Rust Sitemap Crawler

Concurrent web crawler in Rust. 256 concurrent workers, sharded frontier, persistent state with WAL, distributed crawling with Redis.

Available as both a standalone CLI tool and a Python package.

Install

Python Package (Recommended)

pip install rustmapper

From Source (Rust)

cargo build --release

Usage

Python Package

Once installed via pip, use the rustmapper command:

# Basic crawl
rustmapper crawl --start-url example.com

# With options
rustmapper crawl --start-url example.com --workers 128 --timeout 10

# Resume
rustmapper resume --data-dir ./data

# Export sitemap
rustmapper export-sitemap --data-dir ./data --output sitemap.xml

Python API

from rustmapper import Crawler

# Create a crawler instance
crawler = Crawler(
    start_url="https://example.com",
    data_dir="./data",
    workers=256,
    timeout=20,
    ignore_robots=False
)

# Start crawling
results = crawler.crawl()
print(f"Discovered: {results['discovered']}, Processed: {results['processed']}")

# Export to sitemap
crawler.export_sitemap(
    output="sitemap.xml",
    include_lastmod=True,
    include_changefreq=True,
    default_priority=0.5
)

Rust CLI (from source)

# Basic crawl
cargo run --release -- crawl --start-url example.com

# With options
cargo run --release -- crawl --start-url example.com --workers 128 --timeout 10

# Resume
cargo run --release -- resume --data-dir ./data

# Export sitemap
cargo run --release -- export-sitemap --data-dir ./data --output sitemap.xml

Options

Seeding Strategies

• none - Only start URL
• sitemap - Discover from sitemap.xml
• ct - Certificate Transparency logs (finds subdomains)
• commoncrawl - Query Common Crawl index
• all - Use all methods

Performance

Timing breakdown per URL:

• Body download: 700-900ms (70-90%)
• Network fetch: 50-550ms (10-20%)
• Everything else: <50ms (<5%)

Throughput: 50-200 URLs/minute depending on page size. Network I/O bound.

Recommended settings:

# Focused crawl (skip subdomains)
--timeout 10 --seeding-strategy sitemap

# University sites (avoid internal hosts)
--timeout 5 --seeding-strategy sitemap --start-url www.university.edu

# Maximum discovery (all seeders)
--workers 256 --timeout 10 --seeding-strategy all

Output

JSONL (automatic): ./data/sitemap.jsonl

{"url":"https://example.com/","depth":0,"status_code":200,"content_length":1024,"title":"Example","link_count":5}

XML sitemap:

cargo run --release -- export-sitemap --data-dir ./data --output sitemap.xml

Distributed Crawling

# Instance 1
cargo run --release -- crawl --start-url example.com --enable-redis --redis-url redis://localhost:6379

# Instance 2
cargo run --release -- crawl --start-url example.com --enable-redis --redis-url redis://localhost:6379

Automatic URL deduplication, work stealing, distributed locks.

Architecture

• Frontier: Sharded queues (14 shards), bloom filter dedup, per-host politeness
• State: Embedded redb database + WAL for crash recovery
• Governor: Adaptive concurrency control based on commit latency
• Workers: Async task pool with semaphore-based backpressure

Troubleshooting

Testing

cargo test

Docs

• PERFORMANCE_ANALYSIS.md - Detailed timing breakdown
• BOTTLENECK_SUMMARY.md - Where time is spent

License

MIT