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
|
# ---------------------------------------------------------
# Copyright (c) Microsoft Corporation. All rights reserved.
# ---------------------------------------------------------
from unittest.mock import MagicMock
import pytest
from azure.ai.evaluation._evaluators._task_completion import _TaskCompletionEvaluator
from azure.ai.evaluation._exceptions import EvaluationException
# Mock flow side effect that mimics the output of the prompty (the _flow call)
async def flow_side_effect(timeout, **kwargs):
query = kwargs.get("query", "")
response = kwargs.get("response", "")
# Simple logic to determine success based on keywords in query and response
if "complete" in str(response).lower() or "done" in str(response).lower():
success = 1
explanation = "Task was completed successfully with all requirements met."
details = {
"task_requirements": "The requirements for the task.",
"delivered_outcome": "The final deliverable of the task.",
"completion_gaps": "complete",
}
elif "partial" in str(response).lower():
success = 0
explanation = "Task was only partially completed, missing key requirements."
details = {
"task_requirements": "The requirements for the task.",
"delivered_outcome": "The final deliverable of the task.",
"completion_gaps": "incomplete",
}
elif "invalid" in str(response).lower():
# Return invalid output to test error handling
success = "invalid_value"
explanation = "This is an invalid response."
details = {}
else:
success = 0
explanation = "Task was not completed."
details = {
"task_requirements": "The requirements for the task.",
"delivered_outcome": "none",
"completion_gaps": "failed",
}
return {
"llm_output": {
"success": success,
"explanation": explanation,
"details": details,
},
}
@pytest.mark.usefixtures("mock_model_config")
@pytest.mark.unittest
class TestTaskCompletionEvaluator:
def test_task_completed_successfully(self, mock_model_config):
"""Test when task is completed successfully"""
evaluator = _TaskCompletionEvaluator(model_config=mock_model_config)
evaluator._flow = MagicMock(side_effect=flow_side_effect)
query = "Plan a 3-day itinerary for Paris with cultural landmarks and local cuisine."
response = "I have completed a comprehensive 3-day Paris itinerary with cultural landmarks and local cuisine recommendations."
result = evaluator(query=query, response=response)
key = _TaskCompletionEvaluator._RESULT_KEY
assert result is not None
assert key in result
assert result[key] == 1
assert result[f"{key}_result"] == "pass"
assert f"{key}_reason" in result
assert "completed successfully" in result[f"{key}_reason"]
assert f"{key}_details" in result
assert isinstance(result[f"{key}_details"], dict)
def test_task_not_completed(self, mock_model_config):
"""Test when task is not completed"""
evaluator = _TaskCompletionEvaluator(model_config=mock_model_config)
evaluator._flow = MagicMock(side_effect=flow_side_effect)
query = "Write a detailed analysis of market trends."
response = "I cannot provide this analysis at the moment."
result = evaluator(query=query, response=response)
key = _TaskCompletionEvaluator._RESULT_KEY
assert result is not None
assert key in result
assert result[key] == 0
assert result[f"{key}_result"] == "fail"
assert f"{key}_reason" in result
assert f"{key}_details" in result
def test_task_partially_completed(self, mock_model_config):
"""Test when task is only partially completed"""
evaluator = _TaskCompletionEvaluator(model_config=mock_model_config)
evaluator._flow = MagicMock(side_effect=flow_side_effect)
query = "Create a budget plan with income, expenses, and savings goals."
response = "I have provided a partial budget plan with income and expenses only."
result = evaluator(query=query, response=response)
key = _TaskCompletionEvaluator._RESULT_KEY
assert result is not None
assert key in result
assert result[key] == 0
assert result[f"{key}_result"] == "fail"
assert "partial" in result[f"{key}_reason"].lower()
assert f"{key}_details" in result
def test_with_tool_definitions(self, mock_model_config):
"""Test evaluation with tool definitions provided"""
evaluator = _TaskCompletionEvaluator(model_config=mock_model_config)
evaluator._flow = MagicMock(side_effect=flow_side_effect)
query = [{"role": "user", "content": "Find hotels in Paris and book the cheapest one"}]
response = [
{"role": "assistant", "content": "Task is complete. I found hotels and booked the cheapest option."}
]
tool_definitions = [
{
"name": "search_hotels",
"description": "Search for hotels in a location",
"parameters": {
"type": "object",
"properties": {"location": {"type": "string", "description": "City name"}},
},
},
{
"name": "book_hotel",
"description": "Book a hotel",
"parameters": {
"type": "object",
"properties": {"hotel_id": {"type": "string", "description": "Hotel identifier"}},
},
},
]
result = evaluator(query=query, response=response, tool_definitions=tool_definitions)
key = _TaskCompletionEvaluator._RESULT_KEY
assert result is not None
assert key in result
assert result[key] == 1
assert result[f"{key}_result"] == "pass"
def test_with_conversation_history(self, mock_model_config):
"""Test evaluation with conversation history as list of messages"""
evaluator = _TaskCompletionEvaluator(model_config=mock_model_config)
evaluator._flow = MagicMock(side_effect=flow_side_effect)
query = [
{"role": "system", "content": "You are a helpful travel assistant."},
{"role": "user", "content": "I need to book a flight to Tokyo."},
{"role": "assistant", "content": "I can help you with that. What dates?"},
{"role": "user", "content": "December 15-22, 2025"},
]
response = [
{"role": "assistant", "content": "Done! I have booked your flight to Tokyo for December 15-22, 2025."}
]
result = evaluator(query=query, response=response)
key = _TaskCompletionEvaluator._RESULT_KEY
assert result is not None
assert key in result
assert result[key] == 1
assert result[f"{key}_result"] == "pass"
def test_missing_query_and_response(self, mock_model_config):
"""Test that missing both query and response raises an exception"""
evaluator = _TaskCompletionEvaluator(model_config=mock_model_config)
evaluator._flow = MagicMock(side_effect=flow_side_effect)
with pytest.raises(EvaluationException) as exc_info:
evaluator()
assert "Either 'conversation' or individual inputs must be provided" in str(exc_info.value)
def test_string_success_value_true(self, mock_model_config):
"""Test handling of string 'TRUE' as success value"""
evaluator = _TaskCompletionEvaluator(model_config=mock_model_config)
async def string_true_flow(timeout, **kwargs):
return {
"llm_output": {
"success": "TRUE",
"explanation": "Task completed",
"details": {},
}
}
evaluator._flow = MagicMock(side_effect=string_true_flow)
query = "Complete this task"
response = "Task is complete"
result = evaluator(query=query, response=response)
key = _TaskCompletionEvaluator._RESULT_KEY
assert result[key] == 1
assert result[f"{key}_result"] == "pass"
def test_string_success_value_false(self, mock_model_config):
"""Test handling of string 'FALSE' as success value"""
evaluator = _TaskCompletionEvaluator(model_config=mock_model_config)
async def string_false_flow(timeout, **kwargs):
return {
"llm_output": {
"success": "FALSE",
"explanation": "Task not completed",
"details": {},
}
}
evaluator._flow = MagicMock(side_effect=string_false_flow)
query = "Complete this task"
response = "Could not complete"
result = evaluator(query=query, response=response)
key = _TaskCompletionEvaluator._RESULT_KEY
assert result[key] == 0
assert result[f"{key}_result"] == "fail"
def test_invalid_llm_output_format(self, mock_model_config):
"""Test handling when LLM output is not a dictionary"""
evaluator = _TaskCompletionEvaluator(model_config=mock_model_config)
async def invalid_output_flow(timeout, **kwargs):
return {"llm_output": "This is a string, not a dictionary"}
evaluator._flow = MagicMock(side_effect=invalid_output_flow)
query = "Complete this task"
response = "Done"
result = evaluator(query=query, response=response)
key = _TaskCompletionEvaluator._RESULT_KEY
assert result[key] == 0
def test_default_success_value(self, mock_model_config):
"""Test that default success value is 0 when not provided"""
evaluator = _TaskCompletionEvaluator(model_config=mock_model_config)
async def no_success_flow(timeout, **kwargs):
return {
"llm_output": {
"explanation": "No success field provided",
"details": {},
}
}
evaluator._flow = MagicMock(side_effect=no_success_flow)
query = "Complete this task"
response = "Task response"
result = evaluator(query=query, response=response)
key = _TaskCompletionEvaluator._RESULT_KEY
assert result[key] == 0
assert result[f"{key}_result"] == "fail"
def test_complex_response_with_tool_calls(self, mock_model_config):
"""Test evaluation with response containing tool calls"""
evaluator = _TaskCompletionEvaluator(model_config=mock_model_config)
evaluator._flow = MagicMock(side_effect=flow_side_effect)
query = "Search for restaurants and make a reservation"
response = [
{
"role": "assistant",
"content": [{"type": "text", "text": "Let me search for restaurants."}],
"tool_calls": [
{
"type": "function",
"id": "call_1",
"function": {
"name": "search_restaurants",
"arguments": '{"location": "downtown", "cuisine": "Italian"}',
},
}
],
},
{"role": "tool", "tool_call_id": "call_1", "content": "Found 5 Italian restaurants downtown."},
{"role": "assistant", "content": "Task complete! I found restaurants and made a reservation."},
]
tool_definitions = [
{
"name": "search_restaurants",
"type": "function",
"description": "Search for restaurants",
"parameters": {"type": "object", "properties": {}},
}
]
result = evaluator(query=query, response=response, tool_definitions=tool_definitions)
key = _TaskCompletionEvaluator._RESULT_KEY
assert result is not None
assert key in result
assert result[key] == 1
assert result[f"{key}_result"] == "pass"
def test_numeric_success_values(self, mock_model_config):
"""Test that numeric 0 and 1 values work correctly"""
evaluator = _TaskCompletionEvaluator(model_config=mock_model_config)
async def numeric_success_flow(timeout, **kwargs):
# Return 1 for success
return {
"llm_output": {
"success": 1,
"explanation": "Task completed",
"details": {},
}
}
evaluator._flow = MagicMock(side_effect=numeric_success_flow)
query = "Complete this task"
response = "Done"
result = evaluator(query=query, response=response)
key = _TaskCompletionEvaluator._RESULT_KEY
assert result[key] == 1
assert result[f"{key}_result"] == "pass"
def test_empty_details(self, mock_model_config):
"""Test handling of empty details field"""
evaluator = _TaskCompletionEvaluator(model_config=mock_model_config)
async def empty_details_flow(timeout, **kwargs):
return {
"llm_output": {
"success": 1,
"explanation": "Task completed",
# No details field
}
}
evaluator._flow = MagicMock(side_effect=empty_details_flow)
query = "Test"
response = "Complete"
result = evaluator(query=query, response=response)
key = _TaskCompletionEvaluator._RESULT_KEY
assert f"{key}_details" in result
assert result[f"{key}_details"] == ""
def test_empty_explanation(self, mock_model_config):
"""Test handling of missing explanation field"""
evaluator = _TaskCompletionEvaluator(model_config=mock_model_config)
async def no_explanation_flow(timeout, **kwargs):
return {
"llm_output": {
"success": 1,
"details": {},
# No explanation field
}
}
evaluator._flow = MagicMock(side_effect=no_explanation_flow)
query = "Test"
response = "Complete"
result = evaluator(query=query, response=response)
key = _TaskCompletionEvaluator._RESULT_KEY
assert f"{key}_reason" in result
assert result[f"{key}_reason"] == ""
|
