File size: 8,631 Bytes
11bcc35
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
"""
Unit Tests β€” GraphRAG Inference Hackathon
==========================================
Tests for core utility functions across all layers.
Run: python -m pytest tests/ -v
"""
import sys
import os
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))


# ── Layer 1: Graph Layer Tests ─────────────────────────

def test_cosine_similarity_identical():
    from graphrag.layers.graph_layer import cosine_similarity
    assert cosine_similarity([1, 0, 0], [1, 0, 0]) == 1.0

def test_cosine_similarity_orthogonal():
    from graphrag.layers.graph_layer import cosine_similarity
    assert cosine_similarity([1, 0, 0], [0, 1, 0]) == 0.0

def test_cosine_similarity_opposite():
    from graphrag.layers.graph_layer import cosine_similarity
    assert abs(cosine_similarity([1, 0], [-1, 0]) - (-1.0)) < 1e-9

def test_cosine_similarity_zero_vector():
    from graphrag.layers.graph_layer import cosine_similarity
    assert cosine_similarity([0, 0, 0], [1, 2, 3]) == 0.0

def test_cosine_similarity_mismatched_lengths():
    from graphrag.layers.graph_layer import cosine_similarity
    assert cosine_similarity([1, 2], [1, 2, 3]) == 0.0

def test_chunk_text_basic():
    from graphrag.layers.graph_layer import chunk_text
    text = "Hello world. " * 100
    chunks = chunk_text(text, chunk_size=200, overlap=20)
    assert len(chunks) > 1
    assert all(len(c) <= 220 for c in chunks)  # allow slight overshoot for sentence boundary

def test_chunk_text_empty():
    from graphrag.layers.graph_layer import chunk_text
    assert chunk_text("") == []
    assert chunk_text(None) == []

def test_chunk_text_short():
    from graphrag.layers.graph_layer import chunk_text
    result = chunk_text("Short text.", chunk_size=1000)
    assert len(result) == 1
    assert result[0] == "Short text."

def test_chunk_text_overlap():
    from graphrag.layers.graph_layer import chunk_text
    text = "A" * 500 + " " + "B" * 500
    chunks = chunk_text(text, chunk_size=300, overlap=50)
    assert len(chunks) >= 3

def test_generate_entity_id_deterministic():
    from graphrag.layers.graph_layer import generate_entity_id
    id1 = generate_entity_id("Albert Einstein", "PERSON")
    id2 = generate_entity_id("Albert Einstein", "PERSON")
    assert id1 == id2

def test_generate_entity_id_case_insensitive():
    from graphrag.layers.graph_layer import generate_entity_id
    id1 = generate_entity_id("Albert Einstein", "PERSON")
    id2 = generate_entity_id("albert einstein", "person")
    assert id1 == id2

def test_generate_entity_id_different_types():
    from graphrag.layers.graph_layer import generate_entity_id
    id1 = generate_entity_id("Apple", "ORGANIZATION")
    id2 = generate_entity_id("Apple", "PRODUCT")
    assert id1 != id2

def test_generate_chunk_id():
    from graphrag.layers.graph_layer import generate_chunk_id
    assert generate_chunk_id("doc1", 0) == "doc1_chunk_0000"
    assert generate_chunk_id("doc1", 42) == "doc1_chunk_0042"


# ── Layer 4: Evaluation Tests ─────────────────────────

def test_normalize_answer():
    from graphrag.layers.evaluation_layer import normalize_answer
    assert normalize_answer("The Answer") == "answer"
    assert normalize_answer("  a  big   space  ") == "big space"
    assert normalize_answer("Hello, World!") == "hello world"

def test_compute_f1_perfect():
    from graphrag.layers.evaluation_layer import compute_f1
    assert compute_f1("the cat sat", "the cat sat") == 1.0

def test_compute_f1_partial():
    from graphrag.layers.evaluation_layer import compute_f1
    score = compute_f1("the cat sat on the mat", "the cat sat")
    assert 0.5 < score < 1.0

def test_compute_f1_no_overlap():
    from graphrag.layers.evaluation_layer import compute_f1
    assert compute_f1("dogs run fast", "cats sit quietly") == 0.0

def test_compute_f1_empty():
    from graphrag.layers.evaluation_layer import compute_f1
    assert compute_f1("", "") == 1.0
    assert compute_f1("something", "") == 0.0
    assert compute_f1("", "something") == 0.0

def test_compute_exact_match():
    from graphrag.layers.evaluation_layer import compute_exact_match
    assert compute_exact_match("Yes", "yes") == 1.0
    assert compute_exact_match("The answer", "the answer") == 1.0
    assert compute_exact_match("Yes", "No") == 0.0

def test_compute_context_hit_rate():
    from graphrag.layers.evaluation_layer import compute_context_hit_rate
    contexts = ["Einstein was born in Germany.", "He developed relativity."]
    facts = ["Einstein was born in Germany.", "He won Nobel Prize."]
    rate = compute_context_hit_rate(contexts, facts)
    assert rate == 0.5

def test_compute_context_hit_rate_empty():
    from graphrag.layers.evaluation_layer import compute_context_hit_rate
    assert compute_context_hit_rate([], []) == 0.0
    assert compute_context_hit_rate(["something"], []) == 0.0

def test_compute_token_efficiency():
    from graphrag.layers.evaluation_layer import compute_token_efficiency
    assert compute_token_efficiency(100, 250) == 2.5
    assert compute_token_efficiency(100, 50) == 0.5
    assert compute_token_efficiency(0, 100) == 0.0


# ── Universal LLM Tests ──────────────────────────────

def test_provider_registry_completeness():
    from graphrag.layers.universal_llm import PROVIDERS
    expected = {"openai", "anthropic", "gemini", "mistral", "cohere",
                "ollama", "openrouter", "groq", "xai", "together",
                "huggingface", "deepseek"}
    assert set(PROVIDERS.keys()) == expected

def test_provider_has_required_fields():
    from graphrag.layers.universal_llm import PROVIDERS
    for pid, cfg in PROVIDERS.items():
        assert "name" in cfg, f"{pid} missing name"
        assert "default_model" in cfg, f"{pid} missing default_model"
        assert "litellm_prefix" in cfg, f"{pid} missing litellm_prefix"
        assert "cost_input" in cfg, f"{pid} missing cost_input"
        assert "cost_output" in cfg, f"{pid} missing cost_output"

def test_ollama_is_free():
    from graphrag.layers.universal_llm import PROVIDERS
    ollama = PROVIDERS["ollama"]
    assert ollama["cost_input"] == 0
    assert ollama["cost_output"] == 0
    assert ollama.get("is_local") is True

def test_get_available_providers_includes_ollama():
    from graphrag.layers.universal_llm import get_available_providers
    available = get_available_providers()
    assert "ollama" in available  # always included as local


# ── Evaluation Layer Aggregate Tests ──────────────────

def test_evaluation_layer_aggregate():
    from graphrag.layers.evaluation_layer import EvaluationLayer, EvalSample
    evl = EvaluationLayer()
    sample = EvalSample(
        query="test?", reference_answer="yes",
        baseline_answer="yes", graphrag_answer="yes indeed",
        question_type="factoid", difficulty="easy",
    )
    evl.evaluate_sample(sample, baseline_tokens=100, graphrag_tokens=200,
                        baseline_cost=0.001, graphrag_cost=0.002)
    agg = evl.compute_aggregate_metrics()
    assert agg["num_samples"] == 1
    assert agg["baseline"]["avg_f1"] > 0
    assert agg["graphrag"]["avg_f1"] > 0

def test_evaluation_layer_report():
    from graphrag.layers.evaluation_layer import EvaluationLayer, EvalSample
    evl = EvaluationLayer()
    for i in range(3):
        sample = EvalSample(query=f"q{i}?", reference_answer="answer",
                            baseline_answer="answer", graphrag_answer="answer",
                            question_type="bridge" if i % 2 == 0 else "comparison")
        evl.evaluate_sample(sample, baseline_tokens=100+i*10, graphrag_tokens=200+i*20)
    report = evl.generate_report()
    assert "BENCHMARK REPORT" in report
    assert "bridge" in report or "comparison" in report


if __name__ == "__main__":
    # Run all tests
    import traceback
    tests = [v for k, v in sorted(globals().items()) if k.startswith("test_")]
    passed = failed = 0
    for test_fn in tests:
        try:
            test_fn()
            print(f"  βœ… {test_fn.__name__}")
            passed += 1
        except Exception as e:
            print(f"  ❌ {test_fn.__name__}: {e}")
            traceback.print_exc()
            failed += 1
    print(f"\n{'='*50}")
    print(f"Results: {passed} passed, {failed} failed, {passed+failed} total")
    if failed == 0:
        print("πŸŽ‰ ALL TESTS PASSED!")
    else:
        print(f"⚠️  {failed} tests failed")