@wildcard-ai/deepcontext/dist/services/SemanticSubChunker.d.ts

Advanced codebase indexing and semantic search MCP server

/**
 * Semantic Sub-Chunker
 * Intelligently splits large code chunks while preserving semantic structure,
 * imports/exports, symbols, and maintaining searchable context.
 */
import { SymbolInfo } from '../types/core.js';
import { ConfigurationService } from './ConfigurationService.js';
export interface CodeChunk {
    id: string;
    content: string;
    filePath: string;
    relativePath: string;
    startLine: number;
    endLine: number;
    language: string;
    symbols: SymbolInfo[];
    imports: Array<{
        module: string;
        symbols: string[];
        line: number;
    }>;
}
export interface SubChunkContext {
    fileHeader: string;
    globalContext: string;
    localContext: string;
}
export interface SemanticSection {
    content: string;
    type: 'header' | 'import' | 'export' | 'class' | 'function' | 'interface' | 'comment' | 'other';
    startLine: number;
    endLine: number;
    symbols: SymbolInfo[];
    dependencies: string[];
    priority: number;
}
export declare class SemanticSubChunker {
    private configurationService;
    private logger;
    private readonly MIN_OVERLAP_SIZE;
    private readonly CONTEXT_WINDOW;
    constructor(configurationService: ConfigurationService);
    /**
     * Main entry point: Split a large chunk into semantic sub-chunks
     */
    splitLargeChunk(chunk: CodeChunk): Promise<CodeChunk[]>;
    /**
     * Parse content into semantic sections with metadata
     */
    private parseSemanticSections;
    /**
     * Detect what type of semantic section a line represents
     */
    private detectSectionType;
    /**
     * Check if this is a natural boundary for section splitting
     */
    private isNaturalBoundary;
    /**
     * Create a semantic section with metadata
     */
    private createSemanticSection;
    /**
     * Assign priority scores to different section types
     */
    private getSectionPriority;
    /**
     * Extract dependencies from content (imports, function calls, etc.)
     */
    private extractDependencies;
    /**
     * Extract global context that should be preserved across all sub-chunks
     */
    private extractGlobalContext;
    /**
     * Create sub-chunks with preserved context and semantic integrity
     */
    private createSubChunksWithContext;
    /**
     * Create overlap context to maintain semantic continuity
     */
    private createOverlapContext;
    /**
     * Create a single sub-chunk with full context
     */
    private createSubChunk;
    /**
     * Validate that sub-chunks maintain semantic quality
     */
    validateSubChunks(subChunks: CodeChunk[]): Promise<{
        isValid: boolean;
        issues: string[];
        metrics: {
            totalSymbols: number;
            averageSize: number;
            contextPreservation: number;
        };
    }>;
}
//# sourceMappingURL=SemanticSubChunker.d.ts.map