Code chunking is not working due to recent tree-sitter api changes

[fpertl]

https://github.com/docling-project/docling-core/blob/ffe48dbe991bca46c8fb2e3e9b6447dbd4924731/docling_core/transforms/chunker/code_chunking/_utils.py#L157C1-L162C38

patching the _utils file with

...
from tree_sitter import Node, Tree, QueryCursor
...
def _query_tree(language, tree: Tree, query: str):
    """Query a tree-sitter tree with the given query string."""
    if not language:
        return []
    q = language.query(query)
    return QueryCursor(q).captures(tree.root_node)

seems to fix the issue.

[dosubot]

The patch works because recent tree-sitter API changes moved the captures method from the query object to the QueryCursor class. Updating _query_tree to use QueryCursor(q).captures(tree.root_node) restores compatibility and fixes code chunking across supported languages in docling-core. This change only affects code under docling_core/transforms/chunker/code_chunking; other parts of the codebase do not use tree-sitter and are unaffected. Make sure your environment uses tree-sitter <0.25.0 and Python >=3.10 for full compatibility, and consider running the test suite after applying the patch to catch any future API changes.
See details in the code and recent PRs.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[dolfim-ibm]

@fpertl thanks, do you want to submit the patch?

[fpertl]

Thanks, but I don't think I'll get to this any time soon.