Module chunk 

Tree-sitter based code chunking with sliding-window fallback.

Parses source files into ASTs and extracts semantic chunks at function, class, and method boundaries. For files without recognized semantic structure (or very large fallback chunks), splits into overlapping sliding windows for uniform embedding sizes.

Structs

ArchivedCodeChunk

An archived CodeChunk

ChunkConfig

Runtime configuration for the chunking pipeline.

CodeChunk

A semantic chunk extracted from a source file.

CodeChunkResolver

The resolver for an archived CodeChunk

Functions

build_scope_chain

Walk up the AST parent chain collecting structural container names.

chunk_file

Extract semantic chunks from a source file.

chunk_text

Split source text into overlapping sliding windows.

extract_signature

Extract the function/method signature from a definition node.

minify_whitespace

Reduce indentation waste for embedding by normalizing whitespace.