realizar 0.8.5

Pure Rust ML inference engine built from scratch - model serving for GGUF and safetensors
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

//! Mock Backend Tests (PMAT-COV-95)
//!
//! Tests for the mock inference backend that enable testing without disk I/O.
//! These tests exercise the full inference flow: configuration, token counting,
//! timing calculations, and result formatting.

use super::*;
use std::path::PathBuf;

// ============================================================================
// MOCK INFERENCE BASIC TESTS
// ============================================================================

#[test]
fn test_mock_inference_basic() {
    let config = mock_config("Hello world");
    let result = run_inference(&config).unwrap();

    assert_eq!(result.format, "Mock");
    assert!(!result.used_gpu);
    assert!(result.text.contains("mock response for:"));
    assert!(result.text.contains("Hello world"));
}

#[test]
fn test_mock_inference_token_counting() {
    let config = InferenceConfig::new("/dev/null")
        .with_prompt("one two three four five")
        .with_max_tokens(10)
        .with_mock_backend();

    let result = run_inference(&config).unwrap();

    // 5 words = 5 input tokens (mock tokenization)
    assert_eq!(result.input_token_count, 5);
    // Should generate min(max_tokens, 32) = 10 tokens
    assert_eq!(result.generated_token_count, 10);
    // Total tokens = input + generated
    assert_eq!(result.tokens.len(), 15);
}

#[test]
fn test_mock_inference_with_input_tokens() {
    let config = InferenceConfig::new("/dev/null")
        .with_input_tokens(vec![1, 2, 3, 4])
        .with_max_tokens(8)
        .with_mock_backend();

    let result = run_inference(&config).unwrap();

    assert_eq!(result.input_token_count, 4);
    assert_eq!(result.generated_token_count, 8);
    // First 4 tokens should be input tokens
    assert_eq!(&result.tokens[0..4], &[1, 2, 3, 4]);
    // Generated tokens should be [100, 101, 102, ...]
    assert_eq!(result.tokens[4], 100);
    assert_eq!(result.tokens[5], 101);
}

#[test]
fn test_mock_inference_no_prompt() {
    let config = InferenceConfig::new("/dev/null")
        .with_max_tokens(5)
        .with_mock_backend();

    let result = run_inference(&config).unwrap();

    // No prompt = BOS token only
    assert_eq!(result.input_token_count, 1);
    assert_eq!(result.tokens[0], 1); // BOS
    assert!(result.text.contains("(no prompt)"));
}

// ============================================================================
// TIMING CALCULATION TESTS
// ============================================================================

#[test]
fn test_mock_inference_timing_calculations() {
    let config = InferenceConfig::new("/dev/null")
        .with_prompt("test")
        .with_max_tokens(16)
        .with_mock_backend();

    let result = run_inference(&config).unwrap();

    // Load time should be simulated 10ms
    assert!((result.load_ms - 10.0).abs() < 0.01);

    // Inference time should be 50 + (num_tokens * 2)
    let expected_ms = 50.0 + (16.0 * 2.0);
    assert!((result.inference_ms - expected_ms).abs() < 0.01);

    // Tokens per second calculation
    let expected_tps = 16.0 / (expected_ms / 1000.0);
    assert!((result.tok_per_sec - expected_tps).abs() < 1.0);
}

#[test]
fn test_mock_inference_tok_per_sec_positive() {
    let config = mock_config("test prompt");
    let result = run_inference(&config).unwrap();

    assert!(result.tok_per_sec > 0.0);
    assert!(result.inference_ms > 0.0);
}

// ============================================================================
// CONFIGURATION VALIDATION TESTS
// ============================================================================

#[test]
fn test_mock_inference_negative_temperature_error() {
    let mut config = mock_config("test");
    config.temperature = -0.5;

    let result = run_inference(&config);
    assert!(result.is_err());

    let err = result.unwrap_err();
    let err_str = format!("{:?}", err);
    assert!(err_str.contains("temperature"));
}

#[test]
fn test_mock_inference_zero_max_tokens_error() {
    let config = InferenceConfig::new("/dev/null")
        .with_prompt("test")
        .with_max_tokens(0)
        .with_mock_backend();

    let result = run_inference(&config);
    assert!(result.is_err());

    let err = result.unwrap_err();
    let err_str = format!("{:?}", err);
    assert!(err_str.contains("max_tokens"));
}

#[test]
fn test_mock_inference_max_tokens_capped() {
    let config = InferenceConfig::new("/dev/null")
        .with_prompt("test")
        .with_max_tokens(1000) // Way more than cap of 32
        .with_mock_backend();

    let result = run_inference(&config).unwrap();

    // Should be capped at 32
    assert_eq!(result.generated_token_count, 32);
}

// ============================================================================
// CONFIGURATION BUILDER TESTS
// ============================================================================

#[test]
fn test_inference_config_builder() {
    let config = InferenceConfig::new("model.gguf")
        .with_prompt("Hello")
        .with_max_tokens(64)
        .with_temperature(0.7)
        .with_top_k(40)
        .without_gpu()
        .with_verbose(true)
        .with_trace(true)
        .with_trace_output("/tmp/trace.json");

    assert_eq!(config.model_path, PathBuf::from("model.gguf"));
    assert_eq!(config.prompt, Some("Hello".to_string()));
    assert_eq!(config.max_tokens, 64);
    assert!((config.temperature - 0.7).abs() < 0.01);
    assert_eq!(config.top_k, 40);
    assert!(config.no_gpu);
    assert!(config.verbose);
    assert!(config.trace);
    assert_eq!(config.trace_output, Some(PathBuf::from("/tmp/trace.json")));
}

#[test]
fn test_inference_config_with_input_tokens() {
    let config = InferenceConfig::new("model.gguf").with_input_tokens(vec![1, 2, 3]);

    assert_eq!(config.input_tokens, Some(vec![1, 2, 3]));
}

#[test]
fn test_inference_config_defaults() {
    let config = InferenceConfig::new("model.gguf");

    assert_eq!(config.max_tokens, 32);
    assert!((config.temperature - 0.0).abs() < 0.01); // Greedy
    assert_eq!(config.top_k, 1);
    assert!(!config.no_gpu);
    assert!(!config.verbose);
    assert!(!config.trace);
    assert!(!config.use_mock_backend);
}

#[test]
fn test_mock_config_helper() {
    let config = mock_config("test prompt");

    assert_eq!(config.model_path, PathBuf::from("/dev/null"));
    assert_eq!(config.prompt, Some("test prompt".to_string()));
    assert_eq!(config.max_tokens, 16);
    assert!(config.use_mock_backend);
}

// ============================================================================
// TRACE OUTPUT TESTS
// ============================================================================

#[test]
fn test_mock_inference_with_trace_output() {
    let trace_path = std::env::temp_dir().join("mock_trace_test.json");

    let config = InferenceConfig::new("/dev/null")
        .with_prompt("trace test")
        .with_max_tokens(8)
        .with_trace_output(&trace_path)
        .with_mock_backend();

    let result = run_inference(&config).unwrap();
    assert!(result.text.contains("trace test"));

    // Verify trace file was written
    let trace_content = std::fs::read_to_string(&trace_path).unwrap();
    assert!(trace_content.contains("\"mock\": true"));
    assert!(trace_content.contains("\"input_tokens\""));
    assert!(trace_content.contains("\"generated_tokens\""));

    // Cleanup
    let _ = std::fs::remove_file(&trace_path);
}

#[test]
fn test_mock_inference_trace_invalid_path() {
    let config = InferenceConfig::new("/dev/null")
        .with_prompt("test")
        .with_max_tokens(4)
        .with_trace_output("/nonexistent/dir/trace.json")
        .with_mock_backend();

    let result = run_inference(&config);
    assert!(result.is_err());
}

// ============================================================================
// INFERENCE RESULT TESTS
// ============================================================================

#[test]
fn test_inference_result_fields() {
    let config = InferenceConfig::new("/dev/null")
        .with_prompt("a b c")
        .with_max_tokens(5)
        .with_mock_backend();

    let result = run_inference(&config).unwrap();

    // All fields should be populated
    assert!(!result.text.is_empty());
    assert!(!result.tokens.is_empty());
    assert!(result.input_token_count > 0);
    assert!(result.generated_token_count > 0);
    assert!(result.inference_ms > 0.0);
    assert!(result.tok_per_sec > 0.0);
    assert!(result.load_ms > 0.0);
    assert!(!result.format.is_empty());
}

#[test]
fn test_inference_result_deterministic() {
    let config = mock_config("deterministic test");

    let result1 = run_inference(&config).unwrap();
    let result2 = run_inference(&config).unwrap();

    // Mock should produce identical results
    assert_eq!(result1.text, result2.text);
    assert_eq!(result1.tokens, result2.tokens);
    assert_eq!(result1.input_token_count, result2.input_token_count);
    assert_eq!(result1.generated_token_count, result2.generated_token_count);
    assert_eq!(result1.format, result2.format);
}

// ============================================================================
// EDGE CASE TESTS
// ============================================================================

#[test]
fn test_mock_inference_empty_prompt() {
    let config = InferenceConfig::new("/dev/null")
        .with_prompt("")
        .with_max_tokens(4)
        .with_mock_backend();

    let result = run_inference(&config).unwrap();

    // Empty prompt should have 0 input tokens (no words to split)
    assert_eq!(result.input_token_count, 0);
    assert_eq!(result.generated_token_count, 4);
}

#[test]
fn test_mock_inference_single_token_generation() {
    let config = InferenceConfig::new("/dev/null")
        .with_prompt("test")
        .with_max_tokens(1)
        .with_mock_backend();

    let result = run_inference(&config).unwrap();

    assert_eq!(result.generated_token_count, 1);
    assert_eq!(result.tokens.last(), Some(&100)); // First generated token
}

#[test]
fn test_mock_inference_long_prompt() {
    let long_prompt = "word ".repeat(100);
    let config = InferenceConfig::new("/dev/null")
        .with_prompt(&long_prompt)
        .with_max_tokens(8)
        .with_mock_backend();

    let result = run_inference(&config).unwrap();

    assert_eq!(result.input_token_count, 100);
    assert_eq!(result.generated_token_count, 8);
}

#[test]
fn test_mock_inference_with_temperature() {
    // Temperature doesn't affect mock output, but config should accept it
    let config = InferenceConfig::new("/dev/null")
        .with_prompt("test")
        .with_temperature(0.9)
        .with_max_tokens(4)
        .with_mock_backend();

    let result = run_inference(&config).unwrap();
    assert!(result.text.contains("test"));
}

#[test]
fn test_mock_inference_with_top_k() {
    // Top-k doesn't affect mock output, but config should accept it
    let config = InferenceConfig::new("/dev/null")
        .with_prompt("test")
        .with_top_k(50)
        .with_max_tokens(4)
        .with_mock_backend();

    let result = run_inference(&config).unwrap();
    assert!(result.text.contains("test"));
}

// ============================================================================
// CLEAN MODEL OUTPUT TESTS
// ============================================================================

#[test]
fn test_clean_model_output_strips_chatml_markers() {
    let raw = "<|im_start|>assistant\nHello there!<|im_end|>";
    let cleaned = clean_model_output(raw);
    assert_eq!(cleaned, "Hello there!");
}

#[test]
fn test_clean_model_output_strips_multiple_markers() {
    let raw = "<|im_start|>assistant<|im_end|><|endoftext|>Test";
    let cleaned = clean_model_output(raw);
    assert_eq!(cleaned, "Test");
}

#[test]
fn test_clean_model_output_preserves_clean_text() {
    let raw = "This is clean text without markers.";
    let cleaned = clean_model_output(raw);
    assert_eq!(cleaned, raw);
}

#[test]
fn test_clean_model_output_trims_whitespace() {
    let raw = "  \n  text with whitespace  \n  ";
    let cleaned = clean_model_output(raw);
    assert_eq!(cleaned, "text with whitespace");
}

// ============================================================================
// KV-CACHE TESTS (Protocol T-COV-95 Directive 3: Popper Falsification)
// ============================================================================
//
// These tests exercise the KV cache to verify:
// 1. Cache state consistency between consecutive generations
// 2. Cache clear properly resets state
// 3. Append/get operations are correct
// 4. Capacity and length tracking is accurate

use crate::apr_transformer::{AprKVCache, AprTransformerConfig};

/// Helper: create a minimal test config for KV cache
fn create_test_kv_config() -> AprTransformerConfig {
    AprTransformerConfig {
        architecture: "test".to_string(),
        hidden_dim: 64,
        num_layers: 2,
        num_heads: 4,
        num_kv_heads: 4,
        intermediate_dim: 128,
        vocab_size: 100,
        context_length: 32,
        rope_theta: 10000.0,
        eps: 1e-5,
            eos_token_id: None,
    ..Default::default()
    }
}

#[test]
fn test_kv_cache_creation() {
    let config = create_test_kv_config();
    let cache = AprKVCache::new(&config);

    assert_eq!(cache.len(), 0);
    assert!(cache.is_empty());
    assert_eq!(cache.capacity(), 32); // context_length
    assert_eq!(cache.num_kv_heads(), 4);
    assert_eq!(cache.head_dim(), 16); // 64 / 4
}

#[test]
fn test_kv_cache_append_single_position() {
    let config = create_test_kv_config();
    let mut cache = AprKVCache::new(&config);

    let kv_size = 4 * 16; // num_kv_heads * head_dim = 64
    let k = vec![1.0f32; kv_size];
    let v = vec![2.0f32; kv_size];

    // Append to both layers for position 0
    cache.append(0, &k, &v);
    cache.append(1, &k, &v);

    assert_eq!(cache.len(), 1);
    assert!(!cache.is_empty());
}

include!("tests_mock_cache.rs");