File size: 12,729 Bytes
9a8a9c5 | 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 | """Code reasoning / completion benchmark adapters (CRUXEval, SAFIM, HumanEval-X)."""
from __future__ import annotations
import re
from typing import Any
from adapters import DatasetAdapter
# Injected at runtime by _set_helpers()
_highlight_code = None
_code_offset = None
_extract_test_classes = None
# ---------------------------------------------------------------------------
# CRUXEval adapter (HuggingFace: cruxeval-org/cruxeval)
# ---------------------------------------------------------------------------
class CRUXEvalAdapter(DatasetAdapter):
slug = "cruxeval"
display_name = "CRUXEval"
has_ground_truth = False
has_tasks = True
def __init__(self, hf_dataset):
self._ds = hf_dataset
def problem_count(self) -> int:
return len(self._ds)
def get_problem_summary(self, idx: int) -> dict[str, Any]:
row = self._ds[idx]
return {
"idx": idx,
"task_id": row["id"],
"entry_point": "f",
"num_inputs": 1,
"source": "CRUXEval",
}
def get_problem_detail(self, idx: int) -> dict[str, Any]:
row = self._ds[idx]
code = row["code"]
return {
"idx": idx,
"task_id": row["id"],
"entry_point": "f",
"code": code,
"highlighted_code": _highlight_code(code),
"inputs": [row["input"]],
"outputs": [row["output"]],
"test": None,
"tasks": [
{
"name": "Output Prediction",
"description": "Given the code and input, predict the output.",
"given": "input",
"predict": "output",
"input": row["input"],
"output": row["output"],
},
{
"name": "Input Prediction",
"description": "Given the code and output, predict the input.",
"given": "output",
"predict": "input",
"input": row["input"],
"output": row["output"],
},
],
"source": "CRUXEval",
"has_ground_truth": False,
"has_tasks": True,
}
# ---------------------------------------------------------------------------
# SAFIM adapter (HuggingFace: gonglinyuan/safim)
# ---------------------------------------------------------------------------
class SAFIMAdapter(DatasetAdapter):
slug = "safim"
display_name = "SAFIM"
has_ground_truth = False
has_tasks = False
def __init__(self, hf_dataset):
self._ds = hf_dataset
def problem_count(self) -> int:
return len(self._ds)
def get_problem_summary(self, idx: int) -> dict[str, Any]:
row = self._ds[idx]
return {
"idx": idx,
"task_id": row.get("task_id", str(idx)),
"entry_point": row.get("task_id", f"safim_{idx}"),
"num_inputs": 0,
"source": row.get("lang", "unknown"),
}
# Patterns that mark where the completion should be inserted
_HOLE_MARKERS = [
"{{completion}}",
"/* TODO: Your code here */",
"// TODO: Your code here",
"# TODO: Your code here",
]
def _find_hole_marker(self, prompt: str) -> str | None:
"""Return the first matching hole marker found in the prompt, or None."""
for marker in self._HOLE_MARKERS:
if marker in prompt:
return marker
return None
def get_problem_detail(self, idx: int) -> dict[str, Any]:
row = self._ds[idx]
prompt = row.get("prompt", "")
ground_truth = row.get("ground_truth", "")
lang = row.get("lang", "python")
marker = self._find_hole_marker(prompt)
if marker:
display_code = prompt.replace(marker, "/* [HOLE] */")
before_hole = prompt.split(marker)[0]
merged_code = prompt.replace(marker, ground_truth)
else:
display_code = prompt + "\n/* [HOLE] */\n"
before_hole = prompt + "\n"
merged_code = prompt + "\n" + ground_truth + "\n"
# Compute 1-indexed line range of the inserted ground truth
gt_start_line = before_hole.count("\n") + 1
gt_line_count = ground_truth.count("\n") + (1 if ground_truth else 0)
gt_end_line = gt_start_line + gt_line_count - 1
lang_key = {"Python": "python", "Java": "java", "C++": "cpp", "C#": "csharp"}.get(
lang, lang.lower()
)
return {
"idx": idx,
"task_id": row.get("task_id", str(idx)),
"entry_point": row.get("task_id", f"safim_{idx}"),
"code": display_code,
"highlighted_code": _highlight_code(display_code, language=lang_key),
"inputs": [],
"outputs": [],
"test": None,
"tasks": [],
"source": lang,
"has_ground_truth": False,
"has_tasks": False,
"fim_prefix": prompt,
"fim_ground_truth": ground_truth,
"fim_ground_truth_highlighted": _highlight_code(ground_truth, language=lang_key),
"fim_merged_code": merged_code,
"fim_merged_highlighted": _highlight_code(
merged_code,
highlight_lines=list(range(gt_start_line, gt_end_line + 1)),
language=lang_key,
),
"fim_gt_start_line": gt_start_line,
"fim_gt_end_line": gt_end_line,
"language": lang,
}
# ---------------------------------------------------------------------------
# HumanEval-X adapter (HuggingFace: THUDM/humaneval-x)
# ---------------------------------------------------------------------------
def _extract_func_name(declaration: str) -> str:
"""Extract the function/method name from a code declaration string."""
m = re.search(r"def\s+(\w+)\s*\(", declaration)
if m:
return m.group(1)
m = re.search(r"(\w+)\s*\(", declaration)
if m:
return m.group(1)
return ""
# ---------------------------------------------------------------------------
# HumanEvalPack adapter (HuggingFace: bigcode/humanevalpack)
# ---------------------------------------------------------------------------
class HumanEvalPackAdapter(DatasetAdapter):
slug = "humanevalpack"
display_name = "HumanEvalPack"
has_ground_truth = False
has_tasks = False
LANGUAGES = ["python", "js", "cpp", "go", "java", "rust"]
def __init__(self, datasets_by_lang: dict[str, Any]):
self._by_lang = datasets_by_lang
first_lang = next(iter(self._by_lang))
self._count = len(self._by_lang[first_lang])
def problem_count(self) -> int:
return self._count
def get_problem_summary(self, idx: int) -> dict[str, Any]:
first_lang = next(iter(self._by_lang))
row = self._by_lang[first_lang][idx]
return {
"idx": idx,
"task_id": row["task_id"],
"entry_point": row.get("entry_point", f"problem_{idx}"),
"num_inputs": len(self._by_lang),
"source": "HumanEvalPack",
}
def get_problem_detail(self, idx: int) -> dict[str, Any]:
first_lang = next(iter(self._by_lang))
row = self._by_lang[first_lang][idx]
lang_labels = {
"python": "Python",
"js": "JavaScript",
"cpp": "C++",
"go": "Go",
"java": "Java",
"rust": "Rust",
}
lang_pygments = {
"python": "python",
"js": "javascript",
"cpp": "cpp",
"go": "go",
"java": "java",
"rust": "rust",
}
lang_solutions = []
for lang in self.LANGUAGES:
if lang not in self._by_lang:
continue
lrow = self._by_lang[lang][idx]
canonical = lrow.get("prompt", "") + lrow.get("canonical_solution", "")
buggy = lrow.get("prompt", "") + lrow.get("buggy_solution", "")
lang_key = lang_pygments.get(lang, lang)
lang_solutions.append(
{
"language": lang,
"language_label": lang_labels.get(lang, lang),
"code": canonical,
"highlighted_code": _highlight_code(canonical, language=lang_key),
"buggy_code": buggy,
"buggy_highlighted_code": _highlight_code(buggy, language=lang_key),
"test": lrow.get("test", ""),
"example_test": lrow.get("example_test", ""),
"bug_type": lrow.get("bug_type", ""),
"failure_symptoms": lrow.get("failure_symptoms", ""),
}
)
py_row = self._by_lang.get("python", self._by_lang[first_lang])[idx]
default_code = py_row.get("prompt", "") + py_row.get("canonical_solution", "")
return {
"idx": idx,
"task_id": row["task_id"],
"entry_point": row.get("entry_point", f"problem_{idx}"),
"code": default_code,
"highlighted_code": _highlight_code(default_code),
"inputs": [],
"outputs": [],
"test": py_row.get("test", ""),
"tasks": [],
"source": "HumanEvalPack",
"has_ground_truth": False,
"has_tasks": False,
"description": row.get("instruction", row.get("docstring", "")),
"lang_solutions": lang_solutions,
"bug_type": py_row.get("bug_type", ""),
"failure_symptoms": py_row.get("failure_symptoms", ""),
}
# ---------------------------------------------------------------------------
# HumanEval-X adapter (HuggingFace: THUDM/humaneval-x)
# ---------------------------------------------------------------------------
class HumanEvalXAdapter(DatasetAdapter):
slug = "humanevalx"
display_name = "HumanEval-X"
has_ground_truth = False
has_tasks = False
LANGUAGES = ["python", "cpp", "java", "go", "js"]
def __init__(self, datasets_by_lang: dict[str, Any]):
"""datasets_by_lang maps language name -> HF dataset split."""
self._by_lang = datasets_by_lang
first_lang = next(iter(self._by_lang))
self._count = len(self._by_lang[first_lang])
def problem_count(self) -> int:
return self._count
def get_problem_summary(self, idx: int) -> dict[str, Any]:
first_lang = next(iter(self._by_lang))
row = self._by_lang[first_lang][idx]
task_id = row["task_id"].split("/")[-1]
decl = row.get("declaration", row.get("prompt", ""))
entry = _extract_func_name(decl) or f"problem_{task_id}"
return {
"idx": idx,
"task_id": f"HumanEval/{task_id}",
"entry_point": entry,
"num_inputs": len(self._by_lang),
"source": "HumanEval-X",
}
def get_problem_detail(self, idx: int) -> dict[str, Any]:
first_lang = next(iter(self._by_lang))
row = self._by_lang[first_lang][idx]
task_id = row["task_id"].split("/")[-1]
decl = row.get("declaration", row.get("prompt", ""))
entry = _extract_func_name(decl) or f"problem_{task_id}"
lang_solutions = []
for lang in self.LANGUAGES:
if lang not in self._by_lang:
continue
lrow = self._by_lang[lang][idx]
code = lrow["prompt"] + lrow["canonical_solution"]
lang_solutions.append(
{
"language": lang,
"code": code,
"highlighted_code": _highlight_code(code, language=lang),
"test": lrow.get("test", ""),
"example_test": lrow.get("example_test", ""),
}
)
py_row = self._by_lang.get("python", self._by_lang[first_lang])[idx]
default_code = py_row["prompt"] + py_row["canonical_solution"]
return {
"idx": idx,
"task_id": f"HumanEval/{task_id}",
"entry_point": entry,
"code": default_code,
"highlighted_code": _highlight_code(default_code),
"inputs": [],
"outputs": [],
"test": py_row.get("test", ""),
"tasks": [],
"source": "HumanEval-X",
"has_ground_truth": False,
"has_tasks": False,
"lang_solutions": lang_solutions,
}
|