Expand description
Arrow schemas for code graph objects.
Two tables:
- CodeNodes: functions, classes, modules, files — the objects in the code graph
- CodeEdges: calls, imports, inheritance, containment — relationships between objects
§V12a-1 additions (EX-3168)
V12/V13 parity audit: V12/V13 Python had no position metadata and no Rust-specific
node kinds. All position columns and Rust-specific CodeNodeKind variants are new work,
not ports from V12.
Position columns (13–18) are nullable — backward-compatible with existing Parquet files. New Parquet files produced by the Rust tree-sitter parser (EX-3120) will populate them.
Modules§
- edge_
col - Named column indices for the CodeEdges schema.
- node_
col - Named column indices for the CodeNodes schema.
Structs§
Enums§
- Code
Edge Predicate - Kind of relationship between code objects.
- Code
Node Kind - Kind of code object.
Constants§
- CODE_
EMBEDDING_ DIM - Default embedding dimension for code objects.
Functions§
- build_
code_ edges_ batch - Build a RecordBatch of CodeEdges.
- build_
code_ nodes_ batch - Build a RecordBatch of CodeNodes (without embeddings — those are added later).
- code_
edges_ schema - Schema for the CodeEdges table.
- code_
nodes_ schema - Schema for the CodeNodes table.
- extract_
file_ path - Extract file path from a CodeNode ID.