I'm trying to come up with a reusable pattern for updating MongoDB Documents when using Spring Data in conjunction with Spring MVC.
The use case can generally be summarized by:
A document is created in Mongo using repository.save()
Parts of that document are then presented in a Spring MVC editable form.
A user submits updated parts of that document which are then saved.
If I use the repository.save() method in step 3, I will lose any data in the document that was not bound to the form. Making the form responsible for the entire document is fragile so this is where it seems the findAndModify() method of the MongoTemplate comes in handy.
To use the findAndModify() method, I've created Form objects that support a toMap() method which takes the Form object's properties as a Map and removes some of the fields (e.g. class and id). This gets me a Map that contains only the fields that I care about from the Form object. Passing the object ID and this map to an update() method on my customized repository, I build Query and Update objects that I can pass to the findAndModify() method.
Using this approach, I'm able to add fields to my objects easily and only worry about instances when there are fields I don't want to update from a form posting. Document fields not manipulated by the Form should be retained. It still seems slightly convoluted to be using both the Repository and MongoTemplate so I'm wondering if there are better examples for how to handle this. It seems like this should be a consistent pattern when working with Mongo and Spring MVC (at the least).
I've created a sample project showing how I achieve this model on GitHub. The Spock Tests show how "updating" a Document using save() will blow away fields as expected and my update() method.
https://github.com/watchwithmike/diner-data
What are other people doing when dealing with partial updates to Documents using Spring MVC and Spring Data?
If you are taking whatever the user supplies and just shoving that in the database you are running the risk of doing something dangerous like updating or creating data that they shouldn't be able to. Instead, you should first query Mongo to get the most recent version of the document, change any fields (it looks like you are using Groovy so you could loop through all the properties and set them on the new document), and then save the new, complete document.
If you are making small, consistent updates (like increasing the number of votes, or something like that), you could create a custom MongoDB query using the MongoTemplate to do an update to a few fields. Check out the spring-data-mongodb docs for more. You can also add custom methods to the MongoRepository that use the MongoTemplate.
Related
I would like to override the Query Async Method with some Custom code, so I can access an external api, get some data and then use this data to query the db tables to get the results, this should support all the default sync and paging features provided by the base method. I'm using the latest 5.0 DataSync library. Old versions returned a IQueryable so this was easy to do, but now it returns an action result. Is there any solution? Could not find any docs.
e.g. I get a set of guids from api. I query the db tables with these guids and get all the matching rows and return back to client.
I know how to override the method, call external api's, get data and query the db tables, but this provides me with a custom data format not the one that the query async gives by default with paging support.
The way to accomplish this is to create a new repository, swapping out the various CRUD elements with your own implementation. You can use the EntityTableRepository as an example. You will note that the "list" operation just returns an IQueryable, so you can do what you need to do and then return the updated IQueryable.
I have an existing web application that uses EF and POCO objects. I want to improve the client experience by exposing some of my objects through WCF(JSON). I have this working fine but where I am unsure is how to handle derived objects(not sure if that is the correct term) or IEnumerable anonymous objects if you will.
Let's say I have 3 tables structured like so:
Templates
ID
Template
Groups
ID
Group
Instances
ID
TemplateID
GroupID
This is obviously a one-to-many type relationship. I have my navigation properties setup correctly and getting strongly typed object properties works great. However, how do I send serialized anonymous type object(s) over the wire. Like an object that sends all instances that are equal to groupid=1 and include the names of the template and the object.
Am I missing something or do I have to create another class object for WCF that would look like this:
WCF Object
InstanceID
TemplateID
TemplateName
GroupID
GroupName
I guess I could alter my tables to account for this but that seems wrong too. I know that IEnumerable objects can't be serialized and I know that throw away objects are probably not the way to go either. I want to do this the right way but I am not sure how to go about it.
Your suggestions are appreciated.
Regards
Based on what you're doing, I'd suggest looking at OData with WCF Data Services. You state that you want to be able to send all instances where the groupid=1 - OData is great at this type of filtering.
If you're want to stick with your current approach and not use OData, then my first question is why are you sending back anonymous types at all? You can do what you are seeking (all instances with a groupid=1) without sending back an anonymous type. In your select clause you just create new instances of your concrete objects rather than newing up anonymous types. If your query is really just filtering and not executing any meaningful projection with the selct to anonymous type, then I don't see any reason to send back your anonymous type at all.
i'm a total newbie with asp.net mvc and here's my jam:
i have a 3 level list box which selection on box A shows options on box B and selection on box B will show the options for box C.
I'm trying to do the whole thing in asp.net MVC and what i see is that the nerd dinner tutorial uses the ORM method.
so i created a dbml to the database and drag the stored proc inside.
i create a datacontext object but i don't quite know how to connect the result from the stored proce which should be multiple rows of data and make it into a json.
so i can keep all the json data inside the html page and using jquery i could make the selection process faster.
i don't expect the data inside the three boxes to change so often thus i think this method should be quite viable.
Questions:
So how do i get the stored proc part
to return the data as json?
i've noticed some tutorial online
that the json return result part is
at the controller and not at the
model end.
Why is that?
Edit
FYI, i find what i mostly wanted to do here.
For the json part, i referenced here.
Return a JsonResult from your controller action. You may need to coerce the result from your stored procedure into a C# class serializable to Json.
Json conversion should be done in the controller because it's not really part of the domain. More a DTO in the MVVM (Model-View-ViewModel) style.
I want to put context-sensitive, dynamic command options on my asp.net pages.
I tried coding my own command structure but it's not very good, and I'm sure there must be a framework for doing this somewhere I can re-use?
Example:
I have a detailsview for some database object, I want to code in the object class what commands are available, based on the state of the object. I then want a UI object I can place on the webform that will pass commands back to the object when user clicks them, or jump to a different link (e.g. when additional parameters are available).
e.g. form might look like this
Product Details
Name: XXXX product
Price: $1.00
Qty: 1
Commands:
> Edit
> New Stock
> Mark as obsolete
So the commands at the bottom would have very little UI code and pass actions back to the object. For example the New Stock command would jump to a new page to ask for a quantity.
I don't know of the framework, but you could create something yourself. Let's say you are using MVP pattern, and assuming that this is a CRUD application, you could tell each view what type of object it is related to, then annotate you object with operations that are available. Then Presenter could call Service to perform the operation. You could name your methods using some convention so that you can wire it up in a Service. It is a lot of work, and unless you have 100s of views it is not worth while. I am building app that is about that size, and I am in process of creating GenericMVP framework, that would make wiring a breeze.
I have a local SQLite database that contains a tree (as Nested Sets). In an AIR application, I want to display that tree in a tree control and provide means to change the nodes' names and copy, move, add or delete nodes.
Now, I'm hiccupping a little on where to put which code. Obviously, I have a class which will perform operations like load / update / insert / delete against the database. This would load the whole tree into some storage variable and save changes made by the user back to the db.
Should this class be the dataProvider, the dataDescriptor or an extension of the Tree control itself? And when the user requests an operation like adding a node, should that update the dataProvider and let the database handler react on an event, or should it call the database handler's method and then update the dataProvider? I'd say that the latter is better, because it's easier to not update the Tree's data if something goes wrong with the db query.
There's methods to add and remove nodes in the DefaultDataDescriptor and in the Tree class (protected methods in the latter), should I use / extend those or ignore them?
The reason I'm confused about this is that, according to the docs, a Tree control uses the object stored in its 'dataDescriptor' property to parse and manipulate the actual data which is stored inside its 'dataProvider' property.
This seems to make sense, until you realize that unless you subclass it, it's never the Tree control that manipulates data (with the exception of drag&drop, if that's enabled), and it's not the dataDescriptor, either. Rather, in all examples, manipulating data happens directly via the dataProvider object and that triggers event handlers in the Tree control.
What is it I don't get here?
Take a look at mx.controls.treeClasses.HierarchicalCollectionView. It is not part of the public API, but its full source is available as part of Flex. The Tree controller uses this class internally to handle various data sources.