[Modeling-cvs] ProjectModeling/Modeling/doc/UserGuide DefiningaModel.tex,1.35,1.36

[<sbi...@us...>]

Update of /cvsroot/modeling/ProjectModeling/Modeling/doc/UserGuide
In directory sc8-pr-cvs1:/tmp/cvs-serv16069/Modeling/doc/UserGuide

Modified Files:
	DefiningaModel.tex 
Log Message:
Added doc.: how to model many-to-many relationships (to be continued)

Index: DefiningaModel.tex
===================================================================
RCS file: /cvsroot/modeling/ProjectModeling/Modeling/doc/UserGuide/DefiningaModel.tex,v
retrieving revision 1.35
retrieving revision 1.36
diff -C2 -d -r1.35 -r1.36
*** DefiningaModel.tex	30 Aug 2003 16:46:49 -0000	1.35
--- DefiningaModel.tex	31 Aug 2003 17:22:50 -0000	1.36
***************
*** 1853,1856 ****
--- 1853,2008 ----
  
  
+ \subsection{Modeling many-to-many relationships\label{design-rels-many-to-many}}
+ 
+ Many-to-many relationships are not automatically handled by the framework yet,
+ however they can be modeled and used as such.
+ 
+ We'll first examine how many-to-many relationships are mapped in relational
+ database schema. Then we will give a short example demonstrating how to do
+ this, including the code that you'll have to insert in your classes to handle
+ them.
+ 
+ We will use the following model as an example:
+ \begin{verbatim}
+ +--------+ 0,*                  0,*  +---------+
+ | Person |<-persons------addresses-->| Address |
+ |--------|                           |---------|
+ |  name  |                           | street  |
+ +--------+                           | town    |
+                                      +---------+
+ \end{verbatim}
+ 
+ \subsubsection{General principle: the correlation table\label{design-rels-many-to-many-principle}}
+ 
+ Modeling many-to-many relationships in a relational database schema usually
+ involves designing a \emph{correlation table}, a table which holds the
+ informations about the objects in the two tables, \class{Person} and
+ \class{Address}. Each line in the correlation table is basically a tuple of
+ two elements, one from \class{Address} and one from \class{Person}, so that we
+ can tell that if \code{person_1} and \code{address_2} appear in a single row
+ of the correlation table, we know that those two objects are related to each
+ other.
+ 
+ Now let's integrate the correlation table into our model:
+ \begin{verbatim}
+ +--------+ 0,1        0,*  +----------+ 0,*             0,1  +---------+
+ | Person |<-person---pAs->>| PersAddr |<<-pAs------address-->| Address |
+ |--------|                 |----------|                      |---------|
+ |  name  |                 | FKperson |                      | street  |
+ +--------+                 | FKaddress|                      | town    |
+                            +----------+                      +---------+
+ \end{verbatim}
+ 
+ where:
+ \begin{itemize}
+ 
+ \item \class{PersAddr} is the abbreviation for PersonAddress (the correlation
+       table is usually named after the names of the two tables it correlates)
+ 
+ \item relationships named \code{personAddresses} are abbreviated with
+       \code{pAs}
+ 
+ \end{itemize}
+ 
+ To fully understand what's going on here, let's look at the tables themselves.
+ Suppose tables \class{Address} and \class{Person} have 3 rows each:
+ \begin{verbatim}
+ SELECT * FROM Person;
+  id | name 
+ --- +------
+   1 | p1  
+   2 | p2  
+   3 | p3  
+ 
+ SELECT * FROM Address;
+  id |  street   | town 
+ ----+-----------+------
+   1 | street a1 |     
+   2 | street a2 |     
+   3 | street a3 |     
+ \end{verbatim}
+ 
+ Then, for the following situation:
+ \begin{itemize}
+ \item \code{p1}'s addresses are	\code{[a1, a2]}
+ \item \code{p2}'s addresses are	\code{[a1]}
+ \item \code{p3}'s addresses are	\code{[a1, a3]}
+ \end{itemize}
+ 
+ The correlation table looks like:
+ 
+ \begin{verbatim}
+ SELECT * FROM Person_Address;
+  id | fk_person | fk_address 
+ ----+-----------+------------
+   1 |         1 |          1    # p1 <--> a1 
+   2 |         1 |          2	# p1 <--> a2
+   3 |         2 |          1	# p2 <--> a1
+   4 |         3 |          3	# p3 <--> a3
+   5 |         3 |          1    # p3 <--> a1
+ \end{verbatim}
+ 
+ \subsubsection{A short example\label{design-rels-many-to-many-principles}}
+ 
+ Now that we now how many-to-many relationships are handled, we know what we
+ should do:
+ 
+ \begin{enumerate}
+ 
+ \item define the correlation table as an entity in the model,
+ 
+ \item connect the two tables to the correlation table and back,
+ 
+ \item because we do not want to directly manipulate objects coming from the
+       correlation table, we will write some code so that we can directly
+       access persons from a given address, and addresses from a given person;
+ 
+ \end{enumerate}
+ 
+ We've seen in the previous how the model looks like with the correlation
+ table. Here is the full PyModel:
+ 
+ \makeatletter
+ \def\verbatim@font{\footnotesize\ttfamily}
+ \makeatother
+ \begin{verbatim}
+ from Modeling.PyModel import *
+ 
+ # defaults
+ AString.defaults['width'] = 40
+ Entity.defaults['properties'] = [
+   APrimaryKey('id', isClassProperty=0, isRequired=1, doc='PK')
+ ]
+ 
+ ##
+ _connDict = {'database': 'MM1', 'user': 'postgres', 'host': 'localhost',
+              'password': ''}
+ model = Model('MM1',adaptorName='Postgresql',connDict=_connDict)
+ model.version='0.1'
+ model.entities = [
+   Entity('Person',
+          properties=[ AString('name',isRequired=1) ] ),
+   Entity('Address',
+          properties=[  AString('street', isRequired=1),
+                        AString('town') ] ),
+   Entity('PersonAddress'),
+   ]
+ #---
+ model.associations=[
+   Association('PersonAddress','Person',
+        	       relations=['person','personAddresses'],
+        	       delete=['nullify','cascade'] ),
+   Association('PersonAddress','Address',
+       	      relations=['address','personAddresses'],
+       	      delete=['nullify','deny'] ),
+   ]
+ \end{verbatim}
+ 
+ 
+ 
+ 
+ \makeatletter
+ \def\verbatim@font{\normalsize\ttfamily}
+ \makeatother