Module store 

SQLite storage for chunks and embeddings (sqlx async with sync wrappers)

Provides sync methods that internally use tokio runtime to execute async sqlx operations. This allows callers to use the Store synchronously while benefiting from sqlx’s async features.

Module Structure

• helpers - Types and embedding conversion functions
• chunks - Chunk CRUD operations
• notes - Note CRUD and search
• calls - Call graph storage and queries

Structs

CallGraph

In-memory call graph (forward + reverse adjacency lists). In-memory call graph for BFS traversal

CallerInfo

Information about a function caller (from call graph). Caller information from the full call graph

CallerWithContext

Caller with call-site context for impact analysis. Caller with call-site context for impact analysis

ChunkIdentity

Chunk identity for diff comparison (name, file, line, window info). Chunk identity for diff comparison

ChunkSummary

Summary of an indexed code chunk (function, class, etc.). Chunk metadata returned from search results

IndexStats

Statistics about the index (chunk counts, languages, etc.). Index statistics

ModelInfo

Embedding model metadata. Model metadata for index initialization

NoteSearchResult

A note search result with similarity score. A note search result with similarity score

NoteStats

Statistics about indexed notes. Note statistics (total count and categorized counts)

NoteSummary

Summary of a note (text, sentiment, mentions). Note metadata returned from search results

SearchFilter

Filter and scoring options for search. Filter and scoring options for search

SearchResult

A code chunk search result with similarity score. A search result with similarity score

Store

Thread-safe SQLite store for chunks and embeddings

Enums

StoreError

Store operation errors.

UnifiedResult

Unified search result (code chunk or note). Unified search result (code chunk or note)

Constants

CURRENT_SCHEMA_VERSION

Current database schema version. Schema version for database migrations

EXPECTED_DIMENSIONS

Expected embedding dimensions (768 model + 1 sentiment). Expected embedding dimensions — derived from crate::EMBEDDING_DIM

MODEL_NAME

Name of the embedding model used.