raggrep 0.1.5 → 0.1.6

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/README.md CHANGED
@@ -7,7 +7,7 @@ RAGgrep indexes your code and allows semantic search using natural language quer
7
7
  ## Features
8
8
 
9
9
  - **🏠 Local-first** — All indexing and search happens locally. No cloud dependencies.
10
- - **📁 Filesystem-based** — Index stored as readable JSON files alongside your code.
10
+ - **📁 Filesystem-based** — Index stored as readable JSON files in system temp directory.
11
11
  - **⚡ Tiered search** — Fast keyword filtering + semantic search for efficiency.
12
12
  - **🔍 Hybrid scoring** — Combines semantic similarity with BM25 keyword matching.
13
13
  - **🔄 Incremental** — Only re-indexes files that have changed.
@@ -17,9 +17,12 @@ RAGgrep indexes your code and allows semantic search using natural language quer
17
17
  ## Installation
18
18
 
19
19
  ```bash
20
- # Install globally
20
+ # Install globally with npm
21
21
  npm install -g raggrep
22
22
 
23
+ # Or with Bun (recommended)
24
+ bun install -g raggrep
25
+
23
26
  # Or use without installing
24
27
  npx raggrep --help
25
28
  ```
@@ -92,15 +95,26 @@ raggrep status # Show index status
92
95
 
93
96
  ## How It Works
94
97
 
95
- RAGgrep uses a two-tier index system:
98
+ RAGgrep uses a **dual-module architecture** with two complementary index types:
99
+
100
+ ### Core Module
101
+
102
+ - **Language-agnostic** regex-based symbol extraction
103
+ - **BM25 keyword matching** for fast, deterministic search
104
+ - Works on any text file
105
+
106
+ ### TypeScript Module
96
107
 
97
- 1. **Symbolic Index** Lightweight file summaries with extracted keywords. Used for fast BM25 filtering.
98
- 2. **Embedding Index** — Full chunk embeddings for semantic search. Only loaded for relevant files.
108
+ - **AST-based parsing** via TypeScript Compiler API
109
+ - **Semantic embeddings** for natural language understanding
110
+ - **Symbolic index** for fast BM25 candidate filtering
99
111
 
100
- This design keeps memory usage low and enables fast search on large codebases.
112
+ Search combines results from both modules:
101
113
 
102
114
  ```
103
- Query → BM25 filter (symbolic) → Load candidates → Semantic search → Results
115
+ Query → Core (symbol/BM25) ─┐
116
+ ├→ Merge & rank → Results
117
+ Query → TypeScript (BM25 filter → semantic) ─┘
104
118
  ```
105
119
 
106
120
  ## What Gets Indexed
@@ -139,8 +153,8 @@ Query → BM25 filter (symbolic) → Load candidates → Semantic search → Res
139
153
 
140
154
  ## Requirements
141
155
 
142
- - Node.js 18+
143
- - ~50MB disk space for models (cached globally)
156
+ - Node.js 18+ or Bun 1.0+
157
+ - ~50MB disk space for models (cached globally at `~/.cache/raggrep/models/`)
144
158
 
145
159
  ## License
146
160