This project enables z/OS developers to parse JCL and to then walk the parse tree. The default build creates a jcl2jcl program that can be used to just generate JCL from JCL. This is meant for testing and understanding of the code. A real use case would be to walk the parse trees to do some sort of analysis.

The altsrc and alttest and alttestsrc directories are historical and were used for testing the parser was working well. I am changing over to now just generate JCL from JCL so that I can 'diff' the output.

Here is an example of jcl2jcl being used to parse orig.jcl and print it out.

jcl2jcl --input=testsrc/commented.jcl

The code works on z/OS or off. It has been tested on z/OS and on MacOS using clang.

I am not a big fan of JCL, even though I have worked on z/OS forever. I also try to use Unix System Services as much as possible to develop software. jcl2jcl generates JCL from JCL - which can be handy if you want to 'normalize' or 'clean up' JCL.

To install jcl2jcl:

• copy the files to z/OS Unix System Services directory. For this example, we assume it is /u/ibmuser/JCLParser
• cd to the directory (/u/ibmuser/JCLParser)
• install CMake, Ninja, GNU Make, clang if you haven't already and use the provided CMake code.
• Alternately, just compile all the C code in src and link it.
• The resultant program should be called jcl2jcl to use the sample parser.

• Use z/OS Open Tools to download cmake, ninja, gnu make, and install clang.
• export CLANG_ROOT to the root directory where you installed the clang compiler.
• Generate build files if first build:

cd JCLParser
mkdir build
cd build
export CC=${CLANG_ROOT}/usr/lpp/IBM/oelcpp/v2r0/bin/clang
export CXX=${CLANG_ROOT}/usr/lpp/IBM/oelcpp/v2r0/bin/clang++ 
cmake ../

• Re-build code:

cmake --build .

This will create a jcl2jcl binary in your build directory

• Tailor the files in the .vscode directory to suit your needs.
• cd /u/ibmuser/JCLParser
• mkdir build
• cd build
• cmake ..
• cmake --build .

To get started reading the code, begin in jcl2jcl.c, which has main and drives all the functions in the other files.

Basic regression testing can be done after building jcl2jcl with:

• cd tests
• ./regression.sh

There should be no output. As bugs fixed, files with a filetype of skipjcl should be renamed to jcl to get included in regression testing.

The parser is complete as far as I know. Bug reports welcome. Providing interesting tools with the parser is TBD.

Mike Fulton (IBM Canada) is the sole contributor at this point. I am happy to change this :)

The code still needs work. Error messages can (always) be improved and there are missing features people may need. If anyone wants to contribute, please reach out to [email protected] (Mike Fulton)

The code uses the Apache license