dm-tree

What it is and what it does

dm-tree is a library for manipulating nested data structures—dictionaries, lists, tuples, and their combinations—by applying operations uniformly to leaf values while preserving the overall container structure. It generalizes Python's built-in `map` function from flat sequences to arbitrarily nested hierarchies. The core operations are `flatten` (convert a nested structure to a flat list), `map_structure` (apply a function to each leaf), and traversal utilities. The implementation is backed by optimized C++ code designed for performance-critical applications like machine learning pipelines that process large nested tensors or parameter trees.

The package depends on absl-py, attrs, numpy, and wrapt, and targets modern Python (3.10+). It is actively maintained, with recent releases and no known security vulnerabilities. The library is particularly useful in machine learning frameworks where nested structures of tensors or parameters are common, though it applies to any domain working with complex hierarchical data.

Use it for

• Flatten nested parameter dictionaries in machine learning models for batch processing or serialization.
• Apply transformations uniformly across all leaves of a nested tensor structure while preserving hierarchy.
• Traverse and validate deeply nested configuration or data structures in data pipelines.
• Convert between nested and flat representations when interfacing with APIs that expect one or the other.
• Implement recursive operations on tree-like data without manual recursion boilerplate.

Worth the install?

AI-flagged interpretation of the facts on this page. Verify before relying on it.

Install

dm-tree on PyPI

Before you install

Medium install friction due to compiled C++ components requiring platform-specific wheels; however, wheels are available for modern Python versions (3.10–3.14) across macOS, Linux (x86_64 and aarch64), and Windows, and the package is actively maintained with a recent release.

Requires Python 3.10 or later; installation pulls in absl-py, attrs, numpy, and wrapt as runtime dependencies.

License in practice

Licensed under Apache 2.0 (permissive), allowing free use, modification, and distribution in both open-source and commercial projects with minimal restrictions.

Quickstart

pip install dm-tree

import tree
structure = [[1], [[[2, 3]]], [4]]
flat = tree.flatten(structure)  # [1, 2, 3, 4]
mapped = tree.map_structure(lambda v: v**2, structure)  # [[1], [[[4, 9]]], [16]]

Package facts

Evidence: dm_tree-0.1.10-cp310-cp310-macosx_10_9_universal2.whl; dm_tree-0.1.10-cp310-cp310-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl; dm_tree-0.1.10-cp310-cp310-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl; dm_tree-0.1.10-cp310-cp310-win_amd64.whl; dm_tree-0.1.10-cp311-cp311-macosx_10_9_universal2.whl; dm_tree-0.1.10-cp311-cp311-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl; dm_tree-0.1.10-cp311-cp311-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl; dm_tree-0.1.10-cp311-cp311-win_amd64.whl; dm_tree-0.1.10-cp312-cp312-macosx_10_13_universal2.whl; dm_tree-0.1.10-cp312-cp312-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl; dm_tree-0.1.10-cp312-cp312-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl; dm_tree-0.1.10-cp312-cp312-win_amd64.whl; dm_tree-0.1.10-cp313-cp313-macosx_10_13_universal2.whl; dm_tree-0.1.10-cp313-cp313-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl; dm_tree-0.1.10-cp313-cp313-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl; dm_tree-0.1.10-cp313-cp313t-macosx_10_13_universal2.whl; dm_tree-0.1.10-cp313-cp313t-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl; dm_tree-0.1.10-cp313-cp313t-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl; dm_tree-0.1.10-cp313-cp313-win_amd64.whl; dm_tree-0.1.10-cp314-cp314-macosx_10_15_universal2.whl

Tags

More Libraries packages

See also optree · pathable · json-normalize · banal · flatdict · llist · json-flatten · treelib · plucky · glom