Encyclopedia of Computer Science and Technology

awk 33automatic programmingFrom the beginning of the computer age, computer scientistshave grappled with the fact that writing programsin any computer language, even relatively high-level onessuch as FORTRAN or C, requires painstaking attention todetail. While language developers have responded to thischallenge by trying to create more “programmer friendly”languages such as COBOL with its English-like syntax,another approach is to use the capabilities of the computerto automate the task of programming itself. It is truethat any high-level language compiler does this to someextent (by translating program statements into the underlyingmachine instructions), but the more ambitious taskis to create a system where the programmer would specifythe problem and the system would generate the high-levellanguage code. In other words, the task of programming,which had already been abstracted from the machine codelevel to the assembler level and from that level to the highlevellanguage, would be abstracted a step further.During the 1950s, researchers began to apply artificialintelligence principles to automate the solving of mathematicalproblems (see artificial intelligence). For example,in the 1950s Anthony Hoare introduced the definition ofpreconditions and postconditions to specify the states ofthe machine as it proceeds toward an end state (the solutionof the problem). The program Logic Theorist demonstratedthat a computer could use a formal logical calculus to solveproblems from a set of conditions or axioms. Techniquessuch as deductive synthesis (reasoning from a set of programmedprinciples to a solution) and transformation (stepby-steprules for converting statements in a specificationlanguage into the target programming language) allowed forthe creation of automated programming systems, primarilyin mathematical and scientific fields (see also prolog).The development of the expert system (combining aknowledge base and inference rules) offered yet anotherroute toward automated programming (see expert systems).Herbert Simon’s 1963 Heuristic Compiler was anearly demonstration of this approach.ApplicationsSince many business applications are relatively simple inlogical structure, practical automatic principles have beenused in developing application generators that can create,for example, a database management system given adescription of the data structures and the required reports.While some systems output code in a language such as C,others generate scripts to be run by the database managementsoftware itself (for example, Microsoft Access).To simplify the understanding and specification of problems,a visual interface is often used for setting up the applicationrequirements. Onscreen objects can represent itemssuch as data files and records, and arrows or other connectinglinks can be dragged to indicate data relationships.The line between automated program generators andmodern software development environments is blurry. Aprogramming environment such as Visual Basic encapsulatesa great deal of functionality in objects called controls,which can represent menus, lists, buttons, text input boxes,and other features of the Windows interface, as well asother functionalities (such as a Web browser). The VisualBasic programmer can design an application by assemblingthe appropriate interface objects and processing tools, setproperties (characteristics), and write whatever additionalcode is necessary. While not completely automating programming,much of the same effect can be achieved.Further ReadingAndrews, James H. Logic Programming: Operational Semantics andProof Theory. New York: Cambridge University Press, 1992.“Automatic Programming Server.” Available online. URL: http://www.cs.utexas.edu/users/novak/cgi/apserver.cgi. AccessedApril 14, 2007.“Programming and Problem Solving by Connecting Diagrams.”Available online. URL: http://www.cs.utexas.edu/users/novak/cgi/vipdemo.cgi. Accessed April 14, 2007.Tahid, Walid, ed. Semantics, Applications and Implementation ofProgram Generation. New York: Springer-Verlag, 2000.awkThis is a scripting language developed under the UNIXoperating system (see scripting languages) by Alfred V.Aho, Brian W. Kernighan, and Peter J. Weinberger in 1977.(The name is an acronym from their last initials.) The languagebuilds upon many of the pattern matching utilitiesof the operating system and is designed primarily for theextraction and reporting of data from files. A number ofvariants of awk have been developed for other operatingsystems such as DOS.As with other scripting languages, an awk program consistsof a series of commands read from a file by the awkinterpreter. For example the following UNIX commandline:awk -f MyProgram > Reportreads awk statements from the file MyProgram into theawk interpreter and sends the program’s output to the fileReport.Language FeaturesAn awk statement consists of a pattern to match and anaction to be taken with the result (although the pattern canbe omitted if not needed). Here are some examples:{print $1} # prints the first field of every# line of input (since no pattern# is specified)/debit/ {print $2} # print the second field of# every line that contains the# word “debit”if ( Code == 2 ) # if Code equals 2,print $3 # print third field# of each linePattern matching uses a variety of regular expressions familiarto UNIX users. Actions can be specified using a limitedbut adequate assortment of control structures similar to