The last post ended with us just getting started on stumbling into text template file customization, a task that required a Visual Studio extension (Tangible T4 Editor) to even have a chance at completing. Despite
the benefits of
the Tangible T4 Editor, I still had a hard time putting together a solid text template that would be easy to explain. This is mostly due to
the way
the files allow you to mix code (encapsulated in <# #>) with straight-up text to generate. It is effective to be sure, but not very readable. Nevertheless, I will try and explain what was accomplished in my custom tt file, though
the details of which are not really
the point of this article (my way of saying
dont criticize my crappy code, and certainly
dont use it in any somewhat real application. You may become dumber just by looking at this code. You have been warned really
the footnote I should put at
the end of all of my blog posts). To begin with, there were two basic requirements that I needed
the code generator to satisfy: Reading one to many entity framework files, and using
the entities that were found to write one to many class files. Thankfully, using
the Entity Object Generator as a starting point gave us an example on how to do exactly that by using
the MetadataLoader and EntityFrameworkTemplateFileManager you include references to these items and use them like so: // Instantiate an entity framework file reader and file writer MetadataLoader loader = new MetadataLoader(this); EntityFrameworkTemplateFileManager fileManager = EntityFrameworkTemplateFileManager.Create(this); // Load
the entity model metadata workspace MetadataWorkspace metadataWorkspace = null; bool allMetadataLoaded =loader.TryLoadAllMetadata("MFL.tt", out metadataWorkspace); EdmItemCollection ItemCollection = (EdmItemCollection)metadataWorkspace.GetItemCollection(DataSpace.CSpace); // Create an IO class to contain
the 'get' methods for all entities in
the model fileManager.StartNewFile("MFL.IO.gen.cs"); Next, we want to be able to loop through all of
the entities found in
the model, and then each property for each entity so we can generate classes and methods for each.
The code for that is blissfully simple: // Iterate through each entity in
the model foreach (EntityType entity in ItemCollection.GetItems<EntityType>().OrderBy(e => e.Name)) { // Iterate through each primitive property of
the entity foreach (EdmProperty edmProperty in entity.Properties.Where(p => p.TypeUsage.EdmType is PrimitiveType && p.DeclaringType == entity)) { // TODO: Create properties } // Iterate through each relationship of
the entity foreach (NavigationProperty navProperty in entity.NavigationProperties.Where(np => np.DeclaringType == entity)) { // TODO: Create associations } } There really isnt anything more advanced than that going on in
the text template
the only thing I had to blunder through was realizing that if you want
the generator to interpret a line of code (such as our iterations above), you need to enclose
the code in <# and #> while if you want
the generator to interpret
the VALUE of code, such as putting
the entity name into
the class name, you need to enclose
the code in <#= and #> like so: public partial class <#=entity.Name#> To make a long story short, I did a lot of repetition of
the above to come up with a text template that generates a class for each entity based on its properties, and a set of IO methods for each entity based on its relationships.
The two work together to provide lazy-loading for hierarchical data (such getting Team.Players) so it should be pretty intuitive to use on a front-end. This text template is available here you can tweak
the inputFiles array to load one or many different edmx models and generate
the basic xml IO and class files, though it will probably only work correctly in
the simplest of cases, like our MFL model described in
the previous post. Additionally, there is no validation, logging or error handling which is something I want to handle later by stumbling through
the enterprise library 5.0.
The code that gets generated isnt anything special, though using
the LINQ to XML feature was something very new and exciting for me I had only worked with XML in
the past using
the DOM or XML Reader objects along with XPath, and
the LINQ to XML model is just so much more elegant and supposedly efficient (something to test later). For example,
the following code was generated to create a Player object for each Player node in
the XML: return from element in GetXmlData(_PlayerDataFile).Descendants("Player") select new Player { Id = int.Parse(element.Attribute("Id").Value) ,ParentName = element.Parent.Name.LocalName ,ParentId = long.Parse(element.Parent.Attribute("Id").Value) ,Name = element.Attribute("Name").Value ,PositionId = int.Parse(element.Attribute("PositionId").Value) }; It is all done in one line of code, no looping needed. Even though GetXmlData loads
the entire xml file just like
the old XML DOM approach would have, it is supposed to be much less resource intensive. I will definitely put that to
the test after we develop a user interface for getting at this data. Speaking of
the data where IS
the data? Weve put together a pretty model and a bunch of code around it, but we
dont have any data to speak of. We can certainly drop to our favorite XML editor and crank out some data, but if it doesnt totally match our model, it will not load correctly. To help with this, Ive built in a method to generate xml at any given layer in
the hierarchy. So for us to get
the closest possible thing to real data, wed need to invoke MFL.IO.GenerateTeamXML and save
the results to file. Doing so should get us something that looks like this: <Team Id="0" Name="0"> <Player Id="0" Name="0" PositionId="0"> <Statistic Id="0" PassYards="0" RushYards="0" Year="0" /> </Player> </Team> Sadly, it is missing
the Positions node (havent thought of a way to generate lookup xml yet) and
the data itself isnt quite realistic (well, as realistic as MFL data can be anyway). Lets manually remedy that for now to give us a decent starter set of data. Note that this is TWO xml files Lookups.xml and Teams.xml: <Lookups Id=0> <Position Id="0" Name="Quarterback"/> <Position Id="1" Name="Runningback"/> </Lookups> <Teams Id=0> <Team Id="0" Name="Chicago"> <Player Id="0" Name="QB Bears" PositionId="0"> <Statistic Id="0" PassYards="4000" RushYards="120" Year="2008" /> <Statistic Id="1" PassYards="4200" RushYards="180" Year="2009" /> </Player> <Player Id="1" Name="RB Bears" PositionId="1"> <Statistic Id="2" PassYards="0" RushYards="800" Year="2007" /> <Statistic Id="3" PassYards="0" RushYards="1200" Year="2008" /> <Statistic Id="4" PassYards="3" RushYards="1450" Year="2009" /> </Player> </Team> </Teams> Ok, so we have some data, we have a way to read/write that data and we have a friendly way of representing that data. Now, what remains is
the part that I have been looking forward to
the most: present
the data to
the user and give them
the ability to add/update/delete, and doing so in a way that is very intuitive (easy) from a development standpoint.Did you know that DotNetSlackers also publishes .net articles written by top known .net Authors? We already have over 80 articles in several categories including Silverlight. Take a look: here.
