Web Testing Papers

1
Web Testing Papers

• Zhenyu Dai

2
Part A 6 papers related to test case generation
3
Paper 1 Testing Web Applications by Modeling
with FSMs

• Anneliese A. Andrews, School of EE and CS,
  Washington State University, Pullman, WA 99164
• Jeff Offutt, Info and Software Engineering,
  George Mason University, Fairfax, VA 22030
• Roger T. Alexander, Computer Science, Colorado
  State University, Fort Collins, CO 80523
• To appear, SOftware SYstems Modeling, Springer,
  2004

4

• Contributions address the problem of black box
  system or application testing of Web
  applications propose a methodology of using FSMs
  to generate test cases (transaction sequences).
• Defects no validation of the effectiveness of
  this methodology (no case study) value of user
  inputs may affect the execution sequences of
  transactions, but this paper do not provide an
  methodology to generate appropriate inputs for a
  test case.

5
Type of Connections

• Static links (HTML -gt HTML)
• Dynamic links (HTML -gt software)
• Dynamically created HTML (software -gt HTML)
• User/time specific GUIs (softwarestate -gt HTML)
• Operational transitions (user)
• Software connections include connections among
  back-end components, such as method calls and
  forwarding.
• Off-site software connections.
• Dynamic connections Both the J2EE and .NET allow
  new web components to be installed dynamically
  during execution.

6
Phase 1 build the model

• 1. partition the web application into clusters.
  (Its very subjective, and the clusters may have
  several sub-levels)
• 2. logical web pages either an entire physical
  web page or the portion of a web page that
  accepts data from the user through an HTML form
  and then sends the data to a specific software
  module.
• 3. build a FSM for each cluster beginning from
  the bottom-level cluster. (input constraints are
  annotated in FSMS.) (each node is a logical web
  page or a finite state machine from lower level,
  each edge is a connection defined above.)
• 4. build the application FSM.

7
Phase 2 generate test cases

• 1. generate partial transaction sequences for
  each FSM by some graph theory. Then connect
  partial sequences to form aggregate sequences.
  (use each choice or base choice combination
  method)
• 2. randomly select some inputs for each aggregate
  sequence.

8
Paper 2 Testing Web Database Applications

• Yuetang Deng, Phyllis Frankl, Jiong Wang
• Technical report TR-CIS-2004-01 (04/28/2004)
  Polytechnic University

9

• White-box testing.
• Extend a database transaction consistency testing
  tool AGENDA (A Test Generator for Relational
  Database Applications).
• Contribution provide an algorithm to generate
  test cases for web application.
• Defects no faults report of applying this
  methodology value of user inputs may affect the
  execution sequences of transactions, but this
  paper do not provide an methodology to generate
  appropriate inputs for a test case.

10

• This tool (AGENDA) is currently targeted toward
  web applications written as Java Servlets using
  JDBC for database access.
• A test case for a web application is a sequence
  of pages to be visited plus the input values to
  be provided to pages containing forms.
• Data based web pages refer to dynamic web pages
  that derive some or all of their content from
  data files or databases.

11
Steps

• 1. useful information such as URL links and
  inputs for each URL is extracted from the
  application source.
• 2. an application graph, where nodes represent
  URLs and edges represent URL links, is generated
  and then simplified according to URL link types.
• 3. some paths through the graph are selected
  based on the cyclomatic complexity theory. Each
  path corresponds to one or more test case.
• 4. for each path, AGENDA is used to generate
  inputs for each URL. The path, along with these
  inputs constitutes a test case. An XML file
  representing the test case is generated.
• 5. the test case in the XML file is automatically
  executed and AGENDA checks the new database state
  and the output pages. (HttpClient is used to
  execute the test cases.)

12
Case Study

• Use TPC-W benchmark, an online book store.
• No faults report.

13
Paper 3 Modeling and Testing Web-based
Applications

• Ye Wu, Jeff Offutt
• Information and Software Engineering Department,
  George Mason University, Fairfax, VA, 22030
• GMU ISE Technical ISE-TR-02-08, November 2002

14

• Contributions provide a way to model the
  internal structure of a web page and model the
  transitions between web pages.
• Defects do not provide a test case generation
  algorithm.

15
Atomic Section

• Atomic section is a static HTML file or a section
  of a server program that prints HTML.
• It has an all-or-nothing property.
• Atomic sections are combined to form more complex
  units.

16
(No Transcript)
17
Composition Rule

• p is a composite section of a server program P
  if
• 1. Basis p is an atomic section.
• 2. Sequence (p ? p1 p2)
• 3. Selection p ? p1 p2
• 4. Aggregation p1 p2
• the composition rule for the above example is P
  ? p1 (p2 p3) p4
• The above representation can be used to model the
  internal structure of individual server
  components.

18
Transition Rules (transition between components)

• In the following, p and q are composite sections
  and s is a servlet or other software component
• 1. Link Transition (p gt q) Invoking a link in p
  causes a transition to q.
• 2. Composite Transition (s ? p) The execution of
  s causes p to be produced and returned to the
  client.
• 3. Operational Transition (p -gt q) The user can
  inject new transitions out of the softwares
  control by pressing the back button or the
  refresh button.

19
Web Application Model

• A web application W is modeled as a triple S, C,
  T.
• S is the start page.
• C is a set of composition rules for each
  component.
• T is a set of transition rules.

20
Test Case Generation

• A derivation is a sequence of transitions that
  begins at the start page, and uses composition
  and transition rules to reach the desired page.
• Each derivation can be used to create a test
  case.
• But no test case generation algorithm is provided
  in this paper.

21
Paper 4 Web Application Slicing

• Filippo Ricca, Paolo Tonella
• ITC-irst
• Centro per la Ricerca Scientifica e Tecnologica
• 38050 Povo (Trento), Italy
• Proceedings of the IEEE International Conference
  on Software Maintenance (ICSM'01), Nov. 2001,
  Florence, Italy, pp. 148-157

22
(No Transcript)
23
Web Application Slicing

• Definition a web application slice is obtained
  from a given set of web pages and scripts by
  removing HTML and script statements according to
  some criterion, so that part of the behavior of
  the initial web application remains.
• A web application slice is computed as a
  transitive closure over all dependences in the
  associated SDG starting from a point of interest.
• 4 types of dependences control dependences, data
  dependences, call dependences, semantic
  dependences.

24
Control Dependence and Data Dependence

• A control dependence holds between two
  PseudoVBScript statements if the former defines a
  scope which directly includes the latter.
• A data dependence holds between two
  PseudoVBScript server side statements or between
  a server side and an HTML statement if the former
  defines the value of a variable which is used by
  the latter, and a definition clear path exists
  between the two.

25
(No Transcript)
26
Call Dependence and Semantic Dependence

• Call dependence between the HTML statement of
  type submit and the associated dynamic page.
• Semantic dependence holds between an informative
  object and a PseudoVBScript statement if the
  former provides information on the latter.

27
Paper 5 Analysis and Testing of Web Applications

• Filippo Ricca, Paolo Tonella
• ITC-irst
• Centro per la Ricerca Scientifica e Tecnologica
• 38050 Povo (Trento), Italy
• In 23rd International Conference on Software
  Engineering (ICSE 01), pages 25-34, Toronto, CA,
  May 2001

28

• White-box testing.
• Contributions provide a test case generation
  algorithm for web applications.
• Defects find no faults when applying this
  methodology on 2 web site value of user inputs
  may affect the execution sequences of
  transactions, but this paper do not provide an
  methodology to generate appropriate inputs for a
  test case.

29
(No Transcript)
30
Static Verification

• Unreachable pages pages in the server site but
  unreachable.
• Ghost pages a link may reference a non-existing
  page.
• Undesirable reaching frames page is loaded into
  the frame when it is should not.
• Data dependence check check the definition and
  usage of each variable in the input set.

31
Dynamic validation

• Page testing every page should be visited at
  least once.
• Hyperlink testing every hyperlink is traversed
  at least once.
• Definition-use testing all navigation paths from
  every definition of a variable to every use of it
  should be exercised.
• All-use testing at least one navigation path
  from every definition of a variable to every use
  of it is exercised.
• All-paths testing every path is traversed at
  least once.

32
Test Case Generation

• A path expression is an algebraic representation
  of the paths in a graph.
• An example of path expression is (e1e3 e2e4).
  It corresponds to a site where the initial page,
  say p1, is followed either by p2 or p3 (edge e1
  or e2). In turn, pages p2 and p3 are connected to
  p1 via e3 and e4 respectively.
• An algorithm is provided to find the minimum
  number of paths from path expression to satisfy a
  criterion.
• Tools RebWeb and TestWeb. Test 2 web site
  wordnet and amazon. Find no faults in wordnet,
  and one abnormal behavior in amazon.

P1
e4
e3
e1
e2
P2
P3
33
Paper 6 Structural Testing of Web Applications

• Chien-Hung Liu, David C. Kung, Pei Hsia,
• Department of Computer Science and Engineering,
  The University of Texas at Arlington, P.O. Box
  19015,
• Arlington, TX 76019-0015
• Chih-Tung Hsu
• Sun Microsystems, Inc. M/S UNWK03-105
• 8000 Central Ave., Newark, CA 94560
• In Proceedings of the 11th Annual International
  Symposium on Software Reliability Engineering,
  pages 84-96, San Jose CA, October 2000.

34

• Contribution provide a data-flow testing
  methodology for web testing.
• Defects no faults report of applying this
  methodology.

35
Data Flow for HTML and XML

• DOM technology enables HTML and XML documents to
  be manipulated by exposing each HTML/XML document
  element as an object.
• The data flow analysis should include document
  elements of HTML and XML.

36
Web Application Test Model (WATM)

• Each component of a web application is modeled by
  an object consisting attributes and operations.
• 3 types of objects client pages, server pages,
  components.
• A component can be an HTML template, a Java
  applet, an ActiveX control, a Java Bean, or any
  program module that interacts with the client
  pages, server pages or other components.
• Attributes can be either program variables or
  document elements.
• Operations can be functions written in scripting
  or programming languages.
• Object Relation Diagram (ORD). 7 relationship
  type request, response, navigation, redirect,
  inheritance, aggregation, association.

37
Generate Data-Flow Test Cases

• Function level test individual functions for the
  variables that have def-use chain limited to a
  single function.
• Function cluster level test a cluster of
  functions within an object for the variables
  whose def-use chains involve more than one
  function in the cluster. A function cluster of an
  object is a set of functions that are related by
  their calling relations.
• Object level test various function invocation
  sequences within an object for the variables
  whose def-use chains can be changed by different
  invoking sequences.
• Object cluster level test a cluster of objects
  for the variables whose def-use chains cross the
  objects in the cluster. An object cluster is a
  set of objects that are associated via message
  passing.
• Application level test application-scope
  variables whose def-use chains cross different
  clients that can access the application.

38
Part B 3 Web Testing Related Papers (no test
case generation)
39
Paper 7 Constructing an Object-Oriented
Architecture for Web Application Testing

• Ji-Tzay Yang, Jiun-Long Huang, Feng-Jian Wang,
  William. C. Chu
• Department of Computer Science and Information
  Engineering, National Chiao Tung University,
  Hsinchu, 300 Taiwan
• Journal of Information Science and Engineering,
  18(1)59-84, Jan. 2002

40
(No Transcript)
41
Testing Architecture 5 subsystems

• The source document analysis subsystem.
• The test management subsystem.
• The test development subsystem.
• The test execution subsystem.
• The test failure analysis subsystem.

42
(No Transcript)
43
(No Transcript)
44

• Please refer to the paper for the rest of
  subsystems.
• Since this paper does not have an testing
  algorithm, I will just skip it.

45
Paper 8 Rigorous and Automatic Testing of Web
Applications

• Xiaoping Jia, Hongming Liu
• School of Computer Science, Telecommunication and
  Information Systems, Depaul University, Chicago,
  Illinois
• In 6th IASTED International Conference on
  Software Engineering and Applications (SEA 2002),
  pages 280-285, Cambridge, MA, Nov. 2002.

46
The Approach

• Using a formal specification language (in XML) to
  specify the testing process, as well as the
  functionality, security and performance of a web
  application.
• NO test case generation algorithm.
• It assumes the test cases are already generated.
  This approach is just to automatically execute
  the test cases and compare the output.

47
Paper 9 Bypass Testing of Web Applications

• Jeff Offutt, Ye Wu, Xiaochen Du, Hong Huang
• Information and Software Engineering
• George Mason University, Fairfax, VA 22030
• 15th International Symposium on Software
  Reliability Engineering (ISSRE 04), Nov. 2004,
  page 187-197

48

• In web application, input validation is often
  performs with scripting languages such as
  JavaScript.
• But the end users can bypass this validation.
• Bypass testing is to detect this type of security
  faults.

49
Appendix
50
Some Ideas

• We may combine control dependence and data
  dependence.
• Input value from user may affect the sequence of
  web pages. So how to generate value for input of
  a test case may be a problem.