SPRL/I - an interpreter for Informix Stored Procedure regression

Hi all I have written an interpreter for a simple scripting language I call SPRL/I, that will allow one to run a series of canned tests for a group of stored procedures. It is written using GNU flex, bison and ESQL/gcc and is available for download at my web site: http://www.geocities.com/sujitpal/artdb006.html If you are just interested in the source code, there is a listing at: http://www.geocities.com/sujitpal/cdldb008.html The software is free. In case you people think it is useful I will contact Walt to include it in the IIUG software archives. Here is the README file that is included in the tarball, which will probably explain more: The first URL in the mail contains an HTML version of the same thing. ------------------------------------------------8<----------------------------------------------------- SPRL/I - a scripting language for regression testing Informix Stored Procedures -------------------------- INTRODUCTION ------------ SPRL/I is a scripting language which is designed to be used as a scripting language to set up a regression test for a group of stored procedures written in the Informix Stored Procedure Language (SPL). The lexer and parser parse a small grammar which I call SPRL/I (Stored Procedure Regression testing Language for Informix), pronounced sprawl or spiral, depending on your preference. This program takes the tokens and converts them to ESQL/C calls to run the stored procedure, and compare the output with the result specified in the SPRL script file. The SPRL language is described in this document in the Language Specifications section. The rationale for creating a language and a language interpreter to do stored procedure regression testing is to ensure that changes made to a stored procedure to keep it in line with business requirements does not inadverdently break something in some other stored procedure. It is possible to manually check each stored procedure to make sure, but that becomes quite time-consuming as the number of stored procedures grows. SPRL allows you to check each stored procedure by feeding it values and comparing it to expected results, but it does so automatically, thereby resulting in large time-savings. You are also sure that the test plan you created is adhered to each time the regression test occurs. An HTML copy of this document is available at http://www.geocities.com/sujitpal/artdb006.html. LANGUAGE SPECIFICATIONS ----------------------- 1. Comments are indicated by a #. Anything followed by a # sign is ignored till the end of line. This means that you can put inline comments as well as block comments using #. Example: # This is a block comment set $varname = "Value"; # setting varname (inline comment) 2. The SET keyword assigns a value to a variable name. Example: set $varname = "Value"; will assign a value "Value" to the variable $varname. 3. The ECHO keyword prints the string following the ECHO to STDOUT. Example: echo "The quick brown fox jumped over the lazy dog"; will print the string "The quick brown..." to STDOUT. 4. The SYSTEM keyword executes the string following the SYSTEM from the operating system. Example: system "ls -l"; will make the operating system execute the command "ls -l". 5. The DATABASE keyword sets the database to be used for subsequent stored procedure calls. Example: database quix;will make the database quix the current database. It can be overriden by specifying the database from the sprl command line with -d=database_name. 6. The PROC keyword instructs the sprl program to run the stored procedure named in the PROC command. Example: PROC procname ([var1, ...]) [>> $outvar1 [, $outvar2,...]]; where procname is the stored procedure name, [var1, ...] is an optional number of input variables, >> is the assignment to the output variables outvar1 [, outvar2, ...]. Note that you dont need to explicitly name all the output variables. Assume that you have a procedure returning three values, out of which you only need the first variable for later processing, then you would say: PROC procname ([var1, ...]) >> $outvar1; However, if you need only the second output variable, then you would say: PROC procname ([var1, ...]) >> $outvar1, $outvar2; where you would then discard the value of $outvar1. 7. The CMP keyword specifies a string that is expected and which will be compared to the output of the PROC command. It is followed by a string which is a comma-separated sequence of variables and/or values. Example: CMP 'row0:0,$variable,Value 1,Value 2,$variable2'; Note that there are no spaces between the commas. However, if your value contains a space, such as Value 1, then you can put it as shown above. The rown: precedes each result line, where n is the row number. For multiple line results, you need to separate the lines with a \\ (backslash). Example: CMP 'row0:0,$variable1,Value 1\\row1:0,$variable2,Value 2'; 8. A PROC must be followed by one and only one CMP. 9. Variables are indicated by preceding it with a $ (dollar) sign, like Perl or Basic. 10. All commands are terminated by a ; (semi-colon). COMMAND LINE USAGE ------------------ If you type the sprl command with a -h option, it will tell you the command line syntax, like so: prompt> sprl -h Informix Stored Procedure Regression Tester ver. 0.01, written by Sujit Pal, November 1999 Usage: isprt [-h] [-c] [-d=dbname] scriptfile Options: -h : Display this message -c : Check Syntax of file -g=n : Debug mode -d=dbname : Specify database name Using the -c option, you can check the syntax of the SPRL script file without running the stored procedures or doing the compares. Note that this will not check for syntax errors in the stored procedures itself or bad calls (insufficient arguments, etc) to the stored procedures. Using the -g=n option, you can generate a trace and a symbol table dump in case you need to debug the SPRL interpreter. There are 3 debug levels you can specify by setting the value of n. With n=1, the interpreter prints out the stored procedure it is executing currently. With n=2, it prints out a trace of the function names it executes as well as the current stored procedure. With n=3, it prints out the parser trace (same as setting yydebug=1) as well as the above. The default is -g=0, where it does not print out any trace information at all. Using the -d=dbname option, you can override the database specified inside the script file. You can also not have the database specified inside the script file and specify it here instead. The scriptfile name is the scriptfile that you are using to invoke the stored procedures you are testing. ERROR MESSAGES -------------- 1. parse error at line <line_number>, token: <string> This points to a syntax error in the SPRL scriptfile, at the line specified by <line_number>. The token where the parse failed is <string>. 2. Too many input arguments for procedure <procedure_name>, max <m