File size: 15,568 Bytes
5374a2d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import os 
import regex
from typing import Union, Any, List, Callable
from ..core.logging import logger
from .benchmark import CodingBenchmark 
from ..utils.utils import download_file
from ..core.module_utils import load_json
from ..utils.aflow_utils.data_utils import AFLOW_DATASET_FILES_MAP, download_aflow_benchmark_data


def download_raw_mbpp_data(name: str, save_folder: str):
    url = "https://raw.githubusercontent.com/google-research/google-research/master/mbpp/sanitized-mbpp.json"
    logger.info(f"Downloading MBPP data from: {url}")
    download_file(url=url, save_file=os.path.join(save_folder, name))


def load_mbpp_data(data_path: str):

    """
    load MBPP data from the given path and convert to HumanEval format
    """

    def extract_func_name(func_header: str) -> str:
        func_name_pattern = r"def\s+([a-zA-Z_]\w*)\s*\("
        match = regex.search(func_name_pattern, func_header)
        if match:
            return match.group(1)
        else:
            return None

    def extract_func_header(code: str, test_list: List[str]) -> str:
        lines = code.split("\n")
        imports, defs = [], []
        for line in lines:
            if line.startswith("def "):
                break
            imports.append(line)
        for line in lines:
            if line.startswith("def "):
                defs.append(line)
        func_head = None
        for header in defs:
            func_name = extract_func_name(header)
            if func_name is None:
                continue 
            if all(func_name in test for test in test_list):
                func_head = header 
                break 
        if func_head is None:
            logger.warning(f"No function header found for {code}")
        return ("\n".join(imports) + "\n\n" + func_head).strip() 

    data = load_json(data_path, type="json")

    for example in data:
        original_prompt = example["prompt"] 
        code = example["code"]
        test_list = [assert_str.strip() for assert_str in example["test_list"]]
        func_header = extract_func_header(code, test_list)

        if example["task_id"] == 56:
            # change the `check` function to `check_answer`
            func_header = func_header.replace("check", "check_answer")
            code = code.replace("check", "check_answer")
            test_list = [test.replace("check", "check_answer") for test in test_list]

        prompt = example["prompt"] + "\n\n" + func_header + "\n"
        canonical_solution = code 
        test = "def check(candidate):\n    " + "\n    ".join(test_list) + "\n" 
        entry_point = extract_func_name(func_header)

        example["prompt"] = prompt 
        example["entry_point"] = entry_point 
        example["canonical_solution"] = canonical_solution 
        example["test"] = test 
        example["original_prompt"] = original_prompt 
    
    return data


class MBPPPLUS(CodingBenchmark):

    """Benchmark class for evaluating code generation on the MBPP dataset.
    
    MBPP (Mostly Basic Python Programming) is a collection of Python programming 
    problems designed to test a model's ability to generate functionally correct 
    code from natural language descriptions. This class handles loading the dataset, 
    evaluating solutions, and computing metrics such as pass@k.
    
    The original MBPP format is transformed to be compatible with the HumanEval
    benchmark format, allowing for consistent evaluation infrastructure.
    
    Each MBPP example has the following structure:
    {
        "task_id" (int): 2, 
        "prompt" (str): "Write a function to find the shared elements from the given two lists.",
        "code" (str): "def similar_elements(test_tup1, test_tup2):\n  res = tuple(set(test_tup1) & set(test_tup2))\n  return (res) ", 
        "test_imports": [] 
        "test_list" (List[str]): ['assert set(similar_elements((3, 4, 5, 6),(5, 7, 4, 10))) == set((4, 5))', 'assert set(similar_elements((1, 2, 3, 4),(5, 4, 3, 7))) == set((3, 4))', 'assert set(similar_elements((11, 12, 14, 13),(17, 15, 14, 13))) == set((13, 14))']
    }
    
    Attributes:
        k: An integer or list of integers specifying which pass@k metrics to compute
    """

    def __init__(self, path: str = None, mode: str = "all", timeout: int = 60, k: Union[int, list] = 1,**kwargs):
        path = os.path.expanduser(path or "~/.evoagentx/data/mbpp")
        self.k = k 
        self.name = "mbpp"
        super().__init__(name=type(self).__name__, path=path, mode=mode, timeout=timeout, **kwargs)
        
        
    def _load_data_from_file(self, file_name: str):
        if file_name is None:
            return None
        file_path = os.path.join(self.path, file_name)
        if not os.path.exists(file_path):
            download_aflow_benchmark_data(dataset="mbpp", save_folder=self.path)
        
        return load_json(path=file_path, type="jsonl")
#     def _load_data_from_file(self, file_name: str):
# #         if file_name is None:
# #             return None
# #         file_path = os.path.join(self.path, file_name)
# #         if not os.path.exists(file_path):
# #             download_aflow_benchmark_data(dataset="mbpp", save_folder=self.path)
#         if self.mode == "train" or self.mode == "all":
#             logger.info(f"Loading train data from {AFLOW_DATASET_FILES_MAP['mbpp']['train']}")
#             self._train_data = self._load_data_from_file(file_name=AFLOW_DATASET_FILES_MAP["mbpp"]["train"])
#         if self.mode == "dev" or self.mode == "all":
#             logger.info(f"Loading dev data from {AFLOW_DATASET_FILES_MAP['mbpp']['dev']}")
#             self._dev_data = self._load_data_from_file(file_name=AFLOW_DATASET_FILES_MAP["mbpp"]["dev"])
#         if self.mode == "test" or self.mode == "all":
#             logger.info(f"Loading test data from {AFLOW_DATASET_FILES_MAP['mbpp']['test']}")
#             self._test_data = self._load_data_from_file("/home/tl688/pitl688/selfevolve/EvoAgentX/mbppplus_info.jsonl")
        
#         # load test cases 
#         self._test_cases = self._load_data_from_file("/home/tl688/pitl688/selfevolve/EvoAgentX/mbppplus_info.jsonl")
    
    def _load_data(self):
#         print(AFLOW_DATASET_FILES_MAP)
        if self.mode == "train" or self.mode == "all":
            logger.info(f"Loading train data from {AFLOW_DATASET_FILES_MAP['mbpp']['train']}")
            self._train_data = self._load_data_from_file(file_name=AFLOW_DATASET_FILES_MAP["mbpp"]["train"])
        if self.mode == "dev" or self.mode == "all":
            logger.info(f"Loading dev data from {AFLOW_DATASET_FILES_MAP['mbpp']['dev']}")
            self._dev_data = self._load_data_from_file(file_name=AFLOW_DATASET_FILES_MAP["mbpp"]["dev"])
        if self.mode == "test" or self.mode == "all":
            logger.info(f"Loading test data from {AFLOW_DATASET_FILES_MAP['mbpp']['test']}")
            self._test_data = self._load_data_from_file("/home/tl688/pitl688/selfevolve/EvoAgentX/mbppplus_info.jsonl")
        
        # load test cases 
        self._test_cases = self._load_data_from_file("/home/tl688/pitl688/selfevolve/EvoAgentX/mbppplus_info.jsonl")
    
    def _get_id(self, example: Any) -> Any:
        return example["task_id"]

    def _get_label(self, example: Any) -> Any:
        # return the unit test code
        return {
            "task_id": example["task_id"],
            "canonical_solution": example["code"],
            "test": example["test"],
            "entry_point": example["entry_point"]
        }
    
    def evaluate(self, prediction: Any, label: Any) -> dict:
        """
        Evaluate the solution code.

        Args:
            prediction (str | List[str]): The solution code(s).
            label (dict | List[dict]): The unit test code(s).

        Returns:
            dict: The evaluation metrics (pass@k).
        """
        prediction, label = self._check_evaluation_inputs(prediction, label)

        results = []
        for solution in prediction:
            solution_states = []
            for label_data in label:
                task_id = label_data["task_id"]
                prompt = self.get_example_by_id(task_id)["prompt"]
                unit_test = label_data["test"]
                entry_point = label_data["entry_point"]
                state, message = self.check_solution_plus(
                    task_id=task_id, 
                    solution=solution,
                    test=unit_test, 
                    entry_point=entry_point
                )
                if state != self.SUCCESS:
                    break 
                solution_states.append(state)
            results.append(len(solution_states)==len(label) and all(state==self.SUCCESS for state in solution_states))
        
        k_list = [self.k] if isinstance(self.k, int) else self.k
        pass_at_k = self.compute_pass_at_k(results, k_list)
        
        return pass_at_k
    


class AFlowMBPPPLUS(MBPPPLUS):

    """
    AFlow-specific implementation of MBPP benchmark.
    """

    def __init__(self, path: str = None, mode: str = "all", timeout: int = 60, k: Union[int, list] = 1,**kwargs):
        path = os.path.expanduser(path or "~/.evoagentx/data/aflow/mbpp")
        super().__init__(path=path, mode=mode, timeout=timeout, k=k, **kwargs)

    def _load_data_from_file(self, file_name: str):
        if file_name is None:
            return None
        file_path = os.path.join(self.path, file_name)
        if not os.path.exists(file_path):
            download_aflow_benchmark_data(dataset="mbpp", save_folder=self.path)
        
        return load_json(path=file_path, type="jsonl")

    def _load_data(self):
#         AFLOW_DATASET_FILES_MAP["mbpp"]["train"] = None
#         AFLOW_DATASET_FILES_MAP["mbpp"]["dev"] = "/gpfs/radev/home/tl688/pitl688/selfevolve/AFlow/data/datasets/mbpp_validate.jsonl"
#         AFLOW_DATASET_FILES_MAP["mbpp"]["test"] = "/gpfs/radev/home/tl688/pitl688/selfevolve/AFlow/data/datasets/mbpp_test.jsonl"
#         AFLOW_DATASET_FILES_MAP["mbpp"]["test_cases"] = "/gpfs/radev/home/tl688/pitl688/selfevolve/AFlow/data/datasets/mbpp_public_test.jsonl"
        if self.mode == "train" or self.mode == "all":
            logger.info(f"Loading train data from {AFLOW_DATASET_FILES_MAP['mbpp']['train']}")
            self._train_data = self._load_data_from_file(file_name=AFLOW_DATASET_FILES_MAP["mbpp"]["train"])
        if self.mode == "dev" or self.mode == "all":
            logger.info(f"Loading dev data from {AFLOW_DATASET_FILES_MAP['mbpp']['dev']}")
            self._dev_data = self._load_data_from_file(file_name=AFLOW_DATASET_FILES_MAP["mbpp"]["dev"])
        if self.mode == "test" or self.mode == "all":
            logger.info(f"Loading test data from {AFLOW_DATASET_FILES_MAP['mbpp']['test']}")
            self._test_data = self._load_data_from_file("/home/tl688/pitl688/selfevolve/EvoAgentX/mbppplus_info.jsonl")
        
        # load test cases 
        self._test_cases = self._load_data_from_file("/home/tl688/pitl688/selfevolve/EvoAgentX/mbppplus_info.jsonl")
    
    def _get_label(self, example: Any):
        return {
            "task_id": example["task_id"], 
            "canonical_solution": example["code"], 
            "test": example["test"], 
            "entry_point": example["entry_point"]
        }
    
    def extract_test_cases_with_entry_point(self, entry_point: str):

        hardcoded_cases = {
            "remove_odd": "",
            "replace_spaces": "",
            "snake_to_camel": "",
            "Split": "",
            "swap_List": "",
            "square_Sum": "",
            "sort_sublists": "",
            "unique_sublists": "",
        }
        if entry_point in hardcoded_cases:
            return hardcoded_cases[entry_point]
        
        for case in self._test_cases:
            if case["entry_point"] == entry_point:
                return case["test"]
        
        return None
    
    async def async_evaluate(self, graph: Callable, example: Any) -> float:

        # generate solution 
        prompt, entry_point = example["prompt"], example["entry_point"]
        solution = await graph(prompt, entry_point)
        label = self._get_label(example)
        metrics = await super().async_evaluate(prediction=solution, label=label)
        return metrics["pass@1"]
    
    def evaluate(self, prediction: Any, label: Any) -> dict:
        """
        Evaluate the solution code.

        Args:
            prediction (str | List[str]): The solution code(s).
            label (dict | List[dict]): The unit test code(s).

        Returns:
            dict: The evaluation metrics (pass@k).
        """
        prediction, label = self._check_evaluation_inputs(prediction, label)

        results = []
        for solution in prediction:
            solution_states = []
            for label_data in label:
                task_id = label_data["task_id"]
                prompt = self.get_example_by_id(task_id)["prompt"]
                unit_test = label_data["test"]
                entry_point = label_data["entry_point"]
                state, message = self.check_solution_plus(
                    task_id=task_id, 
                    solution=solution,
                    test=unit_test, 
                    entry_point=entry_point,
                    use_entrypoint_as_input=False
                )
#                 print(state)
#                 print(message)
                if state != self.SUCCESS:
                    break 
                solution_states.append(state)
            self.error_list[task_id] = message.split('\n')[0]
            results.append(len(solution_states)==len(label) and all(state==self.SUCCESS for state in solution_states))
        
        k_list = [self.k] if isinstance(self.k, int) else self.k
        pass_at_k = self.compute_pass_at_k(results, k_list)
#         print(self.error_list)
        
        return pass_at_k
    
    def evaluate_mbpp_witherror(self, prediction: Any, label: Any) -> dict:
        """
        Evaluate the solution code.

        Args:
            prediction (str | List[str]): The solution code(s).
            label (dict | List[dict]): The unit test code(s).

        Returns:
            dict: The evaluation metrics (pass@k).
        """
        prediction, label = self._check_evaluation_inputs(prediction, label)

        results = []
        error_list = []
        for solution in prediction:
            solution_states = []
            for label_data in label:
                task_id = label_data["task_id"]
                prompt = self.get_example_by_id(task_id)["prompt"]
                unit_test = label_data["test"]
                entry_point = label_data["entry_point"]
                state, message = self.check_solution_plus(
                    task_id=task_id, 
                    solution=prompt + "\n" + solution,
                    test=unit_test, 
                    entry_point=entry_point,
                    use_entrypoint_as_input=False
                )
                print(state)
                print(message)
                if state != self.SUCCESS:
                    break 
                solution_states.append(state)
                error_list.append(message)
            results.append(len(solution_states)==len(label) and all(state==self.SUCCESS for state in solution_states))
        
        k_list = [self.k] if isinstance(self.k, int) else self.k
        pass_at_k = self.compute_pass_at_k(results, k_list)
        
        return pass_at_k