@vohongtho.infotech/code-intel 0.8.0 → 1.0.0

package/dist/index.d.ts

@@ -66,6 +66,19 @@ declare function createKnowledgeGraph(): KnowledgeGraph;
 declare function generateNodeId(kind: NodeKind, filePath: string, qualifiedName: string): string;
 declare function generateEdgeId(source: string, target: string, kind: string): string;
 
+declare class DbManager {
+    private db;
+    private conn;
+    private dbPath;
+    private readOnly;
+    constructor(dbPath: string, readOnly?: boolean);
+    init(): Promise<void>;
+    query(cypher: string): Promise<Record<string, unknown>[]>;
+    execute(cypher: string): Promise<void>;
+    close(): void;
+    get isOpen(): boolean;
+}
+
 /** Initialize web-tree-sitter (idempotent). */
 declare function initParser(): Promise<void>;
 /**
@@ -206,6 +219,10 @@ interface PhaseResult {
     status: PipelinePhaseStatus;
     duration: number;
     message?: string;
+    /** v1.0.0 profiling — heap MB before phase (set by orchestrator) */
+    memoryBeforeMB?: number;
+    /** v1.0.0 profiling — heap MB after phase (set by orchestrator) */
+    memoryAfterMB?: number;
 }
 interface PipelineContext {
     workspaceRoot: string;
@@ -232,6 +249,8 @@ interface PipelineContext {
     summarize?: boolean;
     /** LLM provider config used by the summarize phase. */
     llmConfig?: LLMConfig;
+    /** v1.0.0: when true, orchestrator captures per-phase memory and writes profile.json */
+    profile?: boolean;
 }
 interface Phase {
     name: string;
@@ -312,6 +331,57 @@ interface SearchResult {
 declare function textSearch(graph: KnowledgeGraph, query: string, limit?: number): SearchResult[];
 declare function reciprocalRankFusion(...rankings: SearchResult[][]): SearchResult[];
 
+/**
+ * bm25-index.ts  — Epic 2: Pre-Built BM25 Inverted Index (v1.0.0)
+ *
+ * Strategy:
+ *  - Built at analysis time (post-pipeline), stored in `.code-intel/bm25.db` (better-sqlite3).
+ *  - Loaded into memory on `serve` startup: replaces linear O(n) scan.
+ *  - Incremental updates: only changed nodes' terms are rewritten.
+ *  - LIMIT pushdown: applies limit before sorting the full score list.
+ *
+ * Tables:
+ *   bm25_index(term TEXT PK, postings TEXT)     — term→[{nodeId,tf}] JSON
+ *   bm25_doclen(node_id TEXT PK, doclen INT)     — document length per node
+ *   bm25_nodemeta(node_id TEXT PK, …)            — name/kind/filePath/snippet
+ *   bm25_meta(key TEXT PK, value TEXT)           — avgdl, docCount
+ */
+
+declare class Bm25Index {
+    private readonly dbPath;
+    /** In-memory inverted index (populated after `load()`). */
+    private readonly invertedIndex;
+    private readonly docLengths;
+    private readonly nodeMeta;
+    private avgdl;
+    private docCount;
+    private _loaded;
+    constructor(dbPath: string);
+    get isLoaded(): boolean;
+    /**
+     * Build the inverted index from a KnowledgeGraph and persist to SQLite.
+     * Called once at analysis time after the main pipeline completes.
+     */
+    build(graph: KnowledgeGraph): void;
+    /**
+     * Load the full inverted index into memory.
+     * Called once on `serve` startup.
+     */
+    load(): void;
+    /**
+     * BM25 search. LIMIT pushdown: scores only the posting lists for query terms,
+     * then partial-sorts to return only the top `limit` results.
+     */
+    search(query: string, limit: number): SearchResult[];
+    /**
+     * Incrementally update index for a set of changed/added nodes.
+     * Only terms that overlap with the changed nodes are rewritten.
+     * Works even if `load()` was not called (reads affected terms directly from DB).
+     */
+    updateNodes(nodes: CodeNode[]): void;
+}
+declare function getBm25DbPath(workspaceRoot: string): string;
+
 declare function createMcpServer(graph: KnowledgeGraph, repoName: string, workspaceRoot?: string): Server;
 declare function startMcpStdio(graph: KnowledgeGraph, repoName: string, workspaceRoot?: string): Promise<void>;
 
@@ -360,18 +430,6 @@ declare function startHttpServer(graph: KnowledgeGraph, repoName: string, port?:
     lastEventAt: number | null;
 }): Promise<HttpServerInstance>;
 
-declare class DbManager {
-    private db;
-    private conn;
-    private dbPath;
-    constructor(dbPath: string);
-    init(): Promise<void>;
-    query(cypher: string): Promise<Record<string, unknown>[]>;
-    execute(cypher: string): Promise<void>;
-    close(): void;
-    get isOpen(): boolean;
-}
-
 /**
  * Load graph into DB using bulk CSV COPY — dramatically faster than
  * individual CREATE statements (10-100× speedup for large repos).
@@ -508,4 +566,4 @@ declare function queryGroup(group: RepoGroup, query: string, limit?: number): Pr
 
 declare function mergeSearchResults(...perRepoResults: SearchResult[][]): SearchResult[];
 
-export { AstCache, BindingTracker, type CallKind, type CallSite, type ClusterResult, type Contract, type ContractLink, DbManager, type EntryPoint, type FileSet, type FlowTrace, type GroupMember, type GroupQueryResult, type GroupSyncResult, type HeritageInfo, type ImportBinding, type ImportInfo, type ImportResolutionResult, type IndexMetadata, type KnowledgeGraph, type LanguageModule, type MroStrategy, type Phase, type PhaseResult, type PipelineContext, type PipelineRunResult, type QueryCapture, type QueryMatch, type RepoEntry, type RepoGroup, type Scope, type ScopeBinding, type SearchResult, addBinding, addClustersToGraph, addMember, buildCallEdges, buildHeritageEdges, classifyCall, clusterPhase, computeMRO, createApp, createKnowledgeGraph, createMcpServer, createScope, deleteGroup, detectCommunities, detectOverrides, findEntryPoints, flowPhase, generateEdgeId, generateNodeId, getAllLanguageModules, getDbPath, getLanguage, getLanguageModule, getParser, groupExists, initParser, isTreeSitterAvailable, listGroups, loadGraphToDB, loadGroup, loadMetadata, loadRegistry, loadSyncResult, mergeSearchResults, parsePhase, parseSource, queryGroup, reciprocalRankFusion, removeMember, removeRepo, resolveBinding, resolveImports, resolvePhase, runPipeline, runQuery, runQueryMatches, saveGroup, saveMetadata, saveSyncResult, scanPhase, startHttpServer, startMcpStdio, structurePhase, syncGroup, textSearch, topologicalSort, traceFlow, upsertRepo, validateDAG };
+export { AstCache, BindingTracker, Bm25Index, type CallKind, type CallSite, type ClusterResult, type Contract, type ContractLink, DbManager, type EntryPoint, type FileSet, type FlowTrace, type GroupMember, type GroupQueryResult, type GroupSyncResult, type HeritageInfo, type ImportBinding, type ImportInfo, type ImportResolutionResult, type IndexMetadata, type KnowledgeGraph, type LanguageModule, type MroStrategy, type Phase, type PhaseResult, type PipelineContext, type PipelineRunResult, type QueryCapture, type QueryMatch, type RepoEntry, type RepoGroup, type Scope, type ScopeBinding, type SearchResult, addBinding, addClustersToGraph, addMember, buildCallEdges, buildHeritageEdges, classifyCall, clusterPhase, computeMRO, createApp, createKnowledgeGraph, createMcpServer, createScope, deleteGroup, detectCommunities, detectOverrides, findEntryPoints, flowPhase, generateEdgeId, generateNodeId, getAllLanguageModules, getBm25DbPath, getDbPath, getLanguage, getLanguageModule, getParser, groupExists, initParser, isTreeSitterAvailable, listGroups, loadGraphToDB, loadGroup, loadMetadata, loadRegistry, loadSyncResult, mergeSearchResults, parsePhase, parseSource, queryGroup, reciprocalRankFusion, removeMember, removeRepo, resolveBinding, resolveImports, resolvePhase, runPipeline, runQuery, runQueryMatches, saveGroup, saveMetadata, saveSyncResult, scanPhase, startHttpServer, startMcpStdio, structurePhase, syncGroup, textSearch, topologicalSort, traceFlow, upsertRepo, validateDAG };