant(propertyValue);
var containsmethod = value.Type.GetMethod("Contains", new[] { typeof(string) });
var call = Expression.Call(property, containsmethod, value);
var lambda = Expression.Lambda<Func<T, bool>>(call, parameter);
return source.Where(lambda);
}
}
And use of it:
var foo = entity.AsQueryable().Like("Name", "bla bla");
If send PropertyName with type of int, the method return original entity that you passed before to it.
Related
I have a property in a class with Column(TypeName) set to VARCHAR but when the linq uses ToList(), the SQL generated by linq converts to NVARCHAR instead. Is there a way to avoid nvarchar conversion which happens when ToList() method is called ?
var codes = xyz.Where(x => x.IsValid).Select(x => x.Code.ToLower()).ToList();
requests = requests.Where(p => codes.Contains(p.Type.ToLower()));
Property(c => c.Type).HasColumnType("varchar").HasMaxLength(3).IsFixedLength();
As shown above, though the Type property has column type set to VARCHAR but linq uses NVARCHAR when ToList() is used.
You can build your own predicate:
public static class PredicateBuilder
{
private static readonly MethodInfo asNonUnicodeMethodInfo =
typeof(EntityFunctions).GetMethod("AsNonUnicode");
private static readonly MethodInfo stringEqualityMethodInfo =
typeof(string).GetMethod("op_Equality");
public static Expression<Func<TEntity, bool>> ContainsNonUnicodeString<TEntity>(
IEnumerable<string> source,
Expression<Func<TEntity, string>> expression)
{
if (source == null) throw new ArgumentNullException("source");
if (expression == null) throw new ArgumentNullException("expression");
Expression predicate = null;
foreach (string value in source)
{
var fragment = Expression.Equal(
expression.Body,
Expression.Call(null,
asNonUnicodeMethodInfo,
Expression.Constant(value, typeof(string))),
false,
stringEqualityMethodInfo);
if (predicate == null)
{
predicate = fragment;
}
else
{
predicate = Expression.OrElse(predicate, fragment);
}
}
return Expression.Lambda<Func<TEntity, bool>>(predicate,
((LambdaExpression)expression).Parameters);
}
}
See more here: https://learn.microsoft.com/da-dk/archive/blogs/diego/workaround-for-performance-with-enumerable-contains-and-non-unicode-columns-against-ef-in-net-4-0
I have this method on my Dao class:
public List<E> search(String key, Object value) {
EntityManager entityManager = getEntityManager();
entityManager.getTransaction().begin();
List result = entityManager.createQuery("SELECT a FROM "+clazz.getSimpleName()+" a WHERE a."+key+" LIKE '"+value+"%'").getResultList();
entityManager.getTransaction().commit();
entityManager.close();
return result;
}
the sql works fine when the attribute is #Column or a #OneToOne`, but when it's something like that:
#OneToMany(fetch = FetchType.EAGER, cascade=CascadeType.ALL)
#OrderColumn
private List<Titulo> nome;
where the class Titulo has this attributes:
#Id
#GeneratedValue(strategy=GenerationType.IDENTITY)
private Integer id;
#Column
private String idioma;
#Column(length=32)
private String conteudo;
causes this error:
message: left and right hand sides of a binary logic operator were incompatible [java.util.List(org.loja.model.categoria.Categoria.nome) : string]; nested exception is org.hibernate.TypeMismatchException: left and right hand sides of a binary logic operator were incompatible [java.util.List(org.loja.model.categoria.Categoria.nome) : string]
How I can change the method to make work for both types of attributes?
I manage to solve this issue with the approach below, using java reflection to detect the type of the field trying to be queried, and using a proper sql command. Don't know how efficient this can be; if anyone have a better solution to this, feel free to add another answer with it.
public List<E> search(String key, Object value) throws NoSuchFieldException {
EntityManager entityManager = getEntityManager();
entityManager.getTransaction().begin();
List result;
Field field = clazz.getDeclaredField(key);
ParameterizedType listType = (ParameterizedType) field.getGenericType();
Class<?> classElement = (Class<?>) listType.getActualTypeArguments()[0];
String nome = classElement.getSimpleName();
Field field2[] = classElement.getDeclaredFields();
String attr = field2[field2.length - 1].getName();
if(field != null) {
if(field2 != null) {
result = entityManager.createQuery("SELECT a FROM "+clazz.getSimpleName()+" a, "+nome+" b WHERE b."+attr+" LIKE '"+value+"%'").getResultList();
} else {
result = entityManager.createQuery("SELECT a FROM "+clazz.getSimpleName()+" a, "+nome+" b WHERE b LIKE '"+value+"%'").getResultList();
}
} else {
result = entityManager.createQuery("SELECT a FROM "+clazz.getSimpleName()+" a WHERE a."+key+" LIKE '"+value+"%'").getResultList();
}
entityManager.getTransaction().commit();
entityManager.close();
return result;
}
UPDATE
I got one issue with the code above: in the first query (of the three in the if/else), it's always returned all the elements of the table, almost if the LIKE was being ignored.
On a project built with .NET 3.5, I am using LINQ expressions to dynamically generate code at runtime. The LINQ expressions are compiled using the Compile method and stored for later use as predicates with LINQ to objects.
The expressions are sometimes quite complicated and difficult to debug.
Below is an example of an expression viewed through the debugger visualizer in Visual Studio.
{request
=> (Invoke(workEnvelopeHead
=> (workEnvelopeHead.Method = value(Wombl.Scenarios.CannedResponses+<>c_DisplayClass58).pipeline),
request.WorkEnvelope.Head)
And Invoke(body =>
Invoke(value(Wombl.Scenarios.CannedResponses+<>c_DisplayClass78).isMatch,
body.SingleOrDefault()),Convert(request.WorkEnvelope.Body.Any)))}
I would like to be able to optimize expressions like the above so that the value(Wombl.Scenarios.CannedResponses+<>c__DisplayClass58).pipeline expression is replaced with a constant that is the variable's value.
In this particular case, value(Wombl.Scenarios.CannedResponses+<>c__DisplayClass58).pipeline is a reference in the lambda to a variable in the parent scope. Something like:
var pipeline = "[My variable's value here]";
// My lambda expression here, which references pipeline
// Func<RequestType, bool> predicate = request => ........ workEnvelopeHead.Method == pipeline ..........
The original expression, optimized ought to look like:
{request => (Invoke(workEnvelopeHead =>
(workEnvelopeHead.Method = "[My variable's value here]",
request.WorkEnvelope.Head) And Invoke(body => > Invoke(value(Wombl.Scenarios.CannedResponses+<>c__DisplayClass78).isMatch,
body.SingleOrDefault()),Convert(request.WorkEnvelope.Body.Any)))}
How can I make such optimizations at runtime to the LINQ expression, before compiling?
So I went ahead and wrote an expression visitor that replaces the variable references with the actual value. It wasn't so hard to do after all.
Usage:
var simplifiedExpression = ExpressionOptimizer.Simplify(complexExpression);
The class:
It inherits from ExpressionVisitor which came from the code samples on this page because in .NET 3.0 it is internal. In .NET 4.0 the class is public but might require some changes to this class.
public sealed class ExpressionOptimizer : ExpressionVisitor
{
private ExpressionOptimizer()
{
}
#region Methods
public static Expression<TDelegate> Simplify<TDelegate>(Expression<TDelegate> expression)
{
return expression == null
? null
: (Expression<TDelegate>) new ExpressionOptimizer().Visit(expression);
}
private static bool IsPrimitive(Type type)
{
return type.IsPrimitive
|| type.IsEnum
|| type == typeof (string)
|| type == typeof (DateTime)
|| type == typeof (TimeSpan)
|| type == typeof (DateTimeOffset)
|| type == typeof (Decimal)
|| typeof(Delegate).IsAssignableFrom(type);
}
protected override Expression VisitMemberAccess(MemberExpression memberExpression)
{
var constantExpression = memberExpression.Expression as ConstantExpression;
if (constantExpression == null || !IsPrimitive(memberExpression.Type))
return base.VisitMemberAccess(memberExpression);
// Replace the MemberExpression with a ConstantExpression
var constantValue = constantExpression.Value;
var propertyInfo = memberExpression.Member as PropertyInfo;
var value = propertyInfo == null
? ((FieldInfo) memberExpression.Member).GetValue(constantValue)
: propertyInfo.GetValue(constantValue, null);
return Expression.Constant(value);
}
#endregion
}
Question
I'm trying to use the Dynamic Linq Sample from Microsoft with BindingList<T> objects. But it looks like the Dynamic Linq will only work with IQueryable. What's the deal here, why doesn't BindingList<T> implement IQueryable. And is there a way around this?
Background Detail: I have many data sets that I need to dynamically filter at run time. Here is an example:
BindingList<MyObject> list = new BindingList<MyObject>();
MyObject selectedObj = list.FirstOrDefault(o => o.Name == "Master P")
// then later ...
MyObject selectedObj = list.FirstOrDefault(o => o.City == "Boston")
I am trying to make these queries dynamic, so the user can choose from all properties of MyObject to use in the query.
There is an Extension method on BindingList; AsQueryable(). So you can use
list.AsQueryable();
But if you want to search on all criteria could you create a search that uses an instance of MyObject as the search criteria and then generated a result set based on the Criteria in the object using standard link.
For example:
public List<MyObject> Search(MyObject SearchCriteria)
{
BindingList<MyObject> list = new BindingList<MyObject>();
list.Add(new MyObject("Test", "Boston"));
list.Add(new MyObject("Test2", "Atlanta"));
IEnumerable<MyObject> results = list.AsEnumerable();
if (!String.IsNullOrEmpty(SearchCriteria.Name))
results = results.Where(l => l.Name.Contains(SearchCriteria.Name));
if (!String.IsNullOrEmpty(SearchCriteria.City))
results = results.Where(l => l.City.Contains(SearchCriteria.City));
return results.ToList();
}
So in the following, Results1 will have 2 results and Results 2 will have only 1.
List<MyObject> results1 = Search(new MyObject("Test", ""));
List<MyObject> results2 = Search(new MyObject("Test", "Boston"));
I used a simple structure for MyObject as an example in this:
public class MyObject
{
public MyObject(string name, string city)
{
this.Name = name;
this.City = city;
}
public string Name { get; set; }
public string City { get; set; }
}
As part of my project, I am trying to build a web UI where user will select a method and pass the values. My program should be able to call the method dynamically and build a parameter list on runtime to pass it to the method.
I have created a comma separated list (string) of key and value pairs. This key/value pair is nothing but the parameter name and value of my method (methodname stored in a variable). Example: string params = "ID:123;Name:Garry;Address:addressObject;AddressLine:108 Plaza Lane;City:Avenel;State:NJ;Zip:07001;". Where ID and Name are simple string varaibles while Address is user defined type. What follows after Address i.e. AddressLine, City, State and Zip is elements of Address object. And my method definition is
public string GetInfo(string ID, string Name, Address addressObject)
{
//return something;
}
I am dynamically calling the method (GetInfo) that is stored in sMethodName variable using DynamicProxy like :
string sMethodName = "GetInfo";
object result = (object) proxy.CallMethod(sMethodName, arguments);
Challenge is how to pass the argument list dynamically? Till now I am just able to extract the values from the csv variable into NamedValueCollection. Here is the code:
public static void StoreParameterValues(string param)
{
nvc = new NameValueCollection();
param = param.TrimEnd(';');
string[] parameters = param.Split(new char[] { ';' });
foreach (string val in parameters)
{
string[] keyValue = val.Split(new char[] { ':' });
nvc.Add(keyValue[0], keyValue[1]);
}
}
..and here is the code that tries to build the parameter:
string methodName = "GetInfo";
DynamicProxyFactory factory = new DynamicProxyFactory("http://../myservice.svc");
string sContract = "";
foreach (ServiceEndpoint endpoint in factory.Endpoints)
{
sContract = endpoint.Contract.Name;
}
DynamicProxy proxy = factory.CreateProxy(sContract);
string[] values = null;
// Create the parameter list
object[] arguments = new object[nvc.Count];
int i = -1;
foreach (string key in nvc.Keys)
{
values = nvc.GetValues(key);
foreach (string value in values)
{
arguments[++i] = value;
}
}
object result = (object) proxy.CallMethod(methodName, arguments);
The above code works if I have simple primitive types but not sure how can I build the logic for any other userdefined types. How can I create a object dynamically of type stored in a variable? Not sure if I was able to put my question correctly. I hope so :)
Edit: 01/19/2011: Applied the suggestion from Chris - using Reflection instead of ComponentModel.
I have converted the code to make it more generic. This works now for all primitive and custom types (resursion). Code snippet below:
private object BuildParameterList(Type type)
{
object item = new object();
item = Activator.CreateInstance(type);
PropertyInfo[] propArray = type.GetProperties(BindingFlags.Public|BindingFlags.Instance);
for (int i = 0; i < propArray.Length; i++)
{
PropertyInfo pi = (PropertyInfo)propArray[i];
////Check for custom type
if (IsCustomType(pi.PropertyType))
{
object item1 = BuildParameterList(pi.PropertyType);
pi.SetValue(item, item1, null);
}
else
{
if (property.ContainsKey(pi.Name))
{
pi.SetValue(item, Convert.ChangeType(property[pi.Name], pi.PropertyType), null);
}
}
}
return item;
}
But if one of the property is Color (I just tested with Color type, will fail with other system types aswell-i guess), then it fails at the following line. Not sure how to handle system types - Color or something similar.
pi.SetValue(item, Convert.ChangeType(property[pi.Name], pi.PropertyType), null);
Can you not find what types are expected by the method, by inspecting its ParameterInfos:
endpoint.Contract.ContractType.GetMethod(methodName).GetParameters();
and then instantiating the custom types using:
Activator.CreateInstance(parameterType);