Datasets:
File size: 4,733 Bytes
e39cfed | 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 | from __future__ import annotations
import json
from pathlib import Path
from typing import Any
ROOT = Path(__file__).resolve().parent
REQUIRED_ROW_FIELDS = {
"id",
"taxonomy_version",
"text",
"language",
"country",
"domain",
"entities",
"source",
"synthetic",
"review_status",
"split",
}
ALLOWED_SPLITS = {"train", "validation", "test"}
def _read_json(path: Path) -> Any:
with path.open(encoding="utf-8") as stream:
return json.load(stream)
def validate() -> list[str]:
errors: list[str] = []
manifest = _read_json(ROOT / "manifest.json")
taxonomy_version = manifest["taxonomy_version"]
files: dict[str, str] = manifest["files"]
expected_paths = {ROOT / name for name in files}
actual_paths = set(ROOT.glob("*.jsonl"))
for path in sorted(expected_paths - actual_paths):
errors.append(f"missing manifest file: {path.name}")
for path in sorted(actual_paths - expected_paths):
errors.append(f"unlisted JSONL file: {path.name}")
seen_ids: set[str] = set()
for filename, primary_type in sorted(files.items()):
path = ROOT / filename
if not path.exists():
continue
primary_found = False
with path.open(encoding="utf-8") as stream:
for line_number, raw_line in enumerate(stream, start=1):
if not raw_line.strip():
continue
location = f"{filename}:{line_number}"
try:
row = json.loads(raw_line)
except json.JSONDecodeError as error:
errors.append(f"{location}: invalid JSON: {error.msg}")
continue
missing = REQUIRED_ROW_FIELDS - set(row)
if missing:
errors.append(f"{location}: missing fields: {', '.join(sorted(missing))}")
continue
if row["id"] in seen_ids:
errors.append(f"{location}: duplicate id: {row['id']}")
seen_ids.add(row["id"])
if row["taxonomy_version"] != taxonomy_version:
errors.append(f"{location}: taxonomy_version must be {taxonomy_version}")
if not isinstance(row["text"], str) or not row["text"]:
errors.append(f"{location}: text must be a non-empty string")
continue
if row["split"] not in ALLOWED_SPLITS:
errors.append(f"{location}: invalid split: {row['split']}")
if not isinstance(row["entities"], list):
errors.append(f"{location}: entities must be an array")
continue
for entity_index, entity in enumerate(row["entities"]):
entity_location = f"{location}:entity[{entity_index}]"
if not isinstance(entity, dict):
errors.append(f"{entity_location}: entity must be an object")
continue
missing_entity = {"start", "end", "type", "subtype"} - set(entity)
if missing_entity:
errors.append(
f"{entity_location}: missing fields: {', '.join(sorted(missing_entity))}"
)
continue
start, end = entity["start"], entity["end"]
if (
not isinstance(start, int)
or isinstance(start, bool)
or not isinstance(end, int)
or isinstance(end, bool)
or not 0 <= start < end <= len(row["text"])
):
errors.append(
f"{entity_location}: invalid span [{start!r}, {end!r}) "
f"for text length {len(row['text'])}"
)
continue
if not row["text"][start:end].strip():
errors.append(f"{entity_location}: span contains only whitespace")
if entity["type"] == primary_type:
primary_found = True
if not primary_found:
errors.append(f"{filename}: no positive {primary_type} entity found")
return errors
if __name__ == "__main__":
validation_errors = validate()
if validation_errors:
print(f"Sample training data validation failed ({len(validation_errors)} errors):")
for validation_error in validation_errors:
print(f"- {validation_error}")
raise SystemExit(1)
print("Sample training data validation passed") |