Crate nusy_codegraph 

nusy-codegraph — Code as a live Arrow object graph.

Parses Python source files via tree-sitter into CodeNodes (functions, classes, modules) and CodeEdges (calls, imports, inheritance, containment). All data stored as Arrow RecordBatches for zero-copy query and versioning via nusy-arrow-git.

Architecture

Python source → tree-sitter CST → CodeNodes + CodeEdges → Arrow RecordBatches
                                    ↓
                             NameResolver (cross-file edge extraction)

Re-exports

pub use edges::NameResolver;

pub use edges::extract_call_edges;

pub use edges::extract_cross_file_edges;

pub use edges::extract_edges;

pub use embeddings::HashEmbeddingProvider;

pub use embeddings::SearchResult;

pub use embeddings::attach_embeddings;

pub use embeddings::embed_nodes;

pub use embeddings::semantic_search;

pub use executor::ExecutionResult;

pub use executor::ExecutorError;

pub use executor::execute_object;

pub use git_tools::CodeConflict;

pub use git_tools::CodeDiffChangeType;

pub use git_tools::CodeDiffEntry;

pub use git_tools::CodeDiffResult;

pub use git_tools::CodeMergeResult;

pub use git_tools::MergeWarning;

pub use git_tools::SmartMergeResult;

pub use git_tools::codegraph_diff;

pub use git_tools::codegraph_merge;

pub use git_tools::smart_merge;

pub use impact::ChangedNode;

pub use impact::ImpactReport;

pub use impact::ImpactStats;

pub use impact::format_impact_report;

pub use impact::impact_analysis;

pub use ingest::IngestResult;

pub use ingest::Language;

pub use ingest::callers_of;

pub use ingest::ingest_directory;

pub use ingest::ingest_files;

pub use ingest::ingest_python_directory;

pub use ingest::nodes_in_file;

pub use ingest_pipeline::GraphViolations;

pub use ingest_pipeline::WorkspaceIngestResult;

pub use ingest_pipeline::discover_workspace_crates;

pub use ingest_pipeline::ingest_workspace;

pub use ingest_pipeline::verify_graph;

pub use ingest_pipeline::write_graph_parquet;

pub use mcp_tools::McpToolError;

pub use mcp_tools::NodeUpdate;

pub use mcp_tools::QueryFilter;

pub use mcp_tools::QueryResult;

pub use mcp_tools::codegraph_add_edge;

pub use mcp_tools::codegraph_query_objects;

pub use mcp_tools::codegraph_remove_edge;

pub use mcp_tools::codegraph_update_object;

pub use metrics::CodebaseMetrics;

pub use metrics::FileCoverage;

pub use metrics::compute_codebase_metrics;

pub use metrics::enrich_with_coverage;

pub use metrics::enrich_with_git_timestamps;

pub use metrics::high_complexity_nodes;

pub use metrics::largest_nodes;

pub use metrics::low_coverage_nodes;

pub use metrics::parse_coverage_json;

pub use module_resolver::RustModuleResolver;

pub use parser::ImportInfo;

pub use parser::ParseError;

pub use parser::ParseResult;

pub use parser::parse_python_file;

pub use python_parser::PythonParseResult;

pub use python_parser::PythonParser;

pub use python_parser::PythonParserError;

pub use python_resolver::PythonModuleResolver;

pub use rename::RenameError;

pub use rename::rename_node;

pub use rust_parser::parse_rust_file;

pub use schema::CodeEdge;

pub use schema::CodeEdgePredicate;

pub use schema::CodeNode;

pub use schema::CodeNodeKind;

pub use schema::build_code_edges_batch;

pub use schema::build_code_nodes_batch;

pub use schema::code_edges_schema;

pub use schema::code_nodes_schema;

pub use schema::extract_file_path;

pub use scip_calls::ScipCallResult;

pub use scip_calls::extract_scip_call_edges;

pub use search::CodeSearch;

pub use search::callees;

pub use search::callers;

pub use search::children_of;

pub use search::find_sources;

pub use search::find_targets;

pub use search::search_nodes;

pub use search::tests_for;

pub use search::transitive_callers;

pub use search::transitive_deps;

pub use semantic_diff::AffectedEdge;

pub use semantic_diff::AffectedReason;

pub use semantic_diff::ChangeClassification;

pub use semantic_diff::DiffStats;

pub use semantic_diff::SemanticDiff;

pub use semantic_diff::SemanticDiffEntry;

pub use semantic_diff::format_semantic_diff;

pub use semantic_diff::semantic_diff;

pub use topo_sort::ParallelismStats;

pub use topo_sort::crate_parallelism_stats;

pub use topo_sort::sort_crates;

pub use topo_sort::sort_crates_parallel;

pub use topo_sort::sort_functions_in_crate;

pub use topo_sort::sort_functions_parallel;

Modules

build_cache

Content-addressed build cache — V12-Spike-2.

cargo_parser

Parse Cargo.toml manifests into CrateManifest structs.

crate_graph

Build a crate-level dependency graph from a Cargo workspace.

crate_schema

Arrow schemas for the crate-level dependency graph.

edges

Edge extraction — build CodeEdges from parsed CodeNodes.

embeddings

Embedding generation and semantic search for code objects.

executor

CodeGraph execution layer — execute code objects from the graph.

git_tools

Git-integrated MCP tools for CodeGraph versioning.

impact

Impact analysis — compute the blast radius of code changes.

ingest

Batch ingestion — walk a directory tree, parse all Python files, build Arrow tables.

ingest_pipeline

High-level ingestion pipeline — workspace-wide self-ingest and graph coherence verification.

mcp_bridge

Generic MCP-to-NATS bridge.

mcp_tools

MCP tool implementations for CodeGraph query and update operations.

metrics

Code metrics computation — LOC, complexity, coverage, git history.

module_resolver

Rust module tree resolution for cross-file edge extraction.

nats_service

CodeGraph NATS service — exposes code graph tools via NatsServiceBuilder.

nats_sync

NATS-based code graph synchronization (EX-3184).

parser

tree-sitter Python parser — extract CodeNodes from Python source files.

python_parser

Python-specific tree-sitter parser — extracts CodeNodes with position metadata.

python_resolver

Python module resolver — maps dotted import paths to file paths.

rename

Graph-native rename — rename a code object, all edges follow.

rust_parser

tree-sitter Rust parser — extract CodeNodes from Rust source files.

schema

Arrow schemas for code graph objects.

scip_calls

SCIP-based call edge extraction.

search

Code search — graph-aware querying of CodeNodes and CodeEdges.

semantic_diff

Semantic diff — enriches object-level diffs with structural context.

test_discovery

Graph-native test discovery — find #[test] functions in the code graph.

topo_sort

Topological sorters for the NuSy crate dependency graph and function call graph.

Traits

EmbeddingProvider

Trait for embedding providers.

Functions

cosine_similarity

Cosine similarity between two vectors.