/**
 * Indexing pipeline orchestrator.
 *
 * Coordinates the full indexing pipeline:
 * 1. Process all configured sources to generate chunks
 * 2. Generate embeddings for all chunks
 * 3. Upload to Vectorize (unless dry-run)
 * 4. Invalidate search cache (after successful upload)
 *
 * This replaces the mixed logic previously in index.ts, providing
 * cleaner separation between orchestration and source-specific processing.
 */
import type { IndexerConfig } from './config/types.js';
import type { Chunk } from './chunking/types.js';
import type { SourceStats } from './sources/types.js';
/**
 * Options for the indexing pipeline.
 */
export interface IndexOptions {
    /** Preview chunks without uploading */
    dryRun?: boolean;
    /** Show detailed progress */
    verbose?: boolean;
    /** Write chunks to JSON file instead of uploading */
    outputPath?: string;
    /** Maximum pages/files to process per source (for preview runs) */
    maxPages?: number;
    /** Skip cache invalidation after upload */
    skipCacheInvalidation?: boolean;
    /** API URL for cache invalidation (defaults to production) */
    apiUrl?: string;
    /** Cache invalidation token (from env if not provided) */
    cacheInvalidateToken?: string;
    /**
     * Continue processing remaining sources if one source fails.
     * When enabled, errors are collected and reported at the end instead of
     * stopping the entire indexing run.
     * @default false
     */
    continueOnError?: boolean;
    /**
     * Process chunks in batches of this size to reduce memory usage.
     * When set, chunks are embedded and uploaded in batches rather than
     * accumulating all chunks in memory before processing.
     * Recommended for large indexing runs (>10k chunks).
     * @default undefined (process all chunks at once)
     */
    chunkBatchSize?: number;
}
/**
 * Statistics for a single source.
 */
export interface SourceStatistics extends SourceStats {
    /** Source name */
    source: string;
    /** Error message if source processing failed */
    error?: string;
}
/**
 * Result from running the indexing pipeline.
 */
export interface IndexResult {
    /** Total chunks generated across all sources */
    chunks: Chunk[];
    /** Whether chunks were uploaded to Vectorize */
    uploaded: boolean;
    /** Processing statistics by source */
    stats: SourceStatistics[];
    /** Errors from sources that failed (when continueOnError is enabled) */
    errors?: Array<{
        source: string;
        error: string;
    }>;
}
/**
 * Run the indexing pipeline.
 *
 * Processes all configured sources, generates embeddings, and uploads
 * to Vectorize. Supports dry-run mode for previewing chunks.
 *
 * @param config - Indexer configuration
 * @param options - Pipeline options
 * @returns Index result with chunks and statistics
 *
 * @example
 * ```typescript
 * const config = await loadConfig({ config: './config.json' });
 * const result = await runIndex(config, { verbose: true });
 * console.log(`Indexed ${result.chunks.length} chunks`);
 * ```
 */
export declare function runIndex(config: IndexerConfig, options?: IndexOptions): Promise<IndexResult>;
//# sourceMappingURL=orchestrator.d.ts.map