Datasets:
File size: 3,552 Bytes
0e239c6 | 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 | """Read a task statement and verify its published record bindings."""
import hashlib
import json
from pathlib import Path
import re
from common import ROOT, contained
_cached_index = None
def _signature(path):
stat = path.stat()
return stat.st_dev, stat.st_ino, stat.st_size, stat.st_mtime_ns, stat.st_ctime_ns
def _index(root):
global _cached_index
root = Path(root).resolve()
names = ['data/crossswe-bench.jsonl', 'data/task-bindings.jsonl']
paths = [root / name for name in names]
checks = root / 'CHECKSUMS.sha256'
signature = tuple(_signature(path) for path in [*paths, checks])
expected = {}
for line in checks.read_text().splitlines():
digest, name = line.split(' ', 1)
if name in names:
if name in expected:
raise ValueError('Duplicate dataset checksum entry')
expected[name] = digest
if set(expected) != set(names):
raise ValueError('Missing dataset checksum entry')
payloads = [path.read_bytes() for path in paths]
digests = [hashlib.sha256(raw).hexdigest() for raw in payloads]
for name, digest in zip(names, digests):
if digest != expected[name]:
raise ValueError('Dataset checksum does not match')
if signature != tuple(_signature(path) for path in [*paths, checks]):
raise ValueError('Dataset files changed while reading')
key = root, *digests
if _cached_index is not None and _cached_index[0] == key:
return _cached_index[1]
bindings = {}
for line in payloads[1].decode('utf-8').splitlines():
item = json.loads(line)
iid = item['instance_id']
if iid in bindings:
raise ValueError('Duplicate task binding')
for field in ['task_sha256', 'source_task_sha256']:
if not isinstance(item[field], str) or not re.fullmatch('[0-9a-f]{64}', item[field]):
raise ValueError('Invalid task binding checksum')
bindings[iid] = item
index = {}
for line in payloads[0].decode('utf-8').splitlines():
row = json.loads(line)
iid = row['instance_id']
text = row['problem_statement']
if not isinstance(iid, str) or not iid or iid in index:
raise ValueError('Invalid or duplicate dataset instance ID')
contained(root / 'instances', iid)
if not isinstance(text, str) or not text.strip():
raise ValueError('Missing problem statement')
if iid not in bindings or bindings[iid]['source_task_sha256'] != row['original_task_sha256']:
raise ValueError('Source task binding does not match')
index[iid] = text, bindings[iid]['task_sha256'], 'statements/' + iid + '.md'
if set(index) != set(bindings):
raise ValueError('Dataset and task-binding IDs differ')
_cached_index = key, index
return index
def read_problem_statement(task, root=ROOT):
root = Path(root).resolve()
iid = task['instance_id']
task_path = contained(root / 'instances', iid) / 'task.json'
try:
text, expected, statement_path = _index(root)[iid]
except KeyError as error:
raise ValueError('Task is absent from the dataset') from error
raw = task_path.read_bytes()
if hashlib.sha256(raw).hexdigest() != expected or json.loads(raw) != task:
raise ValueError('Published task checksum does not match')
if contained(root, statement_path).read_text(encoding='utf-8') != text:
raise ValueError('Statement file does not match the dataset')
return text
|