Complexity
In this workshop, you'll be implementing simple algorithms for checking basic properties of code complexity.
Two design patterns are of importance here:
- A Builder pattern, which is used to build up state and then finally emit.
- A Visitor pattern, which is used to abstract the process of visiting a data structure such as abstract syntax tree (AST). The only input you have to provide is what action to perform at each node.
Esprima
Instead of building a scanner and parser by hand like we demonstrated previously. We will use an existing library, esprima, to parse code and create a static analyzer for basic code complexity metrics.
Video
If you have not already watched this video, you can get more background about Esprima here: Watch 5:00-14:30,28:00-34:00.
Interactive AST
Use the demo page to see what a code snippet looks like, in order to help you navigate the data structure.
For example, the following snippet:
function functionName( node )
{
if( node.id )
{
return node.id.name;
}
return "anon function @" + node.loc.start.line;
}
Will appear as following, and:
Can be interacted with here.
Workshop
The repository contains a stub that parses a javascript file and visits each function.
-
Run the program and print all the tokens in an ast.
npm install node analysis.js -
Do a simple calculation
- ParameterCount: The number of parameters for functions
- PackageComplexity: The number of imports used by code.
-
Statements inside a function.
- SimpleCyclomaticComplexity: The number of if statements/loops + 1. For live example of a code complexity calculator, see jscomplexity.
-
Using multiple visitors.
- MaxConditions: The max number of conditions in one statement.
- MaxNestingDepth: The max depth of scopes (nested ifs, loops, etc) -- this one is hard, only expect a few to get to do finish this one.