Function Repository Resource:

CodeStructure (1.0.1) current version: 1.0.2 »

Source Notebook

Analyze C code in a variety of forms

Contributed by: Wolfram Research

ResourceFunction["CodeStructure"][File[filename]]

gives the syntax tree of the C file filename.

ResourceFunction["CodeStructure"]["string"]

gives the syntax tree of the C program given by string.

ResourceFunction["CodeStructure"]["string", form]

generates a representation of the C program given by string in the format specified by form.

ResourceFunction["CodeStructure"][File[filename],form]

generates a representation of the C file filename in the format specified by form.

Details

The possible types of forms include the following:

"SyntaxTree"

syntax tree of the code as a graph

"SyntaxAnnotation"

annotation of how the syntax tree matches up with a C program

"SourceAnnotation"

annotation of how the syntax tree matches up with the source code of a C program exactly

"TokenAnnotation"

annotation of the tokens of the C program

"CallGraph"

basic call graph of the C program

"FileCallGraph"

basic call graph of the C program where the files are the nodes and the function calls are the edges

The following option can be given for all forms:

"CommandLineArguments"

arguments to give to the compiler

The following option can be given if form is "SyntaxTree" or is absent:

"IncludeTypeInformation"

whether to include type information in the syntax tree

The following option is required if form is "CallGraph" or "FileCallGraph" and filename is a directory:

"BinaryLocation"

location of the binary expected from the build process within the directory

The following options can be given if form is "CallGraph" or "FileCallGraph":

"ClangBinariesDirectory"

location of the directory containing the clang and clang++ binaries (required if they are not on PATH)

"DropExternalFunctions"

whether to drop functions that seem to have been defined outside the project to be analyzed

"NoPostProcessing"

whether to return a raw call graph with no post-processing

"Recursive"

whether to recursively search for C source files when analyzing a directory that does not contain a CMakeLists.txt or Makefile

The following option can be given if form is "CallGraph":

"IncludeFilePrefixes"

whether to include in the label of a function the file it was defined in

The following option can be given if form is "FileCallGraph":

"ShowFunctionCount"

whether to show how many times a file calls a function

If form is "CallGraph" or "FileCallGraph" and filename is a directory, ResourceFunction["CodeStructure"] will attempt to use any build system it can detect. If a CMakeLists.txt is present, ResourceFunction["CodeStructure"] will use filename/codeanalysis-build as the build directory. If a Makefile is present, ResourceFunction["CodeStructure"] will attempt to build the project in the project directory. If neither a CMakeLists.txt nor a Makefile is present, ResourceFunction["CodeStructure"] will attempt to auto-discover C source files and will then attempt to compile them so that they can be analyzed.

Examples

Basic Examples (2)

Create two C source files:

In[1]:=

$Export["main.c", "void func(void);\n\nint main(void)\n{\n\tfunc();\n\treturn 1;\n}\n", "String"]; Export["func.c", "void func(void)\n{\n}\n", "String"];$

Get the call graph of the two C source files and use the directory HeadersFolder as an included directory:

In[2]:=

Out[2]=

Drop the file prefixes by setting the option "IncludeFilePrefixes" to False:

In[3]:=

Out[3]=

Get the call graph of a short C program:

In[4]:=

$ResourceFunction["CodeStructure"]["#include <stdio.h> int main(int argc, char **argv) { (void) printf(\"%s\\n\", \"I am a short function!\"); }", "CallGraph", "DropExternalFunctions" -> False]$

Out[4]=

Scope (6)

Use the form "SyntaxAnnotation" to display how the syntax tree of a C program lines up with its source code:

In[5]:=

$ResourceFunction["CodeStructure", ResourceVersion->"1.0.1"]["/* add.c -- Read a sequence of positive integers and print them \n * out together with their sum. Use a Sentinel value\n * (say 0) to determine when the sequence has terminated.\n */\n\n#include <stdio.h>\n#define SENTINEL 0\n\nint main(void) {\n int sum = 0; /* The sum of numbers already read */\n int current; /* The number just read */\n\n do {\n printf(\"\\nEnter an integer > \");\n scanf(\"%d\", &current);\n if (current > SENTINEL)\n sum = sum + current;\n } while (current > SENTINEL);\n printf(\"\\nThe sum is %d\\n\", sum);\n}\n", "SyntaxAnnotation"]$

Out[5]=

Use the form "CallGraph" to show the call graph of a directory containing a complete C project using its CMakeLists.txt:

In[6]:=

Out[6]=

Use the form "FileCallGraph" to show the call graph of a C project where the files are the nodes and the function calls are the edges:

In[7]:=

Out[7]=

Use Wolfram Language graph functions to further refine the file call graph:

In[8]:=

Out[8]=

Get the syntax tree of a short C program:

In[9]:=

$ResourceFunction["CodeStructure"][ "#include <stdio.h> int main(int argc, char **argv) { (void) printf(\"%s\n\", \"I am a very short program!\"); }"]$

Out[9]=

Display the syntax tree of a C program using the form "SyntaxTree":

In[10]:=

Out[10]=

Display tags for the types of tokens in a C program using the form "TokenAnnotation":

In[11]:=

Out[11]=

Use the form "SourceAnnotation" to display how the syntax tree of a C program lines up with a more exact version of its source code:

In[12]:=

Out[12]=

Possible Issues (3)

Call graph generation will fail if the clang and clang++ binaries are not on PATH, as is typical on Linux, where gcc is the default compiler:

In[13]:=

Out[13]=

Check the value of CodeAnalysis`$BuildError to find the cause of the error:

In[14]:=

Out[14]=

Manually specify the location of the clang and clang++ binaries using the "ClangBinariesDirectory" option:

In[15]:=

Out[15]=

Call graph generation will fail if filename is a directory that contains a CMakeLists.txt that requires certain cache variables to be defined:

In[16]:=

Out[16]=

Check the value of CodeAnalysis`$BuildError to find the cause of the error:

In[17]:=

Out[17]=

To resolve the issue, manually define any required CMake cache variables in the codeanalysis-build subdirectory:

$ cd /Users/wolfram/wstpserver/codenalysis-build

$ cmake -DMATHLINK_INCLUDE_DIRECTORY=/Users/wolfram/mathlink/build/include.

Code analysis can fail or produce a bad result if required command-line arguments are missing:

In[18]:=

$ResourceFunction["CodeStructure"][ "int get_command_line_number(void) { return COMMAND_LINE_NUMBER; }", "CallGraph"]$

Out[18]=

Check the value of CodeAnalysis`$BuildError to find the cause of the error:

In[19]:=

Out[19]=

Pass the definition of the COMMAND_LINE_NUMBER macro to the compiler using the "CommandLineArguments" option:

In[20]:=

$ResourceFunction["CodeStructure"][ "int get_command_line_number(void) { return COMMAND_LINE_NUMBER; }", "CallGraph", "CommandLineArguments" -> {"-DCOMMAND_LINE_NUMBER=42"}]$

Out[20]=

Publisher

Christopher Cooley

Version History

1.0.2 – 07 October 2022
1.0.1 – 21 July 2021

Related Resources

License Information

This work is licensed under a Creative Commons Attribution 4.0 International License

Wolfram Function Repository

CodeStructure (1.0.1) current version: 1.0.2 »

Details