P. Purdom. A sentence generator for testing parsers. BIT, 12(3):366--375, 1972.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....de nitions, attribute grammars, algebraic speci cations, and logic programs. Such a uni ed style could be helpful to compare coverage criteria, and to realise the potential for additional criteria. So far, coverage criteria were de ned in a rather speci c context. The seminal paper by Purdom [Pur72b] de nes a coverage criterion for rule coverage of a context free grammar, and algorithms for test set generation are given. In logic programming, some coverage criteria have been studied (cf. Den91,Jac96] which seem to have a useful interpretation for other forms of speci cations and programs, ....

....e.g. minimality and completeness, ultimately leading to the notion of a proper coverage criterion. As for the e ectiveness of the speci cations, we rely on basic regular language theory. In Section 5, a number of coverage criteria are developed. We start from of an abstraction of rule coverage [Pur72b] Then, a number of more involved criteria are de ned. A more elaborated article (with proofs, more examples, and hints on applications) is under the way. Acknowledgement The authors are indebted to Guido Wachsmuth for helping us with the implementation of the ideas formalised in the article. We ....

[Article contains additional citation context not shown here]

P. Purdom. A sentence generator for testing parsers. BIT, 12(3):366-375, 1972. See [Pur72a].

....path from it based on simple regular language theory. This path in turn can be completed to a full term. Finally, the term can be passed to the pretty printer. Test set generation according to a coverage criterion is an elaboration of this idea. A well known coverage notion is rule coverage [50]. One can also think of more challenging criteria, e.g. an elaboration of rule coverage, where all rules (say constructors) are covered in all occurrences of the corresponding sort. In Figure 17, we list all regular expressions modelling the coverage of all constructors in the context of the ....

P. Purdom. A sentence generator for testing parsers. BIT, 12(3):366--375, 1972.

....There is a lack of formal, standard criteria, general purpose techniques, and tools. Much of this research is from the early 60s, 70s, and 80s. For example, Purdom describes an algorithm for producing a small set of short sentences so that each production of a grammar is used at least once [37]. Bird and Munoz developed a test case generator for PL I that randomly generated executable test cases and predicted their execution [6] Bauer and Finger describe a test plan generator (TPG) for systems that can be specified using an augmented finite state automaton (fsa) model [3] Bazzichi and ....

....technique is applied to input specifications and hence differs from Purdom and Bazzichi. The technique does not require the user to learn a new specification language or technique, and the technique requires minimal interaction from the user. In this way it differs from Purdom and von Mayrhauser [37,4,5,26,40]. Further, this technique can be used for functional testing as well. Functional testing selects test cases to assess the implementation of specific required functions [19] 14 The input validation testing problem has been decomposed into several sub problems. These are: 1) How to specify the ....

Purdom, P. A sentence generator for testing parsers. BIT, Volume 12, 1972, pp. 366 - 375.

....Another approach uses genetic algorithms to extend an initial (manual) plan to generate longer usage paths [17] The resulting test suite depends on the paths taken by the expert user. Other, more traditional approaches to the generation of black box tests include grammar based methods [23, 4, 16, 18] and state machine representations [5, 10] State machine representations work well for generating sensible sequences of command types, but become cumbersome for generation of both sequencing as well as command details of systems with large and intricate command languages. 2 Category partition ....

P. Purdom. "A Sentence Generator for Testing Parsers", BIT, 12(3), 1972, pp. 366-375.

....systems is based on a variety of grammars or state machine representations. 2. 1 Grammar based Test Generation Grammar based test generation represents each command using a grammar, generates sentences (commands) from the grammar, and runs the list of commands as the test case (for early work see [32, 5]) When generating a test case from a context free grammar, the generator has to decide which grammar productions to use and which choices within a production to select in deriving terminal symbols (the test case) 2.1.1 Generation Assuring Production Coverage Purdom [32] resolves this through ....

....(for early work see [32, 5] When generating a test case from a context free grammar, the generator has to decide which grammar productions to use and which choices within a production to select in deriving terminal symbols (the test case) 2.1. 1 Generation Assuring Production Coverage Purdom [32] resolves this through test criteria rules: Each production in the grammar is used at least once. In addition, the algorithm prefers short sentences when there is choice. In generating sentences, the algorithm uses two types of information, ffl the production rule to use next on any of the ....

[Article contains additional citation context not shown here]

Paul Purdom; "A Sentence Generator for Testing Parsers", BIT, 12 (1972), pp. 366 -- 375.

....grammar recovery but also for other grammar dependent problems, e.g. parser testing, grammar maintenance, and automated software modification. 3 Context dependent branch coverage We discuss the notion of context dependent branch coverage obtained as an essential generalisation of rule coverage [13]. Firstly, rule coverage will be revisited. Secondly, a context dependent generalisation is developed. Finally, the use of the coverage notion is illustrated in grammar recovery. 3.1 Rule coverage Rule coverage simply means that a test set explores all rules of a grammar. It is clear that for ....

....and parser testing. Returning to the application scenario of parser testing in Section 3.1, test set generation automates testing the parser P w.r.t. a reference grammar G. The generative application of context free grammars is reasonably understood. Some fundamental algorithms are developed in [13], e.g. shortest derivations to reach a certain nonterminal, and the derivation of tests sets achieving rule coverage. One approach to test set generation is the following. Given a context free rule, a shortest completion is computed. Thereby, it is possible to compute a small test set of words ....

P. Purdom. A sentence generator for testing parsers. BIT, 12(3):366--375, 1972.

....the support of the specication development process. Designing an attribute grammar for a programming language, it is common to make several mistakes in the attribute grammar. Often the mistake can easily be found by reading sentences which use the erroneous productions. 4. 1 Related work Purdom ( Pur72] gives an eOEcient algorithm to produce a small set of short sentences from a context free grammar so that each production of the grammar is used at least once. He argues that a set of sentences which makes use of all the rules of the grammar has a good chance of exercising most of the ....

....the specifying grammar nor to create a special grammar for this purpose as in [CCD 80] and [BS82] Therefore, the test programs are not only useful for compiler testing, but also for supporting the specication design process. The generated test sets meet the coverage criterion suggested in [Pur72] Due to the simple structure of the generation algorithm it should be easily extended to produce test sets meeting more advanced coverage criteria. The search algorithm is to some extend similar to the backtracking algorithm suggested in [Kas80] The termination problem of the backtracking ....

Paul Purdom. A sentence generator for testing parsers. BIT, 12:366375, July 1972.

....obey safety rules for plant employees. 2.2. 2 Background Using Formal Language for Test Generation The idea of using formal language definitions to automatically generate test cases is not new [BS82] BF79] Early research by Purdom shows how to efficiently generate sentences to test parsers [Pur72] His research concentrated on generating sentences from a context free language such that each production in the grammar is used at least once. His algorithms were presented to make sure production coverage is met with a minimal number of sentences. This method can also be used to debug ....

Paul Purdom. A Sentence Generator for Testing Parsers. BIT, 12(3):366--375, 1972.

....based systems is based on a variety of grammars or state machine representations. Automated test generation for systems with a command language interface represents each command using a grammar, generates commands from the grammar, and runs the list of commands as the test case (for early work see (Purdom, 1972), Bauer and Finger, 1979) When using grammars for test case generation, we also need to address command language semantics (Bazzichi and Spadafora, 1982) Celentano et al. 1980) Duncan and Hutchison, 1981) Ince, 1987) Duncan and Hutchison, 1981) von Mayrhauser and Crawford Hines, ....

....whether based on grammars or state machines, requires making choices during the traversal of the representations. The choices are due to ambiguities as well as the purposeful inclusion of options in the representation. Choice is directed by incorporating selection rules of various types. (Purdom, 1972) integrates coverage rules for grammar productions to reduce choice, while Maurer (Maurer, 1990) uses probabilistic context free grammars that are enhanced by selection rules including permutations, combinations, dynamic probabilities, and Poisson distribution. Thus, value selection is based on ....

P. Purdom. "A Sentence Generator for Testing Parsers", BIT, 12(3), 1972, pp. 366-375.

No context found.

P. Purdom. A sentence generator for testing parsers. BIT, 12(3):366--375, 1972.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC