fix: handle missing nodes key in get_leaf_nodes

[fooSynaptic]

Fixes #330

Problem

get_leaf_nodes() crashes with KeyError: 'nodes' when traversing trees built by list_to_tree().

list_to_tree() runs clean_node() on every node. For nodes with no children, clean_node() deletes the nodes key entirely (del node['nodes']), rather than leaving it as an empty list. Later, get_leaf_nodes() checks leaf status via structure['nodes'], which raises KeyError on those nodes.

Fix

Use .get('nodes') instead of direct key access:

if not structure.get('nodes'):

When the key is absent, .get('nodes') returns None (falsy), so the node is correctly treated as a leaf. This matches how other helpers in the same file already access nodes (e.g. format_structure, is_leaf_node).

Tests

Added tests/test_issue_330.py:

• leaf dict without a nodes key is returned as a leaf
• list_to_tree() → get_leaf_nodes() end-to-end traversal does not raise

pytest tests/test_issue_330.py -v — 2 passed locally.

[fooSynaptic]

Hi @KylinMountain — gentle ping when you have a moment.

This PR fixes #330 (KeyError: 'nodes' in get_leaf_nodes after list_to_tree()). I noticed a similar one-line fix was merged in #331, but main still has the direct structure['nodes'] access, so this re-applies the guard and adds regression tests in tests/test_issue_330.py.

Happy to adjust anything per your feedback — thanks for maintaining PageIndex!