Skip to main content

Node

Simple Story

When a tool reads a pupil's source, it does not keep it as text. It turns it into a tree.

The whole file is the trunk, each class is a branch, each method a smaller branch, each statement a twig. A node is any one of those. Working on a tree is far easier than working on text, because the structure is already worked out.

What it is​

A node in an abstract syntax tree is one syntactic construct. The tree is abstract because it drops what does not affect meaning, such as whitespace, and keeps the structure.

The essential difference from the architecture layer: an AST node comes from source, not bytecode. So it can see things the compiler discards, such as which loop construct was written, and it cannot see anything for which no source is available.

In Ares 2​

The api/ast package uses this to make structural assertions about how student code is written, rather than about what it may do at runtime. That answers a different kind of exercise requirement: did they use recursion, rather than may they open this file.

Further reading​