nirbarazida
/
promptfoo
mirror of https://github.com/promptfoo/promptfoo


  
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

	
            import request from 'supertest';
import { runDbMigrations } from '../../src/migrate';
import Eval from '../../src/models/eval';
import EvalResult from '../../src/models/evalResult';
import { createApp } from '../../src/server/server';
import invariant from '../../src/util/invariant';
import EvalFactory from '../factories/evalFactory';

describe('eval routes', () => {
  let app: ReturnType<typeof createApp>;
  const testEvalIds = new Set<string>();

  beforeAll(async () => {
    await runDbMigrations();
  });

  beforeEach(() => {
    app = createApp();
  });

  afterEach(async () => {
    // More robust cleanup with proper error handling
    const cleanupPromises = Array.from(testEvalIds).map(async (evalId) => {
      try {
        const eval_ = await Eval.findById(evalId);
        if (eval_) {
          await eval_.delete();
        }
      } catch (error) {
        // Log the error instead of silently ignoring it
        console.error(`Failed to cleanup eval ${evalId}:`, error);
      }
    });

    // Wait for all cleanups to complete
    await Promise.allSettled(cleanupPromises);
    testEvalIds.clear();
  });

  function createManualRatingPayload(originalResult: any, pass: boolean) {
    const payload = { ...originalResult.gradingResult };
    const score = pass ? 1 : 0;
    payload.componentResults?.push({
      pass,
      score,
      reason: 'Manual result (overrides all other grading results)',
      assertion: { type: 'human' },
    });
    payload.reason = 'Manual result (overrides all other grading results)';
    payload.pass = pass;
    payload.score = score;
    return payload;
  }

  describe('post("/:evalId/results/:id/rating")', () => {
    it('Passing test and the user marked it as passing (no change)', async () => {
      const eval_ = await EvalFactory.create();
      testEvalIds.add(eval_.id);

      const results = await eval_.getResults();
      const result = results[0];
      expect(eval_.prompts[result.promptIdx].metrics?.assertPassCount).toBe(1);
      expect(eval_.prompts[result.promptIdx].metrics?.assertFailCount).toBe(1);
      expect(eval_.prompts[result.promptIdx].metrics?.testPassCount).toBe(1);
      expect(eval_.prompts[result.promptIdx].metrics?.testFailCount).toBe(1);
      expect(eval_.prompts[result.promptIdx].metrics?.score).toBe(1);

      expect(result.gradingResult?.pass).toBe(true);
      expect(result.gradingResult?.score).toBe(1);
      const ratingPayload = createManualRatingPayload(result, true);

      const res = await request(app)
        .post(`/api/eval/${eval_.id}/results/${result.id}/rating`)
        .send(ratingPayload);

      expect(res.status).toBe(200);
      invariant(result.id, 'Result ID is required');
      const updatedResult = await EvalResult.findById(result.id);
      expect(updatedResult?.gradingResult?.pass).toBe(true);
      expect(updatedResult?.gradingResult?.score).toBe(1);
      expect(updatedResult?.gradingResult?.componentResults).toHaveLength(2);
      expect(updatedResult?.gradingResult?.reason).toBe(
        'Manual result (overrides all other grading results)',
      );

      const updatedEval = await Eval.findById(eval_.id);
      invariant(updatedEval, 'Eval is required');
      expect(updatedEval.prompts[result.promptIdx].metrics?.score).toBe(1);
      expect(updatedEval.prompts[result.promptIdx].metrics?.assertPassCount).toBe(2);
      expect(updatedEval.prompts[result.promptIdx].metrics?.assertFailCount).toBe(1);
      expect(updatedEval.prompts[result.promptIdx].metrics?.testPassCount).toBe(1);
      expect(updatedEval.prompts[result.promptIdx].metrics?.testFailCount).toBe(1);
    });

    it('Passing test and the user changed it to failing', async () => {
      const eval_ = await EvalFactory.create();
      testEvalIds.add(eval_.id);

      const results = await eval_.getResults();
      const result = results[0];
      expect(eval_.prompts[result.promptIdx].metrics?.assertPassCount).toBe(1);
      expect(eval_.prompts[result.promptIdx].metrics?.assertFailCount).toBe(1);
      expect(eval_.prompts[result.promptIdx].metrics?.testPassCount).toBe(1);
      expect(eval_.prompts[result.promptIdx].metrics?.testFailCount).toBe(1);
      expect(eval_.prompts[result.promptIdx].metrics?.score).toBe(1);

      expect(result.gradingResult?.pass).toBe(true);
      expect(result.gradingResult?.score).toBe(1);
      const ratingPayload = createManualRatingPayload(result, false);
      const res = await request(app)
        .post(`/api/eval/${eval_.id}/results/${result.id}/rating`)
        .send(ratingPayload);

      expect(res.status).toBe(200);
      invariant(result.id, 'Result ID is required');
      const updatedResult = await EvalResult.findById(result.id);
      expect(updatedResult?.gradingResult?.pass).toBe(false);
      expect(updatedResult?.gradingResult?.score).toBe(0);
      expect(updatedResult?.gradingResult?.componentResults).toHaveLength(2);
      expect(updatedResult?.gradingResult?.reason).toBe(
        'Manual result (overrides all other grading results)',
      );
      const updatedEval = await Eval.findById(eval_.id);
      invariant(updatedEval, 'Eval is required');
      expect(updatedEval.prompts[result.promptIdx].metrics?.assertPassCount).toBe(1);
      expect(updatedEval.prompts[result.promptIdx].metrics?.assertFailCount).toBe(2);
      expect(updatedEval.prompts[result.promptIdx].metrics?.score).toBe(0);
      expect(updatedEval.prompts[result.promptIdx].metrics?.testPassCount).toBe(0);
      expect(updatedEval.prompts[result.promptIdx].metrics?.testFailCount).toBe(2);
    });

    it('Failing test and the user changed it to passing', async () => {
      const eval_ = await EvalFactory.create();
      testEvalIds.add(eval_.id);

      const results = await eval_.getResults();
      const result = results[1];
      expect(eval_.prompts[result.promptIdx].metrics?.assertPassCount).toBe(1);
      expect(eval_.prompts[result.promptIdx].metrics?.assertFailCount).toBe(1);
      expect(eval_.prompts[result.promptIdx].metrics?.testPassCount).toBe(1);
      expect(eval_.prompts[result.promptIdx].metrics?.testFailCount).toBe(1);
      expect(eval_.prompts[result.promptIdx].metrics?.score).toBe(1);

      expect(result.gradingResult?.pass).toBe(false);
      expect(result.gradingResult?.score).toBe(0);

      const ratingPayload = createManualRatingPayload(result, true);

      const res = await request(app)
        .post(`/api/eval/${eval_.id}/results/${result.id}/rating`)
        .send(ratingPayload);

      expect(res.status).toBe(200);

      invariant(result.id, 'Result ID is required');

      const updatedResult = await EvalResult.findById(result.id);
      expect(updatedResult?.gradingResult?.pass).toBe(true);
      expect(updatedResult?.gradingResult?.score).toBe(1);
      expect(updatedResult?.gradingResult?.componentResults).toHaveLength(2);
      expect(updatedResult?.gradingResult?.reason).toBe(
        'Manual result (overrides all other grading results)',
      );

      const updatedEval = await Eval.findById(eval_.id);
      invariant(updatedEval, 'Eval is required');
      expect(updatedEval.prompts[result.promptIdx].metrics?.score).toBe(2);
      expect(updatedEval.prompts[result.promptIdx].metrics?.assertPassCount).toBe(2);
      expect(updatedEval.prompts[result.promptIdx].metrics?.assertFailCount).toBe(1);
      expect(updatedEval.prompts[result.promptIdx].metrics?.testPassCount).toBe(2);
      expect(updatedEval.prompts[result.promptIdx].metrics?.testFailCount).toBe(0);
    });

    it('Failing test and the user marked it as failing (no change)', async () => {
      const eval_ = await EvalFactory.create();
      testEvalIds.add(eval_.id);

      const results = await eval_.getResults();
      const result = results[1];
      expect(eval_.prompts[result.promptIdx].metrics?.assertPassCount).toBe(1);
      expect(eval_.prompts[result.promptIdx].metrics?.assertFailCount).toBe(1);
      expect(eval_.prompts[result.promptIdx].metrics?.testPassCount).toBe(1);
      expect(eval_.prompts[result.promptIdx].metrics?.testFailCount).toBe(1);
      expect(eval_.prompts[result.promptIdx].metrics?.score).toBe(1);

      expect(result.gradingResult?.pass).toBe(false);
      expect(result.gradingResult?.score).toBe(0);
      const ratingPayload = createManualRatingPayload(result, false);
      const res = await request(app)
        .post(`/api/eval/${eval_.id}/results/${result.id}/rating`)
        .send(ratingPayload);

      expect(res.status).toBe(200);
      invariant(result.id, 'Result ID is required');
      const updatedResult = await EvalResult.findById(result.id);
      expect(updatedResult?.gradingResult?.pass).toBe(false);
      expect(updatedResult?.gradingResult?.score).toBe(0);
      expect(updatedResult?.gradingResult?.componentResults).toHaveLength(2);
      expect(updatedResult?.gradingResult?.reason).toBe(
        'Manual result (overrides all other grading results)',
      );
      const updatedEval = await Eval.findById(eval_.id);
      invariant(updatedEval, 'Eval is required');
      expect(updatedEval.prompts[result.promptIdx].metrics?.assertPassCount).toBe(1);
      expect(updatedEval.prompts[result.promptIdx].metrics?.assertFailCount).toBe(2);
      expect(updatedEval.prompts[result.promptIdx].metrics?.score).toBe(1);
      expect(updatedEval.prompts[result.promptIdx].metrics?.testPassCount).toBe(1);
      expect(updatedEval.prompts[result.promptIdx].metrics?.testFailCount).toBe(1);
    });
  });
});