raggrep 0.4.0 → 0.5.0

package/dist/domain/entities/config.d.ts

@@ -35,6 +35,12 @@ export interface Config {
 export declare const DEFAULT_IGNORE_PATHS: string[];
 /**
  * Default file extensions to index.
+ *
+ * Note: Each module filters for its own supported extensions.
+ * - language/typescript: .ts, .tsx, .js, .jsx, .mjs, .cjs, .mts, .cts
+ * - data/json: .json
+ * - docs/markdown: .md
+ * - core: all remaining extensions
  */
 export declare const DEFAULT_EXTENSIONS: string[];
 /**

package/dist/domain/services/chunking.d.ts

@@ -0,0 +1,66 @@
+/**
+ * Text Chunking Service
+ *
+ * Provides generic text chunking strategies for indexing.
+ * These are language-agnostic and work with any text content.
+ */
+import type { ChunkType } from "../entities";
+/**
+ * Represents a chunk of text with location information.
+ */
+export interface TextChunk {
+    /** The text content */
+    content: string;
+    /** 1-based start line number */
+    startLine: number;
+    /** 1-based end line number */
+    endLine: number;
+    /** The type of chunk */
+    type: ChunkType;
+    /** Optional name for the chunk */
+    name?: string;
+}
+/**
+ * Options for line-based chunking.
+ */
+export interface ChunkingOptions {
+    /** Lines per chunk (default: 30) */
+    chunkSize?: number;
+    /** Overlap between chunks (default: 5) */
+    overlap?: number;
+    /** Minimum lines to create multiple chunks (default: chunkSize) */
+    minLinesForMultipleChunks?: number;
+}
+/** Default lines per chunk */
+export declare const DEFAULT_CHUNK_SIZE = 30;
+/** Default overlap between chunks */
+export declare const DEFAULT_OVERLAP = 5;
+/**
+ * Split text into overlapping chunks based on line boundaries.
+ *
+ * This is a generic chunking strategy that works with any text content.
+ * It creates overlapping chunks to ensure context is preserved across
+ * chunk boundaries.
+ *
+ * @param content - The text content to chunk
+ * @param options - Chunking options
+ * @returns Array of text chunks
+ */
+export declare function createLineBasedChunks(content: string, options?: ChunkingOptions): TextChunk[];
+/**
+ * Create a single chunk from entire content.
+ * Useful for small files or when chunking isn't needed.
+ *
+ * @param content - The text content
+ * @returns A single file chunk
+ */
+export declare function createSingleChunk(content: string): TextChunk;
+/**
+ * Generate a unique chunk ID from filepath and line numbers.
+ *
+ * @param filepath - The source file path
+ * @param startLine - Start line number
+ * @param endLine - End line number
+ * @returns Unique chunk identifier
+ */
+export declare function generateChunkId(filepath: string, startLine: number, endLine: number): string;

package/dist/domain/services/index.d.ts

@@ -7,3 +7,5 @@
 export { BM25Index, tokenize, normalizeScore, type BM25Document, type BM25Result, type BM25SerializedData, } from "./bm25";
 export { extractKeywords, extractPathKeywords, parsePathContext, formatPathContextForEmbedding, COMMON_KEYWORDS, type PathContext, } from "./keywords";
 export { cosineSimilarity, euclideanDistance } from "./similarity";
+export { detectQueryIntent, extractQueryTerms, calculateFileTypeBoost, isSourceCodeFile, isDocFile, isDataFile, IMPLEMENTATION_TERMS, DOCUMENTATION_TERMS, SOURCE_CODE_EXTENSIONS, DOC_EXTENSIONS, DATA_EXTENSIONS, type QueryIntent, } from "./queryIntent";
+export { createLineBasedChunks, createSingleChunk, generateChunkId, DEFAULT_CHUNK_SIZE, DEFAULT_OVERLAP, type TextChunk, type ChunkingOptions, } from "./chunking";

package/dist/domain/services/queryIntent.d.ts

@@ -0,0 +1,55 @@
+/**
+ * Query Intent Detection Service
+ *
+ * Detects whether a search query is looking for implementation code
+ * or documentation, and calculates appropriate boosts.
+ */
+/** Implementation-related query terms that boost source code files */
+export declare const IMPLEMENTATION_TERMS: string[];
+/** Documentation-related query terms that boost documentation files */
+export declare const DOCUMENTATION_TERMS: string[];
+/** Source code file extensions */
+export declare const SOURCE_CODE_EXTENSIONS: string[];
+/** Documentation file extensions */
+export declare const DOC_EXTENSIONS: string[];
+/** Data/config file extensions */
+export declare const DATA_EXTENSIONS: string[];
+/** Query intent types */
+export type QueryIntent = "implementation" | "documentation" | "neutral";
+/**
+ * Detect query intent based on terms.
+ * Returns: 'implementation' | 'documentation' | 'neutral'
+ *
+ * @param queryTerms - Array of query terms (lowercase)
+ * @returns The detected intent
+ */
+export declare function detectQueryIntent(queryTerms: string[]): QueryIntent;
+/**
+ * Extract query terms from a search query.
+ *
+ * @param query - The search query string
+ * @returns Array of lowercase terms (length > 2)
+ */
+export declare function extractQueryTerms(query: string): string[];
+/**
+ * Determine if a file is a source code file based on extension.
+ */
+export declare function isSourceCodeFile(filepath: string): boolean;
+/**
+ * Determine if a file is a documentation file based on extension.
+ */
+export declare function isDocFile(filepath: string): boolean;
+/**
+ * Determine if a file is a data/config file based on extension.
+ */
+export declare function isDataFile(filepath: string): boolean;
+/**
+ * Calculate boost based on file type and query context.
+ * Bidirectional: boosts code for implementation queries, docs for documentation queries.
+ * Only applies when query intent is clear.
+ *
+ * @param filepath - The file path
+ * @param queryTerms - Array of query terms (lowercase)
+ * @returns Boost value (0 to ~0.1)
+ */
+export declare function calculateFileTypeBoost(filepath: string, queryTerms: string[]): number;