rx-tool 2.0.1

High-performance file search and analysis tool powered by ripgrep

RX (Regex Tracer)

A high-performance tool for searching and analyzing large files, powered by ripgrep.

Designed for large files.

RX is optimized for processing multi-GB files efficiently through parallel chunking and streaming. If you need to process many files repeatedly, use the API server (rx serve) instead of running CLI commands in a loop. The server mode avoids Python startup overhead on each invocation.

Key Features

• Byte-Offset Based: Returns precise byte offsets for efficient large file processing (line-based indexing available)
• Parallel Processing: Automatic chunking and parallel execution for large files
• Samples output: Can show arbitrary parts of text files with context when you found interested offsets
• REST API Server: All CLI features available via async HTTP API
• File Analysis: Extract metadata, statistics, and metrics from files (including compressed files)
• Regex Complexity Analysis: Detect ReDoS vulnerabilities before production use
• Compressed File Support: Analyze and search gzip, zstd, xz, bzip2 files transparently
• Seekable Zstd: Fast random access to seekable zstd compressed files with automatic indexing
• Analysis Caching: Cache file analysis results for instant repeated access
• Background Tasks: Background compression and indexing with progress tracking
• Security Sandbox: Restrict file access to specific directories in server mode

Prerequisites

ripgrep must be installed:

• macOS: brew install ripgrep
• Ubuntu/Debian: apt install ripgrep
• Windows: choco install ripgrep

Installation

Option 1: Install from PyPI (Recommended)

# Requires Python 3.13+
pip install rx-tool

# Now use the rx command
rx /var/log/app.log "error.*"
rx --version

Note: Requires ripgrep to be installed separately (see Prerequisites).

Option 2: Development with uv

uv sync
uv run rx /var/log/app.log "error.*"

Option 3: Standalone Binary

./build.sh
./dist/rx /var/log/app.log "error.*"

Shell Completion

Enable tab completion for rx commands and options:

Zsh (add to ~/.zshrc):

_RX_COMPLETE=zsh_source rx > ~/.rx-complete.zsh
echo 'source ~/.rx-complete.zsh' >> ~/.zshrc
source ~/.zshrc

Bash (add to ~/.bashrc):

_RX_COMPLETE=bash_source rx > ~/.rx-complete.bash
echo 'source ~/.rx-complete.bash' >> ~/.bashrc
source ~/.bashrc

Fish:

_RX_COMPLETE=fish_source rx > ~/.config/fish/completions/rx.fish

After setup, rx <Tab> will suggest subcommands (analyse, trace, etc.) and options.

Quick Start

Basic Examples

# Search a file (returns byte offsets)
rx /var/log/app.log "error.*"

# Search a directory
rx /var/log/ "error.*"

# Show context lines
rx /var/log/app.log "error" --samples --context=3

# Analyze file metadata
rx analyse /var/log/app.log

# Check regex complexity
rx check "(a+)+"

# Start API server
rx serve --port=8000

Why Byte Offsets?

RX returns byte offsets instead of line numbers for efficiency. Seeking to byte position is O(1), while counting lines is O(n). For large files, this matters significantly.

Need line numbers? Use the indexing feature:

# Create index for a large file
rx index /var/log/huge.log

# Now you can use line-based operations
rx samples /var/log/huge.log -l 1000,2000,3000 --context=5

The index enables fast line-to-offset conversion for files >50MB.

Server Mode (Recommended for Repeated Operations)

The CLI spawns a Python interpreter on each invocation. For processing multiple files or repeated operations, use the API server:

# Start server
rx serve --port=8000

# Use HTTP API (same endpoints as CLI)
curl "http://localhost:8000/v1/trace?path=/var/log/app.log&regexp=error"
curl "http://localhost:8000/v1/analyse?path=/var/log/"

Benefits:

• No Python startup overhead per request
• Async processing with configurable workers
• Webhook support for event notifications
• Security sandbox with --search-root

Security Sandbox

Restrict file access in server mode:

# Only allow access to /var/log
rx serve --search-root=/var/log

# Attempts to access other paths return 403 Forbidden
curl "http://localhost:8000/v1/trace?path=/etc/passwd&regexp=root"
# => 403 Forbidden

Prevents directory traversal (../) and symlink escape attacks.

CLI Commands

rx (search)

Search files for regex patterns.

rx /var/log/app.log "error.*"              # Basic search
rx /var/log/ "error.*"                     # Search directory
rx /var/log/app.log "error" --samples      # Show context lines
rx /var/log/app.log "error" -i             # Case-insensitive (ripgrep flags work)
rx /var/log/app.log "error" --json         # JSON output

rx analyse

Extract file metadata and statistics (works with text and compressed files).

rx analyse /var/log/app.log               # Single file
rx analyse /var/log/app.log.gz            # Compressed file (auto-decompressed)
rx analyse /var/log/                      # Directory
rx analyse /var/log/ --max-workers=20     # Parallel processing
rx analyse /var/log/app.log --json        # JSON output with all metrics

Output includes:

• File size (bytes and human-readable)
• Compression info (format, ratio, decompressed size)
• Index info (if indexed, checkpoint count, validity)
• Line statistics (count, length metrics, endings)
• All results cached for instant repeated access

rx check

Analyze regex complexity and detect ReDoS vulnerabilities.

rx check "(a+)+"                          # Returns risk level and fixes

rx index

Create line-offset index for large files.

rx index /var/log/huge.log                # Create index
rx index /var/log/huge.log --info         # Show index info

rx samples

Extract context lines around byte offsets or line numbers.

rx samples /var/log/app.log -b 12345,67890 --context=3   # Byte offsets
rx samples /var/log/app.log -l 100,200 --context=5       # Line numbers (requires index)

rx serve

Start REST API server.

rx serve                                  # Start on localhost:8000
rx serve --host=0.0.0.0 --port=8080       # Custom host/port
rx serve --search-root=/var/log           # Restrict to directory

API Endpoints

Once the server is running, visit http://localhost:8000/docs for interactive API documentation.

Main Endpoints:

• GET /v1/trace - Search files for patterns
• GET /v1/analyse - File metadata and statistics (including compressed files)
• GET /v1/complexity - Regex complexity analysis
• GET /v1/samples - Extract context lines
• GET /health - Server health and configuration

Background Task Endpoints:

• POST /v1/compress - Start background compression to seekable zstd (returns task_id)
• POST /v1/index - Start background indexing for large files (returns task_id)
• GET /v1/tasks/{task_id} - Check task status (queued, running, completed, failed)

Examples:

# Search
curl "http://localhost:8000/v1/trace?path=/var/log/app.log&regexp=error&max_results=10"

# Analyse (works with compressed files too)
curl "http://localhost:8000/v1/analyse?path=/var/log/app.log.gz"

# Start background compression (returns immediately with task_id)
curl -X POST "http://localhost:8000/v1/compress" \
  -H "Content-Type: application/json" \
  -d '{"input_path": "/var/log/huge.log", "frame_size": "1MB"}'
# Response: {"task_id": "550e8400-e29b-41d4-a716-446655440000", "status": "queued"}

# Check compression progress
curl "http://localhost:8000/v1/tasks/550e8400-e29b-41d4-a716-446655440000"

# With webhooks
curl "http://localhost:8000/v1/trace?path=/var/log/app.log&regexp=error&hook_on_complete=https://example.com/webhook"

Configuration

Environment Variables

Variable	Description	Default
RX_WORKERS	Worker processes for server	1
RX_LOG_LEVEL	Log level (DEBUG, INFO, WARNING, ERROR)	INFO
RX_MAX_SUBPROCESSES	Max parallel workers for file processing	20
RX_MIN_CHUNK_SIZE_MB	Min chunk size for splitting files	20

Server Configuration

# Production example (8-core, 16GB RAM)
RX_WORKERS=17 \
RX_LIMIT_CONCURRENCY=500 \
RX_LIMIT_MAX_REQUESTS=10000 \
rx serve --host=0.0.0.0 --port=8000 --search-root=/data

# Container/Kubernetes (1 worker per pod, scale with replicas)
RX_WORKERS=1 rx serve --host=0.0.0.0 --port=8000

Compressed File Support

RX can search, analyze, and extract samples from compressed files without manual decompression. Supported formats:

• gzip (.gz)
• zstd (.zst) - including seekable zstd
• xz (.xz)
• bzip2 (.bz2)

Searching Compressed Files

# Search a gzip file - works exactly like regular files
rx /var/log/syslog.1.gz "error.*"

# Search with context
rx /var/log/syslog.1.gz "error" --samples --context=3

# All regular options work with compressed files
rx /var/log/app.log.gz "error" -i --json

Analyzing Compressed Files

# Full analysis with automatic decompression
rx analyse /var/log/app.log.gz

# Output includes compression info:
# Compressed: gzip, ratio: 5.2x, decompressed: 2.5 GB
# Lines: 50000000 total, 0 empty
# Line length: max=256, avg=128.5, median=130.0

Extracting Samples from Compressed Files

For compressed files, use line numbers (byte offsets are not meaningful in compressed streams):

# Get lines 100, 200, 300 with 5 lines of context
rx samples /var/log/syslog.1.gz -l 100,200,300 --context=5

Seekable Zstd Performance

For very large compressed files, convert to seekable zstd format for parallel processing:

# Background compression to seekable zstd (returns immediately)
rx compress /var/log/huge.log --frame-size=1MB

# Or via API:
curl -X POST "http://localhost:8000/v1/compress" \
  -H "Content-Type: application/json" \
  -d '{"input_path": "/var/log/huge.log", "frame_size": "1MB"}'

Seekable zstd benefits:

• Fast random access without decompressing entire file
• Automatic frame-based indexing
• Better than regular zstd for large files
• Compressed index cached at ~/.cache/rx/indexes/

Performance Considerations

• Regular compressed files (gzip, xz, bzip2) processed sequentially
• Seekable zstd supports parallel frame access (Tier 2 - coming soon)
• Decompression index cached at ~/.cache/rx/compressed_indexes/
• Analysis results cached at ~/.cache/rx/analyse_cache/ for instant repeated access

API Usage

# Search compressed file
curl "http://localhost:8000/v1/trace?path=/var/log/syslog.1.gz&regexp=error"

# Analyze compressed file (includes compression info)
curl "http://localhost:8000/v1/analyse?path=/var/log/app.log.gz"

# Get samples (use lines parameter, not offsets)
curl "http://localhost:8000/v1/samples?path=/var/log/syslog.1.gz&lines=100,200&context=3"

# Start background compression task
curl -X POST "http://localhost:8000/v1/compress" \
  -H "Content-Type: application/json" \
  -d '{"input_path": "/var/log/huge.log", "build_index": true}'

Roadmap

Completed Features ✅

• Compressed File Support - Analyze and search gzip, zstd, xz, bzip2 files
• File Analysis Caching - Cache analysis results at ~/.cache/rx/analyse_cache/
• Background Tasks - Compression and indexing endpoints with progress tracking
• Seekable Zstd - Frame-based indexing for random access without full decompression
• Enhanced Analysis - Compression info, index info, detailed statistics

In Progress / Coming Soon

• Parallel Seekable Zstd - Parallel frame processing for seekable zstd files (Tier 2)
• Streaming API - WebSocket endpoint for real-time results
• Compression Streaming - Stream compressed output without decompression

Web Frontend

RX includes a web-based file viewer with advanced navigation and analysis capabilities:

# Start server with web UI
rx serve --port=8000

# Open browser to http://localhost:8000

Frontend Features:

• Multi-file viewer with tabs - View multiple files simultaneously with tab management and drag-and-drop reordering

• Advanced search capabilities - In-file text search with match highlighting and navigation; global trace search across configured roots with regex support

• Syntax highlighting and filtering - Toggle-able syntax highlighting; regex-based content filtering with hide/show/highlight modes

• Invisible characters display mode - Toggle to visualize spaces (·), tabs (→), carriage returns (⏎), non-breaking spaces (°), and zero-width characters (|)

• Virtual scrolling for large files - Efficient on-demand content loading with adjustable lines per page; support for compressed files (gz, zst, xz, bz2)

• File tree browser with analysis - Hierarchical directory navigation with context actions to analyze file statistics and create line-offset indexes

• Theme support and customization - Light/dark/system themes; configurable font size, line numbers, and keyboard shortcuts

Development

# Run tests
uv run pytest -v

# Run with coverage
uv run pytest --cov=rx --cov-report=html

# Build binary
uv sync --group build
./build.sh

# Build frontend
make frontend-build

License

MIT