Simple.OData.Client has a typed and dynamic (and basic) syntax.
I like the typed, but I don't want to build out all my types. In the end I really only need two or so types in the results I get.
But my queries need more types to properly filter the results.
So I want to use the dynamic syntax. But I want to cast the results to classes I have.
I can easily do this manually, but I thought I would see if Simple.OData.Client supports this before I go writing up all that conversion code for each query.
Here is some dynamic syntax code that runs without errors:
client.For(x.Client).Top(10).Select(x.ClientId, x.Name).FindEntriesAsync();
Here is an example of what I had hoped would work (selecting into a new Client object)
client.For(x.Client).Top(10).Select(new Client(x.ClientId, x.Name)).FindEntriesAsync();
But that kind of projection is not supported (I get an "has some invalid arguments" error).
Is there a way to support projection into an existing class when using the dynamic syntax of Simple.OData.Client?
EDIT: The code below works. But it's performance is terrible. I decided to abandon it and write hand written mappers for each type I needed.
This is what I came up with:
dynamic results = oDataClient.For(x.Client).Select(x.ClientId, x.Name).FindEntriesAsync().Result;
var listOfClients = SimpleODataToList<Client>(results);
public List<T> SimpleODataToList<T>(dynamic sourceObjects) where T : new()
{
List<T> targetList = new List<T>();
foreach (var sourceObject in sourceObjects)
{
// This is a dictionary with keys (properties) and values. But no
// matter what sourceObject is passed in, the values will always be
// the values of the first entry in the sourceObjects list.
var sourceProperties = ((System.Collections.Generic.IDictionary<string, object>)sourceObject);
var targetProperties = typeof(Client).GetProperties().Where(prop => prop.CanWrite);
var targetObject = new T();
foreach (var targetProperty in targetProperties)
{
if (sourceProperties.ContainsKey(targetProperty.Name))
{
var sourceValue = GetProperty(sourceObject, targetProperty.Name);
targetProperty.SetValue(targetObject, sourceValue, null);
}
}
targetList.Add(targetObject);
}
return targetList;
}
public static object GetProperty(object o, string member)
{
if (o == null) throw new ArgumentNullException("o");
if (member == null) throw new ArgumentNullException("member");
Type scope = o.GetType();
IDynamicMetaObjectProvider provider = o as IDynamicMetaObjectProvider;
if (provider != null)
{
ParameterExpression param = Expression.Parameter(typeof(object));
DynamicMetaObject mobj = provider.GetMetaObject(param);
GetMemberBinder binder = (GetMemberBinder)Microsoft.CSharp.RuntimeBinder.Binder.GetMember(0, member, scope, new CSharpArgumentInfo[] { CSharpArgumentInfo.Create(0, null) });
DynamicMetaObject ret = mobj.BindGetMember(binder);
BlockExpression final = Expression.Block(
Expression.Label(CallSiteBinder.UpdateLabel),
ret.Expression
);
LambdaExpression lambda = Expression.Lambda(final, param);
Delegate del = lambda.Compile();
return del.DynamicInvoke(o);
}
else
{
return o.GetType().GetProperty(member, BindingFlags.Public | BindingFlags.Instance).GetValue(o, null);
}
}
It was made much harder because normal casts and such for the dynamic objects returned would only give the first object in the list over and over. The GetProperty method works around this limitation.
Related
I'm using OrmLite against an existing SQL Server database that has published stored procedures for access. One of these SPs takes 3 int parameters, but expects that one or another will be null. However, none of the parameters are declared optional.
Here's the code I've tried:
using (IDbConnection scon = myFactory.OpenDbConnection())
{
rowCount = scon.SqlScalar<int>("EXEC myProc #FileID, #FileTypeID, #POID",
new
{
FileID = req.FileId,
FileTypeID = (int?)null,
POID = req.PoId,
});
}
But this produces a SqlException: Must declare the scalar variable "#FileTypeID". Examining the SQLParameterCollection under the covers shows that only two parameters are being generated by OrmLite.
Is it possible to call this SP with a null parameter?
It's not supported with SqlScalar. When you look at the code then you can see that SqlScalar methods from class ServiceStack.OrmLite.OrmLiteReadExtensions execute SetParameters method responsible for adding parameters to query with second parameter(excludeNulls) equal true I don't know why- mythz should answer for this ;).
If you want to fix it then you have change all SqlScalar methods to invoke SetParameters with true and SetParameters method should look like following(must support DBNull.Value not null)
private static void SetParameters(this IDbCommand dbCmd, object anonType, bool excludeNulls)
{
dbCmd.Parameters.Clear();
lastQueryType = null;
if (anonType == null) return;
var pis = anonType.GetType().GetSerializableProperties();
foreach (var pi in pis)
{
var mi = pi.GetGetMethod();
if (mi == null) continue;
var value = mi.Invoke(anonType, new object[0]);
if (excludeNulls && value == null) continue;
var p = dbCmd.CreateParameter();
p.ParameterName = pi.Name;
p.DbType = OrmLiteConfig.DialectProvider.GetColumnDbType(pi.PropertyType);
p.Direction = ParameterDirection.Input;
p.Value = value ?? DBNull.Value; // I HAVE CHANGED THAT LINE ONLY
dbCmd.Parameters.Add(p);
}
}
When you change code then you can set null for parameters in the following way:
var result = db.SqlScalar<int>("EXEC DummyScalar #Times", new { Times = (int?)null });
In my opinion you can describe it as a defect on github and I can make pull request.
I've got data returned from my JavaScript client that just includes the data that has changed. That is, I may have an array with each row containing 10 columns of JSON downloaded, but on the Update, only the data that is returned to me is the data that got updated. On my update, I only want to update those columns that are changed (not all of them).
In other words, I have code like below but because I'm passing in an instance of the "President" class, I have no way of knowing what actually came in on the original JSON.
How can I just update what comes into my MVC3 update method and not all columns. That is, 8 of the columns may not come in and will be null in the "data" parameter passed in. I don't want to wipe out all my data because of that.
[HttpPost]
public JsonResult Update(President data)
{
bool success = false;
string message = "no record found";
if (data != null && data.Id > 0)
{
using (var db = new USPresidentsDb())
{
var rec = db.Presidents.FirstOrDefault(a => a.Id == data.Id);
rec.FirstName = data.FirstName;
db.SaveChanges();
success = true;
message = "Update method called successfully";
}
}
return Json(new
{
data,
success,
message
});
}
rec.FirstName = data.FirstName ?? rec.FirstName;
I would use reflection in this case because the code will be too messy like
if (data.FirstName != null)
rec.FirstName = data.FirstName
.
.
.
and so on for all the fields
Using reflection, it would be easier to do this. See this method
public static void CopyOnlyModifiedData<T>(T source, ref T destination)
{
foreach (var propertyInfo in source.GetType().GetProperties())
{
object value = propertyInfo.GetValue(source, null);
if (value!= null && !value.GetType().IsValueType)
{
destination.GetType().GetProperty(propertyInfo.Name, value.GetType()).SetValue(destination, value, null);
}
}
}
USAGE
CopyOnlyModifiedData<President>(data, ref rec);
Please mind that, this won't work for value type properties.
Can anybody say me what is faster: Array or ArrayList? (ActionScript3)
I tried to find a page about this but didn't find anything.
Thank you.
The ArrayList class is a simple implementation of IList that uses a backing Array as the source of the data. Items in the backing Array can be accessed and manipulated using the methods and properties of the IList interface. Operations on an ArrayList instance modify the data source; for example, if you use the removeItemAt() method on an ArrayList, you remove the item from the underlying Array.
Apparently ArrayList class wraps an Array object - hence a plain Array would be faster than an ArrayList object.
As already stated, Array is faster. Actually it is orders of magnitude faster.
The equivalents of array access are getItemAt and setItemAt.
Implementation:
public function getItemAt(index:int, prefetch:int = 0):Object
{
if (index < 0 || index >= length)
{
var message:String = resourceManager.getString(
"collections", "outOfBounds", [ index ]);
throw new RangeError(message);
}
return source[index];
}
and:
public function setItemAt(item:Object, index:int):Object
{
if (index < 0 || index >= length)
{
var message:String = resourceManager.getString(
"collections", "outOfBounds", [ index ]);
throw new RangeError(message);
}
var oldItem:Object = source[index];
source[index] = item;
stopTrackUpdates(oldItem);
startTrackUpdates(item);
//dispatch the appropriate events
if (_dispatchEvents == 0)
{
var hasCollectionListener:Boolean =
hasEventListener(CollectionEvent.COLLECTION_CHANGE);
var hasPropertyListener:Boolean =
hasEventListener(PropertyChangeEvent.PROPERTY_CHANGE);
var updateInfo:PropertyChangeEvent;
if (hasCollectionListener || hasPropertyListener)
{
updateInfo = new PropertyChangeEvent(PropertyChangeEvent.PROPERTY_CHANGE);
updateInfo.kind = PropertyChangeEventKind.UPDATE;
updateInfo.oldValue = oldItem;
updateInfo.newValue = item;
updateInfo.property = index;
}
if (hasCollectionListener)
{
var event:CollectionEvent =
new CollectionEvent(CollectionEvent.COLLECTION_CHANGE);
event.kind = CollectionEventKind.REPLACE;
event.location = index;
event.items.push(updateInfo);
dispatchEvent(event);
}
if (hasPropertyListener)
{
dispatchEvent(updateInfo);
}
}
return oldItem;
}
There's a LOT of calls and checks involved here. Please note, that _dispatchEvents == 0 is true by default (unless you disableEvents), thus writing in fact is an immense operation.
However ArrayList does provide a lot of feature, that are usefull within flex. A good compormise is to grab the underlying Array (accessible as ArrayList::source), peform your operations, and then reassign it (supposing you have listeners observing that Array).
Also, if you go with Flash Player 10, then Vector will outperform Array.
greetz
back2dos
Array is probably slightly faster or they are equal. All an ArrayList is, is an implementation of iList that uses an... Array as a backing object.
I think I've established that in as3corelib JSON.decode I have no choice but to deserialise to a plain old flex object.
var data:Object = JSON.decode(json);
If I then want to get the data contained in the object into another type I can't use type casting. I have to instantiate a new instance and add the properties manually.
var data:Object = JSON.decode(json);
var model:Model = new Model();
model.name = data.name;
model.notes = data.notes;
A pain and a bit ugly, but I'm guessing this is the price to be paid for going from untyped json to a flex type. My first question is whether my assumption is correct and there is no prettier way to create my model instance with the data contained within the json?
My second question, if so then before I write my own method to do this, is there anything inside the flex api that will take the data object and mixin it's values to my model instance?
Cheers,
Chris
the approach I've always used proved to be part of the AMF3 serialization mechanism in ActionScript.
have a look at IExternalizable and registerClassAlias.
now what I use is the following:
interface ISerializable {
public function getRawData():Object;
public function setRawData(param:Object):void;
}
function registerType(id:String, type:Class):void {
//implementation
}
function getTypeByID(id:String):Class {
//implementation
}
function getTypeID(type:Class):String {
//implementation
}
and to the decoder/encoder you register a class alias.
serialization of an object works as follows:
var raw:Object = model.getRawData();
raw[" type"] = getTypeID(model);
var encoded:String = JSON.encode(raw);
decoding works as follows:
var raw:Object = JSON.decode(raw);
var cl:Class = getTypeByID(raw[" type"]);
if (cl == null) throw new Error("no class registered for type: "+raw[" type"]);
delete raw[" type"];
var model:ISerializable = new cl();
model.setRawData(raw);
you will need to do this recursively on the whole deserialized JSON tree, starting at the leafs.
For cyclic reference, you'll need a trick.
I had an implementation of this somewhere, but I can't find it.
You can loop within the field of you json decoded object and assign them into your model:
function json2model(json:String):Model{
var data:Object = JSON.decode(json);
var m:Model=new Model();
for (var field:String in data) {
if (m.hasOwnProperty(field)) {
m[field] = data[field];
}
}
return m;
}
var model:Model=json2model(json)
or add a static function within your Model if you preffer:
public class Model {
//...
public static function fromJSon(json:String):Model {
var data:Object = JSON.decode(json);
var m:Model=new Model();
for (var field:String in data) {
if (m.hasOwnProperty(field)) {
m[field] = data[field];
}
}
return m;
}
}
}
var model:Model=Model.fromJSon(json);
Im trying to extend the flex ArrayCollection to be able to search for an object containing specific data and give it back.
Here is my function:
public function getItemContaining(value: String): Object {
//Loop through the collection
for each(var i: Object in this) {
//Loop through fields
for(var j: String in i) {
//If field value is equal to input value
if(i[j] == value) {
return i;
}
}
}
//If not found
return null;
}
Problem is j is always null so the second loop never works. So I read flex loop descriptions and actually it should work just fine. What can possibly be the problem?
Try it like this:
for (var name:String in myObject){
trace(name + ":" + myObject[name];
}
Okay that was actually the same you were doing. The error must be in this line:
for each(var i: Object in this) {
Try using this:
for each(var i: Object in this.source) {
My first instinct would be to have a look at data type. You're setting up a loop declaring j:String and the symptom is that j is always null. This suggests to me that Flex is failing to interpret the elements of i as strings. If Flex only recognizes the elements of i as Objects (because all Strings are Objects, and Objects are the lowest common denominator), it would return null for j:String.
Try this for your inner loop:
for(var j: Object in i) {
//If field value is equal to input value
if(i[j] is String && (i[j] as String) == value) {
return i;
}
}
if you are using ArrayCollection as your datasource, you should look at using the IViewCursor interface. You can supply a custom compare function, or supply the fields top compare to. This interface is well documented with examples in adobe/livedocs
var _cursor:IViewCursor;
var _idSortField:SortField;
var _idSort:Sort = new Sort();
_idSortField = new SortField();
_idSortField.compareFunction = this.myCompareFunction;
_idSort.fields = [_idSortField];
myArrayCollection.sort = _idSort;
myArrayCollection.refresh();
_cursor = myArrayCollection.createCursor();
if (_cursor.findAny(search))
return _cursor;
if you are search for a value in a specific property, then its even easier. Here's the link to adobe livedocs on this topic