Awx is an acronym for "AWK With eXtensions". Awx is aimed to be a drop in replacement for AWK with some added extensions (this goal is very similar to GNU AWK).

Goals Awx aims to achieve:

• Be fast
• Add features to make AWK more powerful
  • Optional local scoping for variables
  • Default arguments and keyword arguments
  • Negative field references (stealing this from goawk)
  • Add namespaces
• Maintain POSIX AWK compatibility
• Expand on the standard library
• Support reading of common formats
  • CSV
  • JSON
  • XML
  • etc

Awx is currently in active development and not very mature but the following has already been implemented:

• Reading input from standard in and from files
• Passing programs via command line and from files
• Correctly handling the precedence and associativity of expressions
• Operators (unary prefixes, arithmetic, string concatenation, comparisons, match, membership, logical, ternary, assignment, postcrement)
• Recipes (body is executed when the record matches the regex expression)
• Control statements (if, for, while, do while, return, continue, break, next, exit)
• Functions (calling functions as well as defining functions)
• Fieldreferences (currently the field separator is always set to be a space)
• Builtin variables (some)
• Rudimentary benchmarking (measuring total execution time as well as lexing, parsing, execution)
• Profiling with callgrind

Important features that must still be implemented:

• Bug fixes (parsing is still not perfect yet)
  • Empty control statements bodies segfault
• Arrays
• Implementing builtin functions
• POSIX compliance
• Bytecode compilation
• Common optimizations
• Displaying data about the AST in a table (useful for contributors that want to do debugging)

Awx aims to keep dependencies to a minimum. But the following packages are required:

• C++ compiler (clang, g++ or msvc++)
  • Must support C++23
• Make (Used to invoke CMake and scripts)
• Cmake (Main buildsysstem)
• CLI11 (CLI option parsing library)
• tabulate (Text table library)

Tabulate is downloaded using CMake and statically linked but CLI11 is dynamically linked (CMake searches for it on your system). You can install all required dependencies with the following aptitude command:

apt install -y build-essential cmake libcli11-dev

make -j

The binary will then be in build/awx. To get the help manual run awx -h.

I sourced some scripts from Tim Sherwood for the purpose of testing the implementation. I got them from his personal website Some Useful Gawk Scripts. I also took a script from learnxinyminutes;

Awx is not mature at all and I would like to share two other AWK(-like) implementations that inspired me.

1. Frawk is an AWK like language that uses some techniques to increase performance, like static type inference (this means the type is inferred once) and some rudimentary concurrency support. Frawk also has support for reading CSV and TSV files, overall Frawk is a very robust and way more mature implementation.

2. Goawk is a POSIX compliant AWK-interpreter with CSV support. Goawk also supports bytecode compilation as well as interoperability with Go.