uri-register 0.2.0

A high-performance PostgreSQL-backed URI dictionary service for assigning unique integer IDs to URIs
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
# URI Register

<!--
Copyright TELICENT LTD

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->

[![CI](https://github.com/telicent-oss/uri-register/actions/workflows/ci.yml/badge.svg)](https://github.com/telicent-oss/uri-register/actions/workflows/ci.yml)

> **Beta Software**: This library is in active development and the API may change. While it's being used in production environments, you should pin to a specific version and test thoroughly before upgrading.

A caching PostgreSQL-backed URI register service for assigning unique integer IDs to URIs. Perfect for string interning, deduplication, and systems that need consistent global identifier mappings.

**Note:** The Rust library requires an async runtime (tokio). Python bindings support both synchronous and asynchronous usage.

## Overview

The URI Register provides a simple, fast way to assign unique integer IDs to URI strings. Once registered, a URI always returns the same ID, making it ideal for string interning and deduplication in distributed systems.

## Features

- **Simple API**: Just 2 methods - `register_uri()` and `register_uri_batch()`
- **Async + Sync**: Built on tokio for high concurrency, with sync wrappers for Python
- **Batch optimised**: Process thousands of URIs in a single database round-trip
- **Configurable caching**: W-TinyLFU (Moka) or LRU caching for frequently accessed URIs
- **Order preservation**: Batch operations maintain strict order correspondence
- **PostgreSQL backend**: Durable, scalable, with connection pooling
- **Automatic retry logic**: Configurable exponential backoff for transient database errors
- **Thread-safe**: Designed for concurrent access from multiple threads/processes

## Use Cases

- **String interning systems**: Reduce memory footprint by storing strings once and referencing by ID
- **URL deduplication**: Assign unique IDs to URLs across distributed crawlers
- **Global identifier systems**: Centralised ID assignment for URIs/strings in microservices
- **Data warehousing**: Efficient storage of repeated string values
- **Distributed caching**: Consistent ID assignment across cache nodes

## Installation

### Rust

Add to your `Cargo.toml`:

```toml
[dependencies]
uri-register = "0.2.0"
```

Or use as a git dependency:

```toml
[dependencies]
uri-register = { git = "https://github.com/telicent-oss/uri-register" }
```

### Python

Install from TestPyPI (during beta):

```bash
pip install --index-url https://test.pypi.org/simple/ uri-register
```

**Requirements**: Python 3.8+

**Note**: The package is currently published to TestPyPI for testing. Once stable, it will be available on the main PyPI repository.

## Setup

### 1. Database Initialisation

Before using the URI Register service, you must initialise the PostgreSQL schema.

**Run the schema creation script:**

```bash
psql -U username -d database_name -f schema.sql
```

Or execute the SQL directly:

```sql
CREATE TABLE IF NOT EXISTS uri_register (
    id BIGSERIAL PRIMARY KEY,
    uri TEXT NOT NULL,
    uri_hash UUID GENERATED ALWAYS AS (md5(uri)::uuid) STORED UNIQUE
);
```

### 2. Database Configuration

The service requires a PostgreSQL connection string. Set it as an environment variable or pass it directly:

```bash
export DATABASE_URL="postgresql://username:password@localhost:5432/database_name"
```

## Usage

### Rust Example

```rust
use uri_register::{CacheStrategy, PostgresUriRegister, UriService};

#[tokio::main]
async fn main() -> uri_register::Result<()> {
    // Connect to PostgreSQL
    let register = PostgresUriRegister::new(
        "postgres://localhost/mydb",
        "uri_register",  // table name
        20,              // max connections
        10_000           // cache size (uses Moka/W-TinyLFU by default)
    ).await?;

    // Register a single URI
    let id = register.register_uri("http://example.org/resource/1").await?;
    println!("Registered URI with ID: {}", id);

    // Register the same URI again - returns the same ID
    let same_id = register.register_uri("http://example.org/resource/1").await?;
    assert_eq!(id, same_id);

    // Register multiple URIs in batch (much faster!)
    let uris = vec![
        "http://example.org/resource/2".to_string(),
        "http://example.org/resource/3".to_string(),
        "http://example.org/resource/4".to_string(),
    ];
    let ids = register.register_uri_batch(&uris).await?;

    // IDs maintain order: ids[i] corresponds to uris[i]
    for (uri, id) in uris.iter().zip(ids.iter()) {
        println!("{} -> {}", uri, id);
    }

    Ok(())
}
```

### Synchronous Rust API

For synchronous Rust applications that cannot use async/await, use `SyncPostgresUriRegister`:

```rust
use uri_register::SyncPostgresUriRegister;

fn main() -> uri_register::Result<()> {
    // Connect to PostgreSQL
    let register = SyncPostgresUriRegister::new(
        "postgres://localhost/mydb",
        "uri_register",  // table name
        20,              // max connections
        10_000           // cache size (uses Moka/W-TinyLFU by default)
    )?;

    // Register a single URI (blocks until complete)
    let id = register.register_uri("http://example.org/resource/1")?;
    println!("Registered URI with ID: {}", id);

    // Register multiple URIs in batch
    let uris = vec![
        "http://example.org/resource/2".to_string(),
        "http://example.org/resource/3".to_string(),
    ];
    let ids = register.register_uri_batch(&uris)?;

    Ok(())
}
```

The synchronous API wraps the async implementation with a Tokio runtime internally. All methods have identical semantics to their async counterparts but block the calling thread until completion.

### Python Example (Synchronous)

```python
from uri_register import UriRegister

# Connect to PostgreSQL
register = UriRegister(
    "postgres://localhost/mydb",
    "uri_register",  # table name
    20,              # max connections
    10_000,          # cache size
    "moka",          # cache strategy ("moka" is default, or use "lru")
)

# Register a single URI
id = register.register_uri("http://example.org/resource/1")
print(f"Registered URI with ID: {id}")

# Register the same URI again - returns the same ID
same_id = register.register_uri("http://example.org/resource/1")
assert id == same_id

# Register multiple URIs in batch (much faster!)
uris = [
    "http://example.org/resource/2",
    "http://example.org/resource/3",
    "http://example.org/resource/4",
]
ids = register.register_uri_batch(uris)

# IDs maintain order: ids[i] corresponds to uris[i]
for uri, id in zip(uris, ids):
    print(f"{uri} -> {id}")

# Get statistics
stats = register.stats()
print(f"Total URIs: {stats['total_uris']}")
```

### Python Example (Asynchronous)

```python
import asyncio
from uri_register import UriRegister

async def main():
    # Connect to PostgreSQL
    register = await UriRegister.new_async(
        "postgres://localhost/mydb",
        "uri_register",  # table name
        20,              # max connections
        10_000,          # cache size
        "moka",          # cache strategy ("moka" is default, or use "lru")
    )

    # Register a single URI
    id = await register.register_uri_async("http://example.org/resource/1")
    print(f"Registered URI with ID: {id}")

    # Register multiple URIs in batch (much faster!)
    uris = [
        "http://example.org/resource/2",
        "http://example.org/resource/3",
    ]
    ids = await register.register_uri_batch_async(uris)

    # Get statistics
    stats = await register.stats_async()
    print(f"Total URIs: {stats['total_uris']}")

asyncio.run(main())
```

### API Reference

The `UriService` trait provides two methods:

#### `register_uri(uri: &str) -> u64`

Register a single URI and return its ID.

- If the URI exists, returns the existing ID
- If the URI is new, creates a new ID and returns it
- Uses configurable cache (Moka/LRU) for fast repeated lookups

```rust
let id = register.register_uri("http://example.org/page").await?;
```

#### `register_uri_batch(uris: &[String]) -> Vec<u64>`

Register multiple URIs in batch and return their IDs.

- **Order preserved**: `ids[i]` corresponds to `uris[i]`
- Much faster than calling `register_uri()` multiple times
- Handles duplicate URIs in input correctly
- Cache-optimised: only queries database for cache misses

```rust
let uris = vec![
    "http://example.org/page1".to_string(),
    "http://example.org/page2".to_string(),
];
let ids = register.register_uri_batch(&uris).await?;

// Access by index
assert_eq!(ids[0], register.register_uri("http://example.org/page1").await?);
```

### Statistics and Observability

The register exposes comprehensive metrics suitable for OpenTelemetry and Prometheus:

```rust
let stats = register.stats().await?;

// Database metrics
println!("Total URIs: {}", stats.total_uris);
println!("Storage size: {} bytes", stats.size_bytes);

// Cache performance metrics
println!("Cache hits: {}", stats.cache.hits);
println!("Cache misses: {}", stats.cache.misses);
println!("Cache hit rate: {:.2}%", stats.cache.hit_rate());
println!("Cache entries: {}/{}", stats.cache.entry_count, stats.cache.capacity);

// Connection pool metrics
println!("Active connections: {}", stats.pool.connections_active);
println!("Idle connections: {}", stats.pool.connections_idle);
println!("Max connections: {}", stats.pool.connections_max);
```

#### Integration with OpenTelemetry

The statistics are designed for easy integration with observability systems:

```rust
use opentelemetry::metrics::Meter;

let stats = register.stats().await?;

// Report as gauges
meter.u64_gauge("uri_register.cache.hits").record(stats.cache.hits, &[]);
meter.u64_gauge("uri_register.cache.misses").record(stats.cache.misses, &[]);
meter.f64_gauge("uri_register.cache.hit_rate").record(stats.cache.hit_rate(), &[]);
meter.u64_gauge("uri_register.cache.size").record(stats.cache.entry_count, &[]);

meter.u64_gauge("uri_register.pool.active").record(stats.pool.connections_active as u64, &[]);
meter.u64_gauge("uri_register.pool.idle").record(stats.pool.connections_idle as u64, &[]);

meter.u64_gauge("uri_register.total_uris").record(stats.total_uris, &[]);
meter.u64_gauge("uri_register.size_bytes").record(stats.size_bytes, &[]);
```

All metrics are cumulative since process start and safe for concurrent access.

## Cache Strategies

The URI register supports two caching strategies:

### Moka (W-TinyLFU) - Default

**Recommended for most workloads.** W-TinyLFU (Window Tiny Least Frequently Used) combines recency and frequency tracking to provide better cache hit rates than plain LRU, especially for workloads with mixed hot/cold data.

Moka is the default cache strategy, so you don't need to specify it:

```rust
let register = PostgresUriRegister::new(
    db_url,
    "uri_register",
    20,      // max connections
    10_000   // cache size
).await?;
```

To explicitly specify Moka:

```rust
use uri_register::CacheStrategy;

let register = PostgresUriRegister::new_with_cache_strategy(
    db_url,
    "uri_register",
    20,
    10_000,
    Some(CacheStrategy::Moka),  // Explicitly use Moka
    None  // No TLS
).await?;
```

**Python:**
```python
register = UriRegister(
    db_url,
    "uri_register",
    20,
    10_000,
    "moka",  # W-TinyLFU algorithm
)
```

### LRU (Least Recently Used)

Simple eviction based on recency of access. Use this if you have specific requirements or want more predictable behavior.

```rust
use uri_register::CacheStrategy;

let register = PostgresUriRegister::new_with_cache_strategy(
    db_url,
    "uri_register",
    20,
    10_000,
    Some(CacheStrategy::Lru),  // Use LRU instead of default Moka
    None  // No TLS
).await?;
```

**Python:**
```python
register = UriRegister(
    db_url,
    "uri_register",
    20,
    10_000,
    "lru",  # Simple LRU
)
```

**Performance Comparison:**

For most real-world workloads, Moka (W-TinyLFU) provides 10-30% better cache hit rates compared to LRU, especially when:
- Access patterns have varying frequency (some URIs accessed much more than others)
- There are periodic "scans" or one-time accesses that would pollute an LRU cache
- Working set size is close to cache capacity

## Logging

The library uses the `tracing` crate for structured logging. Logs include connection info, cache hit/miss statistics, and batch sizes.

### Rust

Use `tracing-subscriber` to see logs:

```rust
use tracing_subscriber::EnvFilter;

// Initialize logging (typically in main())
tracing_subscriber::fmt()
    .with_env_filter(EnvFilter::from_default_env())
    .init();

// Set RUST_LOG environment variable to control log levels:
// RUST_LOG=uri_register=debug  - see debug logs from uri-register
// RUST_LOG=uri_register=trace  - see trace logs (cache hits/misses)
```

### Python

Logs are automatically bridged to Python's `logging` module:

```python
import logging

# Configure Python logging as usual
logging.basicConfig(
    level=logging.DEBUG,
    format='%(asctime)s %(levelname)s %(name)s: %(message)s'
)

# Logs from uri-register will appear with logger name 'uri_register'
# You can also configure just the uri_register logger:
logging.getLogger('uri_register').setLevel(logging.DEBUG)
```

**Log Levels:**
- `INFO`: Connection events, configuration
- `DEBUG`: Cache statistics, batch sizes, database queries
- `TRACE`: Individual cache hits/misses (verbose)

## Performance

### Logged Tables (Default)

With default logged tables on typical hardware:
- **Single registration**: ~500-1K URIs/sec (with cache: 100K+/sec)
- **Batch registration**: ~10K-50K URIs/sec
- **Batch lookup (cached)**: ~1M+ URIs/sec (no DB round-trip)
- **Batch lookup (uncached)**: ~100K-200K URIs/sec

### Unlogged Tables (Optional)

For 2-3x faster writes at the cost of durability:

```sql
ALTER TABLE uri_register SET UNLOGGED;
```

**Performance with unlogged tables:**
- **Batch registration**: ~30K-150K URIs/sec

**WARNING**: Unlogged tables lose all data if PostgreSQL crashes. Only use this if you can rebuild the register from source data.

To revert back to logged mode:

```sql
ALTER TABLE uri_register SET LOGGED;
```

## Performance Tips

1. **Always use batch operations** when processing multiple URIs
2. **Configure connection pooling** appropriately for your workload (typical: 10-50 connections)
3. **Tune cache size** based on your working set size and available memory (typical: 10,000-100,000 entries)
4. **Batch size**: Optimal batch size is typically 1,000-10,000 URIs per operation
5. **Hash-based indexing**: The compact UUID index on `uri_hash` scales much better than indexing full URIs
6. **Consider unlogged tables** for initial bulk loading, then switch to logged

## Architecture

```
Application
UriService trait (2 methods)
PostgresUriRegister impl
    ↓  ↓
Cache (Moka/LRU)  Connection Pool (20 connections)
    ↓                   ↓
    └───────────────→ PostgreSQL Database
```

## Schema Details

The register uses a three-column table with hash-based indexing:

- `id`: BIGSERIAL primary key (auto-incrementing u64)
- `uri`: TEXT storing the full URI (not indexed)
- `uri_hash`: UUID generated from `md5(uri)::uuid` with UNIQUE constraint (indexed)

### Why Hash-Based Indexing?

In environments with enormous numbers of URIs, maintaining a B-tree index on the full URI text becomes prohibitively expensive - both in storage and maintenance overhead. By hashing the URI to a compact 16-byte UUID, we get:

1. **Compact index**: 16 bytes per entry vs potentially hundreds of bytes for full URIs
2. **Fast lookups**: B-tree operations on fixed-size UUIDs are very efficient
3. **Automatic computation**: PostgreSQL computes the hash via `GENERATED ALWAYS AS`

The hash collision probability with MD5 (128-bit) is vanishingly small - you'd need ~2^64 URIs before expecting a collision. However, for absolute safety, queries should verify the full URI matches when retrieving data:

```sql
SELECT id FROM uri_register
WHERE uri_hash = md5('http://example.com/my-uri')::uuid  -- Fast index lookup
AND uri = 'http://example.com/my-uri';                   -- Collision safety check
```

Inserts use `ON CONFLICT (uri_hash)` to handle duplicates efficiently:

```sql
INSERT INTO uri_register (uri)
VALUES ('http://example.com/my-uri')
ON CONFLICT (uri_hash)
DO UPDATE SET uri = EXCLUDED.uri  -- No-op trick to return existing ID
RETURNING id;
```

## Testing

For testing purposes, an in-memory implementation is available:

```rust
#[cfg(test)]
use uri_register::InMemoryUriRegister;

#[tokio::test]
async fn test_uri_register() {
    let register = InMemoryUriRegister::new();
    let id = register.register_uri("http://example.org").await.unwrap();
    assert_eq!(id, 1); // First URI gets ID 1
}
```

## Error Handling

The library uses structured error types for better error handling and programmatic error inspection:

```rust
use uri_register::{CacheStrategy, ConfigurationError, Error, Result};

// Configuration errors with specific variants
match PostgresUriRegister::new("postgres://localhost/db", "uri_register", 0, 10_000).await {
    Ok(register) => { /* use register */ },
    Err(Error::Configuration(ConfigurationError::InvalidMaxConnections(n))) => {
        eprintln!("Invalid max_connections: {}", n);
    },
    Err(Error::Configuration(ConfigurationError::InvalidCacheSize(n))) => {
        eprintln!("Invalid cache_size: {}", n);
    },
    Err(Error::Configuration(ConfigurationError::InvalidTableName(msg))) => {
        eprintln!("Invalid table_name: {}", msg);
    },
    Err(Error::Configuration(ConfigurationError::InvalidBackoff(msg))) => {
        eprintln!("Invalid backoff configuration: {}", msg);
    },
    Err(e) => eprintln!("Other error: {}", e),
}

// Database errors (connection strings are sanitised to prevent password leaks)
match register.register_uri("http://example.org").await {
    Ok(id) => println!("Registered with ID: {}", id),
    Err(Error::Database(msg)) => eprintln!("Database error: {}", msg),
    Err(Error::InvalidUri(msg)) => eprintln!("Invalid URI: {}", msg),
    Err(e) => eprintln!("Other error: {}", e),
}
```

### Error Types

- **Configuration** - Invalid configuration parameters (structured with specific variants)
- **Database** - Database operation failures (error messages sanitised)
- **ConnectionPool** - Connection pool errors
- **Cache** - Cache operation failures
- **InvalidUri** - URI validation failures (non-RFC 3986 compliant URIs)

## License

Licensed under the Apache License, Version 2.0 ([LICENSE](LICENSE) or http://www.apache.org/licenses/LICENSE-2.0).

## Contributing

Contributions are welcome! Please feel free to submit a Pull Request.