XMI and DOT exercises XML Metadata Interchange (XMI) The XML Metadata Interchange (XMI) is a XML based language that allows interchange of model metadata, in particular interchange of UML models. XMI is an Object Management Group (OMG) standard, and the "OMG XML Metadata Interchange (XMI) Specification version 1.2" on which the XMI SDF Syntax in this exercise is based is available from the OMG web site: http://www.omg.org The open source UML modeling tool ArgoUML version 0.24 can be used to create an UML model and export it in the XMI format. Although not necessary to complete this exercise, the ArgoUML application could be used to read and create XMI files for testing. The ArgoUML web site is: http://argouml.tigris.org/ In this exercise only class diagrams with classes (no interfaces) and generalizations (parent/child relations) are used. Each class can have at most 1 super class. Fields and methods can be present in the class diagrams but are not used in this exercise. Graphviz Graph Visualization Software (Graphviz) is a set of tools that allows visualization of graphs described in the DOT Language. Although not necessary to complete this exercise it is encouraged to install the Graphviz open source tools to visualize the generated DOT files in Exercise 3. The DOT Language Syntax and the Graphviz tools are available from the Graphviz web site: http://www.graphviz.org Preparation Extract the xmi-dot.tar archive in an empty work directory and start the Meta Environment from this work directory. Load the module Exercise.sdf. Observe the import-graph. The XMI SDF Syntax is represented by the module Document and the DOT SDF Syntax is represented by the module Dot. The module Transform contains functions that allow conversions between XMI and DOT sorts. Unless indicated to do so, do not make changes to the XMI syntax, DOT syntax or Transform modules. Submit Submit the modified Exercise.sdf, Exercise.asf and Dot,sdf file before June, 20th via Peach. EXERCISE 1 The purpose of exercise 1 is to measure the complexity of a "class diagram" model specified in the XMI language. A number of the functions created in this exercise are also used in exercise 3. The complexity is defined as: complexity = number of classes + number of relations + 2 * maximum length of relations If class B extends class A and class C extends class B, then the number of classes is 3 (A, B and C). The number of relations is 2 (B extends A and C extends B). The maximum length of relations is 2 (C extends B extends A). To calculate the maximum length of relations the relations taken from the XMI file will be stored in a tree. The longest branch is determined from the tree. The length of the longest branch is the maximum length of relations. The following steps are taken in this exercise to create the tree and to determine the maximum length of relations. Step 1: Parse the XMI file and generate a list of all relations. (This is done in Exercise 1b.) Step 2: Order the elements in the list generated in step 1. This forces the first element to be the root element and guarantees that all the other elements in the list can be added to the tree in the order they appear in the list. (This is done in Exercise 1c.) Step 3: Create the tree from the ordered list of all relations. (This is done in Exercise 1c.) Step 4: Determine the length of the longest branch in the tree. (This is done in Exercise 1d.) Exercise 1a In the XMI syntax each class is referenced by a unique XMI-Id. A class is specified by the element with name UML:Class. The unique XMI-Id is set by the attribute with name xmi.id. See the following XMI fragment: <UML:Class xmi.id = '127-0-0-1-6dba6368:111d7e5ea2b:-8000:000000000000077C' name = 'A' visibility = 'public' isSpecification = 'false' isRoot = 'false' isLeaf = 'false' isAbstract = 'false' isActive = 'false'/> The function getClassId returns a list of all class XMI-Id's. Complete the getClassId function in the Exercise Equations. Test the equations by reducing the term file test01.trm, the result should match the file test01.out. The number of classes is the length of the List[[XMI-Id]] returned by this function. Hint: The XMI model used in the term file test01.trm, test02.trm, test06.trm and test07.trm is represented by Figure 1. Figure 1, Overview of the example XMI model. Exercise 1b In the XMI syntax the relations between two classes are specified by the element with name UML:Generalization. This element contains two elements that represent the parent class and child class of the relation. See the following XMI fragment: <UML:Generalization ...> <UML:Generalization.child> <UML:Class xmi.idref = '127-0-0-1-6dba6368:111d7e5ea2b:-8000:0000000000000782'/> </UML:Generalization.child> <UML:Generalization.parent> <UML:Class xmi.idref = '127-0-0-1-6dba6368:111d7e5ea2b:-8000:000000000000077E'/> </UML:Generalization.parent> </UML:Generalization> The sort XMI-Value-Pair is specified in the Exercise Syntax and is used to store the child-class / parent- class relation. The function getRelations returns a list of class relations. Each pair represents a parent/child class relation. Step 1: Complete the getRelations function in the Exercise Equations. Test the equations by reducing the term file test02.trm, the result should match the file test02.out. The number of relations is the length of the Table[[XMI-Value-Pair]] returned by this function. Exercise 1c To calculate the maximum length of relations, the list of relations returned by the getRelations function is stored in a tree. From this tree the longest branch can be retrieved. The list of relations from which the tree is constructed must be sorted to enforce that the first element is a root element and to guarantee that all remaining elements can be added to the tree in the order they appear in the list. The function orderPairs sorts the list of relations in such a way that an XMI-Value-Pair list element with class X as parent, will only occur in the list after an element with class X as child. Step 2: Complete the orderPairs function in the Exercise Equations. The sort Tree is specified in the Exercise Syntax and is used to represents a tree node: <XMI-Value, List[[Tree]]> -> Tree The first element of the pair contains the id of the tree element. The second element contains the list of sub classes of the tree element. This list is empty if there are no sub classes. To create a tree the following functions could be used: createTree: From the XMI-Value-Pair list, create the root tree element and call the addToTree function with as parameters the created tree and the XMI-Value-Pair list from which the root element has been removed. addToTree: For each element in the XMI-Value-Pair list, call the addToTree1 function with the element as parameter. addToTree1: Check if the XMI-Value-Pair element should be added to this tree element. If not, call the addToTree2 function to recursively call the addToTree1 function for all elements contained in the Tree list of this tree element. addToTree2: Call the addToTree1 function for all elements contained in the Tree list. Store the resulting tree objects in the list as they may have been changed. Return this list when the addToTree1 function has been called for all elements. Step 3: Complete the createTree function in the Exercise Equations. Use the suggested "auxiliary" functions: addToTree, addToTree1 and addToTree2. Or define your own function(s). Test the equations by reducing the term file test03.trm, the result should match the file test03.out. The relations are now stored in a tree. Exercise 1d The tree can be parsed to find the longest branch. The following ASF condition can be used to match on a smaller-than condition: ... $Integer1 < $Integer2 == true, ... ====> ... = ... See also the Integers module for all supported conditions: ASF+SDF Library -> basic -> Integers.sdf. The function maxLength returns the length of the longest branch in the tree. Step 4: Complete the maxLength function in the Exercise Equations. Use the existing "auxiliary" function: maxLength1. Or define your own function(s). Test the equations by reducing the term file test04.trm, the result should match the file test04.out. The maximum length of relations is now available. Exercise 1e The complexity: complexity = number of classes + number of relations + 2 * maximum length of relations can now be calculated. The function complexity returns the complexity of an XMI file. Complete the complexity function in the Exercise Equations. Test the equations by reducing the term file test05.trm, the result should match the file test05.out. In this exercise the Meta Environment was used to parse an XMI document, store this data in a new defined tree sort and perform some simple condition based rules EXERCISE 2 The purpose of exercise 2 is to correct an error in the DOT SDF Syntax. The DOT Language contains node statements that describe a specific node and edge statements that specify the relations between nodes. An example of a Dot file is given below. This example contains 4 node statements and 3 edge statements. 1 digraph "Test" 2 { 3 "identifier77C" [shape="record", label="A" ]; 4 "identifier77E" [shape="record", label="B" ]; 5 "identifier780" [shape="record", label="C" ]; 6 "identifier782" [shape="record", label="D" ]; 7 8 "identifier782" -> "identifier77E"; 9 "identifier77E" -> "identifier77C"; 10 "identifier780" -> "identifier77C"; 11 } The edge statements in this example contain only a single edge operation: B -> A; This is supported by the DOT SDF Syntax.