Module graph_query

Structs§

bfs
communities_from_graph
compute_idf
dfs
filter_graph_by_context
find_node
has_chinese
infer_context_filters
is_searchable
load_graph
load_graph_with_cap
normalize_context_filters
pick_seeds
query_graph_text
query_graph_text_hybrid: Hybrid query: BM25 lexical + optional dense cosine, fused with RRF.
query_terms
query_top_nodes: Returns (node_id, label, source_file) for the top-K seeds from hybrid search.
resolve_context_filters
resolve_query: Load global graph from graph_path, run BM25 query, return formatted source bodies for the top-k matching nodes. No dense embeddings (CLI).
rrf: Reciprocal Rank Fusion over multiple ranked lists. k=60 is standard.
score_nodes
search_tokens
segment_chinese
split_camel: Split camelCase/PascalCase into lowercase tokens. Snake_case returned as-is (already meaningful in BM25). Original token is NOT included — caller already has it in bm25_ranked.
strip_diacritics
subgraph_to_text