A collection of util modules for working with Python abstract syntax trees.

In the current state, the two most useful parts are:

1. source_match.GetSource: Allows for “round tripping” between python source and an AST, preserving all formatting.

In more detail: given a node in a Python AST and the text it should match against, attaches .matcher functions to each node in the tree, which know how to return the source that generated them, with formatting (line breaks, spaces, comments, docstrings) preserved. In other words, if we call this on the _ast.Module node with the text that generated the ast, then call module_node.matcher.GetSource(), we will get the exact copy of the source we passed in.

Example:

some_code = open(‘code.py’).read() import ast module_node = ast.parse(some_code) import source_match source_match.GetSource(module_node, some_code) assert some_code == module_node.matcher.GetSource()

This preservation persists even after new nodes are added (they get default formatting) or nodes are removed.

2. create_node.py: This module contains a bunch of functions for creating nodes with reasonable defaults.

NOTE: This is not an official Google project!