File size: 13,099 Bytes
79d4fd5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
# Tests for intent classifier
import pytest
from unittest.mock import Mock, patch, MagicMock

from src.agent.intent_classifier import (
    IntentClassifier,
    IntentType,
    ClassificationResult,
    get_intent_classifier,
    reset_intent_classifier,
)


class TestIntentType:
    # Tests for IntentType enum
    
    def test_intent_types_exist(self):
        assert IntentType.DOCUMENT_QUERY is not None
        assert IntentType.DATABASE_QUERY is not None
        assert IntentType.HYBRID_QUERY is not None
        assert IntentType.CONVERSATIONAL is not None
        assert IntentType.AMBIGUOUS is not None
    
    def test_intent_type_values(self):
        assert IntentType.DOCUMENT_QUERY.value == "document_query"
        assert IntentType.DATABASE_QUERY.value == "database_query"
        assert IntentType.HYBRID_QUERY.value == "hybrid_query"
        assert IntentType.CONVERSATIONAL.value == "conversational"
        assert IntentType.AMBIGUOUS.value == "ambiguous"


class TestClassificationResult:
    # Tests for ClassificationResult dataclass
    
    def test_classification_result_creation(self):
        result = ClassificationResult(
            intent=IntentType.DATABASE_QUERY,
            confidence=0.95,
            reasoning="Test reasoning",
            suggested_tables=["users", "orders"],
        )
        
        assert result.intent == IntentType.DATABASE_QUERY
        assert result.confidence == 0.95
        assert result.reasoning == "Test reasoning"
        assert result.suggested_tables == ["users", "orders"]
    
    def test_classification_result_default_values(self):
        result = ClassificationResult(
            intent=IntentType.DOCUMENT_QUERY,
            confidence=0.8,
            reasoning="Default test",
        )
        
        assert result.suggested_tables == []
        assert result.requires_clarification is False
    
    def test_classification_result_to_dict(self):
        result = ClassificationResult(
            intent=IntentType.HYBRID_QUERY,
            confidence=0.85,
            reasoning="Both sources needed",
            suggested_tables=["products"],
        )
        
        data = result.to_dict()
        
        assert data["intent"] == "hybrid_query"
        assert data["confidence"] == 0.85
        assert data["reasoning"] == "Both sources needed"
        assert data["suggested_tables"] == ["products"]
    
    def test_is_database_related(self):
        db_result = ClassificationResult(IntentType.DATABASE_QUERY, 0.9, "test")
        hybrid_result = ClassificationResult(IntentType.HYBRID_QUERY, 0.8, "test")
        doc_result = ClassificationResult(IntentType.DOCUMENT_QUERY, 0.9, "test")
        
        assert db_result.is_database_related() is True
        assert hybrid_result.is_database_related() is True
        assert doc_result.is_database_related() is False
    
    def test_is_document_related(self):
        doc_result = ClassificationResult(IntentType.DOCUMENT_QUERY, 0.9, "test")
        hybrid_result = ClassificationResult(IntentType.HYBRID_QUERY, 0.8, "test")
        db_result = ClassificationResult(IntentType.DATABASE_QUERY, 0.9, "test")
        
        assert doc_result.is_document_related() is True
        assert hybrid_result.is_document_related() is True
        assert db_result.is_document_related() is False


class TestIntentClassifier:
    # Tests for IntentClassifier class
    
    @pytest.fixture
    def classifier(self):
        with patch("src.agent.intent_classifier.Groq") as mock_groq_class:
            mock_client = Mock()
            mock_groq_class.return_value = mock_client
            classifier = IntentClassifier(api_key="test_key")
            classifier._client = mock_client
            return classifier
    
    def test_classifier_initialization(self):
        classifier = IntentClassifier(api_key="test", confidence_threshold=0.8)
        assert classifier._confidence_threshold == 0.8
    
    def test_classifier_with_custom_threshold(self):
        classifier = IntentClassifier(
            api_key="test",
            confidence_threshold=0.5,
        )
        assert classifier._confidence_threshold == 0.5
    
    def test_classify_database_query(self, classifier):
        mock_response = Mock()
        mock_response.choices = [
            Mock(message=Mock(content='{"intent": "DATABASE_QUERY", "confidence": 0.9, "reasoning": "SQL needed", "suggested_tables": ["musteriler"]}'))
        ]
        classifier._client.chat.completions.create.return_value = mock_response
        
        result = classifier.classify("Kac tane musterimiz var?")
        
        assert result.intent == IntentType.DATABASE_QUERY
        assert result.confidence == 0.9
        assert "musteriler" in result.suggested_tables
    
    def test_classify_document_query(self, classifier):
        mock_response = Mock()
        mock_response.choices = [
            Mock(message=Mock(content='{"intent": "DOCUMENT_QUERY", "confidence": 0.85, "reasoning": "Document search", "suggested_tables": []}'))
        ]
        classifier._client.chat.completions.create.return_value = mock_response
        
        result = classifier.classify("Sirketi kim kurdu?")
        
        assert result.intent == IntentType.DOCUMENT_QUERY
    
    def test_classify_hybrid_query(self, classifier):
        mock_response = Mock()
        mock_response.choices = [
            Mock(message=Mock(content='{"intent": "HYBRID_QUERY", "confidence": 0.75, "reasoning": "Both needed", "suggested_tables": ["urunler"]}'))
        ]
        classifier._client.chat.completions.create.return_value = mock_response
        
        result = classifier.classify("Urun politikamiz nedir ve kac urunumuz var?")
        
        assert result.intent == IntentType.HYBRID_QUERY
    
    def test_classify_conversational(self, classifier):
        mock_response = Mock()
        mock_response.choices = [
            Mock(message=Mock(content='{"intent": "CONVERSATIONAL", "confidence": 0.95, "reasoning": "Greeting", "suggested_tables": []}'))
        ]
        classifier._client.chat.completions.create.return_value = mock_response
        
        result = classifier.classify("Merhaba!")
        
        assert result.intent == IntentType.CONVERSATIONAL
    
    def test_classify_with_context(self, classifier):
        mock_response = Mock()
        mock_response.choices = [
            Mock(message=Mock(content='{"intent": "DATABASE_QUERY", "confidence": 0.88, "reasoning": "With context", "suggested_tables": []}'))
        ]
        classifier._client.chat.completions.create.return_value = mock_response
        
        chat_history = [{"role": "user", "content": "previous query"}]
        result = classifier.classify("Devam et", chat_history)
        
        assert result is not None
        classifier._client.chat.completions.create.assert_called_once()
    
    def test_classify_invalid_json_response(self, classifier):
        mock_response = Mock()
        mock_response.choices = [Mock(message=Mock(content="Invalid JSON"))]
        classifier._client.chat.completions.create.return_value = mock_response
        
        result = classifier.classify("Test query")
        
        # Implementation defaults to DOCUMENT_QUERY on parse failure
        assert result.intent == IntentType.DOCUMENT_QUERY
        assert result.confidence == 0.5
    
    def test_classify_api_error(self, classifier):
        classifier._client.chat.completions.create.side_effect = Exception("API Error")
        
        result = classifier.classify("Test query")
        
        # Implementation defaults to DOCUMENT_QUERY on API error
        assert result.intent == IntentType.DOCUMENT_QUERY
        assert result.confidence == 0.5


class TestIntentClassifierSingleton:
    # Tests for singleton pattern
    
    def setup_method(self):
        reset_intent_classifier()
    
    def teardown_method(self):
        reset_intent_classifier()
    
    @patch("src.agent.intent_classifier.Groq")
    def test_get_intent_classifier_creates_instance(self, mock_groq):
        classifier = get_intent_classifier()
        assert classifier is not None
        assert isinstance(classifier, IntentClassifier)
    
    @patch("src.agent.intent_classifier.Groq")
    def test_get_intent_classifier_returns_same_instance(self, mock_groq):
        classifier1 = get_intent_classifier()
        classifier2 = get_intent_classifier()
        assert classifier1 is classifier2
    
    @patch("src.agent.intent_classifier.Groq")
    def test_reset_intent_classifier(self, mock_groq):
        classifier1 = get_intent_classifier()
        reset_intent_classifier()
        classifier2 = get_intent_classifier()
        assert classifier1 is not classifier2


class TestIntentClassifierEdgeCases:
    # Tests for edge cases and confidence thresholds

    @pytest.fixture
    def classifier(self):
        with patch("src.agent.intent_classifier.Groq") as mock_groq_class:
            mock_client = Mock()
            mock_groq_class.return_value = mock_client
            classifier = IntentClassifier(api_key="test_key", confidence_threshold=0.7)
            classifier._client = mock_client
            return classifier

    def test_classify_empty_query(self, classifier):
        # Empty query should return AMBIGUOUS with requires_clarification
        result = classifier.classify("")
        assert result.intent == IntentType.AMBIGUOUS
        assert result.requires_clarification is True
        assert result.confidence == 1.0

    def test_classify_whitespace_only_query(self, classifier):
        result = classifier.classify("   ")
        assert result.intent == IntentType.AMBIGUOUS
        assert result.requires_clarification is True

    def test_classify_with_available_tables(self, classifier):
        # Verify available_tables is included in prompt
        mock_response = Mock()
        mock_response.choices = [
            Mock(message=Mock(content='{"intent": "DATABASE_QUERY", "confidence": 0.9, "reasoning": "Tables hint", "suggested_tables": ["CreditFileVersion"]}'))
        ]
        classifier._client.chat.completions.create.return_value = mock_response

        result = classifier.classify(
            "Kac kredi dosyasi var?",
            available_tables=["CreditFileVersion", "CreditFileElement", "Contract"],
        )
        assert result.intent == IntentType.DATABASE_QUERY
        # Verify the LLM was called with tables context
        call_args = classifier._client.chat.completions.create.call_args
        user_msg = call_args[1]["messages"][1]["content"]
        assert "CreditFileVersion" in user_msg

    def test_low_confidence_sets_requires_clarification(self, classifier):
        mock_response = Mock()
        mock_response.choices = [
            Mock(message=Mock(content='{"intent": "DATABASE_QUERY", "confidence": 0.4, "reasoning": "Not sure"}'))
        ]
        classifier._client.chat.completions.create.return_value = mock_response

        result = classifier.classify("bilgi ver")
        assert result.confidence == 0.4
        assert result.requires_clarification is True

    def test_high_confidence_no_clarification(self, classifier):
        mock_response = Mock()
        mock_response.choices = [
            Mock(message=Mock(content='{"intent": "DOCUMENT_QUERY", "confidence": 0.95, "reasoning": "Clear question"}'))
        ]
        classifier._client.chat.completions.create.return_value = mock_response

        result = classifier.classify("Atlas ERP nedir?")
        assert result.confidence == 0.95
        assert result.requires_clarification is False

    def test_classify_json_with_code_fence(self, classifier):
        # Test that JSON wrapped in code fences is parsed correctly
        mock_response = Mock()
        mock_response.choices = [
            Mock(message=Mock(content='```json\n{"intent": "CONVERSATIONAL", "confidence": 0.9, "reasoning": "Greeting"}\n```'))
        ]
        classifier._client.chat.completions.create.return_value = mock_response

        result = classifier.classify("Merhaba!")
        assert result.intent == IntentType.CONVERSATIONAL

    def test_classify_batch(self, classifier):
        mock_response = Mock()
        mock_response.choices = [
            Mock(message=Mock(content='{"intent": "DOCUMENT_QUERY", "confidence": 0.8, "reasoning": "test"}'))
        ]
        classifier._client.chat.completions.create.return_value = mock_response

        results = classifier.classify_batch(["q1", "q2", "q3"])
        assert len(results) == 3
        assert all(isinstance(r, ClassificationResult) for r in results)

    def test_classification_time_recorded(self, classifier):
        mock_response = Mock()
        mock_response.choices = [
            Mock(message=Mock(content='{"intent": "DOCUMENT_QUERY", "confidence": 0.8, "reasoning": "test"}'))
        ]
        classifier._client.chat.completions.create.return_value = mock_response

        result = classifier.classify("test query")
        assert result.classification_time_ms >= 0