Pentru PMKB, http://pmkb.codeplex.com/ am dorit sa muncesc cit mai usor. Si, pentru ca adopt o strategie gen Table per Type am considerat ca designul bazei de date este cel al claselor.

Daca va uitati la proiectul PMKB_DAL o sa vedeti urmatoarele fisiere .tt

1. Model1.Context.tt – genereaza EFContext si genereaza codul pentru Fluent API, ca de exemplu:

   modelBuilder.Entity<Link_Tag>().ToTable(HelperTables.NameTable(Tables_PMKBEntities.Link_Tag));
       		modelBuilder.Entity<Link_Tag>().HasKey(item => item.LinkTagID);
   
       		modelBuilder.Entity<Link_Tag>()
       				.HasRequired(item => item.Link_LinkID_Link_Tag_LinkID)
       				.WithMany(u => u.Link_LinkID_Link_Tag_LinkID)
       				.HasForeignKey(x => x.LinkID)
       				.WillCascadeOnDelete(false);
   

2. Model1.tt – genereaza codul pentru fiecare clasa + o interfata pentru ea:

    public interface I_Link
       {
       long LinkID{get;set;}
       string LinkName{get;set;}
       string LinkDescription{get;set;}
       string LinkUrl{get;set;}
       Nullable<long> CreationUserID{get;set;}
       Nullable<System.DateTime> LinkCreationDate{get;set;}
       string LinkShortURL{get;set;}
       }
   
       public partial class Link: FastDatabaseQuery.IReceiveVisitor, I_Link
       {
   

3. ITList.tt – genereaza codul pentru clasa Lista – pentru fiecare tabela pot sa vreau sa incarc mai multe rinduri sau sa sterg ceva rapid:

   
    [System.Diagnostics.DebuggerDisplayAttribute("Link_List Count={Count}")]
       public partial class Link_List : ColList<Link>
       {
   //code
    public void Delete_LinkID(long value)
           {
                   base.Conection.Database.ExecuteSqlCommand("delete from " + HelperTables.NameTable(Tables_PMKBEntities.Link) + " where LinkID = {0}",value);
           }
   

   Astfel incit codul devine:

   using (Link_List ll = new Link_List(ConnectionName))
               {
   
                   if (string.IsNullOrEmpty(OrderBy))
                       OrderBy = Link_List.FieldNames.LinkID;
   
                   var iq = ll.LoadFromDB.LoadAllQ();
                   var ordered = Link_FindDB.OrderByField(iq, OrderBy);
                   if (!string.IsNullOrEmpty(Search))
                   {
                       ll.LoadFromDB.AddToCustomPredicate(Link_FindDB.fexp_LinkUrlContainsMultipleDef(Search), false);
                       ll.LoadFromDB.AddToCustomPredicate(Link_FindDB.fexp_LinkDescriptionContainsMultipleDef(Search), false);
                       ll.LoadFromDB.AddToCustomPredicate(Link_FindDB.fexp_LinkNameContainsMultipleDef(Search), false);
                       ordered = ll.LoadFromDB.LoadFindCustomPredicateOrderedQ(OrderBy);
                   }
   
                   return ordered.ToPagedList(pageNumber, PageSize);
   
               }
   

4. Find.tt – genereaza codul pentru cautare in BD ( de exemplu, pentru date pot sa am between) . IN acest exemplu arat fexp_LinkUrlContainsMultipleDef care face Like din Sql :

     public static Func<string, Expression<Func<Link, bool>>> fexp_LinkUrlContainsMultipleDef = value => fexp_LinkUrlContainsMultiple(value, "%");
   
            public static Func<string,string, Expression<Func<Link, bool>>> fexp_LinkUrlContainsMultiple = (value,separator) =>
           	{
           		value=value.ToLower();
           		var arr=value.Split(new string[1]{separator}, StringSplitOptions.RemoveEmptyEntries);
           		switch(arr.Length)//TODO: use a better expression here rahter than based on length
           		{
           			case 1:
           				return (item =>  item.LinkUrl != null  &&  item.LinkUrl.ToLower().Contains(value));
           			case 2:
           				 {
           				  string v1=arr[0].Replace(separator,"");
           				  string v2=arr[1].Replace(separator,"");
           				 return (item =>  item.LinkUrl != null  &&  item.LinkUrl.ToLower().Contains(v1)  &&  item.LinkUrl.ToLower().Contains(v2) );
           				}
           			default:
           				 {
           				   string v1=arr[0].Replace(separator,"");
           				  string v2=arr[1].Replace(separator,"");
           				  string v3=arr[2].Replace(separator,"");
           				return (item =>  item.LinkUrl != null  &&  item.LinkUrl.ToLower().Contains(v1) &&  item.LinkUrl.ToLower().Contains(v2) &&  item.LinkUrl.ToLower().Contains(v3) );
           				}
   
           		}
   
           	};	 
   

5. Resource.tt – genereaza codul pentru fisiere resx – de exemplu , pentru fiecare clasa am nevoie de “edit”, “delete”, “new” , nume de label pentru (aproape) fiecare proprietate

    <data name="Accessed_AddNew" xml:space="preserve"><value>Add new Accessed</value></data>
   <data name="Accessed_List" xml:space="preserve"><value>List of Accessed</value></data>
   <data name="Accessed_Found_Multi" xml:space="preserve"><value>found {0} Accesseds</value></data>
   <data name="Accessed_Found_One" xml:space="preserve"><value>One Accessed found</value></data>
   <data name="Accessed_Found_Zero" xml:space="preserve"><value>No Accessed found</value></data>
   <data name="Accessed_Save" xml:space="preserve"><value>Save Accessed</value></data>
   <data name="Accessed_Save_Error" xml:space="preserve"><value>Error saving Accessed</value></data>
   <data name="Accessed_Delete_Error" xml:space="preserve"><value>Error deleting Accessed</value></data>
   

6. Visitor.tt – daca am nevoie sa generez ceva date despre fisiere. Implementeaza Visitor pattern din Design Patterns.

    public partial class Link: FastDatabaseQuery.IReceiveVisitor, I_Link
   

   public partial class Link_Visitor: FastDatabaseQuery.IPropertiesVisitor
       {
   
            public string Visited(FastDatabaseQuery.IReceiveVisitor i)
            {
   

Bineinteles ca fiecare cod are nevoie de imbunatatiri. De aceea imi trebuie cod intercalat – si, de obicei, profit de partial in (aproape ) toate formele sale:

1. la definitia clasei
De exemplu,clasa Link_OLAP trebuia sa implementeze si interfata I_Link . Asa incit am adaugat un fisier .cs in care am scris

partial class Link_OLAP:I_Link
    {
    }

2. la introducerea unor metode pe care pot sau nu sa le definesc in corpul clasei asociate.
De exemplu, pentru definirea Modelului trebuia sa spun ca LinkID nu este autogenerat( o problema stupida a EF, care incearca sa fie destept)
Cind am facut override la OnModelCreating am definit si urmatoarele:

partial void OnBeginModelCreating(DbModelBuilder modelBuilder,ref bool Continue);
    	partial void OnFinishModelCreating (DbModelBuilder modelBuilder);
    	protected override void OnModelCreating(DbModelBuilder modelBuilder)
    	{
modelBuilder.Conventions.Remove<IncludeMetadataConvention>();
    		bool Continue=true;
    		OnBeginModelCreating(modelBuilder, ref Continue);
    		if(!Continue)
    			return;
//construct default
modelBuilder.Entity<Accessed>().ToTable(HelperTables.NameTable(Tables_PMKBEntities.Accessed));
    		modelBuilder.Entity<Accessed>().HasKey(item => item.AccessedID);
//other code
OnFinishModelCreating(modelBuilder);

    	}

Asa ca am adaugat un nou fisier, am pus partial class si am definit OnFinishModelCreating ( Nu era obligatoriu – de ex., nu am definit OnBeginModelCreating)

public partial class PMKBEntities
    {
        partial void OnFinishModelCreating(System.Data.Entity.DbModelBuilder modelBuilder)
        {
            modelBuilder.Entity<Link_OLAP>().Property(item => item.LinkID).HasDatabaseGeneratedOption(DatabaseGeneratedOption.None);
        }
    }

Deci, daca nu stiti ce sunt fisierele .tt, atunci cititi de la Hanselmann

http://www.hanselman.com/blog/T4TextTemplateTransformationToolkitCodeGenerationBestKeptVisualStudioSecret.aspx

si mergeti mai departe.

Puteti downloada solutia de la http://pmkb.codeplex.com/

Rezumat

Fisierele .tt va fac viata mai usoara, daca aveti definita o structura de baza. Folositi-le!