Parse the output of the linux `tree` command to json.

tree_to_list.py

"""
Parse the output of the linux `tree` command to a flat list of files

Example::

    python tree_to_list.py tree.txt
"""

import argparse
import logging
import re
from dataclasses import dataclass, field
from functools import lru_cache
from pathlib import Path
from typing import Optional, List, Tuple

logging.basicConfig(level=logging.INFO)


@dataclass
class Node:
    """
    A node of a directory tree with references to parent and children as well as
    the path and depth.
    """
    path: Optional[Path] = None
    parent: Optional["Node"] = None
    children: List["Node"] = field(default_factory=list)
    depth: int = 0

    def to_dict(self):
        """
        Convert the tree under the current node to a dict mapping paths to lists
        of children.
        """
        if len(self.children) == 0:
            return str(self.path)
        return {str(self.path): [child.to_dict() for child in self.children]}

    lru_cache()
    def full_path(self):
        """
        Get the full path for the node.
        """
        if self.parent is None:
            return self.path
        return self.parent.full_path() / self.path


def parse_tree_output(tree_path: str) -> Tuple[Node, List[Path]]:
    """
    Parse the output of the linux `tree` command stored in `tree_path` and
    return a `Node` representing the parsed tree and a list of paths.
    """
    paths = []
    with open(tree_path) as f:
        # Assume the root directory is on the first line
        root = f.readline().strip()
        tree = parent = node = Node(path=Path(root))

        # Parse lines one by one
        for line in f.readlines():
            # Split the tree formatting prefix and the path for lines like:
            # │   │       │   ├── 1.51.51.5_rotated.4dfp.ifh
            # TODO: any corner cases missed with this oversimple regex?
            match = re.match("(.*?── )(.*)", line)
            if match is None:
                logging.warning(f"Line {repr(line)} didn't match")
                continue

            prefix, path = match.groups()
            path = Path(path.strip())
            depth = len(prefix) // 4

            # Determine nesting level relative to previous node
            if depth > node.depth:
                parent = node
            elif depth < node.depth:
                for _ in range(depth, node.depth):
                    parent = parent.parent

            # Append to tree at the appropriate level
            node = Node(path, parent=parent, depth=depth)
            parent.children.append(node)

            # Append full path to list
            paths.append(node.full_path())
    return tree, paths


if __name__ == "__main__":
    parser = argparse.ArgumentParser()
    parser.add_argument(
        "tree_text",
        type=str,
        help="path to text file containing tree output",
    )
    args = parser.parse_args()
    tree, paths = parse_tree_output(args.tree_text)
    print("\n".join([str(p) for p in paths]))

note that tree 1.7 and up have json output built in: tree -J will generate output that can be directly loaded by anything that knows how to parse JSON.