CodeParser

CodeParser is a package for parsing Wolfram Language source code as abstract syntax trees (ASTs) or concrete syntax trees (CSTs). CodeParser is useful for inspecting code, formatting code, and instrumenting code (for e.g., coverage reporting or profiling), and much more!

CodeParser has many key features:

• Understands practically entire Wolfram Language syntax.
• Fast native library implementation.
• Tested with combination of suite of hand-written tests and fuzz testing.
• Gracious error handling and recovery

Needs["CodeParser`"]

CodeParse["1+1"]

Out[2]= ContainerNode[String, {CallNode[LeafNode[Symbol, "Plus", <||>], {LeafNode[Integer, "1", <|Source -> {{1, 1}, {1, 2}}|>], LeafNode[Integer, "1", <|Source -> {{1, 3}, {1, 4}}|>]}, <|Source -> {{1, 1}, {1, 4}}|>]}, <||>]

"CodeParser and CodeInspector" on community.wolfram.com

Parsing the Wolfram Language from WTC 2019: Watch Video (youtube)

Parsing the Wolfram Language from WTC 2019: Watch Video (wolfram.com)

Parsing the Wolfram Language from WTC 2019: Download Presentation

Setup

CodeParser is included in Mathematica 12.2 and above.

For older versions, install CodeParser paclet from the public paclet server:

PacletInstall["CodeParser"]

Build and install the CodeParser paclet locally

Using CodeParser

After CodeParser is installed, it can be used.

Needs["CodeParser`"]

CodeParse["1+1"]

Out[2]= ContainerNode[String, {CallNode[LeafNode[Symbol, "Plus", <||>], {LeafNode[Integer, "1", <|Source -> {{1, 1}, {1, 2}}|>], LeafNode[Integer, "1", <|Source -> {{1, 3}, {1, 4}}|>]}, <|Source -> {{1, 1}, {1, 4}}|>]}, <||>]

The input to CodeParse may be a string, a File, or a list of bytes.

Command-line tool (Optional)

An optional codeparser command-line tool is also built and can be used.

cmake -DBUILD_EXE=ON ..
cmake --build . --target codeparser-exe

$cpp/src/exe/codeparser
>>> 1+1
InfixNode[Plus, {LeafNode[Integer, "1", <|Source->{{1, 2}, {1, 2}}|>], LeafNode[Integer, 1, <|Source->{{1, 3}, {1, 4}}|>]}, <|Source->{{1, 1}, {1, 4}}|>]

>>>

Troubleshooting

Make sure that the CodeParser can be found on your system:

Needs["CodeParser`"]

and try a basic example:

CodeParse["1+1"]

You may get LibraryFunction messages:

In[1]:= Needs["CodeParser`"]

In[2]:= CodeParse["1+1"]

During evaluation of In[2]:= LibraryFunction::version: The version number 7 of the library is not consistent with the current or any previous WolframLibraryVersion.

During evaluation of In[2]:= LibraryFunction::initerr: A nonzero error code 7 was returned during the initialization of the library /Users/user/Library/Mathematica/Paclets/Repository/CodeParser-1.6/LibraryResources/MacOSX-x86-64/CodeParser.dylib.

During evaluation of In[2]:= LibraryFunction::libload: The function ConcreteParseBytes_Listable_LibraryLink was not loaded from the file /Users/user/Library/Mathematica/Paclets/Repository/CodeParser-1.6/LibraryResources/MacOSX-x86-64/CodeParser.dylib.

Out[2]= $Failed

This means that CodeParser was built with a newer version of Wolfram System than your system supports.

To fix this, build CodeParser from source with the version of Wolfram System that you will use.

Benchmarks

Some of the benchmarks test large data files. Those files are tracked in this repository to ensure that benchmarks are always run against identical input. Git LFS is used to ensure that a basic checkout of this repository remains small, which is important in CI/CD builds.

To run the benchmarks, first ensure that the large benchmark files have been checked out locally using:

$ git lfs pull --exclude="" --include="*"

This will override the default settings in .lfsconfig.

Then, to begin running the benchmarks, execute:

$ cargo bench

File Overview

• Tests/files/large/ contains files managed by Git LFS. The files in this directory are used by the benchmarks. These files should never be modified, to ensure that benchmark comparisions between different revisions of this repository can be meaningfully compared.